Split (1)

train · 1.18k rows

title stringlengths 28 102	full_text stringlengths 97 102k	review stringlengths 97 102k	source stringclasses 1 value	review_src stringlengths 18 25	criteria_count dict	dimension_scores dict	overall_score float64 0 3	conversations listlengths 2 2	year int64 2.01k 2.02k
ABHD4-dependent developmental anoikis safeguards the embryonic brain	Reviewers' Comments: Reviewer #1: Remarks to the Author: Prof. Katona and colleagues proposed developmental anoikis and the molecular player that controls this process. Developmental anoikis was triggered as a result of exposure to different types of insults. One is forced delamination of neural progenitor cells by overexpression of Cdh2 mutant. Another is prenatal alcohol exposure. Concept of “developmental anoikis” that reduces errors of development that is elicited by environmental insults is novel. Based on the known function in cancer anoikis, and the specific expression in developmental stage, the authors picked up Abhd4 as a candidate gene of a controller of this event. The knockout animal did not show obvious phenotypes in normal condition. However, under exposure to insults, the aniokis is significantly reduced in the knockout. The authors also provide good discussion of potential mechanisms how Abhd4 controls developmental aniokis. This article provides very interesting model, and I believe the model will be further strengthened by answering the questions below. Abhd4 expression is remarkably specific to radial glial progenitor cells. Did authors observe reduction of cell death only in that cell population (for example, no changes in cortical plate?) in the knockout cortex of Abhd4 in cases of delta cdh2 overexpression and FASD model? A related comment; is Abhd4 expression altered by prenatal alcohol exposure? For example, do cells other than radial glial progenitor cells increase the expression of Abhd4, so that reduction of cell death is not limited to radial glial progenitor cells in the knockout? Depending on answers to those questions, proposed model may also require revision. The effects on migration and cell death are obvious in embryonic stages in Adhd4 manipulation. Are those effects kept until adulthood and make abnormal adult brain structure, or do cells catch up migration? This information is important for discussion about significance of developmental aniokis. Other comments: Most of TUNEL labelling did not overlap with GFP. Does cell death happen in cell non autonomous manner? Reviewer #2: Remarks to the Author: With this paper the Katona’s lab seeks to introduce a new concept in the field of developmental neuroscience: the developmental anoikis as a safeguarding mechanism responsible for the elimination of abnormally delaminated apical progenitors (RGPCs) during brain development. The authors identified (i) ABDH4 as a crucial molecular player in developmental anoikis and (ii) developmental anoikis as one of the possible mechanisms responsible for the elimination of abnormally delaminated cells in a pathological condition, fetal alcohol syndrome. The paper is a pleasant read and the ideas illustrated by the authors bring a fresh perspective in the neurogenesis field. I acknowledge that the authors address an issue that is under-represented in our field: how robustness is achieved in the context of cortical development and brain morphogenesis. Given the potential of their work, the authors must address the following points to make their paper suitable for acceptance and publication in Nature Communication. MINOR POINTS - A general remark. Throughout the text, the authors refer to abnormal RGPC delamination and AJs disassembly. Strictly speaking, delamination is always paralleled by the cell’s disengagement from the AJs belt. What abnormal exactly means? What is different between a normal and an abnormal AJ disassembly/delamination? What is then setting the difference? The same apply to the use of the word “premature”, referred to delamination. Premature compared to what exactly? Do the authors mean a delamination that is not paralleled by a fate transition? Or a delamination that is not sync with cell cycle progression? The authors must clarify and elaborate conceptually on that. - Page 5: the authors write “spatially limited destruction…” upon nCadh20-GFP manipulation referring to Figure1 a-d. However, the panels show a deep and generalized modification of the AJs. The authors must clarify this point and make clear if the image is representative of their conclusions. Is the manipulation really causing a limited change in the AJ? If so, then the authors should choose a picture reflecting this claim. Otherwise, the authors should change the sentence to reflect what the picture showed: a pronounced and general change in the AJs. The same apply to the phalloidin staining shown in Supplementary Figure 1a-f. The phalloidin staining at the apical surface is clearly less intense in the nCadh20-GFP -electroporated area compared to control electroporation area. This is of course compatible with a junctional perturbation, but it suggests that the perturbation is generalized, and not is “spatially limited”. - Regarding the effect of the manipulation on the basal endfeet. I believe this data is crucial in the clarification of the specificity of the cell biological effects of the experimental manipulation. The authors must show the data now parked in Supplementary Figure 1j-k in the main figure, possibly in Figure 1. Furthermore, given previous literature on the dynamics of the basal endfeet (Yokota et al, 2010), it would be important to define if any morphological change happens at the basal endfoot. I am not asking the author to generate new data, but rather to show the data they already have in a quantitative manner. For example, the authors should show a classification/quantification of the structure of the basal end feet as in Yokota et al, 2010, Figure 2 (club like vs branched end feet). In addition, the authors should add a quantification of the data shown in Supplementary Figure 1m-n. - Page 5. The delaminated progenitors generated upon nCadh20-GFP retain the basal process, Pax6 expression and show an increased apoptosis (i) which is the % of delaminated cells that are also apoptotic? (ii) For those cells that are not apoptotic: do these delaminated cells correspond to bRG/oRG cells? The authors must address this point. I would suggest doing 3D reconstruction of their STORM data (or other suitable data sets) to highlight the complete morphology of the manipulated cells in the SVZ. Additionally, I recommend a staining with p-Vim to understand if there are more cells in SVZ retaining the basal process in mitosis, as hallmark of bRGs. - As for scRNAseq public libraries: it would be informative for the readers to show the actual data/plots for the ABHD4 expression in different cell types in the mentioned libraries. These data can be added as or in a supplementary figure. - Based on the data shown by the authors, the main role of ABDH4 seems to be associated with its downregulation. Based in the author’s claim, ABDH4 downregulation takes place immediately after delamination. This point is interesting and it is crucial in defining the mechanism of action of ABDH4. However, I think the claim is not fully supported by the data. It would be very informative to show an immnunostaining to define the localization of ABDH4. I am actually surprised that, given their experience with high resolution microscopy, the authors limit their attention the mRNAs, without investigating the protein localization. Please clarify if this is technical issue related to the Ab availability. Is ABDH4 -both at the mRNa and protein level- associated with the AJs or to the asymmetric partitioning of ABDH4 during RGPC mitosis? That could provide a cellular mechanism for the fast downregulation of ABDH4 in delaminated cells. Investigating this point would bring the entire paper to a higher level. - Page 7-8: the observation of the inverse correlation of ABHD4 with Tbr2 mRNA is interesting. However, I urge the authors to consider their data in light of previously published work. Of note, it has been demonstrated that aRGs cells already express Tbr2 mRNA (see Florio et al, 2015). So, Tbr2 mRNA is not a specific marker for delaminated IPs, as it is present also in not a delaminated IPs, and more importantly also in aRG cells (see Wilsch-Brauninger et al, 2012 and Florio et al, 2015). In addition, the authors mentioned that they consider cells in the VZ. Again, the VZ localization is not indicative of a certain fate, as VZ contain APs, and IPs, both non-delaminated and delaminated. It is not clear if the cells analyzed by the authors are retaining or not the apical contact. I think the authors should run an additional quantification in which they score separately cells with and without the apical contact. That will help with the interpretation of the graphs shown in Figure panel X, that in its present state is hard to interpret and not particularly convincing. MINOR POINTS - Page 65-66: the model should be added as last panel to the main figure. - Page 3: the word INTRODUCTION is missing. - the paper address the issue of how robustness is achieved in the context of cortical development and brain morphogenesis. I would find extremely interesting if the authors could elaborate more on that in the discussion part.	Reviewers' Comments: Reviewer #1: Remarks to the Author: Prof. Katona and colleagues proposed developmental anoikis and the molecular player that controls this process. Developmental anoikis was triggered as a result of exposure to different types of insults. One is forced delamination of neural progenitor cells by overexpression of Cdh2 mutant. Another is prenatal alcohol exposure. Concept of “developmental anoikis” that reduces errors of development that is elicited by environmental insults is novel. Based on the known function in cancer anoikis, and the specific expression in developmental stage, the authors picked up Abhd4 as a candidate gene of a controller of this event. The knockout animal did not show obvious phenotypes in normal condition. However, under exposure to insults, the aniokis is significantly reduced in the knockout. The authors also provide good discussion of potential mechanisms how Abhd4 controls developmental aniokis. This article provides very interesting model, and I believe the model will be further strengthened by answering the questions below. Abhd4 expression is remarkably specific to radial glial progenitor cells. Did authors observe reduction of cell death only in that cell population (for example, no changes in cortical plate?) in the knockout cortex of Abhd4 in cases of delta cdh2 overexpression and FASD model? A related comment; is Abhd4 expression altered by prenatal alcohol exposure? For example, do cells other than radial glial progenitor cells increase the expression of Abhd4, so that reduction of cell death is not limited to radial glial progenitor cells in the knockout? Depending on answers to those questions, proposed model may also require revision. The effects on migration and cell death are obvious in embryonic stages in Adhd4 manipulation. Are those effects kept until adulthood and make abnormal adult brain structure, or do cells catch up migration? This information is important for discussion about significance of developmental aniokis. Other comments: Most of TUNEL labelling did not overlap with GFP. Does cell death happen in cell non autonomous manner? Reviewer #2: Remarks to the Author: With this paper the Katona’s lab seeks to introduce a new concept in the field of developmental neuroscience: the developmental anoikis as a safeguarding mechanism responsible for the elimination of abnormally delaminated apical progenitors (RGPCs) during brain development. The authors identified (i) ABDH4 as a crucial molecular player in developmental anoikis and (ii) developmental anoikis as one of the possible mechanisms responsible for the elimination of abnormally delaminated cells in a pathological condition, fetal alcohol syndrome. The paper is a pleasant read and the ideas illustrated by the authors bring a fresh perspective in the neurogenesis field. I acknowledge that the authors address an issue that is under-represented in our field: how robustness is achieved in the context of cortical development and brain morphogenesis. Given the potential of their work, the authors must address the following points to make their paper suitable for acceptance and publication in Nature Communication. MINOR POINTS - A general remark. Throughout the text, the authors refer to abnormal RGPC delamination and AJs disassembly. Strictly speaking, delamination is always paralleled by the cell’s disengagement from the AJs belt. What abnormal exactly means? What is different between a normal and an abnormal AJ disassembly/delamination? What is then setting the difference? The same apply to the use of the word “premature”, referred to delamination. Premature compared to what exactly? Do the authors mean a delamination that is not paralleled by a fate transition? Or a delamination that is not sync with cell cycle progression? The authors must clarify and elaborate conceptually on that. - Page 5: the authors write “spatially limited destruction…” upon nCadh20-GFP manipulation referring to Figure1 a-d. However, the panels show a deep and generalized modification of the AJs. The authors must clarify this point and make clear if the image is representative of their conclusions. Is the manipulation really causing a limited change in the AJ? If so, then the authors should choose a picture reflecting this claim. Otherwise, the authors should change the sentence to reflect what the picture showed: a pronounced and general change in the AJs. The same apply to the phalloidin staining shown in Supplementary Figure 1a-f. The phalloidin staining at the apical surface is clearly less intense in the nCadh20-GFP -electroporated area compared to control electroporation area. This is of course compatible with a junctional perturbation, but it suggests that the perturbation is generalized, and not is “spatially limited”. - Regarding the effect of the manipulation on the basal endfeet. I believe this data is crucial in the clarification of the specificity of the cell biological effects of the experimental manipulation. The authors must show the data now parked in Supplementary Figure 1j-k in the main figure, possibly in Figure 1. Furthermore, given previous literature on the dynamics of the basal endfeet (Yokota et al, 2010), it would be important to define if any morphological change happens at the basal endfoot. I am not asking the author to generate new data, but rather to show the data they already have in a quantitative manner. For example, the authors should show a classification/quantification of the structure of the basal end feet as in Yokota et al, 2010, Figure 2 (club like vs branched end feet). In addition, the authors should add a quantification of the data shown in Supplementary Figure 1m-n. - Page 5. The delaminated progenitors generated upon nCadh20-GFP retain the basal process, Pax6 expression and show an increased apoptosis (i) which is the % of delaminated cells that are also apoptotic? (ii) For those cells that are not apoptotic: do these delaminated cells correspond to bRG/oRG cells? The authors must address this point. I would suggest doing 3D reconstruction of their STORM data (or other suitable data sets) to highlight the complete morphology of the manipulated cells in the SVZ. Additionally, I recommend a staining with p-Vim to understand if there are more cells in SVZ retaining the basal process in mitosis, as hallmark of bRGs. - As for scRNAseq public libraries: it would be informative for the readers to show the actual data/plots for the ABHD4 expression in different cell types in the mentioned libraries. These data can be added as or in a supplementary figure. - Based on the data shown by the authors, the main role of ABDH4 seems to be associated with its downregulation. Based in the author’s claim, ABDH4 downregulation takes place immediately after delamination. This point is interesting and it is crucial in defining the mechanism of action of ABDH4. However, I think the claim is not fully supported by the data. It would be very informative to show an immnunostaining to define the localization of ABDH4. I am actually surprised that, given their experience with high resolution microscopy, the authors limit their attention the mRNAs, without investigating the protein localization. Please clarify if this is technical issue related to the Ab availability. Is ABDH4 -both at the mRNa and protein level- associated with the AJs or to the asymmetric partitioning of ABDH4 during RGPC mitosis? That could provide a cellular mechanism for the fast downregulation of ABDH4 in delaminated cells. Investigating this point would bring the entire paper to a higher level. - Page 7-8: the observation of the inverse correlation of ABHD4 with Tbr2 mRNA is interesting. However, I urge the authors to consider their data in light of previously published work. Of note, it has been demonstrated that aRGs cells already express Tbr2 mRNA (see Florio et al, 2015). So, Tbr2 mRNA is not a specific marker for delaminated IPs, as it is present also in not a delaminated IPs, and more importantly also in aRG cells (see Wilsch-Brauninger et al, 2012 and Florio et al, 2015). In addition, the authors mentioned that they consider cells in the VZ. Again, the VZ localization is not indicative of a certain fate, as VZ contain APs, and IPs, both non-delaminated and delaminated. It is not clear if the cells analyzed by the authors are retaining or not the apical contact. I think the authors should run an additional quantification in which they score separately cells with and without the apical contact. That will help with the interpretation of the graphs shown in Figure panel X, that in its present state is hard to interpret and not particularly convincing. MINOR POINTS - Page 65-66: the model should be added as last panel to the main figure. - Page 3: the word INTRODUCTION is missing. - the paper address the issue of how robustness is achieved in the context of cortical development and brain morphogenesis. I would find extremely interesting if the authors could elaborate more on that in the discussion part.	TPR	NC_BS_BIOCH_001_RR1.txt	{ "criticism": 6, "example": 9, "importance_and_relevance": 8, "materials_and_methods": 39, "praise": 7, "presentation_and_reporting": 18, "results_and_discussion": 7, "suggestion_and_solution": 24, "total": 83 }	{ "criticism": 0.07228915662650602, "example": 0.10843373493975904, "importance_and_relevance": 0.0963855421686747, "materials_and_methods": 0.46987951807228917, "praise": 0.08433734939759036, "presentation_and_reporting": 0.21686746987951808, "results_and_discussion": 0.08433734939759036, "suggestion_and_solution": 0.2891566265060241 }	1.421687	[ { "content": "Please review this paper.", "role": "user" }, { "content": "Reviewers' Comments: \n \nReviewer #1: \nRemarks to the Author: \nProf. Katona and colleagues proposed developmental anoikis and the molecular player that \ncontrols this process. Developmental anoikis was triggered as a result of exposure to different \ntypes of insults. One is forced delamination of neural progenitor cells by overexpression of Cdh2 \nmutant. Another is prenatal alcohol exposure. Concept of “developmental anoikis” that reduces \nerrors of development that is elicited by environmental insults is novel. Based on the known \nfunction in cancer anoikis, and the specific expression in developmental stage, the authors picked \nup Abhd4 as a candidate gene of a controller of this event. The knockout animal did not show \nobvious phenotypes in normal condition. However, under exposure to insults, the aniokis is \nsignificantly reduced in the knockout. The authors also provide good discussion of potential \nmechanisms how Abhd4 controls developmental aniokis. This article provides very interesting \nmodel, and I believe the model will be further strengthened by answering the questions below. \n \nAbhd4 expression is remarkably specific to radial glial progenitor cells. Did authors observe \nreduction of cell death only in that cell population (for example, no changes in cortical plate?) in \nthe knockout cortex of Abhd4 in cases of delta cdh2 overexpression and FASD model? A related \ncomment; is Abhd4 expression altered by prenatal alcohol exposure? For example, do cells other \nthan radial glial progenitor cells increase the expression of Abhd4, so that reduction of cell death is \nnot limited to radial glial progenitor cells in the knockout? Depending on answers to those \nquestions, proposed model may also require revision. \n \nThe effects on migration and cell death are obvious in embryonic stages in Adhd4 manipulation. \nAre those effects kept until adulthood and make abnormal adult brain structure, or do cells catch \nup migration? This information is important for discussion about significance of developmental \naniokis. \n \nOther comments: \nMost of TUNEL labelling did not overlap with GFP. Does cell death happen in cell non autonomous \nmanner? \n \n \n \nReviewer #2: \nRemarks to the Author: \nWith this paper the Katona’s lab seeks to introduce a new concept in the field of developmental \nneuroscience: the developmental anoikis as a safeguarding mechanism responsible for the \nelimination of abnormally delaminated apical progenitors (RGPCs) during brain development. The \nauthors identified (i) ABDH4 as a crucial molecular player in developmental anoikis and (ii) \ndevelopmental anoikis as one of the possible mechanisms responsible for the elimination of \nabnormally delaminated cells in a pathological condition, fetal alcohol syndrome. \n \nThe paper is a pleasant read and the ideas illustrated by the authors bring a fresh perspective in \nthe neurogenesis field. I acknowledge that the authors address an issue that is under-represented \nin our field: how robustness is achieved in the context of cortical development and brain \nmorphogenesis. \n \nGiven the potential of their work, the authors must address the following points to make their \npaper suitable for acceptance and publication in Nature Communication. \n \nMINOR POINTS \n \n- A general remark. Throughout the text, the authors refer to abnormal RGPC delamination and \n\n\fAJs disassembly. Strictly speaking, delamination is always paralleled by the cell’s disengagement \nfrom the AJs belt. What abnormal exactly means? What is different between a normal and an \nabnormal AJ disassembly/delamination? What is then setting the difference? \nThe same apply to the use of the word “premature”, referred to delamination. Premature \ncompared to what exactly? Do the authors mean a delamination that is not paralleled by a fate \ntransition? Or a delamination that is not sync with cell cycle progression? The authors must clarify \nand elaborate conceptually on that. \n \n- Page 5: the authors write “spatially limited destruction…” upon nCadh20-GFP manipulation \nreferring to Figure1 a-d. \nHowever, the panels show a deep and generalized modification of the AJs. The authors must clarify \nthis point and make clear if the image is representative of their conclusions. Is the manipulation \nreally causing a limited change in the AJ? \nIf so, then the authors should choose a picture reflecting this claim. \nOtherwise, the authors should change the sentence to reflect what the picture showed: a \npronounced and general change in the AJs. \nThe same apply to the phalloidin staining shown in Supplementary Figure 1a-f. The phalloidin \nstaining at the apical surface is clearly less intense in the nCadh20-GFP -electroporated area \ncompared to control electroporation area. This is of course compatible with a junctional \nperturbation, but it suggests that the perturbation is generalized, and not is “spatially limited”. \n \n- Regarding the effect of the manipulation on the basal endfeet. I believe this data is crucial in the \nclarification of the specificity of the cell biological effects of the experimental manipulation. The \nauthors must show the data now parked in Supplementary Figure 1j-k in the main figure, possibly \nin Figure 1. \nFurthermore, given previous literature on the dynamics of the basal endfeet (Yokota et al, 2010), \nit would be important to define if any morphological change happens at the basal endfoot. I am \nnot asking the author to generate new data, but rather to show the data they already have in a \nquantitative manner. For example, the authors should show a classification/quantification of the \nstructure of the basal end feet as in Yokota et al, 2010, Figure 2 (club like vs branched end feet). \nIn addition, the authors should add a quantification of the data shown in Supplementary Figure \n1m-n. \n \n- Page 5. The delaminated progenitors generated upon nCadh20-GFP retain the basal process, \nPax6 expression and show an increased apoptosis \n(i) which is the % of delaminated cells that are also apoptotic? \n(ii) For those cells that are not apoptotic: do these delaminated cells correspond to bRG/oRG cells? \nThe authors must address this point. \nI would suggest doing 3D reconstruction of their STORM data (or other suitable data sets) to \nhighlight the complete morphology of the manipulated cells in the SVZ. Additionally, I recommend \na staining with p-Vim to understand if there are more cells in SVZ retaining the basal process in \nmitosis, as hallmark of bRGs. \n \n- As for scRNAseq public libraries: it would be informative for the readers to show the actual \ndata/plots for the ABHD4 expression in different cell types in the mentioned libraries. These data \ncan be added as or in a supplementary figure. \n \n- Based on the data shown by the authors, the main role of ABDH4 seems to be associated with its \ndownregulation. Based in the author’s claim, ABDH4 downregulation takes place immediately after \ndelamination. This point is interesting and it is crucial in defining the mechanism of action of \nABDH4. However, I think the claim is not fully supported by the data. It would be very informative \nto show an immnunostaining to define the localization of ABDH4. I am actually surprised that, \ngiven their experience with high resolution microscopy, the authors limit their attention the \nmRNAs, without investigating the protein localization. Please clarify if this is technical issue related \nto the Ab availability. Is ABDH4 -both at the mRNa and protein level- associated with the AJs or to \n\n\fthe asymmetric partitioning of ABDH4 during RGPC mitosis? That could provide a cellular \nmechanism for the fast downregulation of ABDH4 in delaminated cells. Investigating this point \nwould bring the entire paper to a higher level. \n \n- Page 7-8: the observation of the inverse correlation of ABHD4 with Tbr2 mRNA is interesting. \nHowever, I urge the authors to consider their data in light of previously published work. Of note, it \nhas been demonstrated that aRGs cells already express Tbr2 mRNA (see Florio et al, 2015). So, \nTbr2 mRNA is not a specific marker for delaminated IPs, as it is present also in not a delaminated \nIPs, and more importantly also in aRG cells (see Wilsch-Brauninger et al, 2012 and Florio et al, \n2015). In addition, the authors mentioned that they consider cells in the VZ. Again, the VZ \nlocalization is not indicative of a certain fate, as VZ contain APs, and IPs, both non-delaminated \nand delaminated. It is not clear if the cells analyzed by the authors are retaining or not the apical \ncontact. I think the authors should run an additional quantification in which they score separately \ncells with and without the apical contact. That will help with the interpretation of the graphs shown \nin Figure panel X, that in its present state is hard to interpret and not particularly convincing. \n \n \nMINOR POINTS \n \n- Page 65-66: the model should be added as last panel to the main figure. \n \n- Page 3: the word INTRODUCTION is missing. \n \n- the paper address the issue of how robustness is achieved in the context of cortical development \nand brain morphogenesis. I would find extremely interesting if the authors could elaborate more \non that in the discussion part. \n \n \n \n\n\f\f\n", "role": "assistant" } ]	2,020
ABHD4-dependent developmental anoikis safeguards the embryonic brain	Reviewers' Comments: Reviewer #1: Remarks to the Author: The authors addressed all of points and the manuscript reads well. Reviewer #2: Remarks to the Author: The revised version of László et al. addresses the points raised during the previous round of review. I appreciate the fact that in the present version the authors provide additional data and quantifications corroborating and strengthening their original claims. It is my opinion that the paper in its present form should be accepted for publication. Only one point should be addressed: there are here and there several typos, that must be corrected and edited before publication.	Reviewers' Comments: Reviewer #1: Remarks to the Author: The authors addressed all of points and the manuscript reads well. Reviewer #2: Remarks to the Author: The revised version of László et al. addresses the points raised during the previous round of review. I appreciate the fact that in the present version the authors provide additional data and quantifications corroborating and strengthening their original claims. It is my opinion that the paper in its present form should be accepted for publication. Only one point should be addressed: there are here and there several typos, that must be corrected and edited before publication.	TPR	NC_BS_BIOCH_001_RR2.txt	{ "criticism": 0, "example": 0, "importance_and_relevance": 0, "materials_and_methods": 1, "praise": 3, "presentation_and_reporting": 1, "results_and_discussion": 0, "suggestion_and_solution": 3, "total": 6 }	{ "criticism": 0, "example": 0, "importance_and_relevance": 0, "materials_and_methods": 0.16666666666666666, "praise": 0.5, "presentation_and_reporting": 0.16666666666666666, "results_and_discussion": 0, "suggestion_and_solution": 0.5 }	1.333333	[ { "content": "Please review this paper.", "role": "user" }, { "content": "\n \n\n\fReviewers' Comments: \n \nReviewer #1: \nRemarks to the Author: \nThe authors addressed all of points and the manuscript reads well. \n \n \n \nReviewer #2: \nRemarks to the Author: \nThe revised version of László et al. addresses the points raised during the previous round of \nreview. \nI appreciate the fact that in the present version the authors provide additional data and \nquantifications corroborating and strengthening their original claims. \n \nIt is my opinion that the paper in its present form should be accepted for publication. \n \nOnly one point should be addressed: there are here and there several typos, that must be \ncorrected and edited before publication. \n \n \n\n\n \n\n\f\n", "role": "assistant" } ]	2,020
Dynamic remodelling of the human host cell proteome and phosphoproteome upon	Reviewers' comments: Reviewer #1 (Remarks to the Author): Giansanti et al. - Dynamic remodeling of the human host cell (phospho)proteome upon enterovirus infection GENERAL COMMENTS This manuscript explores the altered proteome in cells infected with Coxsackievirus B3. The authors use MS proteomics to assess the change in the proteome and phosphoproteome in infected cells. They find phosphoproteomic changes early in infection, while proteomic changes are only detected in later timepoints (after ~8 hrs). The authors use the patterns of phosphorylation to predict the kinases involved and investigate their effects on the mTOR pathway specifically. Through investigation of the downstream targets of mTOR, they show that eEF2 phosphorylation is decreased over the course of infection. The authors also discuss the role of mTOR target TFEB, a regulator of lysosomal and autophagy. They find that TFEB is activated by infection and that knockdown of this factor inhibits virus production, but not replication, suggesting a role for lysosomal/autophagic pathways in egress. The work reported in the manuscript is well done and describes the phosphoproteomic flux associated with infection, which is sure to be a useful reference and source for hyptheses for others in the field. The paper itself focuses on mTOR and its downstream targets. These are known to be involved in enterovirus biology, as the authors note, which makes their observation less novel, but still provides more context for mTOR involvement. The authors suggest one downstream target (TFEB) of mTOR that effects virus production, but again this observation lacks mechanistic insight beyond the genetic experiment. Because of its quality, I believe the paper does merit acceptance and publication after the authors address some of the comments below. Specific Comments: ll. 152-5. “While a decrease in protein levels may be explained by the virus-induced shutdown of protein translation, the observation that certain protein 154 levels increased suggests that a subset of cellular proteins escape the bulk translational shutdown induced by viral cleavage of eIF4G.“ Can the author distinguish between altered turnover vs. EIF4G independent translation? Any analysis that could be performed to confirm this hypothesis? Supp. Table 4.3: Why do the authors include Apoptotic signaling in the translation and transcription? Perhaps labeling the top enriched categories on the plots would be useful for transparency. Have the authors considered a gene set enrichment approach based on fold-change of factors in given pathways. e.g. Do genes in this pathway have a greater fold-change than expected by chance, this might better reflect the phenotype of the infected cells. Small Notes: I recommend renaming the section called “Regulation of viral protein levels during CVB3 expression” as the section does not discuss regulation perse. Coxsackievirus is capitalized as it is named after a city “Coxsackie, NY, USA” Figures: The Blue gradient background is distracting and unnecessary, makes the grey lines and blue boxes hard to visualize. I’d encourage the authors to include some of the more “systems level” figures in the main text to give a general sense of the hits on the screen (S2C specifically). Reviewer #2 (Remarks to the Author): In the manuscript “Dynamic remodeling of the human host cell (phospho)proteome Upon enterovirus infection” Giansanti et al present a comprehensive proteomic and phospoproteomic dataset across a time course of enterovirus infection. The extensive datasets are analyzed to identify changes on the protein and phosphorylation levels and described globally. To interpret findings in the phosphoproteomic dataset further analysis was performed to predict kinases that are active during enterovirus infection. The authors then pick the mTORC1 signaling pathways for more in depth description of its possible regulation during infection and provide limited follow up studies, which do not lead to conclusive biological insights in enterovirus infection. My main concerns regarding the current version of the manuscript are as follows: While the manuscript describes an extensive dataset that seems to be of high quality, the manuscript is mainly focused on describing the dataset rather than deriving biological insights into enterovirus infection and identifying novel host factors/signaling pathways relevant for infection. The majority of the paragraphs are highly descriptive, but the observations are not really linked back to the biological question. The authors derive some hypotheses from the dataset, but they either do not provide any functional follow up study to validate the hypothesis and the follow up studies they performed do not provide conclusive results. Some concerns/questions that I have regarding the proteomic datasets: Currently the authors analyze the proteome and phosphoproteome datasets separately. It would we important to determine whether observed changes in phosphorylation are in fact changing because of different levels of phosphorylation or because of protein abundance changes. This is especially important at later stages of infection where more changes on the proteome level are observed. Some of the earlier time points for the proteome analysis do not cluster at all with the replicates. Is this a reason why there are little significant protein abundance changes observed at earlier timepoints of infection? Were the proteome and phosphoproteome datasets generated from the same samples? In summary, the current version of the manuscript lacks novel biological insights into CVBB3 infection and revisions are necessary to (1) put results in a biological context rather than just providing a detailed description, (2) provide additional functional follow up studies demonstrating the validity of the datasets, and (3) address the specific concerns listed below. Specific concerns: Page 4, Introduction The authors should provide a few sentence summary of the major findings of their study at the end of the introduction. What was the specific goal/motivation of the study? What was the motivation to use CVB3? Page 5, Results, Analysis of the proteome of infected cells • What is the biological motivation to use CVB3 in the study? Why was an MOI of 10 used and what was the infection rate obtained with this MOI? How reproducible was the infection? • Reproducibility of the infection seem concerning, since the clustering of the overall proteome data is variable and shows that replicates of earlier timepoint of the infection do not cluster at all with their respective biological replicates. • How were the clusters defined in Figure 2a (no dendogram provided)? • Sections of this paragraphs sound just like a description of the method and could be shortened. • I would suggest to restructure the manuscript. The flow of the manuscript feels interrupted, since the overall proteome dataset is descried in the first paragraph, but then not followed by the analysis of the proteome changes in the course of infection. Page 6, Results, Analysis of the phosphoproteome of infected cells • See above, sections of this paragraphs sound just like a description of the method and could be shortened. • How were the clusters defined in Figure 2b (no dendogram provided)? • It seems surprising to me that 85% of the phosphoproteome is changing in the course of infection. How does this compare to other phosphoproteomic datasets generated in the course of viral infection? How was the data normalized? Often in proteomic datasets normalization strategies are employed which assume that the majority of the sites are not changing. How did the authors address this in their data analysis? More details need to be provided in the method section? • See above, the manuscript feels interrupted, since the description of the phosphoproteomic dataset is not followed by the analysis of the phosphorylation changes. Page 6, Results, Regulation of viral protein levels during CVB3 infection • This paragraph is purely descriptive and out of the context in this part of the manuscript. • I would suggest to link it ack to CVB3 biology, otherwise I would remove it completely. Page 7, Results, Changes in host protein levels during CVB3 infection • I think for a better reading of the manuscript the authors should consider to restructure the result section and move this section after describing the proteome dataset. • See above, it is concerning to me that replicates of some timepoints do not cluster at all and look very different in the heatmap. What is the reason for this? Was there a quality control for the reproducibility of infection? • Paragraph misses more biology, what are these proteins that escape virus-induced translational shutdown? • It would also be highly interesting if the authors could perform an integrative analysis of their data with the CVB3 host-pathogen interaction network that was recently published in Nature Microbiology. Page 8, Results, Regulation of protein phosphorylation levels during CVB3 infection • I think for a better reading of the manuscript the authors should consider to restructure the result section and move this section after describing the phosphoproteome dataset. • At the end of the section, changes in phosphoproteome and total proteome are compared, I am wondering what the fraction of the phosphoproteome changes can in fact be explained by protein abundance changes. What is the fraction of phosphosites that change while there is no change observed on the protein level? Page 8, Results, Identification of kinases that are activated by CVB3 infection It is stated in the text that three findings of this analysis are of particular interest, such as overrepresentation of ATM/ATR motifs. Why are these findings of particular of particular interest? What does this mean for the infection process? What is the role of ATM/ATR or c PKC and GRKs in CVB3 infection? Page 9, Results, Phosphorylation of host factors involved in formation of viral replication organelles The authors describe in this section the phosphorylation of know proteins that might play a role in replication organelle formation and suggest that the specific sites might be relevant for the organelle formation, but again this is only descriptive and no biological follow up is provided. What is the possible impact of knowing the phosphorylation sites of these specific factors? Does mutating these sides lead to a reduction in viral replication? Page 11, Results, Signaling pathways affected during CVB3 infection More details need to be provided how the pathway analysis was performed. What are the predefined pathways known to be affected by CVB3? Where are they taken from? Was the proteome and phosphoproteome data just mapped onto pathways or was an enrichment analysis performed? Which input data was provided in for the pathway analysis, quantitative data across all time points? Page 11, Results, Inactivation of mTORC1 signaling during CVB3 infection Figure 3 associated with this section is very difficult to read: I would suggest now to use the blue background, which reduces the contrast in the figure. +/- symbols are so small that they are hard to see. Also it is not clear what the different colored arrows and lines mean. Why was a higher MOI of 50 used for validating the phosphoproteomic data by Western Blot, which itself was generated at an MOI of 10. Page 14, Results, Activation of the mTORC1-regulated protein transcription factor EB How was the knockout of TFEB performed? There is no information provided in the method section. Additionally, validation of the ko is missing. What is the explanation that 8hpi a significant decrease in virus titers is observed upon ko of TFEB, however there is no significant difference 10hpi? Does this makes sense? Page 18, Experimental procedures As mentioned in different comments above, a more detailed description of some of the methods is missing and should be provided.	Reviewers' comments: Reviewer #1 (Remarks to the Author): Giansanti et al. - Dynamic remodeling of the human host cell (phospho)proteome upon enterovirus infection GENERAL COMMENTS This manuscript explores the altered proteome in cells infected with Coxsackievirus B3. The authors use MS proteomics to assess the change in the proteome and phosphoproteome in infected cells. They find phosphoproteomic changes early in infection, while proteomic changes are only detected in later timepoints (after ~8 hrs). The authors use the patterns of phosphorylation to predict the kinases involved and investigate their effects on the mTOR pathway specifically. Through investigation of the downstream targets of mTOR, they show that eEF2 phosphorylation is decreased over the course of infection. The authors also discuss the role of mTOR target TFEB, a regulator of lysosomal and autophagy. They find that TFEB is activated by infection and that knockdown of this factor inhibits virus production, but not replication, suggesting a role for lysosomal/autophagic pathways in egress. The work reported in the manuscript is well done and describes the phosphoproteomic flux associated with infection, which is sure to be a useful reference and source for hyptheses for others in the field. The paper itself focuses on mTOR and its downstream targets. These are known to be involved in enterovirus biology, as the authors note, which makes their observation less novel, but still provides more context for mTOR involvement. The authors suggest one downstream target (TFEB) of mTOR that effects virus production, but again this observation lacks mechanistic insight beyond the genetic experiment. Because of its quality, I believe the paper does merit acceptance and publication after the authors address some of the comments below. Specific Comments: ll. 152-5. “While a decrease in protein levels may be explained by the virus-induced shutdown of protein translation, the observation that certain protein 154 levels increased suggests that a subset of cellular proteins escape the bulk translational shutdown induced by viral cleavage of eIF4G.“ Can the author distinguish between altered turnover vs. EIF4G independent translation? Any analysis that could be performed to confirm this hypothesis? Supp. Table 4.3: Why do the authors include Apoptotic signaling in the translation and transcription? Perhaps labeling the top enriched categories on the plots would be useful for transparency. Have the authors considered a gene set enrichment approach based on fold-change of factors in given pathways. e.g. Do genes in this pathway have a greater fold-change than expected by chance, this might better reflect the phenotype of the infected cells. Small Notes: I recommend renaming the section called “Regulation of viral protein levels during CVB3 expression” as the section does not discuss regulation perse. Coxsackievirus is capitalized as it is named after a city “Coxsackie, NY, USA” Figures: The Blue gradient background is distracting and unnecessary, makes the grey lines and blue boxes hard to visualize. I’d encourage the authors to include some of the more “systems level” figures in the main text to give a general sense of the hits on the screen (S2C specifically). Reviewer #2 (Remarks to the Author): In the manuscript “Dynamic remodeling of the human host cell (phospho)proteome Upon enterovirus infection” Giansanti et al present a comprehensive proteomic and phospoproteomic dataset across a time course of enterovirus infection. The extensive datasets are analyzed to identify changes on the protein and phosphorylation levels and described globally. To interpret findings in the phosphoproteomic dataset further analysis was performed to predict kinases that are active during enterovirus infection. The authors then pick the mTORC1 signaling pathways for more in depth description of its possible regulation during infection and provide limited follow up studies, which do not lead to conclusive biological insights in enterovirus infection. My main concerns regarding the current version of the manuscript are as follows: While the manuscript describes an extensive dataset that seems to be of high quality, the manuscript is mainly focused on describing the dataset rather than deriving biological insights into enterovirus infection and identifying novel host factors/signaling pathways relevant for infection. The majority of the paragraphs are highly descriptive, but the observations are not really linked back to the biological question. The authors derive some hypotheses from the dataset, but they either do not provide any functional follow up study to validate the hypothesis and the follow up studies they performed do not provide conclusive results. Some concerns/questions that I have regarding the proteomic datasets: Currently the authors analyze the proteome and phosphoproteome datasets separately. It would we important to determine whether observed changes in phosphorylation are in fact changing because of different levels of phosphorylation or because of protein abundance changes. This is especially important at later stages of infection where more changes on the proteome level are observed. Some of the earlier time points for the proteome analysis do not cluster at all with the replicates. Is this a reason why there are little significant protein abundance changes observed at earlier timepoints of infection? Were the proteome and phosphoproteome datasets generated from the same samples? In summary, the current version of the manuscript lacks novel biological insights into CVBB3 infection and revisions are necessary to (1) put results in a biological context rather than just providing a detailed description, (2) provide additional functional follow up studies demonstrating the validity of the datasets, and (3) address the specific concerns listed below. Specific concerns: Page 4, Introduction The authors should provide a few sentence summary of the major findings of their study at the end of the introduction. What was the specific goal/motivation of the study? What was the motivation to use CVB3? Page 5, Results, Analysis of the proteome of infected cells • What is the biological motivation to use CVB3 in the study? Why was an MOI of 10 used and what was the infection rate obtained with this MOI? How reproducible was the infection? • Reproducibility of the infection seem concerning, since the clustering of the overall proteome data is variable and shows that replicates of earlier timepoint of the infection do not cluster at all with their respective biological replicates. • How were the clusters defined in Figure 2a (no dendogram provided)? • Sections of this paragraphs sound just like a description of the method and could be shortened. • I would suggest to restructure the manuscript. The flow of the manuscript feels interrupted, since the overall proteome dataset is descried in the first paragraph, but then not followed by the analysis of the proteome changes in the course of infection. Page 6, Results, Analysis of the phosphoproteome of infected cells • See above, sections of this paragraphs sound just like a description of the method and could be shortened. • How were the clusters defined in Figure 2b (no dendogram provided)? • It seems surprising to me that 85% of the phosphoproteome is changing in the course of infection. How does this compare to other phosphoproteomic datasets generated in the course of viral infection? How was the data normalized? Often in proteomic datasets normalization strategies are employed which assume that the majority of the sites are not changing. How did the authors address this in their data analysis? More details need to be provided in the method section? • See above, the manuscript feels interrupted, since the description of the phosphoproteomic dataset is not followed by the analysis of the phosphorylation changes. Page 6, Results, Regulation of viral protein levels during CVB3 infection • This paragraph is purely descriptive and out of the context in this part of the manuscript. • I would suggest to link it ack to CVB3 biology, otherwise I would remove it completely. Page 7, Results, Changes in host protein levels during CVB3 infection • I think for a better reading of the manuscript the authors should consider to restructure the result section and move this section after describing the proteome dataset. • See above, it is concerning to me that replicates of some timepoints do not cluster at all and look very different in the heatmap. What is the reason for this? Was there a quality control for the reproducibility of infection? • Paragraph misses more biology, what are these proteins that escape virus-induced translational shutdown? • It would also be highly interesting if the authors could perform an integrative analysis of their data with the CVB3 host-pathogen interaction network that was recently published in Nature Microbiology. Page 8, Results, Regulation of protein phosphorylation levels during CVB3 infection • I think for a better reading of the manuscript the authors should consider to restructure the result section and move this section after describing the phosphoproteome dataset. • At the end of the section, changes in phosphoproteome and total proteome are compared, I am wondering what the fraction of the phosphoproteome changes can in fact be explained by protein abundance changes. What is the fraction of phosphosites that change while there is no change observed on the protein level? Page 8, Results, Identification of kinases that are activated by CVB3 infection It is stated in the text that three findings of this analysis are of particular interest, such as overrepresentation of ATM/ATR motifs. Why are these findings of particular of particular interest? What does this mean for the infection process? What is the role of ATM/ATR or c PKC and GRKs in CVB3 infection? Page 9, Results, Phosphorylation of host factors involved in formation of viral replication organelles The authors describe in this section the phosphorylation of know proteins that might play a role in replication organelle formation and suggest that the specific sites might be relevant for the organelle formation, but again this is only descriptive and no biological follow up is provided. What is the possible impact of knowing the phosphorylation sites of these specific factors? Does mutating these sides lead to a reduction in viral replication? Page 11, Results, Signaling pathways affected during CVB3 infection More details need to be provided how the pathway analysis was performed. What are the predefined pathways known to be affected by CVB3? Where are they taken from? Was the proteome and phosphoproteome data just mapped onto pathways or was an enrichment analysis performed? Which input data was provided in for the pathway analysis, quantitative data across all time points? Page 11, Results, Inactivation of mTORC1 signaling during CVB3 infection Figure 3 associated with this section is very difficult to read: I would suggest now to use the blue background, which reduces the contrast in the figure. +/- symbols are so small that they are hard to see. Also it is not clear what the different colored arrows and lines mean. Why was a higher MOI of 50 used for validating the phosphoproteomic data by Western Blot, which itself was generated at an MOI of 10. Page 14, Results, Activation of the mTORC1-regulated protein transcription factor EB How was the knockout of TFEB performed? There is no information provided in the method section. Additionally, validation of the ko is missing. What is the explanation that 8hpi a significant decrease in virus titers is observed upon ko of TFEB, however there is no significant difference 10hpi? Does this makes sense? Page 18, Experimental procedures As mentioned in different comments above, a more detailed description of some of the methods is missing and should be provided.	TPR	NC_BS_BIOCH_002_RR1.txt	{ "criticism": 16, "example": 14, "importance_and_relevance": 5, "materials_and_methods": 48, "praise": 3, "presentation_and_reporting": 19, "results_and_discussion": 18, "suggestion_and_solution": 19, "total": 93 }	{ "criticism": 0.17204301075268819, "example": 0.15053763440860216, "importance_and_relevance": 0.053763440860215055, "materials_and_methods": 0.5161290322580645, "praise": 0.03225806451612903, "presentation_and_reporting": 0.20430107526881722, "results_and_discussion": 0.1935483870967742, "suggestion_and_solution": 0.20430107526881722 }	1.526882	[ { "content": "Please review this paper.", "role": "user" }, { "content": "Reviewers' comments: \n\n \n\nReviewer #1 (Remarks to the Author): \n\n \n\nGiansanti et al. - Dynamic remodeling of the human host cell (phospho)proteome upon enterovirus \ninfection \n\n \n\nGENERAL COMMENTS \n\n \n\nThis manuscript explores the altered proteome in cells infected with Coxsackievirus B3. The authors \nuse MS proteomics to assess the change in the proteome and phosphoproteome in infected cells. \nThey find phosphoproteomic changes early in infection, while proteomic changes are only detected \nin later timepoints (after ~8 hrs). The authors use the patterns of phosphorylation to predict the \nkinases involved and investigate their effects on the mTOR pathway specifically. Through \ninvestigation of the downstream targets of mTOR, they show that eEF2 phosphorylation is decreased \nover the course of infection. The authors also discuss the role of mTOR target TFEB, a regulator of \nlysosomal and autophagy. They find that TFEB is activated by infection and that knockdown of this \nfactor inhibits virus production, but not replication, suggesting a role for lysosomal/autophagic \npathways in egress. \n\n \n\nThe work reported in the manuscript is well done and describes the phosphoproteomic flux \nassociated with infection, which is sure to be a useful reference and source for hyptheses for others \nin the field. The paper itself focuses on mTOR and its downstream targets. These are known to be \ninvolved in enterovirus biology, as the authors note, which makes their observation less novel, but \nstill provides more context for mTOR involvement. The authors suggest one downstream target \n(TFEB) of mTOR that effects virus production, but again this observation lacks mechanistic insight \nbeyond the genetic experiment. \n\n \n\nBecause of its quality, I believe the paper does merit acceptance and publication after the authors \naddress some of the comments below. \n\n \n\nSpecific Comments: \n\n \n\n\fll. 152-5. “While a decrease in protein levels may be explained by the virus-induced shutdown of \nprotein translation, the observation that certain protein \n\n154 levels increased suggests that a subset of cellular proteins escape the bulk translational \nshutdown induced by viral cleavage of eIF4G.“ Can the author distinguish between altered turnover \nvs. EIF4G independent translation? Any analysis that could be performed to confirm this hypothesis? \n\n \n\nSupp. Table 4.3: Why do the authors include Apoptotic signaling in the translation and transcription? \nPerhaps labeling the top enriched categories on the plots would be useful for transparency. \n\n \n\nHave the authors considered a gene set enrichment approach based on fold-change of factors in \ngiven pathways. e.g. Do genes in this pathway have a greater fold-change than expected by chance, \nthis might better reflect the phenotype of the infected cells. \n\n \n\nSmall Notes: \n\n \n\nI recommend renaming the section called “Regulation of viral protein levels during CVB3 expression” \nas the section does not discuss regulation perse. \n\n \n\nCoxsackievirus is capitalized as it is named after a city “Coxsackie, NY, USA” \n\n \n\nFigures: The Blue gradient background is distracting and unnecessary, makes the grey lines and blue \nboxes hard to visualize. \n\n \n\nI’d encourage the authors to include some of the more “systems level” figures in the main text to \ngive a general sense of the hits on the screen (S2C specifically). \n\n \n\n \n\nReviewer #2 (Remarks to the Author): \n\n \n\nIn the manuscript “Dynamic remodeling of the human host cell (phospho)proteome Upon \nenterovirus infection” Giansanti et al present a comprehensive proteomic and phospoproteomic \ndataset across a time course of enterovirus infection. The extensive datasets are analyzed to identify \n\n\fchanges on the protein and phosphorylation levels and described globally. To interpret findings in \nthe phosphoproteomic dataset further analysis was performed to predict kinases that are active \nduring enterovirus infection. The authors then pick the mTORC1 signaling pathways for more in \ndepth description of its possible regulation during infection and provide limited follow up studies, \nwhich do not lead to conclusive biological insights in enterovirus infection. \n\n \n\nMy main concerns regarding the current version of the manuscript are as follows: While the \nmanuscript describes an extensive dataset that seems to be of high quality, the manuscript is mainly \nfocused on describing the dataset rather than deriving biological insights into enterovirus infection \nand identifying novel host factors/signaling pathways relevant for infection. The majority of the \nparagraphs are highly descriptive, but the observations are not really linked back to the biological \nquestion. The authors derive some hypotheses from the dataset, but they either do not provide any \nfunctional follow up study to validate the hypothesis and the follow up studies they performed do \nnot provide conclusive results. \n\n \n\nSome concerns/questions that I have regarding the proteomic datasets: \n\nCurrently the authors analyze the proteome and phosphoproteome datasets separately. It would we \nimportant to determine whether observed changes in phosphorylation are in fact changing because \nof different levels of phosphorylation or because of protein abundance changes. This is especially \nimportant at later stages of infection where more changes on the proteome level are observed. \n\nSome of the earlier time points for the proteome analysis do not cluster at all with the replicates. Is \nthis a reason why there are little significant protein abundance changes observed at earlier \ntimepoints of infection? Were the proteome and phosphoproteome datasets generated from the \nsame samples? \n\n \n\nIn summary, the current version of the manuscript lacks novel biological insights into CVBB3 \ninfection and revisions are necessary to (1) put results in a biological context rather than just \nproviding a detailed description, (2) provide additional functional follow up studies demonstrating \nthe validity of the datasets, and (3) address the specific concerns listed below. \n\n \n\nSpecific concerns: \n\n \n\nPage 4, Introduction \n\nThe authors should provide a few sentence summary of the major findings of their study at the end \nof the introduction. What was the specific goal/motivation of the study? What was the motivation to \nuse CVB3? \n\n\f \n\nPage 5, Results, Analysis of the proteome of infected cells \n\n• What is the biological motivation to use CVB3 in the study? Why was an MOI of 10 used and what \nwas the infection rate obtained with this MOI? How reproducible was the infection? \n\n• Reproducibility of the infection seem concerning, since the clustering of the overall proteome data \nis variable and shows that replicates of earlier timepoint of the infection do not cluster at all with \ntheir respective biological replicates. \n\n• How were the clusters defined in Figure 2a (no dendogram provided)? \n\n• Sections of this paragraphs sound just like a description of the method and could be shortened. \n\n• I would suggest to restructure the manuscript. The flow of the manuscript feels interrupted, since \nthe overall proteome dataset is descried in the first paragraph, but then not followed by the analysis \nof the proteome changes in the course of infection. \n\n \n\nPage 6, Results, Analysis of the phosphoproteome of infected cells \n\n• See above, sections of this paragraphs sound just like a description of the method and could be \nshortened. \n\n• How were the clusters defined in Figure 2b (no dendogram provided)? \n\n• It seems surprising to me that 85% of the phosphoproteome is changing in the course of infection. \nHow does this compare to other phosphoproteomic datasets generated in the course of viral \ninfection? How was the data normalized? Often in proteomic datasets normalization strategies are \nemployed which assume that the majority of the sites are not changing. How did the authors \naddress this in their data analysis? More details need to be provided in the method section? \n\n• See above, the manuscript feels interrupted, since the description of the phosphoproteomic \ndataset is not followed by the analysis of the phosphorylation changes. \n\n \n\nPage 6, Results, Regulation of viral protein levels during CVB3 infection \n\n• This paragraph is purely descriptive and out of the context in this part of the manuscript. \n\n• I would suggest to link it ack to CVB3 biology, otherwise I would remove it completely. \n\n \n\nPage 7, Results, Changes in host protein levels during CVB3 infection \n\n• I think for a better reading of the manuscript the authors should consider to restructure the result \nsection and move this section after describing the proteome dataset. \n\n\f• See above, it is concerning to me that replicates of some timepoints do not cluster at all and look \nvery different in the heatmap. What is the reason for this? Was there a quality control for the \nreproducibility of infection? \n\n• Paragraph misses more biology, what are these proteins that escape virus-induced translational \nshutdown? \n\n• It would also be highly interesting if the authors could perform an integrative analysis of their data \nwith the CVB3 host-pathogen interaction network that was recently published in Nature \nMicrobiology. \n\n \n\nPage 8, Results, Regulation of protein phosphorylation levels during CVB3 infection \n\n• I think for a better reading of the manuscript the authors should consider to restructure the result \nsection and move this section after describing the phosphoproteome dataset. \n\n• At the end of the section, changes in phosphoproteome and total proteome are compared, I am \nwondering what the fraction of the phosphoproteome changes can in fact be explained by protein \nabundance changes. What is the fraction of phosphosites that change while there is no change \nobserved on the protein level? \n\n \n\nPage 8, Results, Identification of kinases that are activated by CVB3 infection \n\nIt is stated in the text that three findings of this analysis are of particular interest, such as \noverrepresentation of ATM/ATR motifs. Why are these findings of particular of particular interest? \nWhat does this mean for the infection process? What is the role of ATM/ATR or c PKC and GRKs in \nCVB3 infection? \n\n \n\n \n\nPage 9, Results, Phosphorylation of host factors involved in formation of viral replication organelles \n\nThe authors describe in this section the phosphorylation of know proteins that might play a role in \nreplication organelle formation and suggest that the specific sites might be relevant for the organelle \nformation, but again this is only descriptive and no biological follow up is provided. What is the \npossible impact of knowing the phosphorylation sites of these specific factors? Does mutating these \nsides lead to a reduction in viral replication? \n\n \n\nPage 11, Results, Signaling pathways affected during CVB3 infection \n\nMore details need to be provided how the pathway analysis was performed. What are the \npredefined pathways known to be affected by CVB3? Where are they taken from? Was the \n\n\fproteome and phosphoproteome data just mapped onto pathways or was an enrichment analysis \nperformed? Which input data was provided in for the pathway analysis, quantitative data across all \ntime points? \n\n \n\nPage 11, Results, Inactivation of mTORC1 signaling during CVB3 infection \n\nFigure 3 associated with this section is very difficult to read: \n\nI would suggest now to use the blue background, which reduces the contrast in the figure. +/- \nsymbols are so small that they are hard to see. Also it is not clear what the different colored arrows \nand lines mean. \n\nWhy was a higher MOI of 50 used for validating the phosphoproteomic data by Western Blot, which \nitself was generated at an MOI of 10. \n\n \n\nPage 14, Results, Activation of the mTORC1-regulated protein transcription factor EB \n\nHow was the knockout of TFEB performed? There is no information provided in the method section. \nAdditionally, validation of the ko is missing. \n\nWhat is the explanation that 8hpi a significant decrease in virus titers is observed upon ko of TFEB, \nhowever there is no significant difference 10hpi? Does this makes sense? \n\n \n\nPage 18, Experimental procedures \n\nAs mentioned in different comments above, a more detailed description of some of the methods is \nmissing and should be provided. \n\n\f", "role": "assistant" } ]	2,020
Dynamic remodelling of the human host cell proteome and phosphoproteome upon	REVIEWERS' COMMENTS: Reviewer #1 (Remarks to the Author): The revised version of this manuscript address all the questions and suggestions by the authors and in my opinion the majority of the concerns describe by the other reviewer. I believe that the inclusion of of additional experimental validations on GBF1 and PI4KB phosphorylation in the formation of the replication organelles and the connection of TFEB and secretory autophagy during CVB3 infection is interesting finding. I believe the paper has significantly increased in quality and I have not additional comments or concerns. Reviewer #2 (Remarks to the Author): For the revised version of the manuscript entitled “Dynamic remodeling of the host cell (phospho)proteome upon enterovirus infection” the authors addressed the majority of my comments. The main concerns that I raised included the focus on the previous version on the description of the dataset rather than emphasizing what the data tells us about the biology of enterovirus infection, the lack of functional follow up studies, and restructuring of the manuscript. The authors restructured the manuscript substantially as well as shorten some of the descriptive sections, so that it is more intuitive for the reader to follow the major findings of the study. In order to address the lack of biological insights extracted from the large scale dataset the authors added functional data characterizing the role of mTORC1 inactivation leading to increased TFEB activity as well as investigating the role of phosphorylation of host factors involved in the formation of viral replication organelles. I have a few remaining concerns/questions regarding the functional follow up findings: 1. The authors hypothesize that TFEB activation downstream of mTORC1 affect non-lytic virus release vis extracellular vesicles (EV). While the authors show a significantly reduced EV-enclosed virus release, but at the same time TFEB ko also reduces the naked virus, even though it is not significant reduced, which is likely due to variability in the measurements/experiment. Does this result imply that TFEB does not only affect the non-lytic but also the lytic virus release? How do the authors explain ths oberservation? 2. The last sentence of the Result section “These data lend further support to the idea that (entero)viruses not only hijack cellular pathways to build their replication organelles but that they also rewire them from their regulation in non-infected cells.” seems unclear to me. Please clarify the meaning of “…rewire them from their regulation in non-infected cells.”	REVIEWERS' COMMENTS: Reviewer #1 (Remarks to the Author): The revised version of this manuscript address all the questions and suggestions by the authors and in my opinion the majority of the concerns describe by the other reviewer. I believe that the inclusion of of additional experimental validations on GBF1 and PI4KB phosphorylation in the formation of the replication organelles and the connection of TFEB and secretory autophagy during CVB3 infection is interesting finding. I believe the paper has significantly increased in quality and I have not additional comments or concerns. Reviewer #2 (Remarks to the Author): For the revised version of the manuscript entitled “Dynamic remodeling of the host cell (phospho)proteome upon enterovirus infection” the authors addressed the majority of my comments. The main concerns that I raised included the focus on the previous version on the description of the dataset rather than emphasizing what the data tells us about the biology of enterovirus infection, the lack of functional follow up studies, and restructuring of the manuscript. The authors restructured the manuscript substantially as well as shorten some of the descriptive sections, so that it is more intuitive for the reader to follow the major findings of the study. In order to address the lack of biological insights extracted from the large scale dataset the authors added functional data characterizing the role of mTORC1 inactivation leading to increased TFEB activity as well as investigating the role of phosphorylation of host factors involved in the formation of viral replication organelles. I have a few remaining concerns/questions regarding the functional follow up findings: 1. The authors hypothesize that TFEB activation downstream of mTORC1 affect non-lytic virus release vis extracellular vesicles (EV). While the authors show a significantly reduced EV-enclosed virus release, but at the same time TFEB ko also reduces the naked virus, even though it is not significant reduced, which is likely due to variability in the measurements/experiment. Does this result imply that TFEB does not only affect the non-lytic but also the lytic virus release? How do the authors explain ths oberservation? 2. The last sentence of the Result section “These data lend further support to the idea that (entero)viruses not only hijack cellular pathways to build their replication organelles but that they also rewire them from their regulation in non-infected cells.” seems unclear to me. Please clarify the meaning of “…rewire them from their regulation in non-infected cells.”	TPR	NC_BS_BIOCH_002_RR2.txt	{ "criticism": 3, "example": 0, "importance_and_relevance": 0, "materials_and_methods": 4, "praise": 1, "presentation_and_reporting": 3, "results_and_discussion": 4, "suggestion_and_solution": 2, "total": 15 }	{ "criticism": 0.2, "example": 0, "importance_and_relevance": 0, "materials_and_methods": 0.26666666666666666, "praise": 0.06666666666666667, "presentation_and_reporting": 0.2, "results_and_discussion": 0.26666666666666666, "suggestion_and_solution": 0.13333333333333333 }	1.133333	[ { "content": "Please review this paper.", "role": "user" }, { "content": "\n\n\n\fREVIEWERS' COMMENTS: \n\n \n\nReviewer #1 (Remarks to the Author): \n\n \n\nThe revised version of this manuscript address all the questions and suggestions by the authors and \nin my opinion the majority of the concerns describe by the other reviewer. I believe that the \ninclusion of of additional experimental validations on GBF1 and PI4KB phosphorylation in the \nformation of the replication organelles and the connection of TFEB and secretory autophagy during \nCVB3 infection is interesting finding. I believe the paper has significantly increased in quality and I \nhave not additional comments or concerns. \n\n \n\n \n\nReviewer #2 (Remarks to the Author): \n\n \n\nFor the revised version of the manuscript entitled “Dynamic remodeling of the host cell \n(phospho)proteome upon enterovirus infection” the authors addressed the majority of my \ncomments. The main concerns that I raised included the focus on the previous version on the \ndescription of the dataset rather than emphasizing what the data tells us about the biology of \nenterovirus infection, the lack of functional follow up studies, and restructuring of the manuscript. \n\nThe authors restructured the manuscript substantially as well as shorten some of the descriptive \nsections, so that it is more intuitive for the reader to follow the major findings of the study. \n\nIn order to address the lack of biological insights extracted from the large scale dataset the authors \nadded functional data characterizing the role of mTORC1 inactivation leading to increased TFEB \nactivity as well as investigating the role of phosphorylation of host factors involved in the formation \nof viral replication organelles. \n\n \n\nI have a few remaining concerns/questions regarding the functional follow up findings: \n\n1. The authors hypothesize that TFEB activation downstream of mTORC1 affect non-lytic virus \nrelease vis extracellular vesicles (EV). While the authors show a significantly reduced EV-enclosed \nvirus release, but at the same time TFEB ko also reduces the naked virus, even though it is not \nsignificant reduced, which is likely due to variability in the measurements/experiment. Does this \nresult imply that TFEB does not only affect the non-lytic but also the lytic virus release? How do the \nauthors explain ths oberservation? \n\n2. The last sentence of the Result section “These data lend further support to the idea that \n(entero)viruses not only hijack cellular pathways to build their replication organelles but that they \n\n\falso rewire them from their regulation in non-infected cells.” seems unclear to me. Please clarify the \nmeaning of “…rewire them from their regulation in non-infected cells.” \n\n\f", "role": "assistant" } ]	2,020
Visualizing the enzyme mechanism of mevalonate diphosphate decarboxylase	REVIEWERS' COMMENTS: Reviewer #1 (Remarks to the Author): I am totally satisfied with the revisions made by the authors, in which all my comments have been addressed. Now the reaction mechanism of MDD proposed by the authors is updated appropriately and agrees well with recent knowledges of the enzyme. The comparison of the structures of MDD solved in the study with the previously known structures of the enzyme and its orthologs, considering the effect of crystal packing, makes sense. I still hope the catalytic mechanism of this interesting decarboxylase is elucidated by QM/MM study, but it is not necessary in this paper. The elucidation will be made in future probably by the authors or other groups. Reviewer #2 (Remarks to the Author): In the manuscript entitled "Visualizing the Enzyme Mechanism of Mevalonate Diphosphate Decarboxylase" by Chen et al. (Manuscript NCOMMS-19-27996A) the authors submitted a revised version of their manuscript. The revised version has substantially improved. (1) The structural comparison of different MDD structures available in the PDB improves the understanding of the differences of the structures for example different crystal packing effects and hence restricted conformational flexibility of the elements. See also the novel Supplementary Figure 7. Here, I just have a minor point: I personally find it useful if the PDB-ID for the displayed structure (panel c to e) is provided and/or a reference to the primary citation. This suggestion applies as well to Supplementary figures. The functional implications are now much better to follow. In particular, the discussion of the putative function of Asp282 is now better to understand. By the discussion of the other two recent publications, as suggested by the other reviewer, the authors were able to discard the contested hypothesis section on the mechanism. It is a very beautiful example how structural biology can provide insightful structural information to elucidate an enzymatic mechanism. I recommend to accept the current version of the manuscript. Minor points: - Page 12, line 11: Names of organism should be in italic - Supplementary Figure 5: I suggest deleting the primer sequences. They are given in the Materials and Methods section.	REVIEWERS' COMMENTS: Reviewer #1 (Remarks to the Author): I am totally satisfied with the revisions made by the authors, in which all my comments have been addressed. Now the reaction mechanism of MDD proposed by the authors is updated appropriately and agrees well with recent knowledges of the enzyme. The comparison of the structures of MDD solved in the study with the previously known structures of the enzyme and its orthologs, considering the effect of crystal packing, makes sense. I still hope the catalytic mechanism of this interesting decarboxylase is elucidated by QM/MM study, but it is not necessary in this paper. The elucidation will be made in future probably by the authors or other groups. Reviewer #2 (Remarks to the Author): In the manuscript entitled "Visualizing the Enzyme Mechanism of Mevalonate Diphosphate Decarboxylase" by Chen et al. (Manuscript NCOMMS-19-27996A) the authors submitted a revised version of their manuscript. The revised version has substantially improved. (1) The structural comparison of different MDD structures available in the PDB improves the understanding of the differences of the structures for example different crystal packing effects and hence restricted conformational flexibility of the elements. See also the novel Supplementary Figure 7. Here, I just have a minor point: I personally find it useful if the PDB-ID for the displayed structure (panel c to e) is provided and/or a reference to the primary citation. This suggestion applies as well to Supplementary figures. The functional implications are now much better to follow. In particular, the discussion of the putative function of Asp282 is now better to understand. By the discussion of the other two recent publications, as suggested by the other reviewer, the authors were able to discard the contested hypothesis section on the mechanism. It is a very beautiful example how structural biology can provide insightful structural information to elucidate an enzymatic mechanism. I recommend to accept the current version of the manuscript. Minor points: - Page 12, line 11: Names of organism should be in italic - Supplementary Figure 5: I suggest deleting the primer sequences. They are given in the Materials and Methods section.	TPR	NC_BS_BIOCH_003_RR2.txt	{ "criticism": 0, "example": 3, "importance_and_relevance": 1, "materials_and_methods": 6, "praise": 6, "presentation_and_reporting": 6, "results_and_discussion": 3, "suggestion_and_solution": 4, "total": 19 }	{ "criticism": 0, "example": 0.15789473684210525, "importance_and_relevance": 0.05263157894736842, "materials_and_methods": 0.3157894736842105, "praise": 0.3157894736842105, "presentation_and_reporting": 0.3157894736842105, "results_and_discussion": 0.15789473684210525, "suggestion_and_solution": 0.21052631578947367 }	1.526316	[ { "content": "Please review this paper.", "role": "user" }, { "content": "\n\n\fREVIEWERS' COMMENTS: \n\nReviewer #1 (Remarks to the Author): \n\nI am totally satisfied with the revisions made by the authors, in which all my comments have been \naddressed. Now the reaction mechanism of MDD proposed by the authors is updated appropriately \nand agrees well with recent knowledges of the enzyme. The comparison of the structures of MDD \nsolved in the study with the previously known structures of the enzyme and its orthologs, \nconsidering the effect of crystal packing, makes sense. I still hope the catalytic mechanism of this \ninteresting decarboxylase is elucidated by QM/MM study, but it is not necessary in this paper. The \nelucidation will be made in future probably by the authors or other groups. \n\nReviewer #2 (Remarks to the Author): \n\nIn the manuscript entitled \"Visualizing the Enzyme Mechanism of Mevalonate Diphosphate \nDecarboxylase\" by Chen et al. (Manuscript NCOMMS-19-27996A) the authors submitted a revised \nversion of their manuscript. \n\nThe revised version has substantially improved. \n(1) The structural comparison of different MDD structures available in the PDB improves the \nunderstanding of the differences of the structures for example different crystal packing effects and \nhence restricted conformational flexibility of the elements. See also the novel Supplementary \nFigure 7. Here, I just have a minor point: I personally find it useful if the PDB-ID for the displayed \nstructure (panel c to e) is provided and/or a reference to the primary citation. This suggestion \napplies as well to Supplementary figures. \n\nThe functional implications are now much better to follow. In particular, the discussion of the \nputative function of Asp282 is now better to understand. By the discussion of the other two recent \npublications, as suggested by the other reviewer, the authors were able to discard the contested \nhypothesis section on the mechanism. It is a very beautiful example how structural biology can \nprovide insightful structural information to elucidate an enzymatic mechanism. \n\nI recommend to accept the current version of the manuscript. \n\nMinor points: \n- Page 12, line 11: Names of organism should be in italic \n- Supplementary Figure 5: I suggest deleting the primer sequences. They are given in the \nMaterials and Methods section. \n\n\f", "role": "assistant" } ]	2,020
Visualizing the enzyme mechanism of mevalonate diphosphate decarboxylase	Reviewers' comments: Reviewer #1 (Remarks to the Author): The manuscript “Visualizing the enzyme mechanism of mevalonate diphosphate decarboxylase” by C.-L. Chen et al. reports the structural studies on mevalonate decarboxylase from a vancomycin resistant enterococcus, Enterococcus faecalis (MDDEF). The authors succeeded in solving not only the apo-structure but also the structures with the substrate mevalonate diphosphate (MVAPP), MVAPP and an ATP analog AMPPCP, MVAPP and ADP, or MVAPP and ADP/BeF3-. As observed with the structures of other MDDs, the MDDEF structures took open or closed conformations, which allowed the authors to suppose the reaction mechanisms of the enzyme through substrate binding and phosphoryl-transfer, which is the first half of the MDD reaction. It is quite impressive that the authors could verify the binding positions of magnesium ions in the catalytic center of the enzyme. To my knowledge, this is the first discovery for this enzyme family and also the information needed to reveal detailed catalytic mechanism of MDDEF and to develop inhibitor molecules specific to the enzyme. Because human also possesses MDD, only such specific inhibitors can be utilized as antibiotics for VREs. The metal binding sites are confirmed with anomalous dispersion observed in the analysis of the crystals soaked with MVAPP and ADP in the presence of cobalt ion, instead of magnesium ion. This study is basically an extension of their previous studies on the structures and biochemical properties of MDDEF, but it clearly made a substantial progress. The manuscript is readable, and all the data look sound and persuasive. The reviewer, however, still has several concerns with the manuscript. First, the authors proposed the substrate binding mechanism based on the open-close change observed in the enzyme structures through the bindings of different substrates or their analogs. The proposed mechanism fits well the structural changes observed with MDDEF, but not with other MDD structures previously reported. For example, the alfa10-beta4-loop was in the closed position in the MDDEF structure with MVAPP/ADP/BeF3-/Mg2+, but in the open position with MVAPP/AMPPCP/Mg2+, thus the authors concluded that the former structure represents the pre-phosphoryl transfer state, while the latter does an earlier state. In contrast, the loop, and also the Lys residue corresponding to Lys187, were reported to be at the same positions in MDD from an archaeon Sulfolobus solfataricus (MDDSS) not only in similar substrate-binding structures but also in its apo form (pdb: 5gmd, 5gme, 4z7c). The discrepancy casts question if the induced-fit mechanism proposed in this paper is general for MDDs or not. Additional discussion based on the comparison with the previously reported MDD structures is needed. Secondly, the mechanism of the ATP-dependent decarboxylation of MVAPP described in the manuscript, in which Asp282 acts as a base that deprotonate 3’-OH of MVAPP, has been controverted by recent finding of a MDD-homolog kinase that lacks the corresponding Asp residue but can catalyze 3’-phosphorylation of mevalonate (Vinokur J.M. et al., 2015, Protein Sci.), and also by a mutagenic study on MDDSS, which led the proposal of a hypothesis that Asp residue is required in the decarboxylation step (Motoyama K. et al., 2017, J. Biol. Chem.). If the authors want to refute the new hypothesis, they should perform QM/MM calculation study based on the structure of MDDEF binding MVAPP/ADP/BeF3-/Mg2+ or MVAPP/AMPPCP/Mg2+, which will clarify the protonation states of the substrates and catalytic residues and might support the less-persuasive discussion in page 15-16. Minor concerns: 1) Page 2, line 7: large conformational “change”? 2) Page 3, the 1st paragraph: The MEP pathway is found primarily in bacteria, not in prokaryotes. Archaea, another prokaryotes lineage, have the MVA pathway. 3) Page 3, line 8: privarily > primarily 4) Page 17, lines 19-: The method utilized for site-directed mutagenesis is obviously based on the QuickChange method. Appropriate references such as Zheng L. et al. (2004) Nucleic Acid Res. are needed. 5) Page 19, line 10: Instead of 100 nM MDDEF, a higher amount of the mutant enzyme was considered to be used. Please show the amount. 6) Page 21, lines 19-20: ligands “were” manually placed 7) Supplementary Table 4: The table seems not showing the angles of alfa-helices, but their 3D positions. I am afraid that the title and legend are confused with those in Supplementary Table 3. Reviewer #2 (Remarks to the Author): In the manuscript entitled "Visualizing the Enzyme Mechanism of Mevalonate Diphosphate Decarboxylase" by Chen et al. (Manuscript NCOMMS-19-27996-T) the authors describe that two Mg2+ ions are bound to the enzyme preloaded with substrate and ATP. By this sequential binding, the active site is shaped for catalysis. The authors used X-ray crystallography to visualize the Mg2+ ions plus structures with transition state analogues. The resolution of the diffraction data is very high and the quality of the refined structures is very good. Surprised by the localization of the two Mg2+ ions, the authors generated structure-based variants of the MDD protein. To the reviewer’s point of view the current version of the manuscript is very well written and comprehensive. Nevertheless, I have the impression that most of the techniques described were already very well established in the lab and for instance described in earlier publications (Chen et al. (2017) JBC. For instance, the crystallization system has been described in the JBC publication. Probably in many instances, the described methods could be drastically shortened. Most of the designed variants have already been shown, to have functional implications. It is not proven, that the proposed mechanism is valid in other organisms than S. faecalis as well. In particular, the beta10-alpha4 loops show very little conservation. What is the function of R192? It seems to be conserved in bacteria, where as in higher organisms it seems to be a strict threonine residue. Is the catalytic side in both monomers identical? Minor points: - Throughout the manuscript, there is a mixture of one-letter and three-letter code for amino acids. - Page 5, line 13: Space is missing 1.6 M - Page 7, line 7: “extra density”. Please be more precise “extra electron density” - Page 7, line 19: probably “…next section” is meant - Page 19, line 8: Space is missing after 30 - Page 19, line 18: please remove dot after Mus - Figure 1, panel a: I do not think that the topology diagram is necessary. The overall structure of MDDs is very well known - Figure 6, panel a: please carefully check the numbering of the alignment. I see some inconsistencies between the numbers of amino acids mentioned in the body of the text the provided numbering in the figure. Names of organism should be in italic.	Reviewers' comments: Reviewer #1 (Remarks to the Author): The manuscript “Visualizing the enzyme mechanism of mevalonate diphosphate decarboxylase” by C.-L. Chen et al. reports the structural studies on mevalonate decarboxylase from a vancomycin resistant enterococcus, Enterococcus faecalis (MDDEF). The authors succeeded in solving not only the apo-structure but also the structures with the substrate mevalonate diphosphate (MVAPP), MVAPP and an ATP analog AMPPCP, MVAPP and ADP, or MVAPP and ADP/BeF3-. As observed with the structures of other MDDs, the MDDEF structures took open or closed conformations, which allowed the authors to suppose the reaction mechanisms of the enzyme through substrate binding and phosphoryl-transfer, which is the first half of the MDD reaction. It is quite impressive that the authors could verify the binding positions of magnesium ions in the catalytic center of the enzyme. To my knowledge, this is the first discovery for this enzyme family and also the information needed to reveal detailed catalytic mechanism of MDDEF and to develop inhibitor molecules specific to the enzyme. Because human also possesses MDD, only such specific inhibitors can be utilized as antibiotics for VREs. The metal binding sites are confirmed with anomalous dispersion observed in the analysis of the crystals soaked with MVAPP and ADP in the presence of cobalt ion, instead of magnesium ion. This study is basically an extension of their previous studies on the structures and biochemical properties of MDDEF, but it clearly made a substantial progress. The manuscript is readable, and all the data look sound and persuasive. The reviewer, however, still has several concerns with the manuscript. First, the authors proposed the substrate binding mechanism based on the open-close change observed in the enzyme structures through the bindings of different substrates or their analogs. The proposed mechanism fits well the structural changes observed with MDDEF, but not with other MDD structures previously reported. For example, the alfa10-beta4-loop was in the closed position in the MDDEF structure with MVAPP/ADP/BeF3-/Mg2+, but in the open position with MVAPP/AMPPCP/Mg2+, thus the authors concluded that the former structure represents the pre-phosphoryl transfer state, while the latter does an earlier state. In contrast, the loop, and also the Lys residue corresponding to Lys187, were reported to be at the same positions in MDD from an archaeon Sulfolobus solfataricus (MDDSS) not only in similar substrate-binding structures but also in its apo form (pdb: 5gmd, 5gme, 4z7c). The discrepancy casts question if the induced-fit mechanism proposed in this paper is general for MDDs or not. Additional discussion based on the comparison with the previously reported MDD structures is needed. Secondly, the mechanism of the ATP-dependent decarboxylation of MVAPP described in the manuscript, in which Asp282 acts as a base that deprotonate 3’-OH of MVAPP, has been controverted by recent finding of a MDD-homolog kinase that lacks the corresponding Asp residue but can catalyze 3’-phosphorylation of mevalonate (Vinokur J.M. et al., 2015, Protein Sci.), and also by a mutagenic study on MDDSS, which led the proposal of a hypothesis that Asp residue is required in the decarboxylation step (Motoyama K. et al., 2017, J. Biol. Chem.). If the authors want to refute the new hypothesis, they should perform QM/MM calculation study based on the structure of MDDEF binding MVAPP/ADP/BeF3-/Mg2+ or MVAPP/AMPPCP/Mg2+, which will clarify the protonation states of the substrates and catalytic residues and might support the less-persuasive discussion in page 15-16. Minor concerns: 1) Page 2, line 7: large conformational “change”? 2) Page 3, the 1st paragraph: The MEP pathway is found primarily in bacteria, not in prokaryotes. Archaea, another prokaryotes lineage, have the MVA pathway. 3) Page 3, line 8: privarily > primarily 4) Page 17, lines 19-: The method utilized for site-directed mutagenesis is obviously based on the QuickChange method. Appropriate references such as Zheng L. et al. (2004) Nucleic Acid Res. are needed. 5) Page 19, line 10: Instead of 100 nM MDDEF, a higher amount of the mutant enzyme was considered to be used. Please show the amount. 6) Page 21, lines 19-20: ligands “were” manually placed 7) Supplementary Table 4: The table seems not showing the angles of alfa-helices, but their 3D positions. I am afraid that the title and legend are confused with those in Supplementary Table 3. Reviewer #2 (Remarks to the Author): In the manuscript entitled "Visualizing the Enzyme Mechanism of Mevalonate Diphosphate Decarboxylase" by Chen et al. (Manuscript NCOMMS-19-27996-T) the authors describe that two Mg2+ ions are bound to the enzyme preloaded with substrate and ATP. By this sequential binding, the active site is shaped for catalysis. The authors used X-ray crystallography to visualize the Mg2+ ions plus structures with transition state analogues. The resolution of the diffraction data is very high and the quality of the refined structures is very good. Surprised by the localization of the two Mg2+ ions, the authors generated structure-based variants of the MDD protein. To the reviewer’s point of view the current version of the manuscript is very well written and comprehensive. Nevertheless, I have the impression that most of the techniques described were already very well established in the lab and for instance described in earlier publications (Chen et al. (2017) JBC. For instance, the crystallization system has been described in the JBC publication. Probably in many instances, the described methods could be drastically shortened. Most of the designed variants have already been shown, to have functional implications. It is not proven, that the proposed mechanism is valid in other organisms than S. faecalis as well. In particular, the beta10-alpha4 loops show very little conservation. What is the function of R192? It seems to be conserved in bacteria, where as in higher organisms it seems to be a strict threonine residue. Is the catalytic side in both monomers identical? Minor points: - Throughout the manuscript, there is a mixture of one-letter and three-letter code for amino acids. - Page 5, line 13: Space is missing 1.6 M - Page 7, line 7: “extra density”. Please be more precise “extra electron density” - Page 7, line 19: probably “…next section” is meant - Page 19, line 8: Space is missing after 30 - Page 19, line 18: please remove dot after Mus - Figure 1, panel a: I do not think that the topology diagram is necessary. The overall structure of MDDs is very well known - Figure 6, panel a: please carefully check the numbering of the alignment. I see some inconsistencies between the numbers of amino acids mentioned in the body of the text the provided numbering in the figure. Names of organism should be in italic.	TPR	NC_BS_BIOCH_003_RR1.txt	{ "criticism": 6, "example": 11, "importance_and_relevance": 2, "materials_and_methods": 27, "praise": 6, "presentation_and_reporting": 13, "results_and_discussion": 3, "suggestion_and_solution": 11, "total": 56 }	{ "criticism": 0.10714285714285714, "example": 0.19642857142857142, "importance_and_relevance": 0.03571428571428571, "materials_and_methods": 0.48214285714285715, "praise": 0.10714285714285714, "presentation_and_reporting": 0.23214285714285715, "results_and_discussion": 0.05357142857142857, "suggestion_and_solution": 0.19642857142857142 }	1.410714	[ { "content": "Please review this paper.", "role": "user" }, { "content": "Reviewers' comments: \n\nReviewer #1 (Remarks to the Author): \n\nThe manuscript “Visualizing the enzyme mechanism of mevalonate diphosphate decarboxylase” by \nC.-L. Chen et al. reports the structural studies on mevalonate decarboxylase from a vancomycin \nresistant enterococcus, Enterococcus faecalis (MDDEF). The authors succeeded in solving not only \nthe apo-structure but also the structures with the substrate mevalonate diphosphate (MVAPP), \nMVAPP and an ATP analog AMPPCP, MVAPP and ADP, or MVAPP and ADP/BeF3-. As observed with \nthe structures of other MDDs, the MDDEF structures took open or closed conformations, which \nallowed the authors to suppose the reaction mechanisms of the enzyme through substrate binding \nand phosphoryl-transfer, which is the first half of the MDD reaction. It is quite impressive that the \nauthors could verify the binding positions of magnesium ions in the catalytic center of the enzyme. \nTo my knowledge, this is the first discovery for this enzyme family and also the information \nneeded to reveal detailed catalytic mechanism of MDDEF and to develop inhibitor molecules \nspecific to the enzyme. Because human also possesses MDD, only such specific inhibitors can be \nutilized as antibiotics for VREs. The metal binding sites are confirmed with anomalous dispersion \nobserved in the analysis of the crystals soaked with MVAPP and ADP in the presence of cobalt ion, \ninstead of magnesium ion. This study is basically an extension of their previous studies on the \nstructures and biochemical properties of MDDEF, but it clearly made a substantial progress. The \nmanuscript is readable, and all the data look sound and persuasive. The reviewer, however, still \nhas several concerns with the manuscript. First, the authors proposed the substrate binding \nmechanism based on the open-close change observed in the enzyme structures through the \nbindings of different substrates or their analogs. The proposed mechanism fits well the structural \nchanges observed with MDDEF, but not with other MDD structures previously reported. For \nexample, the alfa10-beta4-loop was in the closed position in the MDDEF structure with \nMVAPP/ADP/BeF3-/Mg2+, but in the open position with MVAPP/AMPPCP/Mg2+, thus the authors \nconcluded that the former structure represents the pre-phosphoryl transfer state, while the latter \ndoes an earlier state. In contrast, the loop, and also the Lys residue corresponding to Lys187, \nwere reported to be at the same positions in MDD from an archaeon Sulfolobus solfataricus \n(MDDSS) not only in similar substrate-binding structures but also in its apo form (pdb: 5gmd, \n5gme, 4z7c). The discrepancy casts question if the induced-fit mechanism proposed in this paper \nis general for MDDs or not. Additional discussion based on the comparison with the previously \nreported MDD structures is needed. Secondly, the mechanism of the ATP-dependent \ndecarboxylation of MVAPP described in the manuscript, in which Asp282 acts as a base that \ndeprotonate 3’-OH of MVAPP, has been controverted by recent finding of a MDD-homolog kinase \nthat lacks the corresponding Asp residue but can catalyze 3’-phosphorylation of mevalonate \n(Vinokur J.M. et al., 2015, Protein Sci.), and also by a mutagenic study on MDDSS, which led the \nproposal of a hypothesis that Asp residue is required in the decarboxylation step (Motoyama K. et \nal., 2017, J. Biol. Chem.). If the authors want to refute the new hypothesis, they should perform \nQM/MM calculation study based on the structure of MDDEF binding MVAPP/ADP/BeF3-/Mg2+ or \nMVAPP/AMPPCP/Mg2+, which will clarify the protonation states of the substrates and catalytic \nresidues and might support the less-persuasive discussion in page 15-16. \n\nMinor concerns: \n1) Page 2, line 7: large conformational “change”? \n2) Page 3, the 1st paragraph: The MEP pathway is found primarily in bacteria, not in prokaryotes. \nArchaea, another prokaryotes lineage, have the MVA pathway. \n3) Page 3, line 8: privarily > primarily \n4) Page 17, lines 19-: The method utilized for site-directed mutagenesis is obviously based on the \nQuickChange method. Appropriate references such as Zheng L. et al. (2004) Nucleic Acid Res. are \nneeded. \n5) Page 19, line 10: Instead of 100 nM MDDEF, a higher amount of the mutant enzyme was \nconsidered to be used. Please show the amount. \n6) Page 21, lines 19-20: ligands “were” manually placed \n\n\f7) Supplementary Table 4: The table seems not showing the angles of alfa-helices, but their 3D \npositions. I am afraid that the title and legend are confused with those in Supplementary Table 3. \n\nReviewer #2 (Remarks to the Author): \n\nIn the manuscript entitled \"Visualizing the Enzyme Mechanism of Mevalonate Diphosphate \nDecarboxylase\" by Chen et al. (Manuscript NCOMMS-19-27996-T) the authors describe that two \nMg2+ ions are bound to the enzyme preloaded with substrate and ATP. By this sequential binding, \nthe active site is shaped for catalysis. \nThe authors used X-ray crystallography to visualize the Mg2+ ions plus structures with transition \nstate analogues. The resolution of the diffraction data is very high and the quality of the refined \nstructures is very good. Surprised by the localization of the two Mg2+ ions, the authors generated \nstructure-based variants of the MDD protein. \nTo the reviewer’s point of view the current version of the manuscript is very well written and \ncomprehensive. Nevertheless, I have the impression that most of the techniques described were \nalready very well established in the lab and for instance described in earlier publications (Chen et \nal. (2017) JBC. For instance, the crystallization system has been described in the JBC publication. \nProbably in many instances, the described methods could be drastically shortened. Most of the \ndesigned variants have already been shown, to have functional implications. It is not proven, that \nthe proposed mechanism is valid in other organisms than S. faecalis as well. In particular, the \nbeta10-alpha4 loops show very little conservation. What is the function of R192? It seems to be \nconserved in bacteria, where as in higher organisms it seems to be a strict threonine residue. Is \nthe catalytic side in both monomers identical? \n\nMinor points: \n- Throughout the manuscript, there is a mixture of one-letter and three-letter code for amino \nacids. \n- Page 5, line 13: Space is missing 1.6 M \n- Page 7, line 7: “extra density”. Please be more precise “extra electron density” \n- Page 7, line 19: probably “…next section” is meant \n- Page 19, line 8: Space is missing after 30 \n- Page 19, line 18: please remove dot after Mus \n- Figure 1, panel a: I do not think that the topology diagram is necessary. The overall structure of \nMDDs is very well known \n- Figure 6, panel a: please carefully check the numbering of the alignment. I see some \ninconsistencies between the numbers of amino acids mentioned in the body of the text the \nprovided numbering in the figure. Names of organism should be in italic. \n\n\f\t\n\n\f\n", "role": "assistant" } ]	2,020
Distinct Cdk9-phosphatase switches act at the beginning and end of elongation by RNA	Reviewer #1 (Remarks to the Author): The work by Parua et all demonstrates a conserved role of distinct phosphatases that work on SPT5 at distinct steps of the transcription cycle. By combining in vitro phosphatase assays with in vivo distribution of transcriptional components following perturbation (by mutation or drug), this work makes a compelling case for further dissection of transcription elongation. As much as genomics is powerful, it is Figure 1 that is the highlight of the manuscript. Overall, it is a well thought out and well controlled piece that merits publication once minor questions are addressed. Minor points. 55-74. The introduction is hard to follow in places in the last two paragraphs, but especially the one containing lines 55-74. I read it several times and could not quite grasp it. Is there a way to shorten it? 84. Introduce the abbreviation in the Abstract for consistency? 128. Higher sensitivity of Spt5 “CTD” phosphorylation compared to RPB1’s is intriguing. Flavopiridol has been shown in several systems to lead to a decrease in Ser2 phosphorylation. This is also stated in lines 194-195. Is this discrepancy due to technicalities (such as timing of treatment and the use of a more specific CDK9 inhibitor) or something else such as functional heterogeneity of Pol II Ser2p populations, differential requirements for CDK9 for Ser2 phosphorylation in different cells, etc? 166-167. The sentence “Reactivity with Spt5…” may be correct technically, but it is not clear what to make of it in Supplementary Fig 2B. It seems to me that the effect of 806 substitution is not only modest (and may be should be quantified), but the IP datapoints do not appear to add useful information. Supplementary 2d and 2e. These appear to be shown in linear scale. Would it be better to show these in log scale? Secondly, I could not find information (which I may have missed) what signal was quantified for these plots: signal around genes or gene-agnostic data centric peaks? 182. Principal component analysis is highly sensitive to normalization. Normalization is impossible to obtain for ChIP comparing different antibodies. PCA has limited use for samples (such as RNA-seq) that are highly similar and where normalization is straightforward, but it has no use for ChIP-seq. This panel should be removed. Perhaps clustering analysis that is less sensitive to normalization could be used? Overall, this panel does not add much to the manuscript anyway so deleting this will not harm it. 212-214. Heatmaps of different versions of Spt5 are disjointed between figures 3 and 4. I understand that this is done to preserve the flow of the manuscript, but it possible to add a metagene plot overlapping the profiles of wild type and two mutants? It may be worthwhile to make the overlap plot a new 4A and move the orphan heatmap to supplementary, for example. Lines 221-230. Authors mention transcription induction of CDKN1A. Is transcriptional induction shown somewhere by RT-qPCR? I apologize in advance if I’d missed it. Lines 233-240. Along the same lines, what is happening to expression and ChIP-seq signal on genes that are not induced by P53 stabilization? Figure 5B. Metagene plots in 5B are not convincing along the vertical axis. Spike-in based normalization for ChIP-seq is impossible to very difficult with a notable exception of recombinant nucleosomes. A better figure would be to show the same metaplots, but for genes that are not upregulated by Nutlin. These must show less to no increase of light blue signal relatively to genes shown in 5b. This control is also essential. Figure 7a-c. The gene identity (MYC) is better to be labeled in the figure itself rather than only in a legend to 7B. It takes a moment to decipher that 7C probably shows MYC based on matching primer numbers. 326. Should it be ‘marks’ (plural)? 380-390. Because these are important statements, these should almost be placed to the beginning of discussion? 392. Delete ‘Therefore’. 512. Was Fisher formaldehyde regular or methanol-free that is sold in ampules? 516. Specify the power number for Bioruptor? The high setting for ChIP is considered to be 140 if I recall correctly, but the instrument can go higher. Data handling. Which samples contained spike-ins? Data handling. How Sacharomyces spike-ins are helpful when added to immunoprecipitation reactions? The antibodies used will not recognize yeast proteins and will result in unknown and highly variable signal recovery for yeast read among samples that is likely driven by noise. Reviewer #2 (Remarks to the Author): Papua et al. address an interesting and important question in gene expression regulation by investigating how SPT5 phosphorylation varies across gene bodies. The authors previously showed in fission yeast that P-TEFb dependent SPT5 phosphorylations were lost at transcription termination sites through competition between the P-TEFb kinase and PP1 phosphatase. Here they have extended these studies to look at the role of various phosphatases during transcription elongation in human cells. The authors probe two different SPT5 phosphorylation sites, Ser666 and Thr806. They find that these phosphorylations are added and lost at different points during transcription. PP1 phosphatases are responsible for removing the phosphorylation on CTR-I at termination sites and PP4 phosphatase removes Ser666 phosphorylation at promoter-proximal sites. This paper is highly important because it shows that in addition to RNA polymerase II, specific sites on SPT5 are phosphorylated in response to its spatial location on a gene body. The study poses an enticing question as to whether other transcription elongation factors are regulated in a similar way. If true, this would provide an additional mechanism to carefully regulate spatial recruitment of additional factors to the elongation machinery. Overall, the manuscript is well written, and the data are clear. If some specific issues are addressed (see below), the reviewer fully supports publication of this important work in Nature Communications. Specific comments -Line 49: P-TEFb phosphorylates components of elongation and pausing complexes. The reviewer does not know of additional pausing factors that P-TEFb phosphorylates (NELF and DSIF are already mentioned). The authors should change “the paused complex” to “elongation factors”. -Line 128: The authors should cite the work of Matthias Geyer (2012 Czudnochowski et al., Nature Communications) where it was shown that P-TEFb is not a Ser2 kinase in vitro, specifically because the authors use P-TEFb in vitro in this work. -Supplementary Fig 1e: Not obvious that chromatin bound Thr806 phosphorylation level is changing in the PP1 depleted sample. It is also not clear that this blot is necessary given the other data in the manuscript. -Supplementary Fig 2a- Loading controls suggest that there is more GST-CTR1 input than the other peptides (significantly more). The authors comment in the legend that this is due to incomplete stripping of the membrane. This is a really important control, so it is advised the experiment is repeated and the sample is evenly split onto two gels/membranes to show the same amount of sample was indeed used, specifically because some faint bands appear on the phosphoblot for the other CTR constructs. The blots with the full-length protein look good. -Figure 2B: Are the ChIP plots scaled the same way? Add Y value to plots or state in figure legend that the scaling is the same. -Figure 2C, 3C: Plot the curves with the same Y axis boundaries to show the extent of differences. -Figure 3A, 4D, 7A S3A, S4 C, D: Mark termination sites or CPS sites on gene overview. There are probably multiple sites, but it helps to know where transcription may stop on a particular gene. -Figure 5C: Is it possible to comment on why Nutlin has a greater effect on non-paused genes? The model posited by the authors suggests that Ser666 phosphorylation follows pause release, and it would be naively thought that the effect of Nutlin would be greater on such genes. -Figure 7e: The authors see increased Pol II in the gene body when PP4R2 is depleted. This can come from enhanced transcription but can also arise from slow Pol II. If at all possible, the authors should investigate Pol II elongation rates in PP4R2 depleted cells using Gro-seq or TT-seq. This would confirm that enhanced Spt5 Ser666 phosphorylation leads to pause release and faster elongation rates. Minor comment -Supplementary Figure 3: Flip panel order so they match Figure 3.	Reviewer #1 (Remarks to the Author): The work by Parua et all demonstrates a conserved role of distinct phosphatases that work on SPT5 at distinct steps of the transcription cycle. By combining in vitro phosphatase assays with in vivo distribution of transcriptional components following perturbation (by mutation or drug), this work makes a compelling case for further dissection of transcription elongation. As much as genomics is powerful, it is Figure 1 that is the highlight of the manuscript. Overall, it is a well thought out and well controlled piece that merits publication once minor questions are addressed. Minor points. 55-74. The introduction is hard to follow in places in the last two paragraphs, but especially the one containing lines 55-74. I read it several times and could not quite grasp it. Is there a way to shorten it? 84. Introduce the abbreviation in the Abstract for consistency? 128. Higher sensitivity of Spt5 “CTD” phosphorylation compared to RPB1’s is intriguing. Flavopiridol has been shown in several systems to lead to a decrease in Ser2 phosphorylation. This is also stated in lines 194-195. Is this discrepancy due to technicalities (such as timing of treatment and the use of a more specific CDK9 inhibitor) or something else such as functional heterogeneity of Pol II Ser2p populations, differential requirements for CDK9 for Ser2 phosphorylation in different cells, etc? 166-167. The sentence “Reactivity with Spt5…” may be correct technically, but it is not clear what to make of it in Supplementary Fig 2B. It seems to me that the effect of 806 substitution is not only modest (and may be should be quantified), but the IP datapoints do not appear to add useful information. Supplementary 2d and 2e. These appear to be shown in linear scale. Would it be better to show these in log scale? Secondly, I could not find information (which I may have missed) what signal was quantified for these plots: signal around genes or gene-agnostic data centric peaks? 182. Principal component analysis is highly sensitive to normalization. Normalization is impossible to obtain for ChIP comparing different antibodies. PCA has limited use for samples (such as RNA-seq) that are highly similar and where normalization is straightforward, but it has no use for ChIP-seq. This panel should be removed. Perhaps clustering analysis that is less sensitive to normalization could be used? Overall, this panel does not add much to the manuscript anyway so deleting this will not harm it. 212-214. Heatmaps of different versions of Spt5 are disjointed between figures 3 and 4. I understand that this is done to preserve the flow of the manuscript, but it possible to add a metagene plot overlapping the profiles of wild type and two mutants? It may be worthwhile to make the overlap plot a new 4A and move the orphan heatmap to supplementary, for example. Lines 221-230. Authors mention transcription induction of CDKN1A. Is transcriptional induction shown somewhere by RT-qPCR? I apologize in advance if I’d missed it. Lines 233-240. Along the same lines, what is happening to expression and ChIP-seq signal on genes that are not induced by P53 stabilization? Figure 5B. Metagene plots in 5B are not convincing along the vertical axis. Spike-in based normalization for ChIP-seq is impossible to very difficult with a notable exception of recombinant nucleosomes. A better figure would be to show the same metaplots, but for genes that are not upregulated by Nutlin. These must show less to no increase of light blue signal relatively to genes shown in 5b. This control is also essential. Figure 7a-c. The gene identity (MYC) is better to be labeled in the figure itself rather than only in a legend to 7B. It takes a moment to decipher that 7C probably shows MYC based on matching primer numbers. 326. Should it be ‘marks’ (plural)? 380-390. Because these are important statements, these should almost be placed to the beginning of discussion? 392. Delete ‘Therefore’. 512. Was Fisher formaldehyde regular or methanol-free that is sold in ampules? 516. Specify the power number for Bioruptor? The high setting for ChIP is considered to be 140 if I recall correctly, but the instrument can go higher. Data handling. Which samples contained spike-ins? Data handling. How Sacharomyces spike-ins are helpful when added to immunoprecipitation reactions? The antibodies used will not recognize yeast proteins and will result in unknown and highly variable signal recovery for yeast read among samples that is likely driven by noise. Reviewer #2 (Remarks to the Author): Papua et al. address an interesting and important question in gene expression regulation by investigating how SPT5 phosphorylation varies across gene bodies. The authors previously showed in fission yeast that P-TEFb dependent SPT5 phosphorylations were lost at transcription termination sites through competition between the P-TEFb kinase and PP1 phosphatase. Here they have extended these studies to look at the role of various phosphatases during transcription elongation in human cells. The authors probe two different SPT5 phosphorylation sites, Ser666 and Thr806. They find that these phosphorylations are added and lost at different points during transcription. PP1 phosphatases are responsible for removing the phosphorylation on CTR-I at termination sites and PP4 phosphatase removes Ser666 phosphorylation at promoter-proximal sites. This paper is highly important because it shows that in addition to RNA polymerase II, specific sites on SPT5 are phosphorylated in response to its spatial location on a gene body. The study poses an enticing question as to whether other transcription elongation factors are regulated in a similar way. If true, this would provide an additional mechanism to carefully regulate spatial recruitment of additional factors to the elongation machinery. Overall, the manuscript is well written, and the data are clear. If some specific issues are addressed (see below), the reviewer fully supports publication of this important work in Nature Communications. Specific comments -Line 49: P-TEFb phosphorylates components of elongation and pausing complexes. The reviewer does not know of additional pausing factors that P-TEFb phosphorylates (NELF and DSIF are already mentioned). The authors should change “the paused complex” to “elongation factors”. -Line 128: The authors should cite the work of Matthias Geyer (2012 Czudnochowski et al., Nature Communications) where it was shown that P-TEFb is not a Ser2 kinase in vitro, specifically because the authors use P-TEFb in vitro in this work. -Supplementary Fig 1e: Not obvious that chromatin bound Thr806 phosphorylation level is changing in the PP1 depleted sample. It is also not clear that this blot is necessary given the other data in the manuscript. -Supplementary Fig 2a- Loading controls suggest that there is more GST-CTR1 input than the other peptides (significantly more). The authors comment in the legend that this is due to incomplete stripping of the membrane. This is a really important control, so it is advised the experiment is repeated and the sample is evenly split onto two gels/membranes to show the same amount of sample was indeed used, specifically because some faint bands appear on the phosphoblot for the other CTR constructs. The blots with the full-length protein look good. -Figure 2B: Are the ChIP plots scaled the same way? Add Y value to plots or state in figure legend that the scaling is the same. -Figure 2C, 3C: Plot the curves with the same Y axis boundaries to show the extent of differences. -Figure 3A, 4D, 7A S3A, S4 C, D: Mark termination sites or CPS sites on gene overview. There are probably multiple sites, but it helps to know where transcription may stop on a particular gene. -Figure 5C: Is it possible to comment on why Nutlin has a greater effect on non-paused genes? The model posited by the authors suggests that Ser666 phosphorylation follows pause release, and it would be naively thought that the effect of Nutlin would be greater on such genes. -Figure 7e: The authors see increased Pol II in the gene body when PP4R2 is depleted. This can come from enhanced transcription but can also arise from slow Pol II. If at all possible, the authors should investigate Pol II elongation rates in PP4R2 depleted cells using Gro-seq or TT-seq. This would confirm that enhanced Spt5 Ser666 phosphorylation leads to pause release and faster elongation rates. Minor comment -Supplementary Figure 3: Flip panel order so they match Figure 3.	TPR	NC_BS_BIOCH_004_RR1.txt	{ "criticism": 12, "example": 20, "importance_and_relevance": 3, "materials_and_methods": 39, "praise": 4, "presentation_and_reporting": 24, "results_and_discussion": 3, "suggestion_and_solution": 23, "total": 99 }	{ "criticism": 0.12121212121212122, "example": 0.20202020202020202, "importance_and_relevance": 0.030303030303030304, "materials_and_methods": 0.3939393939393939, "praise": 0.04040404040404041, "presentation_and_reporting": 0.24242424242424243, "results_and_discussion": 0.030303030303030304, "suggestion_and_solution": 0.23232323232323232 }	1.292929	[ { "content": "Please review this paper.", "role": "user" }, { "content": "Reviewer #1 (Remarks to the Author): \n\nThe work by Parua et all demonstrates a conserved role of distinct phosphatases that work on SPT5 at \ndistinct steps of the transcription cycle. By combining in vitro phosphatase assays with in vivo \ndistribution of transcriptional components following perturbation (by mutation or drug), this work \nmakes a compelling case for further dissection of transcription elongation. As much as genomics is \npowerful, it is Figure 1 that is the highlight of the manuscript. Overall, it is a well thought out and well \ncontrolled piece that merits publication once minor questions are addressed. \n\nMinor points. \n55-74. The introduction is hard to follow in places in the last two paragraphs, but especially the one \ncontaining lines 55-74. I read it several times and could not quite grasp it. Is there a way to shorten \nit? \n\n84. Introduce the abbreviation in the Abstract for consistency? \n\n128. Higher sensitivity of Spt5 “CTD” phosphorylation compared to RPB1’s is intriguing. Flavopiridol \nhas been shown in several systems to lead to a decrease in Ser2 phosphorylation. This is also stated \nin lines 194-195. Is this discrepancy due to technicalities (such as timing of treatment and the use of \na more specific CDK9 inhibitor) or something else such as functional heterogeneity of Pol II Ser2p \npopulations, differential requirements for CDK9 for Ser2 phosphorylation in different cells, etc? \n\n166-167. The sentence “Reactivity with Spt5…” may be correct technically, but it is not clear what to \nmake of it in Supplementary Fig 2B. It seems to me that the effect of 806 substitution is not only \nmodest (and may be should be quantified), but the IP datapoints do not appear to add useful \ninformation. \n\nSupplementary 2d and 2e. These appear to be shown in linear scale. Would it be better to show these \nin log scale? Secondly, I could not find information (which I may have missed) what signal was \nquantified for these plots: signal around genes or gene-agnostic data centric peaks? \n\n182. Principal component analysis is highly sensitive to normalization. Normalization is impossible to \nobtain for ChIP comparing different antibodies. PCA has limited use for samples (such as RNA-seq) \nthat are highly similar and where normalization is straightforward, but it has no use for ChIP-seq. This \npanel should be removed. Perhaps clustering analysis that is less sensitive to normalization could be \nused? Overall, this panel does not add much to the manuscript anyway so deleting this will not harm \nit. \n\n212-214. Heatmaps of different versions of Spt5 are disjointed between figures 3 and 4. I understand \nthat this is done to preserve the flow of the manuscript, but it possible to add a metagene plot \noverlapping the profiles of wild type and two mutants? It may be worthwhile to make the overlap plot \na new 4A and move the orphan heatmap to supplementary, for example. \n\nLines 221-230. Authors mention transcription induction of CDKN1A. Is transcriptional induction shown \nsomewhere by RT-qPCR? I apologize in advance if I’d missed it. \n\nLines 233-240. Along the same lines, what is happening to expression and ChIP-seq signal on genes \nthat are not induced by P53 stabilization? \n\nFigure 5B. Metagene plots in 5B are not convincing along the vertical axis. Spike-in based \nnormalization for ChIP-seq is impossible to very difficult with a notable exception of recombinant \nnucleosomes. A better figure would be to show the same metaplots, but for genes that are not \nupregulated by Nutlin. These must show less to no increase of light blue signal relatively to genes \nshown in 5b. This control is also essential. \n\n\fFigure 7a-c. The gene identity (MYC) is better to be labeled in the figure itself rather than only in a \nlegend to 7B. It takes a moment to decipher that 7C probably shows MYC based on matching primer \nnumbers. \n\n326. Should it be ‘marks’ (plural)? \n\n380-390. Because these are important statements, these should almost be placed to the beginning of \ndiscussion? \n\n392. Delete ‘Therefore’. \n\n512. Was Fisher formaldehyde regular or methanol-free that is sold in ampules? \n\n516. Specify the power number for Bioruptor? The high setting for ChIP is considered to be 140 if I \nrecall correctly, but the instrument can go higher. \n\nData handling. Which samples contained spike-ins? \n\nData handling. How Sacharomyces spike-ins are helpful when added to immunoprecipitation \nreactions? The antibodies used will not recognize yeast proteins and will result in unknown and highly \nvariable signal recovery for yeast read among samples that is likely driven by noise. \n\nReviewer #2 (Remarks to the Author): \n\nPapua et al. address an interesting and important question in gene expression regulation by \ninvestigating how SPT5 phosphorylation varies across gene bodies. The authors previously showed in \nfission yeast that P-TEFb dependent SPT5 phosphorylations were lost at transcription termination sites \nthrough competition between the P-TEFb kinase and PP1 phosphatase. Here they have extended these \nstudies to look at the role of various phosphatases during transcription elongation in human cells. The \nauthors probe two different SPT5 phosphorylation sites, Ser666 and Thr806. They find that these \nphosphorylations are added and lost at different points during transcription. PP1 phosphatases are \nresponsible for removing the phosphorylation on CTR-I at termination sites and PP4 phosphatase \nremoves Ser666 phosphorylation at promoter-proximal sites. This paper is highly important because it \nshows that in addition to RNA polymerase II, specific sites on SPT5 are phosphorylated in response to \nits spatial location on a gene body. The study poses an enticing question as to whether other \ntranscription elongation factors are regulated in a similar way. If true, this would provide an additional \nmechanism to carefully regulate spatial recruitment of additional factors to the elongation machinery. \nOverall, the manuscript is well written, and the data are clear. If some specific issues are addressed \n(see below), the reviewer fully supports publication of this important work in Nature Communications. \n\nSpecific comments \n-Line 49: P-TEFb phosphorylates components of elongation and pausing complexes. The reviewer does \nnot know of additional pausing factors that P-TEFb phosphorylates (NELF and DSIF are already \nmentioned). The authors should change “the paused complex” to “elongation factors”. \n\n-Line 128: The authors should cite the work of Matthias Geyer (2012 Czudnochowski et al., Nature \nCommunications) where it was shown that P-TEFb is not a Ser2 kinase in vitro, specifically because \nthe authors use P-TEFb in vitro in this work. \n\n\f-Supplementary Fig 1e: Not obvious that chromatin bound Thr806 phosphorylation level is changing in \nthe PP1 depleted sample. It is also not clear that this blot is necessary given the other data in the \nmanuscript. \n\n-Supplementary Fig 2a- Loading controls suggest that there is more GST-CTR1 input than the other \npeptides (significantly more). The authors comment in the legend that this is due to incomplete \nstripping of the membrane. This is a really important control, so it is advised the experiment is \nrepeated and the sample is evenly split onto two gels/membranes to show the same amount of \nsample was indeed used, specifically because some faint bands appear on the phosphoblot for the \nother CTR constructs. The blots with the full-length protein look good. \n\n-Figure 2B: Are the ChIP plots scaled the same way? Add Y value to plots or state in figure legend that \nthe scaling is the same. \n\n-Figure 2C, 3C: Plot the curves with the same Y axis boundaries to show the extent of differences. \n\n-Figure 3A, 4D, 7A S3A, S4 C, D: Mark termination sites or CPS sites on gene overview. There are \nprobably multiple sites, but it helps to know where transcription may stop on a particular gene. \n\n-Figure 5C: Is it possible to comment on why Nutlin has a greater effect on non-paused genes? The \nmodel posited by the authors suggests that Ser666 phosphorylation follows pause release, and it \nwould be naively thought that the effect of Nutlin would be greater on such genes. \n\n-Figure 7e: The authors see increased Pol II in the gene body when PP4R2 is depleted. This can come \nfrom enhanced transcription but can also arise from slow Pol II. If at all possible, the authors should \ninvestigate Pol II elongation rates in PP4R2 depleted cells using Gro-seq or TT-seq. This would confirm \nthat enhanced Spt5 Ser666 phosphorylation leads to pause release and faster elongation rates. \n\nMinor comment \n-Supplementary Figure 3: Flip panel order so they match Figure 3. \n\n\f", "role": "assistant" } ]	2,020
Ion mobility collision cross-section atlas for known and unknown metabolite annotation in	REVIEWERS' COMMENTS: Reviewer #1 (Remarks to the Author): The authors have done a very thorough job of responding to my comments. I particularly welcome the additional results, especially those that show the improved performance with real MS/MS data (which is, I think, a more common use-case). I would be happy to see the manuscript published now.	REVIEWERS' COMMENTS: Reviewer #1 (Remarks to the Author): The authors have done a very thorough job of responding to my comments. I particularly welcome the additional results, especially those that show the improved performance with real MS/MS data (which is, I think, a more common use-case). I would be happy to see the manuscript published now.	TPR	NC_BS_BIOCH_005_RR2.txt	{ "criticism": 0, "example": 0, "importance_and_relevance": 0, "materials_and_methods": 1, "praise": 3, "presentation_and_reporting": 0, "results_and_discussion": 1, "suggestion_and_solution": 0, "total": 3 }	{ "criticism": 0, "example": 0, "importance_and_relevance": 0, "materials_and_methods": 0.3333333333333333, "praise": 1, "presentation_and_reporting": 0, "results_and_discussion": 0.3333333333333333, "suggestion_and_solution": 0 }	1.666667	[ { "content": "Please review this paper.", "role": "user" }, { "content": "\n \n\n\fREVIEWERS' COMMENTS: \n\n \n\nReviewer #1 (Remarks to the Author): \n\n \n\nThe authors have done a very thorough job of responding to my comments. I particularly welcome \nthe additional results, especially those that show the improved performance with real MS/MS data \n(which is, I think, a more common use-case). \n\nI would be happy to see the manuscript published now. \n\n \n\n\f\n", "role": "assistant" } ]	2,020
Ion mobility collision cross-section atlas for known and unknown metabolite annotation in	REVIEWER COMMENTS Reviewer #1 (Remarks to the Author): This paper described the creation and validation of an atlas of CCS values. The atlas is created using a combination of experimental CCS (merged from multiple databases) and in-silico predicted CCS. The authors show that they are able to predict CCS better than current systems, and that combining predicted CCS with other data (m/z, predicted MS/MS) improved candidate ranking. The atlas looks to be a resource that has practical utility and the authors do a good job showing that it performs well. 1. Overall I enjoyed reading this paper. I think it does a good job in setting out the system, evaluating it, and describing the benefits it brings. There are quite a few places in the text though in which more thorough proof-reading is required (too numerous to list here). 2. The improvement in CCS prediction is good, and the ML methodology looks sound. I would welcome a little more discussion on where the authors think this improvement comes from. Is it simply the larger unified database? 3. The SVM regression for CCS prediction: the gamma parameters (note: radial basis, not radio basis) seem extraordinarily high (2^8 and 2^13). Depending on the way the RBF function is parameterised, this effectively becomes either a linear regression, or something akin to a nearest neighbours scheme. Not necessarily a problem, but something that if I saw it in an analysis, would ring some alarm bells for me. Have the authors looked into this? 4. When combining with MS/MS, the authors use only in-silico MS/MS on their side. Why is this? in- silico MS/MS prediction is still not great. For the "unknown" experiments, predicted MS/MS is obviously the only route. However, could the authors not have done some experiments with real MS/MS library spectra? Playing devils advocate: perhaps the improvement CCS gives about in-silico MS/MS is because in-silico MS/MS is not reliable. Would we see the same improvements with real MS/MS libraries? There are plenty in the public domain (Massbank, GNPS, HMDB, etc). 5. Reproducibility: the authors do a good job of providing the supplementary material to allow another to produce their results. One thing that I cannot find are the in-silico predicted spectra that were used in the experimental section. Recreating these would be challenging (e.g. CFM-ID takes a long time to run) and could prohibit reproduction. I think they should be included as, e.g., one or more .mgf files. Reviewer #2 (Remarks to the Author): The manuscript by Zhou and colleagues describes a software tool for the large scale deployment of ion mobility collision cross-section data for metabolite identification of both known and unknown metabolites. While a number of tools have been created previously, including by the senior author, to search databases of known standards and also predict CCS values for metabolites without CCS measures, AllCCS dramatically extends the use of CCS values and ion mobility for metabolite identification and annotation by unifying a number of experimental CCS values to produce consensus values and predicting CCS values for an impressive database of putative metabolites. The authors demonstrate the use of AllCCS for a number of biological applications and I was impressed with the identification of novel metabolites. My major comment is that if I have understood AllCCS it is available through a webserver maintained by the group and people need to register for access. While the authors make some data available a lot of what is generated is behind the webserver interface. What are the plans to making this available for the wider community? The database of putative metabolites looks particularly useful. Also, if the tools are made freely available to the community they can be greatly improved in terms of versatility – as an example I think about the developments of xcms by the community. Have the authors considered this? Also, how does the server cope with large datasets as can be generated by ion mobility data? As a follow up comment while its impressive how well CCS values can be estimated – in many cases its less than 2% - can the authors discuss examples where this is not enough. Some discussion of what can be separated in terms of structural isomers would help. I am thinking of metabolites such as glucose-6-P, glucose-1-P and fructose-6-P which are poorly separated by both chromatography and ion mobility. Some examples of where an annotation can be made but not an identification would be useful. In addition, I do have some specific comments. Abstract, line 17. The first sentence should be “the metabolome” rather than just metabolome. I think the paper would benefit from a careful read to fix some of the grammar errors that creep in. Introduction, line 32. This first sentence is also odd and should be re-written. Similarly for the sentences on lines 54, 58 and 66. Results. Line 94 – should be laboratories. Line 98 “As a result, a total of 3,539 unified CCS values were calculated for 2,193 compounds…” I think from what the authors say elsewhere the discrepancy in numbers is down to ionisation mode rather than different conformers – its probably worth stating this for the reader and the authors are largely annotating one conformer. Lines 138-141 also need re-writing. Line 283. Is KEGG the best choice for metabolic reconstructions? Is RECON a better alternative for human metabolism? A minor point – Student’s t-test and student’s t-test are written. I think the correct form is with a capital S but needs to be consistent one way or the other depending on whether you believe Student was a genuine author or not! “All other data supporting the findings of this study are available from the corresponding author on reasonable 602 request.” Should the data be made openly available? Something is wrong with reference 9. Reviewer #3 (Remarks to the Author): The manuscript by Zhu et al. describes a curated atlas of ion mobility collision cross sections (CCS). These collision cross sections are being increasingly used for the purpose of metabolite annotation as a complementary parameter by which unknown molecules, as particularly metabolites in untargeted metabolomics studies, can be better annotated. Although the number of databases with collision cross sections is growing steadily both in number and in size, the dearth of comprehensive CCS databases has prompted many scientists to also look into predicting such CCS values using machine learning approaches. The authors of this manuscript have, in the past, presented several aspects of the work covered in this new manuscript in a number of articles. These include the following: Anal. Chem. 2016, 88, 11084−11091 (MetCCS); Anal. Chem. 2017, 89, 9559−9566 (LipidCCS); Bioinformatics, 33(14), 2017, 2235–2237 (MetCCS) etc. Similar work has also been published by other teams, including: Anal. Chem. 2020, 92, 1720−1729; Anal. Chem. 2019, 91, 5191−5199; Anal. Chem. 2017, 89, 6583−6589; Chem. Commun., 2017,53, 7624-7627; Journal of Chromatography A, 1542 (2018) 82–88; Analytica Chimica Acta 924 (2016) 68e76. In that sense, and despite the interesting nature of what is being described in the manuscript, the work is seen as incremental, but not transformative, so publication in Nature Communications is not seen as a good option, and a more specialized Journal such as Analytical Chemistry or The Analyst is recommended.	REVIEWER COMMENTS Reviewer #1 (Remarks to the Author): This paper described the creation and validation of an atlas of CCS values. The atlas is created using a combination of experimental CCS (merged from multiple databases) and in-silico predicted CCS. The authors show that they are able to predict CCS better than current systems, and that combining predicted CCS with other data (m/z, predicted MS/MS) improved candidate ranking. The atlas looks to be a resource that has practical utility and the authors do a good job showing that it performs well. 1. Overall I enjoyed reading this paper. I think it does a good job in setting out the system, evaluating it, and describing the benefits it brings. There are quite a few places in the text though in which more thorough proof-reading is required (too numerous to list here). 2. The improvement in CCS prediction is good, and the ML methodology looks sound. I would welcome a little more discussion on where the authors think this improvement comes from. Is it simply the larger unified database? 3. The SVM regression for CCS prediction: the gamma parameters (note: radial basis, not radio basis) seem extraordinarily high (2^8 and 2^13). Depending on the way the RBF function is parameterised, this effectively becomes either a linear regression, or something akin to a nearest neighbours scheme. Not necessarily a problem, but something that if I saw it in an analysis, would ring some alarm bells for me. Have the authors looked into this? 4. When combining with MS/MS, the authors use only in-silico MS/MS on their side. Why is this? in- silico MS/MS prediction is still not great. For the "unknown" experiments, predicted MS/MS is obviously the only route. However, could the authors not have done some experiments with real MS/MS library spectra? Playing devils advocate: perhaps the improvement CCS gives about in-silico MS/MS is because in-silico MS/MS is not reliable. Would we see the same improvements with real MS/MS libraries? There are plenty in the public domain (Massbank, GNPS, HMDB, etc). 5. Reproducibility: the authors do a good job of providing the supplementary material to allow another to produce their results. One thing that I cannot find are the in-silico predicted spectra that were used in the experimental section. Recreating these would be challenging (e.g. CFM-ID takes a long time to run) and could prohibit reproduction. I think they should be included as, e.g., one or more .mgf files. Reviewer #2 (Remarks to the Author): The manuscript by Zhou and colleagues describes a software tool for the large scale deployment of ion mobility collision cross-section data for metabolite identification of both known and unknown metabolites. While a number of tools have been created previously, including by the senior author, to search databases of known standards and also predict CCS values for metabolites without CCS measures, AllCCS dramatically extends the use of CCS values and ion mobility for metabolite identification and annotation by unifying a number of experimental CCS values to produce consensus values and predicting CCS values for an impressive database of putative metabolites. The authors demonstrate the use of AllCCS for a number of biological applications and I was impressed with the identification of novel metabolites. My major comment is that if I have understood AllCCS it is available through a webserver maintained by the group and people need to register for access. While the authors make some data available a lot of what is generated is behind the webserver interface. What are the plans to making this available for the wider community? The database of putative metabolites looks particularly useful. Also, if the tools are made freely available to the community they can be greatly improved in terms of versatility – as an example I think about the developments of xcms by the community. Have the authors considered this? Also, how does the server cope with large datasets as can be generated by ion mobility data? As a follow up comment while its impressive how well CCS values can be estimated – in many cases its less than 2% - can the authors discuss examples where this is not enough. Some discussion of what can be separated in terms of structural isomers would help. I am thinking of metabolites such as glucose-6-P, glucose-1-P and fructose-6-P which are poorly separated by both chromatography and ion mobility. Some examples of where an annotation can be made but not an identification would be useful. In addition, I do have some specific comments. Abstract, line 17. The first sentence should be “the metabolome” rather than just metabolome. I think the paper would benefit from a careful read to fix some of the grammar errors that creep in. Introduction, line 32. This first sentence is also odd and should be re-written. Similarly for the sentences on lines 54, 58 and 66. Results. Line 94 – should be laboratories. Line 98 “As a result, a total of 3,539 unified CCS values were calculated for 2,193 compounds…” I think from what the authors say elsewhere the discrepancy in numbers is down to ionisation mode rather than different conformers – its probably worth stating this for the reader and the authors are largely annotating one conformer. Lines 138-141 also need re-writing. Line 283. Is KEGG the best choice for metabolic reconstructions? Is RECON a better alternative for human metabolism? A minor point – Student’s t-test and student’s t-test are written. I think the correct form is with a capital S but needs to be consistent one way or the other depending on whether you believe Student was a genuine author or not! “All other data supporting the findings of this study are available from the corresponding author on reasonable 602 request.” Should the data be made openly available? Something is wrong with reference 9. Reviewer #3 (Remarks to the Author): The manuscript by Zhu et al. describes a curated atlas of ion mobility collision cross sections (CCS). These collision cross sections are being increasingly used for the purpose of metabolite annotation as a complementary parameter by which unknown molecules, as particularly metabolites in untargeted metabolomics studies, can be better annotated. Although the number of databases with collision cross sections is growing steadily both in number and in size, the dearth of comprehensive CCS databases has prompted many scientists to also look into predicting such CCS values using machine learning approaches. The authors of this manuscript have, in the past, presented several aspects of the work covered in this new manuscript in a number of articles. These include the following: Anal. Chem. 2016, 88, 11084−11091 (MetCCS); Anal. Chem. 2017, 89, 9559−9566 (LipidCCS); Bioinformatics, 33(14), 2017, 2235–2237 (MetCCS) etc. Similar work has also been published by other teams, including: Anal. Chem. 2020, 92, 1720−1729; Anal. Chem. 2019, 91, 5191−5199; Anal. Chem. 2017, 89, 6583−6589; Chem. Commun., 2017,53, 7624-7627; Journal of Chromatography A, 1542 (2018) 82–88; Analytica Chimica Acta 924 (2016) 68e76. In that sense, and despite the interesting nature of what is being described in the manuscript, the work is seen as incremental, but not transformative, so publication in Nature Communications is not seen as a good option, and a more specialized Journal such as Analytical Chemistry or The Analyst is recommended.	TPR	NC_BS_BIOCH_005_RR1.txt	{ "criticism": 7, "example": 8, "importance_and_relevance": 5, "materials_and_methods": 29, "praise": 10, "presentation_and_reporting": 17, "results_and_discussion": 8, "suggestion_and_solution": 16, "total": 83 }	{ "criticism": 0.08433734939759036, "example": 0.0963855421686747, "importance_and_relevance": 0.060240963855421686, "materials_and_methods": 0.3493975903614458, "praise": 0.12048192771084337, "presentation_and_reporting": 0.20481927710843373, "results_and_discussion": 0.0963855421686747, "suggestion_and_solution": 0.1927710843373494 }	1.204819	[ { "content": "Please review this paper.", "role": "user" }, { "content": "REVIEWER COMMENTS \n\n \n\nReviewer #1 (Remarks to the Author): \n\n \n\nThis paper described the creation and validation of an atlas of CCS values. The atlas is created using a \ncombination of experimental CCS (merged from multiple databases) and in-silico predicted CCS. The \nauthors show that they are able to predict CCS better than current systems, and that combining \npredicted CCS with other data (m/z, predicted MS/MS) improved candidate ranking. The atlas looks \nto be a resource that has practical utility and the authors do a good job showing that it performs \nwell. \n\n \n\n1. Overall I enjoyed reading this paper. I think it does a good job in setting out the system, evaluating \nit, and describing the benefits it brings. There are quite a few places in the text though in which \nmore thorough proof-reading is required (too numerous to list here). \n\n \n\n2. The improvement in CCS prediction is good, and the ML methodology looks sound. I would \nwelcome a little more discussion on where the authors think this improvement comes from. Is it \nsimply the larger unified database? \n\n \n\n3. The SVM regression for CCS prediction: the gamma parameters (note: radial basis, not radio basis) \nseem extraordinarily high (2^8 and 2^13). Depending on the way the RBF function is parameterised, \nthis effectively becomes either a linear regression, or something akin to a nearest neighbours \nscheme. Not necessarily a problem, but something that if I saw it in an analysis, would ring some \nalarm bells for me. Have the authors looked into this? \n\n \n\n4. When combining with MS/MS, the authors use only in-silico MS/MS on their side. Why is this? in-\nsilico MS/MS prediction is still not great. For the \"unknown\" experiments, predicted MS/MS is \nobviously the only route. However, could the authors not have done some experiments with real \nMS/MS library spectra? Playing devils advocate: perhaps the improvement CCS gives about in-silico \nMS/MS is because in-silico MS/MS is not reliable. Would we see the same improvements with real \nMS/MS libraries? There are plenty in the public domain (Massbank, GNPS, HMDB, etc). \n\n \n\n5. Reproducibility: the authors do a good job of providing the supplementary material to allow \nanother to produce their results. One thing that I cannot find are the in-silico predicted spectra that \nwere used in the experimental section. Recreating these would be challenging (e.g. CFM-ID takes a \n\n\flong time to run) and could prohibit reproduction. I think they should be included as, e.g., one or \nmore .mgf files. \n\n \n\n \n\nReviewer #2 (Remarks to the Author): \n\n \n\nThe manuscript by Zhou and colleagues describes a software tool for the large scale deployment of \nion mobility collision cross-section data for metabolite identification of both known and unknown \nmetabolites. While a number of tools have been created previously, including by the senior author, \nto search databases of known standards and also predict CCS values for metabolites without CCS \nmeasures, AllCCS dramatically extends the use of CCS values and ion mobility for metabolite \nidentification and annotation by unifying a number of experimental CCS values to produce \nconsensus values and predicting CCS values for an impressive database of putative metabolites. The \nauthors demonstrate the use of AllCCS for a number of biological applications and I was impressed \nwith the identification of novel metabolites. \n\nMy major comment is that if I have understood AllCCS it is available through a webserver maintained \nby the group and people need to register for access. While the authors make some data available a \nlot of what is generated is behind the webserver interface. What are the plans to making this \navailable for the wider community? The database of putative metabolites looks particularly useful. \nAlso, if the tools are made freely available to the community they can be greatly improved in terms \nof versatility – as an example I think about the developments of xcms by the community. Have the \nauthors considered this? Also, how does the server cope with large datasets as can be generated by \nion mobility data? \n\nAs a follow up comment while its impressive how well CCS values can be estimated – in many cases \nits less than 2% - can the authors discuss examples where this is not enough. Some discussion of \nwhat can be separated in terms of structural isomers would help. I am thinking of metabolites such \nas glucose-6-P, glucose-1-P and fructose-6-P which are poorly separated by both chromatography \nand ion mobility. Some examples of where an annotation can be made but not an identification \nwould be useful. \n\nIn addition, I do have some specific comments. \n\nAbstract, line 17. The first sentence should be “the metabolome” rather than just metabolome. I \nthink the paper would benefit from a careful read to fix some of the grammar errors that creep in. \n\nIntroduction, line 32. This first sentence is also odd and should be re-written. Similarly for the \nsentences on lines 54, 58 and 66. \n\nResults. Line 94 – should be laboratories. \n\nLine 98 “As a result, a total of 3,539 unified CCS values were calculated for 2,193 compounds…” I \nthink from what the authors say elsewhere the discrepancy in numbers is down to ionisation mode \n\n\frather than different conformers – its probably worth stating this for the reader and the authors are \nlargely annotating one conformer. \n\nLines 138-141 also need re-writing. \n\nLine 283. Is KEGG the best choice for metabolic reconstructions? Is RECON a better alternative for \nhuman metabolism? \n\nA minor point – Student’s t-test and student’s t-test are written. I think the correct form is with a \ncapital S but needs to be consistent one way or the other depending on whether you believe Student \nwas a genuine author or not! \n\n“All other data supporting the findings of this study are available from the corresponding author on \nreasonable 602 request.” Should the data be made openly available? \n\nSomething is wrong with reference 9. \n\n \n\n \n\n \n\nReviewer #3 (Remarks to the Author): \n\n \n\nThe manuscript by Zhu et al. describes a curated atlas of ion mobility collision cross sections (CCS). \nThese collision cross sections are being increasingly used for the purpose of metabolite annotation \nas a complementary parameter by which unknown molecules, as particularly metabolites in \nuntargeted metabolomics studies, can be better annotated. Although the number of databases with \ncollision cross sections is growing steadily both in number and in size, the dearth of comprehensive \nCCS databases has prompted many scientists to also look into predicting such CCS values using \nmachine learning approaches. The authors of this manuscript have, in the past, presented several \naspects of the work covered in this new manuscript in a number of articles. These include the \nfollowing: Anal. Chem. 2016, 88, 11084−11091 (MetCCS); Anal. Chem. 2017, 89, 9559−9566 \n(LipidCCS); Bioinformatics, 33(14), 2017, 2235–2237 (MetCCS) etc. Similar work has also been \npublished by other teams, including: Anal. Chem. 2020, 92, 1720−1729; Anal. Chem. 2019, 91, \n5191−5199; Anal. Chem. 2017, 89, 6583−6589; Chem. Commun., 2017,53, 7624-7627; Journal of \nChromatography A, 1542 (2018) 82–88; Analytica Chimica Acta 924 (2016) 68e76. In that sense, and \ndespite the interesting nature of what is being described in the manuscript, the work is seen as \nincremental, but not transformative, so publication in Nature Communications is not seen as a good \noption, and a more specialized Journal such as Analytical Chemistry or The Analyst is recommended. \n\n\f", "role": "assistant" } ]	2,020
Complete chemical structures of human mitochondrial tRNAs	Reviewers' comments: Reviewer #1 (Remarks to the Author): tRNAs are central to the decoding process and the post-transcriptional modifications of tRNA are critical for tRNA function including stability, folding and decoding. It is known that disorders of mitochondrial translation may lead to the mitochondria dysfunction and some point mutations in mt-tRNA have been reported to be associated with mitochondrial diseases. But the molecular pathogenesis of human mitochondrial diseases caused by aberrant tRNA modifications has lagged behind due to the lack of the complete landscape of modifications in mt-tRNA and the list of responsible modifying enzymes. In this manuscript, Suzuki and his colleagues identified the majority of modifications in all 22 human mt-tRNA using mass spectrometric analyses. Further, a comprehensive analysis including CE-Ψ-MS, CMC-PE and tRNA-Ψ-seq were employed to detect Ψ (a mass-silent modification) in mt-tRNA. In addition, the modifying enzymes responsible for these mt-tRNA are also listed. Thus, the collective results represent a very comprehensive resource that will be a key reference for the field. The study will certainly benefit future molecular mechanism studies of human mitochondrial diseases caused by mt-tRNA modifications. The study is also very well designed, performed and written. Hence, I strongly recommend the publication of this paper at Nature communications. I only have a few minor points as listed. 1. In this study, tRNA-Ψ-seq was used to detect Ψ sites in mt-tRNA. Recently a similar approach named DM–Ψ-seq is also developed and identified Ψ sites in cytoplasmic and mitochondrial tRNA (Song et al., 2020, Nat Chem Biol 16, 160-169). It would be nice to also compare the psiU sites (since it is a mass- silent modification) identified in this study with the reference. 2. The authors have listed RNase T1- or A-digested fragments of 22 human mt-tRNAs in supplementary table 1; all the original fragment analysis maps (as Figure 2C) are also desired to be included, which can be used to evaluate the stoichiometry of modified nucleosides. 3. 2D-TLC cannot discriminate Ψ in different positions of mt-tRNAGlu and hence the label in Supplementary 4 is not accurate. Reviewer #2 (Remarks to the Author): In this study, Suzuki et al, performed a comprehensive analysis of post-transcriptional modifications of all human mt-tRNAs. In a previous analysis, the authors mapped 15 types of modified bases at 118 positions in all species of bovine mt-tRNAs. Here, the authors isolated 22 species of human mt-tRNAs and analyzed the primary sequence and post-transcriptional modifications for each tRNA using mass spectrometry in combination with biochemical analyses to map pseudouridines. In total, the authors found 16 kinds of RNA modifications at 136 positions in 22 species of human mt-tRNAs. The results provide a complete list of modifications in human mt-tRNA. This is an interesting study and should facilitate investigations of their physiological roles in the future. 1. The authors isolated 22 species of human mt-tRNAs from human placenta, but only several species of mt-tRNA from HeLa cells. Why not isolate the all 22 species of human mt-tRNA from HeLa cells? I suggest the authors to compare the modifications in mt-tRNA between human placenta and HeLa cells. 2. How about the frequency (or stoichiometry) of specific modifications in given sites of human mt- tRNA? I suggest the authors to provide this information for each identified modification in mt-tRNA. 3. How to confirm the purity of the isolated individual mt-tRNA? The lengths of mt-tRNA are similar and they also show similar gel-shift on the gel. Single band in gel may not indicate the single mt-tRNA. The impure mt-tRNA may affect the determination of the modifications in mt-tRNA. 4. There are some modifications that have the same molecular weights, such as m5C and m3C, m1G and m2G, m1A and m6A. How to confirm these modifications? The authors need to provide detailed experimental evidence. 5. The MS fragment ions of 15 species of mt-tRNA were listed in Table S1. But in this study the authors measured 22 species of mt-tRNA. The authors need to supply the rest part of mt-tRNA of the MS fragment ions. 6. The authors need to provide the sequence and Western blotting data for supporting the knockout experiments. 7. In Figure 3b, the Ψs at positions 66 and 67 in human mt-tRNAPro were identified. If these two sites are both Ψs, why there are two possibilities, i.e. the existence of c6 and y6 ions? 8. Figure 4 and Figure S5 are somehow similar. Figure 4 includes 14 species of mt-tRNA and Figure S5 includes 22. The authors may only provide the full 22 species of mt-tRNA in main text. Reviewer #3 (Remarks to the Author): In this manuscript by Suzuki et al., the authors map the chemical structures of the remaining uncharacterized tRNAs encoded by the human mitochondrial genome. Using previously-developed tRNA purification and mass spectrometry techniques, the authors identify the repertoire of modifications in human mitochondrial tRNAs. The data is technically sound and the authors provide independent assays to validate a subset of modifications. Moreover, the authors provide evidence that the QTRT1 and QTRT2 proteins are required for the queuosine modification in certain human mitochondrial tRNAs. The manuscript is clearly written and have provided extensive Supplemental Data to support their findings. Unfortunately, there is a lack of novelty in this manuscript since no new types of modifications were found. The approach and results are similar to the conclusions published by the group using bovine mitochondrial tRNAs in their 2014 paper in Nucleic Acids Research. Thus, many of the modifications in mitochondrial tRNA could have been predicted and this manuscript serves as validation rather than discovery. Moreover, there is little functional characterization of the modifications and their biological roles. Table 1 is just a list of codons and the cognate tRNA anticodons while Table 2 is a list of the known and confirmed tRNA modifications and their cognate enzymes. While the results represent a complete inventory of human mitochondrial tRNA modifications, no new insight is gained that changes how we think about mitochondrial tRNA modifications or mitochondrial tRNA biology. To elevate this manuscript beyond just simply mapping tRNA modifications, the authors should characterize the consequences of lacking a mitochondrial tRNA modification. For example, the Authors have made a knockout of QTRT1 or QTRT2 but there is no follow-up on the function of queuosine modification in mitochondrial tRNAs. Biological insight could also be gained by monitoring the function of the numerous pseudouridine modifications in mitochondrial tRNAs. Currently, the manuscript describes a list of modifications without providing any biological relevance or significance of the modifications. Minor points: • Figure 1 shows a denaturing gel of isolated tRNAs. Since the RNAs are imaged using a general nucleic acid stain, there is no follow-up on the relative purity of the isolated tRNA fractions. The Authors could perform Northern blotting on an aliquot of the isolated tRNAs to determine whether there are contaminating tRNAs in the fractions. This could be important since any contaminating tRNAs could contribute significantly to the modification signal detected by mass spectrometry. • As mentioned above, there is no follow-up on the QTRT1 or QTRT2-knockout (KO) cell lines. The addition of molecular or biological effects caused by lack of Q modification would have greatly strengthened the manuscript and provided insight into the function of the Q modification in mitochondrial tRNAs.	Reviewers' comments: Reviewer #1 (Remarks to the Author): tRNAs are central to the decoding process and the post-transcriptional modifications of tRNA are critical for tRNA function including stability, folding and decoding. It is known that disorders of mitochondrial translation may lead to the mitochondria dysfunction and some point mutations in mt-tRNA have been reported to be associated with mitochondrial diseases. But the molecular pathogenesis of human mitochondrial diseases caused by aberrant tRNA modifications has lagged behind due to the lack of the complete landscape of modifications in mt-tRNA and the list of responsible modifying enzymes. In this manuscript, Suzuki and his colleagues identified the majority of modifications in all 22 human mt-tRNA using mass spectrometric analyses. Further, a comprehensive analysis including CE-Ψ-MS, CMC-PE and tRNA-Ψ-seq were employed to detect Ψ (a mass-silent modification) in mt-tRNA. In addition, the modifying enzymes responsible for these mt-tRNA are also listed. Thus, the collective results represent a very comprehensive resource that will be a key reference for the field. The study will certainly benefit future molecular mechanism studies of human mitochondrial diseases caused by mt-tRNA modifications. The study is also very well designed, performed and written. Hence, I strongly recommend the publication of this paper at Nature communications. I only have a few minor points as listed. 1. In this study, tRNA-Ψ-seq was used to detect Ψ sites in mt-tRNA. Recently a similar approach named DM–Ψ-seq is also developed and identified Ψ sites in cytoplasmic and mitochondrial tRNA (Song et al., 2020, Nat Chem Biol 16, 160-169). It would be nice to also compare the psiU sites (since it is a mass- silent modification) identified in this study with the reference. 2. The authors have listed RNase T1- or A-digested fragments of 22 human mt-tRNAs in supplementary table 1; all the original fragment analysis maps (as Figure 2C) are also desired to be included, which can be used to evaluate the stoichiometry of modified nucleosides. 3. 2D-TLC cannot discriminate Ψ in different positions of mt-tRNAGlu and hence the label in Supplementary 4 is not accurate. Reviewer #2 (Remarks to the Author): In this study, Suzuki et al, performed a comprehensive analysis of post-transcriptional modifications of all human mt-tRNAs. In a previous analysis, the authors mapped 15 types of modified bases at 118 positions in all species of bovine mt-tRNAs. Here, the authors isolated 22 species of human mt-tRNAs and analyzed the primary sequence and post-transcriptional modifications for each tRNA using mass spectrometry in combination with biochemical analyses to map pseudouridines. In total, the authors found 16 kinds of RNA modifications at 136 positions in 22 species of human mt-tRNAs. The results provide a complete list of modifications in human mt-tRNA. This is an interesting study and should facilitate investigations of their physiological roles in the future. 1. The authors isolated 22 species of human mt-tRNAs from human placenta, but only several species of mt-tRNA from HeLa cells. Why not isolate the all 22 species of human mt-tRNA from HeLa cells? I suggest the authors to compare the modifications in mt-tRNA between human placenta and HeLa cells. 2. How about the frequency (or stoichiometry) of specific modifications in given sites of human mt- tRNA? I suggest the authors to provide this information for each identified modification in mt-tRNA. 3. How to confirm the purity of the isolated individual mt-tRNA? The lengths of mt-tRNA are similar and they also show similar gel-shift on the gel. Single band in gel may not indicate the single mt-tRNA. The impure mt-tRNA may affect the determination of the modifications in mt-tRNA. 4. There are some modifications that have the same molecular weights, such as m5C and m3C, m1G and m2G, m1A and m6A. How to confirm these modifications? The authors need to provide detailed experimental evidence. 5. The MS fragment ions of 15 species of mt-tRNA were listed in Table S1. But in this study the authors measured 22 species of mt-tRNA. The authors need to supply the rest part of mt-tRNA of the MS fragment ions. 6. The authors need to provide the sequence and Western blotting data for supporting the knockout experiments. 7. In Figure 3b, the Ψs at positions 66 and 67 in human mt-tRNAPro were identified. If these two sites are both Ψs, why there are two possibilities, i.e. the existence of c6 and y6 ions? 8. Figure 4 and Figure S5 are somehow similar. Figure 4 includes 14 species of mt-tRNA and Figure S5 includes 22. The authors may only provide the full 22 species of mt-tRNA in main text. Reviewer #3 (Remarks to the Author): In this manuscript by Suzuki et al., the authors map the chemical structures of the remaining uncharacterized tRNAs encoded by the human mitochondrial genome. Using previously-developed tRNA purification and mass spectrometry techniques, the authors identify the repertoire of modifications in human mitochondrial tRNAs. The data is technically sound and the authors provide independent assays to validate a subset of modifications. Moreover, the authors provide evidence that the QTRT1 and QTRT2 proteins are required for the queuosine modification in certain human mitochondrial tRNAs. The manuscript is clearly written and have provided extensive Supplemental Data to support their findings. Unfortunately, there is a lack of novelty in this manuscript since no new types of modifications were found. The approach and results are similar to the conclusions published by the group using bovine mitochondrial tRNAs in their 2014 paper in Nucleic Acids Research. Thus, many of the modifications in mitochondrial tRNA could have been predicted and this manuscript serves as validation rather than discovery. Moreover, there is little functional characterization of the modifications and their biological roles. Table 1 is just a list of codons and the cognate tRNA anticodons while Table 2 is a list of the known and confirmed tRNA modifications and their cognate enzymes. While the results represent a complete inventory of human mitochondrial tRNA modifications, no new insight is gained that changes how we think about mitochondrial tRNA modifications or mitochondrial tRNA biology. To elevate this manuscript beyond just simply mapping tRNA modifications, the authors should characterize the consequences of lacking a mitochondrial tRNA modification. For example, the Authors have made a knockout of QTRT1 or QTRT2 but there is no follow-up on the function of queuosine modification in mitochondrial tRNAs. Biological insight could also be gained by monitoring the function of the numerous pseudouridine modifications in mitochondrial tRNAs. Currently, the manuscript describes a list of modifications without providing any biological relevance or significance of the modifications. Minor points: • Figure 1 shows a denaturing gel of isolated tRNAs. Since the RNAs are imaged using a general nucleic acid stain, there is no follow-up on the relative purity of the isolated tRNA fractions. The Authors could perform Northern blotting on an aliquot of the isolated tRNAs to determine whether there are contaminating tRNAs in the fractions. This could be important since any contaminating tRNAs could contribute significantly to the modification signal detected by mass spectrometry. • As mentioned above, there is no follow-up on the QTRT1 or QTRT2-knockout (KO) cell lines. The addition of molecular or biological effects caused by lack of Q modification would have greatly strengthened the manuscript and provided insight into the function of the Q modification in mitochondrial tRNAs.	TPR	NC_BS_BIOCH_006_RR1.txt	{ "criticism": 7, "example": 6, "importance_and_relevance": 4, "materials_and_methods": 45, "praise": 7, "presentation_and_reporting": 10, "results_and_discussion": 4, "suggestion_and_solution": 15, "total": 76 }	{ "criticism": 0.09210526315789473, "example": 0.07894736842105263, "importance_and_relevance": 0.05263157894736842, "materials_and_methods": 0.5921052631578947, "praise": 0.09210526315789473, "presentation_and_reporting": 0.13157894736842105, "results_and_discussion": 0.05263157894736842, "suggestion_and_solution": 0.19736842105263158 }	1.289474	[ { "content": "Please review this paper.", "role": "user" }, { "content": "Reviewers' comments: \n\nReviewer #1 (Remarks to the Author): \n\ntRNAs are central to the decoding process and the post-transcriptional modifications of tRNA are critical \nfor tRNA function including stability, folding and decoding. It is known that disorders of mitochondrial \ntranslation may lead to the mitochondria dysfunction and some point mutations in mt-tRNA have been \nreported to be associated with mitochondrial diseases. But the molecular pathogenesis of human \nmitochondrial diseases caused by aberrant tRNA modifications has lagged behind due to the lack of the \ncomplete landscape of modifications in mt-tRNA and the list of responsible modifying enzymes. In this \nmanuscript, Suzuki and his colleagues identified the majority of modifications in all 22 human mt-tRNA \nusing mass spectrometric analyses. Further, a comprehensive analysis including CE-Ψ-MS, CMC-PE and \ntRNA-Ψ-seq were employed to detect Ψ (a mass-silent modification) in mt-tRNA. In addition, the \nmodifying enzymes responsible for these mt-tRNA are also listed. Thus, the collective results represent a \nvery comprehensive resource that will be a key reference for the field. The study will certainly benefit \nfuture molecular mechanism studies of human mitochondrial diseases caused by mt-tRNA \nmodifications. The study is also very well designed, performed and written. Hence, I strongly \nrecommend the publication of this paper at Nature communications. I only have a few minor points as \nlisted. \n\n1. In this study, tRNA-Ψ-seq was used to detect Ψ sites in mt-tRNA. Recently a similar approach named \nDM–Ψ-seq is also developed and identified Ψ sites in cytoplasmic and mitochondrial tRNA (Song et al., \n2020, Nat Chem Biol 16, 160-169). It would be nice to also compare the psiU sites (since it is a mass-\nsilent modification) identified in this study with the reference. \n\n2. The authors have listed RNase T1- or A-digested fragments of 22 human mt-tRNAs in supplementary \ntable 1; all the original fragment analysis maps (as Figure 2C) are also desired to be included, which can \nbe used to evaluate the stoichiometry of modified nucleosides. \n\n3. 2D-TLC cannot discriminate Ψ in different positions of mt-tRNAGlu and hence the label in \nSupplementary 4 is not accurate. \n\nReviewer #2 (Remarks to the Author): \n\nIn this study, Suzuki et al, performed a comprehensive analysis of post-transcriptional modifications of \nall human mt-tRNAs. In a previous analysis, the authors mapped 15 types of modified bases at 118 \npositions in all species of bovine mt-tRNAs. Here, the authors isolated 22 species of human mt-tRNAs \nand analyzed the primary sequence and post-transcriptional modifications for each tRNA using mass \n\n\fspectrometry in combination with biochemical analyses to map pseudouridines. In total, the authors \nfound 16 kinds of RNA modifications at 136 positions in 22 species of human mt-tRNAs. The results \nprovide a complete list of modifications in human mt-tRNA. This is an interesting study and should \nfacilitate investigations of their physiological roles in the future. \n\n1. The authors isolated 22 species of human mt-tRNAs from human placenta, but only several species of \nmt-tRNA from HeLa cells. Why not isolate the all 22 species of human mt-tRNA from HeLa cells? I \nsuggest the authors to compare the modifications in mt-tRNA between human placenta and HeLa cells. \n\n2. How about the frequency (or stoichiometry) of specific modifications in given sites of human mt-\ntRNA? I suggest the authors to provide this information for each identified modification in mt-tRNA. \n\n3. How to confirm the purity of the isolated individual mt-tRNA? The lengths of mt-tRNA are similar and \nthey also show similar gel-shift on the gel. Single band in gel may not indicate the single mt-tRNA. The \nimpure mt-tRNA may affect the determination of the modifications in mt-tRNA. \n\n4. There are some modifications that have the same molecular weights, such as m5C and m3C, m1G and \nm2G, m1A and m6A. How to confirm these modifications? The authors need to provide detailed \nexperimental evidence. \n\n5. The MS fragment ions of 15 species of mt-tRNA were listed in Table S1. But in this study the authors \nmeasured 22 species of mt-tRNA. The authors need to supply the rest part of mt-tRNA of the MS \nfragment ions. \n\n6. The authors need to provide the sequence and Western blotting data for supporting the knockout \nexperiments. \n\n7. In Figure 3b, the Ψs at positions 66 and 67 in human mt-tRNAPro were identified. If these two sites \nare both Ψs, why there are two possibilities, i.e. the existence of c6 and y6 ions? \n\n8. Figure 4 and Figure S5 are somehow similar. Figure 4 includes 14 species of mt-tRNA and Figure S5 \nincludes 22. The authors may only provide the full 22 species of mt-tRNA in main text. \n\nReviewer #3 (Remarks to the Author): \n\nIn this manuscript by Suzuki et al., the authors map the chemical structures of the remaining \nuncharacterized tRNAs encoded by the human mitochondrial genome. Using previously-developed tRNA \npurification and mass spectrometry techniques, the authors identify the repertoire of modifications in \nhuman mitochondrial tRNAs. The data is technically sound and the authors provide independent assays \nto validate a subset of modifications. Moreover, the authors provide evidence that the QTRT1 and \nQTRT2 proteins are required for the queuosine modification in certain human mitochondrial tRNAs. The \n\n\fmanuscript is clearly written and have provided extensive Supplemental Data to support their findings. \n\nUnfortunately, there is a lack of novelty in this manuscript since no new types of modifications were \nfound. The approach and results are similar to the conclusions published by the group using bovine \nmitochondrial tRNAs in their 2014 paper in Nucleic Acids Research. Thus, many of the modifications in \nmitochondrial tRNA could have been predicted and this manuscript serves as validation rather than \ndiscovery. Moreover, there is little functional characterization of the modifications and their biological \nroles. Table 1 is just a list of codons and the cognate tRNA anticodons while Table 2 is a list of the known \nand confirmed tRNA modifications and their cognate enzymes. While the results represent a complete \ninventory of human mitochondrial tRNA modifications, no new insight is gained that changes how we \nthink about mitochondrial tRNA modifications or mitochondrial tRNA biology. \n\nTo elevate this manuscript beyond just simply mapping tRNA modifications, the authors should \ncharacterize the consequences of lacking a mitochondrial tRNA modification. For example, the Authors \nhave made a knockout of QTRT1 or QTRT2 but there is no follow-up on the function of queuosine \nmodification in mitochondrial tRNAs. Biological insight could also be gained by monitoring the function \nof the numerous pseudouridine modifications in mitochondrial tRNAs. Currently, the manuscript \ndescribes a list of modifications without providing any biological relevance or significance of the \nmodifications. \n\nMinor points: \n\n• Figure 1 shows a denaturing gel of isolated tRNAs. Since the RNAs are imaged using a general nucleic \nacid stain, there is no follow-up on the relative purity of the isolated tRNA fractions. The Authors could \nperform Northern blotting on an aliquot of the isolated tRNAs to determine whether there are \ncontaminating tRNAs in the fractions. This could be important since any contaminating tRNAs could \ncontribute significantly to the modification signal detected by mass spectrometry. \n• As mentioned above, there is no follow-up on the QTRT1 or QTRT2-knockout (KO) cell lines. The \naddition of molecular or biological effects caused by lack of Q modification would have greatly \nstrengthened the manuscript and provided insight into the function of the Q modification in \nmitochondrial tRNAs. \n\n\f\n\n \n\n", "role": "assistant" } ]	2,020
Complete chemical structures of human mitochondrial tRNAs	REVIEWERS' COMMENTS: Reviewer #1 (Remarks to the Author): The authors have addressed all my previous concerns and the revised manuscript has been improved. I thus strongly support the publication in Nature Communications. Reviewer #2 (Remarks to the Author): The authors addressed my questions. I have no further questions. Reviewer #3 (Remarks to the Author): The Authors have addressed all concerns from the original review. The authors have provided a reason for why a comprehensive map of human tRNA modifications is important since bovine and human mt- tRNAs differ fairly significantly in sequence so simple prediction is not sufficient. In addition, the authors provide evidence for new modifications, cmnm5U and cmnm5s2U. The authors also add ribosome profiling data and protein folding reporter plasmids to provide functional relevance for the findings. Based upon these clarifications and additions, the paper has provided novelty and biological insight that would be of broad interest in the field of RNA biology. Thus, I recommend the manuscript for publication.	REVIEWERS' COMMENTS: Reviewer #1 (Remarks to the Author): The authors have addressed all my previous concerns and the revised manuscript has been improved. I thus strongly support the publication in Nature Communications. Reviewer #2 (Remarks to the Author): The authors addressed my questions. I have no further questions. Reviewer #3 (Remarks to the Author): The Authors have addressed all concerns from the original review. The authors have provided a reason for why a comprehensive map of human tRNA modifications is important since bovine and human mt- tRNAs differ fairly significantly in sequence so simple prediction is not sufficient. In addition, the authors provide evidence for new modifications, cmnm5U and cmnm5s2U. The authors also add ribosome profiling data and protein folding reporter plasmids to provide functional relevance for the findings. Based upon these clarifications and additions, the paper has provided novelty and biological insight that would be of broad interest in the field of RNA biology. Thus, I recommend the manuscript for publication.	TPR	NC_BS_BIOCH_006_RR2.txt	{ "criticism": 1, "example": 0, "importance_and_relevance": 2, "materials_and_methods": 2, "praise": 4, "presentation_and_reporting": 0, "results_and_discussion": 2, "suggestion_and_solution": 1, "total": 10 }	{ "criticism": 0.1, "example": 0, "importance_and_relevance": 0.2, "materials_and_methods": 0.2, "praise": 0.4, "presentation_and_reporting": 0, "results_and_discussion": 0.2, "suggestion_and_solution": 0.1 }	1.2	[ { "content": "Please review this paper.", "role": "user" }, { "content": "\n\n\fREVIEWERS' COMMENTS: \n\nReviewer #1 (Remarks to the Author): \n\nThe authors have addressed all my previous concerns and the revised manuscript has been improved. I \nthus strongly support the publication in Nature Communications. \n\nReviewer #2 (Remarks to the Author): \n\nThe authors addressed my questions. I have no further questions. \n\nReviewer #3 (Remarks to the Author): \n\nThe Authors have addressed all concerns from the original review. The authors have provided a reason \nfor why a comprehensive map of human tRNA modifications is important since bovine and human mt-\ntRNAs differ fairly significantly in sequence so simple prediction is not sufficient. In addition, the authors \nprovide evidence for new modifications, cmnm5U and cmnm5s2U. The authors also add ribosome \nprofiling data and protein folding reporter plasmids to provide functional relevance for the findings. \nBased upon these clarifications and additions, the paper has provided novelty and biological insight that \nwould be of broad interest in the field of RNA biology. Thus, I recommend the manuscript for \npublication. \n\n\f\n", "role": "assistant" } ]	2,020
Warsaw Breakage Syndrome associated DDX11 helicase resolves G-quadruplex structures to	REVIEWERS' COMMENTS: Reviewer #1 (Remarks to the Author): Having read the revised version, we are convinced that the manuscript warrants publication in nature communications. This study addresses a key issue in the cohesin field. There have been conflicting views about the role DDX11 plays in promoting cohesion establishment during replication. The detailed characterisation of the newly identified mutations currently presented in figure 8 makes a strong case for the requirement of the helicase activity of DDX11 in order to promote cohesion establishment. This is an important observation that alters the outlook for future understanding of the molecular mechanism of cohesion establishment during replication. Our concerns were not intended to belittle the authors efforts in any way, we understand that the study is also of interest to clinicians. Comprehensive analysis of independently identified mutations associated with WABS and clarification of conflicting effects reported in literature is very important. Our view was that the way the manuscript was initially presented was distracting for a general reader and more importantly, not enough emphasis was given to the main message of the manuscript, namely the importance of the helicase activity of DDX11 in promoting the establishment of sister chromatid cohesion. Both our concerns are satisfactorily addressed in the revised manuscript and we fully support the immediate publication of the manuscript in nature communications. Reviewer #2 (Remarks to the Author): I have carefully read the responses to the Reviewers' comments, including my own. It is my belief that the authors have addressed the great majority of the comments in a satisfactory manner. Importantly, the revised manuscript is significantly improved. Minor comment: References 19 and 79 in the reference list are the same. This mistake should be corrected. Minor comment: In addition, Reference 22 has not spelled out correctly the name of the late Johan de Winter, a leading pioneer in the discovery and characterization of Warsaw Breakage syndrome. This error should be corrected. Reference 22 also lacks volume and page numbers. Minor comment: Reference 23 lacks volume and page numbers. Minor comment: Reference 38 lacks volume and page numbers. Minor Comment: Reference 75 lacks volume and page numbers. Minor comment: Reference 87 lacks volume and page numbers. There may be additional errors in the reference list. The reference list needs to be carefully examined and corrections made. Reviewer #3 (Remarks to the Author): I have looked at the revisions and the justifications provided, including issues raised by other reviewers. My criticism that there is still lack of mechanistic insight as to how DDX11 functions in cohesion, which the authors do acknowledge. Nevertheless, this is a decent piece of work with something of interest to say and the authors have added additional experiments to support their case. In this respect they have addressed my concerns satisfactorily.	REVIEWERS' COMMENTS: Reviewer #1 (Remarks to the Author): Having read the revised version, we are convinced that the manuscript warrants publication in nature communications. This study addresses a key issue in the cohesin field. There have been conflicting views about the role DDX11 plays in promoting cohesion establishment during replication. The detailed characterisation of the newly identified mutations currently presented in figure 8 makes a strong case for the requirement of the helicase activity of DDX11 in order to promote cohesion establishment. This is an important observation that alters the outlook for future understanding of the molecular mechanism of cohesion establishment during replication. Our concerns were not intended to belittle the authors efforts in any way, we understand that the study is also of interest to clinicians. Comprehensive analysis of independently identified mutations associated with WABS and clarification of conflicting effects reported in literature is very important. Our view was that the way the manuscript was initially presented was distracting for a general reader and more importantly, not enough emphasis was given to the main message of the manuscript, namely the importance of the helicase activity of DDX11 in promoting the establishment of sister chromatid cohesion. Both our concerns are satisfactorily addressed in the revised manuscript and we fully support the immediate publication of the manuscript in nature communications. Reviewer #2 (Remarks to the Author): I have carefully read the responses to the Reviewers' comments, including my own. It is my belief that the authors have addressed the great majority of the comments in a satisfactory manner. Importantly, the revised manuscript is significantly improved. Minor comment: References 19 and 79 in the reference list are the same. This mistake should be corrected. Minor comment: In addition, Reference 22 has not spelled out correctly the name of the late Johan de Winter, a leading pioneer in the discovery and characterization of Warsaw Breakage syndrome. This error should be corrected. Reference 22 also lacks volume and page numbers. Minor comment: Reference 23 lacks volume and page numbers. Minor comment: Reference 38 lacks volume and page numbers. Minor Comment: Reference 75 lacks volume and page numbers. Minor comment: Reference 87 lacks volume and page numbers. There may be additional errors in the reference list. The reference list needs to be carefully examined and corrections made. Reviewer #3 (Remarks to the Author): I have looked at the revisions and the justifications provided, including issues raised by other reviewers. My criticism that there is still lack of mechanistic insight as to how DDX11 functions in cohesion, which the authors do acknowledge. Nevertheless, this is a decent piece of work with something of interest to say and the authors have added additional experiments to support their case. In this respect they have addressed my concerns satisfactorily.	TPR	NC_BS_BIOCH_007_RR2.txt	{ "criticism": 11, "example": 2, "importance_and_relevance": 8, "materials_and_methods": 3, "praise": 10, "presentation_and_reporting": 11, "results_and_discussion": 1, "suggestion_and_solution": 5, "total": 27 }	{ "criticism": 0.4074074074074074, "example": 0.07407407407407407, "importance_and_relevance": 0.2962962962962963, "materials_and_methods": 0.1111111111111111, "praise": 0.37037037037037035, "presentation_and_reporting": 0.4074074074074074, "results_and_discussion": 0.037037037037037035, "suggestion_and_solution": 0.18518518518518517 }	1.888889	[ { "content": "Please review this paper.", "role": "user" }, { "content": "\n \n\n\fREVIEWERS' COMMENTS: \n \nReviewer #1 (Remarks to the Author): \n \nHaving read the revised version, we are convinced that the manuscript warrants publication in \nnature communications. This study addresses a key issue in the cohesin field. There have been \nconflicting views about the role DDX11 plays in promoting cohesion establishment during \nreplication. The detailed characterisation of the newly identified mutations currently presented in \nfigure 8 makes a strong case for the requirement of the helicase activity of DDX11 in order to \npromote cohesion establishment. This is an important observation that alters the outlook for future \nunderstanding of the molecular mechanism of cohesion establishment during replication. \n \nOur concerns were not intended to belittle the authors efforts in any way, we understand that the \nstudy is also of interest to clinicians. Comprehensive analysis of independently identified mutations \nassociated with WABS and clarification of conflicting effects reported in literature is very important. \nOur view was that the way the manuscript was initially presented was distracting for a general \nreader and more importantly, not enough emphasis was given to the main message of the \nmanuscript, namely the importance of the helicase activity of DDX11 in promoting the \nestablishment of sister chromatid cohesion. Both our concerns are satisfactorily addressed in the \nrevised manuscript and we fully support the immediate publication of the manuscript in nature \ncommunications. \n \n \n \nReviewer #2 (Remarks to the Author): \n \nI have carefully read the responses to the Reviewers' comments, including my own. It is my belief \nthat the authors have addressed the great majority of the comments in a satisfactory manner. \nImportantly, the revised manuscript is significantly improved. \nMinor comment: References 19 and 79 in the reference list are the same. This mistake should be \ncorrected. \nMinor comment: In addition, Reference 22 has not spelled out correctly the name of the late Johan \nde Winter, a leading pioneer in the discovery and characterization of Warsaw Breakage syndrome. \nThis error should be corrected. Reference 22 also lacks volume and page numbers. \nMinor comment: Reference 23 lacks volume and page numbers. \nMinor comment: Reference 38 lacks volume and page numbers. \nMinor Comment: Reference 75 lacks volume and page numbers. \nMinor comment: Reference 87 lacks volume and page numbers. \nThere may be additional errors in the reference list. The reference list needs to be carefully \nexamined and corrections made. \n \n \nReviewer #3 (Remarks to the Author): \n \nI have looked at the revisions and the justifications provided, including issues raised by other \nreviewers. My criticism that there is still lack of mechanistic insight as to how DDX11 functions in \ncohesion, which the authors do acknowledge. Nevertheless, this is a decent piece of work with \nsomething of interest to say and the authors have added additional experiments to support their \ncase. In this respect they have addressed my concerns satisfactorily. \n\n", "role": "assistant" } ]	2,010
Warsaw Breakage Syndrome associated DDX11 helicase resolves G-quadruplex structures to	Reviewers' comments: Reviewer #1 (Remarks to the Author): DDX11/ChlR1 and its yeast orthologue Chl1 are ATP-dependent DNA helicases related to Xpd and FANCJ, that unwinds DNA with 5’-3’ directionality. Mutations in DDX11 are linked with Warsaw breakage syndrome (WABS) which is characterised by high incidence of chromosomes with centromeric cohesion defects (“railroads”) and “premature chromatid separation” (PCS). It has previously been documented that yeast Chl1 and mammalian DDX11 are associated with the replisome and promote establishment of sister chromatid cohesion during S phase. How they are recruited to the replisome is not fully understood. DDX11 associates with Timeless via a highly conserved amino acid motif embedded in a large CHL1/DDX11-specific loop inserted into the helicase domain (Cortone et al PLOS gen 2018). Mutations in DDX11 that abolish this interaction show cohesion defects similar to DDX11 depletion. It was recently shown that DDX11 and Timeless collaborate in order to detect and process G4 structures during replication (Lerner, LK et al bioRxiv). How DDX11 contributes to cohesion establishment and DNA replication is currently unclear. There is some conflicting information about the requirement of the helicase activity of DDX11/Chl1 in promoting cohesion establishment. Studies of the yeast and Mammalian Chl1 proteins claim that the Walker A mutant (K48 in yeast and K50 in vertebrates) as well as the Q loop mutant Q23, both of which abolish helicase activity are fully capable of promoting cohesion establishment (Samora et al 2016 and Cortone et al 2018). This has led to the notion that the helicase activity is not required and that a hitherto mysterious activity that requires ‘engagement’ with cohesin and/or Scc2 is instead the mechanism by which Chl1 promotes cohesion establishment. However, analysis of the same Walker mutant in yeast Chl1 or chicken DDX11 by different groups showed the opposite, namely that the helicase activity was critical for cohesion (Holloway S 2000 and Abe T et al 2016). A recent report also showed that the DDX11 Q23A mutant is incapable of rescuing the cohesion defects in WABS fibroblasts, suggesting that the helicase activity is required. Strangely, this paper is from the same lab represented in this manuscript (Faramarz A et al PLOS one 2020). This issue is an important one, as a requirement for the helicase activity would imply that the helicase promotes cohesion establishment by altering the conformation of DNA. The role of Chl1/DDX11’s helicase activity therefore is of substantial interest for the cohesin field. In the current study, the authors present clinical features of seven new WABS patients and characterise the new cases initially by identifying bi-allelic mutations in these patients. They show cells derived from patients display significant cohesion defects and importantly these defects can be corrected by expression of wild type DDX11. A few of the mutations identified are hypomorphic, presumably because of their low protein stability (which they measure). Two of the mutations (G57R and C705Y) lie in the helicase domain of DDX11, the C705Y mutant is shown to abolish DDX11 helicase activity. The paper describes the generation of DDX11 deletion cell lines (with and without p53) and shows that this is accompanied by severe cohesion defects. They show that overexpression of wild type and the hypomorphic mutants rescue the cohesion defects of the DDX11 knockout lines. In contrast, overexpression of the G57R and C705Y mutants as well as the K50A Walker A mutant fail to rescue the cohesion defects. These observations are striking and support the notion that the helicase activity is necessary for its role in DNA replication/repair and in cohesion establishment. Having said this, would it not have been better to ‘knock in’ the mutant proteins? If the manuscript had been presented in this manner, it would have warranted publication in Nature communications. However, in its current state we cannot recommend publication of this manuscript. The main problem the manuscript suffers from is that too many unnecessary details are presented. Most of these (Sensitivity to various drugs, modest reduction in fork speeds, accumulation of damage foci) were already know for WABS cells and the observations presented here do not greatly alter our understanding. Large sections Figures 1-7 could easily be presented as supplements and each discussed with a few lines. The main message of the paper (as prominently stated in the title) appears in Figures 7& 8. Any general reader would have to navigate through all the details presented with different cells and drugs and redundant messages to get to the heart of the story causing them to lose interest. Inexplicably, the evidence that the newly identified mutation is defective in helicase activity is presented supplementary Figure S9. We do not see the point of cramming all these details into the manuscript just for the sake of adding ‘substance’ to the paper, it is very distracting and makes it difficult to read. Another major issue is the fact that the same group of authors have very recently published a paper with all the same assays and show in that manuscript that the helicase activity of DDX11 (based on the analysis of a Q loop mutant Q23A) is required for cohesion and DNA replication/repair (Faramarz A et al PLOS one 2020). Why make two papers out of it? This is a matter for consideration by the editor. The manuscript does have an important message. If this manuscript is to be considered further, it needs serious rearrangement and rewriting in emphasizing the main message, namely the requirement of the helicase activity for cohesion establishment. The mutant proteins could be characterised biochemically to show the effect on ATP binding and hydrolysis in order to make a strong case that the helicase activity of DDX11 is important for promoting cohesion establishment. Reviewer #2 (Remarks to the Author): Summary: Researchers identified new compound heterozygous mutations linked to Warsaw Breakage Syndrome. The DDX11 proteins encoded by the variants display reduced protein stability; the corresponding cells are hyper-sensitive to topoisomerase and PARP inhibitors, show defects in sister chromatid cohesion, and reduced replication fork speed. DDX11 depletion causes chromosomal instability. Treatment of DDX11-deficient cells with G4-interactive compounds was reported to impairs growth induce chromosomal instability and cohesion defect to a more significant extent than loss of FANCJ, a previously determined FANCJ G4-resolving helicase sharing sequence homology with DDX11 in its helicase core domain. DDX11 helicase core domain is required for G4 ligand resistance. Based on their experimental results, authors conclude that DDX11 is important for protection against G4-induced DSBs as well as cohesion defects that are linked to DNA replication. Critical Comments: As detailed in the Summary, this is an investigation of the molecular defects of new patients afflicted by Warsaw Breakage Syndrome, a DDX11 DNA helicase disorder. The new mutations identified help to solidify the molecular pathology underlying the chromosomal instability disorder. The evidence supports the conclusions made by the authors, and the experimental studies are solid. The results will provide value to researchers and clinicians alike. Statistical analyses are appropriate. Experimental procedures and rational are explained well. A number of concerns are raised below that need to be carefully considered and addressed by the authors: Introduction: Suggest that phrasing “putatively moderate symptoms of human WABS patients” be changed as this is subjective and not well defended. I am a bit confused over the diagnosis of WABS03 as “undoubtedly a WABS patient” (Discussion). Based on protein expression it is stated that the corresponding mutant protein R140Q encoded by a mutant allele shows a small increase and that evidence it is the disease-causing mutation remain limited (paraphrase, page 7 lines 215-219). Further, immunofluorescence studies suggested that the R140Q mutant localized to nucleus similar to wt DDX11. Please clarify the basis for assigning WABS03 as a patient with Warsaw Breakage Syndrome so that it is easier for the reader to understand. Perhaps I am missing it, but were all the DDX11 missense mutant proteins, (e.g., R140Q) tested for helicase activity in vitro? I see in Figure S9 that C705Y was tested. The authors should test R140Q. On page 6, the authors suggest that the rationale to assess if WABS cells show a replication defect as assessed by DNA fiber assays was that DDX11 is known to interact with replication factors. While this may be true, a number of these proteins are also implicated in DNA repair pathways. Moreover, a previous paper published in 2016 demonstrated that DDX11-deficient cells display reduced replication fork progression in cells exposed to the nucleotide-depleting agent hydroxyurea compared to DDX11-proficient cells (as demonstrated by DNA fiber studies). I found this reference (39) in the Discussion, but perhaps it should be mentioned earlier, as this reference (39) provides a stronger rationale for investigating if the lymphoblasts of newly identified DDX11 patients display replication defects. Two G4-stabilizing compounds are used for cellular studies: Quarfloxin and CX5461 (reference 53). I do not find in reference 53 any mention of Quarfloxin. Perhaps I am missing it. That paper does list CX3543 in addition to CX5461. Further reading in the literature revealed to me that Quarfloxin is indeed CX3543. Perhaps this could be better spelled out for the reader. The greater Quarfloxin sensitivity of WABS cells compared to FANCJ-deficient cells is interesting but seems under-explored in this study. Although one could argue that the focus of the current work is on DDX11, I found the analysis by comparison to be perfunctory. For example, why not compare the sensitivity of FANCJ- versus DDX11-deficient cells to telomestatin as well? The suggestion that the difference in Quarfloxin sensitivity is based on the observation that “different G4 interacting compounds appear to selectively bind different types of G-quadruplexes” (demonstrated in vitro) may or may not be relevant in vivo and at first glance seems to be speculative. From an experimental standpoint, measuring drug sensitivity by colony survival and acute DNA damage induction by an appropriate immunofluorescent marker (e.g., gamma-H2AX; 53BP1) may provide greater insight than the IncuCyte assay of cell growth for differential effects of the G4 stabilizer as it relates to DDX11 versus FANCJ status. Returning to the point made by the authors that the apparently differential effects of Quarfloxin versus Telomestatin on DDX11-deficient cells versus FANCJ-deficient cells may reflect differences in G4 binding specificity, the authors should address this directly in the Discussion by reporting what has been published in the literature on this issue. The mention in the Discussion that TMS stabilizes formation of telomeric G4 structures whereas Quarafloxin selectively targets a G4 structure commonly found in ribosomal DNA and in promoter sequences, and that quarfloxin accumulates in nucleoli. These points raise the question if a particular G4 architecture is favored for the ribosomal DNA repeats housed in the nucleoli versus the telomeric repeats of telomeres? Given that DDX11 was previously reported to efficiently resolve 5’ ssDNA tailed two-stranded anti- parallel G4 substrates in vitro but act poorly on unimolecular G4 substrate which FANCJ is capable of resolving efficiently, it raises the question following the Reviewer’s comments if Quarafloxin preferentially binds bimolecular G4 substrates versus unimolecular G4 substrates, or (as the authors also suggest) that the preferential subcellular localization of Quarafloxin to nucleoli plays a role and suggests a more prominent nucleolar role of DDX11. Is it known what is the predominant G4 architecture of ribosomal DNA repeats? Given 1) the potential effects of G4-stabilizing compounds on nuclear DNA replication, 2) the observation made in the current study that DDX11-deficient cells show reduced replication speed, and 3) the previously published result that forks are preferentially slowed by hydroxyurea (an agent that like G4 binding cmpds induces replication stress but by a distinct mechanism) in DDX11-deficient cells compared to Wild-type cells, I would have expected the authors to perform additional DNA fiber analyses to assess effects of the G4-stabilizing agents on not only replication fork speed, but other fork dynamic parameters, i.e., fork asymmetry, nascent strand degradation, fork restart, etc. These additional studies may have yielded insight to the relationships among DDX11 status, G-quadruplexes, chromosomal instability and sister chromatid cohesion, as suggested by the authors. Moreover, the DNA fiber studies with DDX11-deficient and proficient cells exposed to G4-binding compounds need to be done to support the conclusions made in the study and provide greater insight to the apparent G4 ligand sensitivity reported in this manuscript. In Figure 2 (panels C and D), the authors show a single image of gamma H2AX and DNA fiber assays with double label (ongoing replication fork), respectively. The authors should show representative images for both panels C and D from the various cell lines in which quantitative data are shown. Similarly, in Figure 3H, representative metaphase spreads for the analyzed cell lines in which quantitative data are shown should be depicted. Reviewer #3 (Remarks to the Author): Faramarz and colleagues describe molecular and cellular defects in cells derived from patients with Warsaw Breakage Syndrome (WBS). Previous studies (as well as the current work) link defects in DDX11 to the development of WBS. While mRNA levels are unaffected WBS cells exhibit DDX11 protein instability. Mutations in some patients are also likely to impair the helicase activity of DDX11. In addition to chromosome breakage WBS cells exhibit impaired chromosome cohesion, hyper-sensitivity to DNA damaging agents such as camptothecin and PARP inhibitors, decreased DNA synthesis and elevated chromosomal aberrations. RPE1 cells knocked out for DDX11 also exhibit impaired cell proliferation that is dependent on P53. Importantly they discount the contribution of the pseudogene DDX12p to these phenotypes. Faramarz et al go on to show that DDX11 defective cells are hyper-sensitive to G4-DNA stabilizing compounds which inhibit cell proliferation, induced loss of cohesion and caused increased chromosomal breaks. The same is true for WBS lymphoblasts and the defects were reversed in cells expressing wild type DDX11. This is a nice piece of work, particularly strong when describing the defects associated with loss of DDX11 and with WBS. However, it is much less strong on mechanism. It documents defects in cohesion and in replication but reports little as to why this is. The biggest limitation is that the manuscript highlights the role of DDX11 in protecting against G-quadrupled induced DNA damage but little about the relevance of this to normal metabolism. A major limitation is the use of G4- stabilizing ligands such as pyridostatin. This compound has recently been shown to trapping of Topoisomerase 2 on DNA but it is not clear whether this is linked to G4 stabilization (https://www.biorxiv.org/content/10.1101/845446v1). So it is possible that the DDX11 defective cells are mainly sensitive to trapping of topoisomerases rather than DNA damage induced by persistent G4-DNA structures. Even, so the investigation into the contribution of DDX11 to DNA repair and the DNA damage response is quite superficial, particularly as the title emphasizes this aspect. It would be useful if the authors looked in more detail at the genetic relationship between DDX11 and genes shown to affect response to G4-DNA such as FANCJ, REV1, BLM and WRN. Ultimately the reader would like to understand wto which physiological pathway(s) DDX11 contributes.	Reviewers' comments: Reviewer #1 (Remarks to the Author): DDX11/ChlR1 and its yeast orthologue Chl1 are ATP-dependent DNA helicases related to Xpd and FANCJ, that unwinds DNA with 5’-3’ directionality. Mutations in DDX11 are linked with Warsaw breakage syndrome (WABS) which is characterised by high incidence of chromosomes with centromeric cohesion defects (“railroads”) and “premature chromatid separation” (PCS). It has previously been documented that yeast Chl1 and mammalian DDX11 are associated with the replisome and promote establishment of sister chromatid cohesion during S phase. How they are recruited to the replisome is not fully understood. DDX11 associates with Timeless via a highly conserved amino acid motif embedded in a large CHL1/DDX11-specific loop inserted into the helicase domain (Cortone et al PLOS gen 2018). Mutations in DDX11 that abolish this interaction show cohesion defects similar to DDX11 depletion. It was recently shown that DDX11 and Timeless collaborate in order to detect and process G4 structures during replication (Lerner, LK et al bioRxiv). How DDX11 contributes to cohesion establishment and DNA replication is currently unclear. There is some conflicting information about the requirement of the helicase activity of DDX11/Chl1 in promoting cohesion establishment. Studies of the yeast and Mammalian Chl1 proteins claim that the Walker A mutant (K48 in yeast and K50 in vertebrates) as well as the Q loop mutant Q23, both of which abolish helicase activity are fully capable of promoting cohesion establishment (Samora et al 2016 and Cortone et al 2018). This has led to the notion that the helicase activity is not required and that a hitherto mysterious activity that requires ‘engagement’ with cohesin and/or Scc2 is instead the mechanism by which Chl1 promotes cohesion establishment. However, analysis of the same Walker mutant in yeast Chl1 or chicken DDX11 by different groups showed the opposite, namely that the helicase activity was critical for cohesion (Holloway S 2000 and Abe T et al 2016). A recent report also showed that the DDX11 Q23A mutant is incapable of rescuing the cohesion defects in WABS fibroblasts, suggesting that the helicase activity is required. Strangely, this paper is from the same lab represented in this manuscript (Faramarz A et al PLOS one 2020). This issue is an important one, as a requirement for the helicase activity would imply that the helicase promotes cohesion establishment by altering the conformation of DNA. The role of Chl1/DDX11’s helicase activity therefore is of substantial interest for the cohesin field. In the current study, the authors present clinical features of seven new WABS patients and characterise the new cases initially by identifying bi-allelic mutations in these patients. They show cells derived from patients display significant cohesion defects and importantly these defects can be corrected by expression of wild type DDX11. A few of the mutations identified are hypomorphic, presumably because of their low protein stability (which they measure). Two of the mutations (G57R and C705Y) lie in the helicase domain of DDX11, the C705Y mutant is shown to abolish DDX11 helicase activity. The paper describes the generation of DDX11 deletion cell lines (with and without p53) and shows that this is accompanied by severe cohesion defects. They show that overexpression of wild type and the hypomorphic mutants rescue the cohesion defects of the DDX11 knockout lines. In contrast, overexpression of the G57R and C705Y mutants as well as the K50A Walker A mutant fail to rescue the cohesion defects. These observations are striking and support the notion that the helicase activity is necessary for its role in DNA replication/repair and in cohesion establishment. Having said this, would it not have been better to ‘knock in’ the mutant proteins? If the manuscript had been presented in this manner, it would have warranted publication in Nature communications. However, in its current state we cannot recommend publication of this manuscript. The main problem the manuscript suffers from is that too many unnecessary details are presented. Most of these (Sensitivity to various drugs, modest reduction in fork speeds, accumulation of damage foci) were already know for WABS cells and the observations presented here do not greatly alter our understanding. Large sections Figures 1-7 could easily be presented as supplements and each discussed with a few lines. The main message of the paper (as prominently stated in the title) appears in Figures 7& 8. Any general reader would have to navigate through all the details presented with different cells and drugs and redundant messages to get to the heart of the story causing them to lose interest. Inexplicably, the evidence that the newly identified mutation is defective in helicase activity is presented supplementary Figure S9. We do not see the point of cramming all these details into the manuscript just for the sake of adding ‘substance’ to the paper, it is very distracting and makes it difficult to read. Another major issue is the fact that the same group of authors have very recently published a paper with all the same assays and show in that manuscript that the helicase activity of DDX11 (based on the analysis of a Q loop mutant Q23A) is required for cohesion and DNA replication/repair (Faramarz A et al PLOS one 2020). Why make two papers out of it? This is a matter for consideration by the editor. The manuscript does have an important message. If this manuscript is to be considered further, it needs serious rearrangement and rewriting in emphasizing the main message, namely the requirement of the helicase activity for cohesion establishment. The mutant proteins could be characterised biochemically to show the effect on ATP binding and hydrolysis in order to make a strong case that the helicase activity of DDX11 is important for promoting cohesion establishment. Reviewer #2 (Remarks to the Author): Summary: Researchers identified new compound heterozygous mutations linked to Warsaw Breakage Syndrome. The DDX11 proteins encoded by the variants display reduced protein stability; the corresponding cells are hyper-sensitive to topoisomerase and PARP inhibitors, show defects in sister chromatid cohesion, and reduced replication fork speed. DDX11 depletion causes chromosomal instability. Treatment of DDX11-deficient cells with G4-interactive compounds was reported to impairs growth induce chromosomal instability and cohesion defect to a more significant extent than loss of FANCJ, a previously determined FANCJ G4-resolving helicase sharing sequence homology with DDX11 in its helicase core domain. DDX11 helicase core domain is required for G4 ligand resistance. Based on their experimental results, authors conclude that DDX11 is important for protection against G4-induced DSBs as well as cohesion defects that are linked to DNA replication. Critical Comments: As detailed in the Summary, this is an investigation of the molecular defects of new patients afflicted by Warsaw Breakage Syndrome, a DDX11 DNA helicase disorder. The new mutations identified help to solidify the molecular pathology underlying the chromosomal instability disorder. The evidence supports the conclusions made by the authors, and the experimental studies are solid. The results will provide value to researchers and clinicians alike. Statistical analyses are appropriate. Experimental procedures and rational are explained well. A number of concerns are raised below that need to be carefully considered and addressed by the authors: Introduction: Suggest that phrasing “putatively moderate symptoms of human WABS patients” be changed as this is subjective and not well defended. I am a bit confused over the diagnosis of WABS03 as “undoubtedly a WABS patient” (Discussion). Based on protein expression it is stated that the corresponding mutant protein R140Q encoded by a mutant allele shows a small increase and that evidence it is the disease-causing mutation remain limited (paraphrase, page 7 lines 215-219). Further, immunofluorescence studies suggested that the R140Q mutant localized to nucleus similar to wt DDX11. Please clarify the basis for assigning WABS03 as a patient with Warsaw Breakage Syndrome so that it is easier for the reader to understand. Perhaps I am missing it, but were all the DDX11 missense mutant proteins, (e.g., R140Q) tested for helicase activity in vitro? I see in Figure S9 that C705Y was tested. The authors should test R140Q. On page 6, the authors suggest that the rationale to assess if WABS cells show a replication defect as assessed by DNA fiber assays was that DDX11 is known to interact with replication factors. While this may be true, a number of these proteins are also implicated in DNA repair pathways. Moreover, a previous paper published in 2016 demonstrated that DDX11-deficient cells display reduced replication fork progression in cells exposed to the nucleotide-depleting agent hydroxyurea compared to DDX11-proficient cells (as demonstrated by DNA fiber studies). I found this reference (39) in the Discussion, but perhaps it should be mentioned earlier, as this reference (39) provides a stronger rationale for investigating if the lymphoblasts of newly identified DDX11 patients display replication defects. Two G4-stabilizing compounds are used for cellular studies: Quarfloxin and CX5461 (reference 53). I do not find in reference 53 any mention of Quarfloxin. Perhaps I am missing it. That paper does list CX3543 in addition to CX5461. Further reading in the literature revealed to me that Quarfloxin is indeed CX3543. Perhaps this could be better spelled out for the reader. The greater Quarfloxin sensitivity of WABS cells compared to FANCJ-deficient cells is interesting but seems under-explored in this study. Although one could argue that the focus of the current work is on DDX11, I found the analysis by comparison to be perfunctory. For example, why not compare the sensitivity of FANCJ- versus DDX11-deficient cells to telomestatin as well? The suggestion that the difference in Quarfloxin sensitivity is based on the observation that “different G4 interacting compounds appear to selectively bind different types of G-quadruplexes” (demonstrated in vitro) may or may not be relevant in vivo and at first glance seems to be speculative. From an experimental standpoint, measuring drug sensitivity by colony survival and acute DNA damage induction by an appropriate immunofluorescent marker (e.g., gamma-H2AX; 53BP1) may provide greater insight than the IncuCyte assay of cell growth for differential effects of the G4 stabilizer as it relates to DDX11 versus FANCJ status. Returning to the point made by the authors that the apparently differential effects of Quarfloxin versus Telomestatin on DDX11-deficient cells versus FANCJ-deficient cells may reflect differences in G4 binding specificity, the authors should address this directly in the Discussion by reporting what has been published in the literature on this issue. The mention in the Discussion that TMS stabilizes formation of telomeric G4 structures whereas Quarafloxin selectively targets a G4 structure commonly found in ribosomal DNA and in promoter sequences, and that quarfloxin accumulates in nucleoli. These points raise the question if a particular G4 architecture is favored for the ribosomal DNA repeats housed in the nucleoli versus the telomeric repeats of telomeres? Given that DDX11 was previously reported to efficiently resolve 5’ ssDNA tailed two-stranded anti- parallel G4 substrates in vitro but act poorly on unimolecular G4 substrate which FANCJ is capable of resolving efficiently, it raises the question following the Reviewer’s comments if Quarafloxin preferentially binds bimolecular G4 substrates versus unimolecular G4 substrates, or (as the authors also suggest) that the preferential subcellular localization of Quarafloxin to nucleoli plays a role and suggests a more prominent nucleolar role of DDX11. Is it known what is the predominant G4 architecture of ribosomal DNA repeats? Given 1) the potential effects of G4-stabilizing compounds on nuclear DNA replication, 2) the observation made in the current study that DDX11-deficient cells show reduced replication speed, and 3) the previously published result that forks are preferentially slowed by hydroxyurea (an agent that like G4 binding cmpds induces replication stress but by a distinct mechanism) in DDX11-deficient cells compared to Wild-type cells, I would have expected the authors to perform additional DNA fiber analyses to assess effects of the G4-stabilizing agents on not only replication fork speed, but other fork dynamic parameters, i.e., fork asymmetry, nascent strand degradation, fork restart, etc. These additional studies may have yielded insight to the relationships among DDX11 status, G-quadruplexes, chromosomal instability and sister chromatid cohesion, as suggested by the authors. Moreover, the DNA fiber studies with DDX11-deficient and proficient cells exposed to G4-binding compounds need to be done to support the conclusions made in the study and provide greater insight to the apparent G4 ligand sensitivity reported in this manuscript. In Figure 2 (panels C and D), the authors show a single image of gamma H2AX and DNA fiber assays with double label (ongoing replication fork), respectively. The authors should show representative images for both panels C and D from the various cell lines in which quantitative data are shown. Similarly, in Figure 3H, representative metaphase spreads for the analyzed cell lines in which quantitative data are shown should be depicted. Reviewer #3 (Remarks to the Author): Faramarz and colleagues describe molecular and cellular defects in cells derived from patients with Warsaw Breakage Syndrome (WBS). Previous studies (as well as the current work) link defects in DDX11 to the development of WBS. While mRNA levels are unaffected WBS cells exhibit DDX11 protein instability. Mutations in some patients are also likely to impair the helicase activity of DDX11. In addition to chromosome breakage WBS cells exhibit impaired chromosome cohesion, hyper-sensitivity to DNA damaging agents such as camptothecin and PARP inhibitors, decreased DNA synthesis and elevated chromosomal aberrations. RPE1 cells knocked out for DDX11 also exhibit impaired cell proliferation that is dependent on P53. Importantly they discount the contribution of the pseudogene DDX12p to these phenotypes. Faramarz et al go on to show that DDX11 defective cells are hyper-sensitive to G4-DNA stabilizing compounds which inhibit cell proliferation, induced loss of cohesion and caused increased chromosomal breaks. The same is true for WBS lymphoblasts and the defects were reversed in cells expressing wild type DDX11. This is a nice piece of work, particularly strong when describing the defects associated with loss of DDX11 and with WBS. However, it is much less strong on mechanism. It documents defects in cohesion and in replication but reports little as to why this is. The biggest limitation is that the manuscript highlights the role of DDX11 in protecting against G-quadrupled induced DNA damage but little about the relevance of this to normal metabolism. A major limitation is the use of G4- stabilizing ligands such as pyridostatin. This compound has recently been shown to trapping of Topoisomerase 2 on DNA but it is not clear whether this is linked to G4 stabilization (https://www.biorxiv.org/content/10.1101/845446v1). So it is possible that the DDX11 defective cells are mainly sensitive to trapping of topoisomerases rather than DNA damage induced by persistent G4-DNA structures. Even, so the investigation into the contribution of DDX11 to DNA repair and the DNA damage response is quite superficial, particularly as the title emphasizes this aspect. It would be useful if the authors looked in more detail at the genetic relationship between DDX11 and genes shown to affect response to G4-DNA such as FANCJ, REV1, BLM and WRN. Ultimately the reader would like to understand wto which physiological pathway(s) DDX11 contributes.	TPR	NC_BS_BIOCH_007_RR1.txt	{ "criticism": 21, "example": 8, "importance_and_relevance": 9, "materials_and_methods": 45, "praise": 10, "presentation_and_reporting": 17, "results_and_discussion": 11, "suggestion_and_solution": 20, "total": 105 }	{ "criticism": 0.2, "example": 0.0761904761904762, "importance_and_relevance": 0.08571428571428572, "materials_and_methods": 0.42857142857142855, "praise": 0.09523809523809523, "presentation_and_reporting": 0.1619047619047619, "results_and_discussion": 0.10476190476190476, "suggestion_and_solution": 0.19047619047619047 }	1.342857	[ { "content": "Please review this paper.", "role": "user" }, { "content": "Reviewers' comments: \n \nReviewer #1 (Remarks to the Author): \n \nDDX11/ChlR1 and its yeast orthologue Chl1 are ATP-dependent DNA helicases related to Xpd and \nFANCJ, that unwinds DNA with 5’-3’ directionality. Mutations in DDX11 are linked with Warsaw \nbreakage syndrome (WABS) which is characterised by high incidence of chromosomes with \ncentromeric cohesion defects (“railroads”) and “premature chromatid separation” (PCS). It has \npreviously been documented that yeast Chl1 and mammalian DDX11 are associated with the \nreplisome and promote establishment of sister chromatid cohesion during S phase. How they are \nrecruited to the replisome is not fully understood. DDX11 associates with Timeless via a highly \nconserved amino acid motif embedded in a large CHL1/DDX11-specific loop inserted into the \nhelicase domain (Cortone et al PLOS gen 2018). Mutations in DDX11 that abolish this interaction \nshow cohesion defects similar to DDX11 depletion. It was recently shown that DDX11 and Timeless \ncollaborate in order to detect and process G4 structures during replication (Lerner, LK et al \nbioRxiv). How DDX11 contributes to cohesion establishment and DNA replication is currently \nunclear. \nThere is some conflicting information about the requirement of the helicase activity of DDX11/Chl1 \nin promoting cohesion establishment. Studies of the yeast and Mammalian Chl1 proteins claim that \nthe Walker A mutant (K48 in yeast and K50 in vertebrates) as well as the Q loop mutant Q23, both \nof which abolish helicase activity are fully capable of promoting cohesion establishment (Samora et \nal 2016 and Cortone et al 2018). This has led to the notion that the helicase activity is not required \nand that a hitherto mysterious activity that requires ‘engagement’ with cohesin and/or Scc2 is \ninstead the mechanism by which Chl1 promotes cohesion establishment. However, analysis of the \nsame Walker mutant in yeast Chl1 or chicken DDX11 by different groups showed the opposite, \nnamely that the helicase activity was critical for cohesion (Holloway S 2000 and Abe T et al 2016). \nA recent report also showed that the DDX11 Q23A mutant is incapable of rescuing the cohesion \ndefects in WABS fibroblasts, suggesting that the helicase activity is required. Strangely, this paper \nis from the same lab represented in this manuscript (Faramarz A et al PLOS one 2020). This issue \nis an important one, as a requirement for the helicase activity would imply that the helicase \npromotes cohesion establishment by altering the conformation of DNA. The role of Chl1/DDX11’s \nhelicase activity therefore is of substantial interest for the cohesin field. \n \nIn the current study, the authors present clinical features of seven new WABS patients and \ncharacterise the new cases initially by identifying bi-allelic mutations in these patients. They show \ncells derived from patients display significant cohesion defects and importantly these defects can \nbe corrected by expression of wild type DDX11. A few of the mutations identified are hypomorphic, \npresumably because of their low protein stability (which they measure). Two of the mutations \n(G57R and C705Y) lie in the helicase domain of DDX11, the C705Y mutant is shown to abolish \nDDX11 helicase activity. The paper describes the generation of DDX11 deletion cell lines (with and \nwithout p53) and shows that this is accompanied by severe cohesion defects. They show that \noverexpression of wild type and the hypomorphic mutants rescue the cohesion defects of the \nDDX11 knockout lines. In contrast, overexpression of the G57R and C705Y mutants as well as the \nK50A Walker A mutant fail to rescue the cohesion defects. These observations are striking and \nsupport the notion that the helicase activity is necessary for its role in DNA replication/repair and \nin cohesion establishment. Having said this, would it not have been better to ‘knock in’ the mutant \nproteins? If the manuscript had been presented in this manner, it would have warranted \npublication in Nature communications. However, in its current state we cannot recommend \npublication of this manuscript. \n \nThe main problem the manuscript suffers from is that too many unnecessary details are presented. \nMost of these (Sensitivity to various drugs, modest reduction in fork speeds, accumulation of \ndamage foci) were already know for WABS cells and the observations presented here do not \ngreatly alter our understanding. Large sections Figures 1-7 could easily be presented as \nsupplements and each discussed with a few lines. The main message of the paper (as prominently \n\n\fstated in the title) appears in Figures 7& 8. Any general reader would have to navigate through all \nthe details presented with different cells and drugs and redundant messages to get to the heart of \nthe story causing them to lose interest. Inexplicably, the evidence that the newly identified \nmutation is defective in helicase activity is presented supplementary Figure S9. We do not see the \npoint of cramming all these details into the manuscript just for the sake of adding ‘substance’ to \nthe paper, it is very distracting and makes it difficult to read. \n \nAnother major issue is the fact that the same group of authors have very recently published a \npaper with all the same assays and show in that manuscript that the helicase activity of DDX11 \n(based on the analysis of a Q loop mutant Q23A) is required for cohesion and DNA \nreplication/repair (Faramarz A et al PLOS one 2020). Why make two papers out of it? This is a \nmatter for consideration by the editor. \n \nThe manuscript does have an important message. If this manuscript is to be considered further, it \nneeds serious rearrangement and rewriting in emphasizing the main message, namely the \nrequirement of the helicase activity for cohesion establishment. The mutant proteins could be \ncharacterised biochemically to show the effect on ATP binding and hydrolysis in order to make a \nstrong case that the helicase activity of DDX11 is important for promoting cohesion establishment. \n \n \n \nReviewer #2 (Remarks to the Author): \n \nSummary: \nResearchers identified new compound heterozygous mutations linked to Warsaw Breakage \nSyndrome. The DDX11 proteins encoded by the variants display reduced protein stability; the \ncorresponding cells are hyper-sensitive to topoisomerase and PARP inhibitors, show defects in \nsister chromatid cohesion, and reduced replication fork speed. DDX11 depletion causes \nchromosomal instability. Treatment of DDX11-deficient cells with G4-interactive compounds was \nreported to impairs growth induce chromosomal instability and cohesion defect to a more \nsignificant extent than loss of FANCJ, a previously determined FANCJ G4-resolving helicase sharing \nsequence homology with DDX11 in its helicase core domain. DDX11 helicase core domain is \nrequired for G4 ligand resistance. Based on their experimental results, authors conclude that \nDDX11 is important for protection against G4-induced DSBs as well as cohesion defects that are \nlinked to DNA replication. \n \n \nCritical Comments: \nAs detailed in the Summary, this is an investigation of the molecular defects of new patients \nafflicted by Warsaw Breakage Syndrome, a DDX11 DNA helicase disorder. The new mutations \nidentified help to solidify the molecular pathology underlying the chromosomal instability disorder. \nThe evidence supports the conclusions made by the authors, and the experimental studies are \nsolid. The results will provide value to researchers and clinicians alike. Statistical analyses are \nappropriate. Experimental procedures and rational are explained well. \n \nA number of concerns are raised below that need to be carefully considered and addressed by the \nauthors: \n \nIntroduction: Suggest that phrasing “putatively moderate symptoms of human WABS patients” be \nchanged as this is subjective and not well defended. \n \nI am a bit confused over the diagnosis of WABS03 as “undoubtedly a WABS patient” (Discussion). \nBased on protein expression it is stated that the corresponding mutant protein R140Q encoded by \na mutant allele shows a small increase and that evidence it is the disease-causing mutation remain \nlimited (paraphrase, page 7 lines 215-219). Further, immunofluorescence studies suggested that \n\n\fthe R140Q mutant localized to nucleus similar to wt DDX11. Please clarify the basis for assigning \nWABS03 as a patient with Warsaw Breakage Syndrome so that it is easier for the reader to \nunderstand. \n \nPerhaps I am missing it, but were all the DDX11 missense mutant proteins, (e.g., R140Q) tested \nfor helicase activity in vitro? I see in Figure S9 that C705Y was tested. The authors should test \nR140Q. \n \nOn page 6, the authors suggest that the rationale to assess if WABS cells show a replication defect \nas assessed by DNA fiber assays was that DDX11 is known to interact with replication factors. \nWhile this may be true, a number of these proteins are also implicated in DNA repair pathways. \nMoreover, a previous paper published in 2016 demonstrated that DDX11-deficient cells display \nreduced replication fork progression in cells exposed to the nucleotide-depleting agent \nhydroxyurea compared to DDX11-proficient cells (as demonstrated by DNA fiber studies). I found \nthis reference (39) in the Discussion, but perhaps it should be mentioned earlier, as this reference \n(39) provides a stronger rationale for investigating if the lymphoblasts of newly identified DDX11 \npatients display replication defects. \n \nTwo G4-stabilizing compounds are used for cellular studies: Quarfloxin and CX5461 (reference 53). \nI do not find in reference 53 any mention of Quarfloxin. Perhaps I am missing it. That paper does \nlist CX3543 in addition to CX5461. Further reading in the literature revealed to me that Quarfloxin \nis indeed CX3543. Perhaps this could be better spelled out for the reader. \n \nThe greater Quarfloxin sensitivity of WABS cells compared to FANCJ-deficient cells is interesting \nbut seems under-explored in this study. Although one could argue that the focus of the current \nwork is on DDX11, I found the analysis by comparison to be perfunctory. For example, why not \ncompare the sensitivity of FANCJ- versus DDX11-deficient cells to telomestatin as well? The \nsuggestion that the difference in Quarfloxin sensitivity is based on the observation that “different \nG4 interacting compounds appear to selectively bind different types of G-quadruplexes” \n(demonstrated in vitro) may or may not be relevant in vivo and at first glance seems to be \nspeculative. From an experimental standpoint, measuring drug sensitivity by colony survival and \nacute DNA damage induction by an appropriate immunofluorescent marker (e.g., gamma-H2AX; \n53BP1) may provide greater insight than the IncuCyte assay of cell growth for differential effects \nof the G4 stabilizer as it relates to DDX11 versus FANCJ status. \n \nReturning to the point made by the authors that the apparently differential effects of Quarfloxin \nversus Telomestatin on DDX11-deficient cells versus FANCJ-deficient cells may reflect differences \nin G4 binding specificity, the authors should address this directly in the Discussion by reporting \nwhat has been published in the literature on this issue. The mention in the Discussion that TMS \nstabilizes formation of telomeric G4 structures whereas Quarafloxin selectively targets a G4 \nstructure commonly found in ribosomal DNA and in promoter sequences, and that quarfloxin \naccumulates in nucleoli. These points raise the question if a particular G4 architecture is favored \nfor the ribosomal DNA repeats housed in the nucleoli versus the telomeric repeats of telomeres? \nGiven that DDX11 was previously reported to efficiently resolve 5’ ssDNA tailed two-stranded anti-\nparallel G4 substrates in vitro but act poorly on unimolecular G4 substrate which FANCJ is capable \nof resolving efficiently, it raises the question following the Reviewer’s comments if Quarafloxin \npreferentially binds bimolecular G4 substrates versus unimolecular G4 substrates, or (as the \nauthors also suggest) that the preferential subcellular localization of Quarafloxin to nucleoli plays a \nrole and suggests a more prominent nucleolar role of DDX11. Is it known what is the predominant \nG4 architecture of ribosomal DNA repeats? \n \nGiven 1) the potential effects of G4-stabilizing compounds on nuclear DNA replication, 2) the \nobservation made in the current study that DDX11-deficient cells show reduced replication speed, \nand 3) the previously published result that forks are preferentially slowed by hydroxyurea (an \nagent that like G4 binding cmpds induces replication stress but by a distinct mechanism) in \n\n\fDDX11-deficient cells compared to Wild-type cells, I would have expected the authors to perform \nadditional DNA fiber analyses to assess effects of the G4-stabilizing agents on not only replication \nfork speed, but other fork dynamic parameters, i.e., fork asymmetry, nascent strand degradation, \nfork restart, etc. These additional studies may have yielded insight to the relationships among \nDDX11 status, G-quadruplexes, chromosomal instability and sister chromatid cohesion, as \nsuggested by the authors. Moreover, the DNA fiber studies with DDX11-deficient and proficient \ncells exposed to G4-binding compounds need to be done to support the conclusions made in the \nstudy and provide greater insight to the apparent G4 ligand sensitivity reported in this manuscript. \n \nIn Figure 2 (panels C and D), the authors show a single image of gamma H2AX and DNA fiber \nassays with double label (ongoing replication fork), respectively. The authors should show \nrepresentative images for both panels C and D from the various cell lines in which quantitative \ndata are shown. \n \nSimilarly, in Figure 3H, representative metaphase spreads for the analyzed cell lines in which \nquantitative data are shown should be depicted. \n \n \n \n \nReviewer #3 (Remarks to the Author): \n \nFaramarz and colleagues describe molecular and cellular defects in cells derived from patients with \nWarsaw Breakage Syndrome (WBS). Previous studies (as well as the current work) link defects in \nDDX11 to the development of WBS. While mRNA levels are unaffected WBS cells exhibit DDX11 \nprotein instability. Mutations in some patients are also likely to impair the helicase activity of \nDDX11. In addition to chromosome breakage WBS cells exhibit impaired chromosome cohesion, \nhyper-sensitivity to DNA damaging agents such as camptothecin and PARP inhibitors, decreased \nDNA synthesis and elevated chromosomal aberrations. RPE1 cells knocked out for DDX11 also \nexhibit impaired cell proliferation that is dependent on P53. Importantly they discount the \ncontribution of the pseudogene DDX12p to these phenotypes. Faramarz et al go on to show that \nDDX11 defective cells are hyper-sensitive to G4-DNA stabilizing compounds which inhibit cell \nproliferation, induced loss of cohesion and caused increased chromosomal breaks. The same is \ntrue for WBS lymphoblasts and the defects were reversed in cells expressing wild type DDX11. \n \nThis is a nice piece of work, particularly strong when describing the defects associated with loss of \nDDX11 and with WBS. However, it is much less strong on mechanism. It documents defects in \ncohesion and in replication but reports little as to why this is. The biggest limitation is that the \nmanuscript highlights the role of DDX11 in protecting against G-quadrupled induced DNA damage \nbut little about the relevance of this to normal metabolism. A major limitation is the use of G4-\nstabilizing ligands such as pyridostatin. This compound has recently been shown to trapping of \nTopoisomerase 2 on DNA but it is not clear whether this is linked to G4 stabilization \n(https://www.biorxiv.org/content/10.1101/845446v1). So it is possible that the DDX11 defective \ncells are mainly sensitive to trapping of topoisomerases rather than DNA damage induced by \npersistent G4-DNA structures. Even, so the investigation into the contribution of DDX11 to DNA \nrepair and the DNA damage response is quite superficial, particularly as the title emphasizes this \naspect. It would be useful if the authors looked in more detail at the genetic relationship between \nDDX11 and genes shown to affect response to G4-DNA such as FANCJ, REV1, BLM and WRN. \nUltimately the reader would like to understand wto which physiological pathway(s) DDX11 \ncontributes. \n\n\f\n", "role": "assistant" } ]	2,010
Isobutanol production freed from biological limits using synthetic biochemistry	REVIEWERS' COMMENTS: Reviewer #1 (Remarks to the Author): I was reviewer #1 in the previous iteration. The authors have done an excellent job addressing the concerns and comments made. I believe this manuscript is ready for publication and will be widely cited as a pivotal study as the cell-free field continues to develop. Reviewer #2 (Remarks to the Author): The manuscript “Isobutanol production freed from biological limits using synthetic biochemistry” described about production of isobutanol by multiple enzymatic reactions without cells. The authors responded to all the issues pointed out in the peer review process [redacted] and it is considered that the current version of the manuscript is suitable for publication in the journal Nature Communications. Reviewer #3 (Remarks to the Author): Reviewer:3 Recommendation: Publish after minor revisions noted. Comments: As the author commented, so far, the cell-free production of isobutanol is not cost competitive compared to cell-based fermentation and further studies are required for real application (listed in Response to reviewers). However, I agree that this work can be a milestone towards cell-free biomanufacturing and thus I feels that the manuscript should attract much attention to the broad reader of Nature Communications. Minor point. In vitro production of isobutanol was performed at 30 °C in both small scale reaction and bioreactor. According to the supplementary Fig.2, the activity of SmilvD gradually decreased at 25 °C and SmilvD showed almost no activity over 37 °C within 12 h. Based on this thermal tolerance results, I expect that SmilvD has low thermal stability and it will lose its activity at 30 °C more rapidly than that at 25 °C. However, according to the Fig.5c, the residual activity of SmilvD is almost 45 %. In case of KivD- S, although it had higher thermal tolerance than SmilvD, it showed almost no activity after 70 hours of incubation (Fig.5c). In addition, AlsS-p also almost lost its activity after 70 hours of incubation (Fig.5c) despite of its high thermal tolerance.	REVIEWERS' COMMENTS: Reviewer #1 (Remarks to the Author): I was reviewer #1 in the previous iteration. The authors have done an excellent job addressing the concerns and comments made. I believe this manuscript is ready for publication and will be widely cited as a pivotal study as the cell-free field continues to develop. Reviewer #2 (Remarks to the Author): The manuscript “Isobutanol production freed from biological limits using synthetic biochemistry” described about production of isobutanol by multiple enzymatic reactions without cells. The authors responded to all the issues pointed out in the peer review process [redacted] and it is considered that the current version of the manuscript is suitable for publication in the journal Nature Communications. Reviewer #3 (Remarks to the Author): Reviewer:3 Recommendation: Publish after minor revisions noted. Comments: As the author commented, so far, the cell-free production of isobutanol is not cost competitive compared to cell-based fermentation and further studies are required for real application (listed in Response to reviewers). However, I agree that this work can be a milestone towards cell-free biomanufacturing and thus I feels that the manuscript should attract much attention to the broad reader of Nature Communications. Minor point. In vitro production of isobutanol was performed at 30 °C in both small scale reaction and bioreactor. According to the supplementary Fig.2, the activity of SmilvD gradually decreased at 25 °C and SmilvD showed almost no activity over 37 °C within 12 h. Based on this thermal tolerance results, I expect that SmilvD has low thermal stability and it will lose its activity at 30 °C more rapidly than that at 25 °C. However, according to the Fig.5c, the residual activity of SmilvD is almost 45 %. In case of KivD- S, although it had higher thermal tolerance than SmilvD, it showed almost no activity after 70 hours of incubation (Fig.5c). In addition, AlsS-p also almost lost its activity after 70 hours of incubation (Fig.5c) despite of its high thermal tolerance.	TPR	NC_BS_BIOCH_008_RR1.txt	{ "criticism": 0, "example": 3, "importance_and_relevance": 2, "materials_and_methods": 4, "praise": 4, "presentation_and_reporting": 0, "results_and_discussion": 1, "suggestion_and_solution": 2, "total": 14 }	{ "criticism": 0, "example": 0.21428571428571427, "importance_and_relevance": 0.14285714285714285, "materials_and_methods": 0.2857142857142857, "praise": 0.2857142857142857, "presentation_and_reporting": 0, "results_and_discussion": 0.07142857142857142, "suggestion_and_solution": 0.14285714285714285 }	1.142857	[ { "content": "Please review this paper.", "role": "user" }, { "content": "\n\nREVIEWERS' COMMENTS: \n\nReviewer #1 (Remarks to the Author): \n\nI was reviewer #1 in the previous iteration. The authors have done an excellent job addressing the \nconcerns and comments made. I believe this manuscript is ready for publication and will be widely \ncited as a pivotal study as the cell-free field continues to develop. \n\nReviewer #2 (Remarks to the Author): \n\nThe manuscript “Isobutanol production freed from biological limits using synthetic biochemistry” \ndescribed about production of isobutanol by multiple enzymatic reactions without cells. \nThe authors responded to all the issues pointed out in the peer review process [redacted] and it is \nconsidered that the current version of the manuscript is suitable for publication in the journal Nature \nCommunications. \n\nReviewer #3 (Remarks to the Author): \n\nReviewer:3 \nRecommendation: Publish after minor revisions noted. \n\nComments: \nAs the author commented, so far, the cell-free production of isobutanol is not cost competitive \ncompared to cell-based fermentation and further studies are required for real application (listed in \nResponse to reviewers). However, I agree that this work can be a milestone towards cell-free \nbiomanufacturing and thus I feels that the manuscript should attract much attention to the broad \nreader of Nature Communications. \n\nMinor point. \nIn vitro production of isobutanol was performed at 30 °C in both small scale reaction and bioreactor. \nAccording to the supplementary Fig.2, the activity of SmilvD gradually decreased at 25 °C and SmilvD \nshowed almost no activity over 37 °C within 12 h. Based on this thermal tolerance results, I expect \nthat SmilvD has low thermal stability and it will lose its activity at 30 °C more rapidly than that at 25 \n°C. However, according to the Fig.5c, the residual activity of SmilvD is almost 45 %. In case of KivD-\nS, although it had higher thermal tolerance than SmilvD, it showed almost no activity after 70 hours \nof incubation (Fig.5c). In addition, AlsS-p also almost lost its activity after 70 hours of incubation \n(Fig.5c) despite of its high thermal tolerance.\n\n\f\f\n", "role": "assistant" } ]	2,020
Caveolin-1-mediated sphingolipid oncometabolism underlies a metabolic vulnerability of	Reviewers' comments: Reviewer #1 (Remarks to the Author): This manuscript reports the association between plasma sphingolipid profile and disease progression in low risk localized prostate cancer, and attempts to link this observation to a targetable biology in the cancer cells involving scavenging and metabolism of extracellular sphingolipids. The manuscript features some interesting lipidomic profiling and sphingolipid biochemistry, but is ultimately limited by several experimental concerns and the use of laboratory models that poorly recapitulate clinical disease, particularly for early stage localized prostate cancer which is the relevant patient group in active surveillance cohorts. The cell line models used in this manuscript are not typical of active surveillance patients or indeed most clinical prostate tumors. The only lines that express CAV1 are PC-3, RM9 and DU145, all of which are androgen receptor negative and reflect a very small proportion of very advanced or neuroendocrine prostate tumors. Virtually all early stage and low risk primary tumors express the androgen receptor and all of the AR-positive cell lines here (and seemingly most clinical tumor epithelium on Protein atlas) appear to express little to no CAV1, which casts some doubt on the mechanistic links being drawn by the authors between the plasma and cancer cell data. Major comments: 1. Lipidomics data is not quantitative in this manuscript but rather expressed as HRs relative to historical plasma values- this is not explained. Can the authors please provide further information about how specific metabolites/lipids were identified and which standards were used? Was any interference from the presence of EDTA and/or age of the plasma samples evident in the samples detected for metabolomics? Were the plasma samples from fasted individuals? 2. I have a concern over some of the clinical terminology used in the manuscript- when considering upgrading in an active surveillance cohort, this should not be described as aggressive versus indolent prostate cancer. There is a wealth of evidence that the vast majority of even those men experiencing upgrading or other measures of disease progression will not ultimately develop clinical metastases or die from prostate cancer. Even biochemical relapse following prostatectomy has been shown to be a poor surrogate for overall survival. This aspect of the manuscript and its title needs to be described more accurately, and the results interpreted with this in mind- the signatures found are really prognostic for disease progression at best, not aggressive disease. 3. Please justify the 25%:75% cutpoint for the Kaplan-Meier survival curve in Figure 1B. Importantly, when considering an active surveillance cohort, “Survival probability” in this context (Y-axis) is misleading and should more clearly relate to disease progression or upgrading. Figure 1C is not well explained in the text and it is not evident why pathways involved in sphingolipid metabolism were not included considering the focus of the rest of Figure 1. Can the authors please also include a figure presenting the absolute expression data for CAV1 in the TCGA samples rather than relative levels? Alternatively IHC staining data from their own cohorts. This is critical to support the link between the plasma data and the mechanistic data in the rest of the manuscript. My quick impression is that the abundance is rather low in most localized tumor samples. 4. For all of the CAV1 overexpression (LNCaP) and knockdown (PC-3) data in the manuscript, the degree of overexpression or knockdown is not shown anywhere in the manuscript, figures, legends or Supplemental data. This is unacceptable and needs to be included for all major experiments. 5. Figure 2E lacks clarity and could be annotated to emphasize the main points regarding the biochemical pathways being highlighted by the authors. 6. Figure 3: CAV1 staining should be included in all of these experiments to show the extent and efficiency of knockdown in the cell populations shown- particularly important when highlighting individual cells. This may also underpin the fact that there appear to be differences in the appearance of the mitochondria in the PC-3M cells in panel A versus Panel C. 7. Figure 4: How does the total number of EVs change in response to altering CAV1 levels? 8. Figure 6: There is no information in the manuscript or figure legends regarding the numbers of mice used in the experiments. The IHC data on single tumors is inadequate in the absence of any quantitative data on the remainder of the mice in the study. It is not evident whether the effects of eliglustat on tumor growth are due to systemic effects versus a targeted effect on the tumors- was any toxicity noted? A parallel experiment to confirm this using stable knockdown of CAV1 in the RM9 cells would help distinguish between these two mechanisms. Minor comments: 1. Certain figures within the manuscript lack clarity or are illegible. Please consider increasing font sizes and image quality for Fig 1B, 2E, 5B at a minimum. Reviewer #2 (Remarks to the Author): The authors of the manuscript entitled “Cav-1-Mediated Sphingolipid Oncometabolism Underlies a Metabolic Vulnerability of Aggressive Prostate Cancer” report on caveolin-1-dependent effects on the metabolome in prostate cancer by plasma from two independent cohorts of impressive size. Moreover, CCLE and TCGA transcriptomic data are incorporated throughout the study partially supporting findings. Mechanistically, three standard prostate cancer cell lines are applied throughout the study using transient silencing and overexpression of Cav-1 as well as inhibitors/ Cav- 1 specific antibody. The investigation of metabolic rewiring in prostate cancer seems to be novel. The manuscript is well written, however, there are quite a few major points (especially in regards to investigations in potential mechanisms) which require a more detailed determination for current conclusions made by the authors. 1. In regards to the volcano plot shown in Figure 1B, there seems to be a general trend towards higher HR in “early disease progression” indicating that there is no single metabolite detected suggesting a smaller risk. Is this the correct interpretation and if yes, does it mean there is a generally higher abundance of detected metabolites in the DP group? 2. The authors utilize a “sphingolipid signature” to determine the prediction power for progression- free survival separating the cohort into top 25th percentile versus bottom 75th percentile, unfortunately, without providing detailed information how this arbitrary threshold was selected. Beside further information needed and without knowing the rational for an optimal threshold, it is suggested to split the entire cohort into quartiles which might reflect prediction power closer to what has been shown in the manuscript. How does a quartile split look like? 3. The authors have accessed the TCGA prostate cancer transcriptomic data to provide further evidence of CAV-1 in prostate cancer and to study the relationship to lipid metabolism. Here, clinicopathological data accompanied with the data sets should be thoroughly investigated in regards to CAV-1/ gene sets. Has this been done? Neither the results nor the figure legends provide details on the generation of “iCluster-1 to 3”. An appropriate dendrogram on the patients and possible prostate cancer subgrouping (if there are any) might be of benefit. 4. The Western blot Figure 2B (bottom of the images) has to be improved. Here, at least a repetitive Western blot with reference protein should be shown for all three cell lines together. In addition, an appropriate quantification should be done on the images. Stating “…exhibited substantially higher fluorescence accumulation” is insufficient. In addition, the scale bar in cell culture images is missing. 5. There is no clear evidence that neither overexpression or silencing of Cav-1 in selected cell lines is functional. The Cav-1 antibody seems to work well in Western blot (Figure 2A). Thus, the overexpression and silencing could be easily demonstrated in all cell lines used throughout this study. 6. CAV1-dependent investigation of the CCLE and TCGA in Figure 2 has been done in order to study the possible association of CAV1 and metabolite enzyme-encoding gene expression. Here, the description in the text (row 189 to 196) is not obvious in the related figures. It might be true for glycosphingolipids, but not for ceramides. A more sophisticated analysis should be done. Moreover, the stratification into high or low CAV1 is insufficiently explained. Why were only the highest and lowest quartiles chosen? What are the remaining patient numbers in both groups? Is this statistically tested? 7. The authors utilize LC-MS for tracing uptake of extracellular sphingomyelins and subsequent synthesis of ceramides and glycosphingolipids in the three prostate cancer cell lines. Drawing conclusions from the cell lines based on their endogenous Cav-1 expression doesn’t seem to be appropriate. Thus, the authors should show in at least 2 cell lines that their results are Cav-1 (in- )dependent by either silencing or overexpressing Cav-1. In general, comparing cell lines based on their Cav-1 expression only, does not adequately allow to address the function of Cav-1 as differences might be unrelated to the protein. Here, additional functional tests as already performed at the beginning should be consequently applied in almost all the experiments. 8. For entire figure 3, scale bar is missing and statistical evaluation is needed. The number of experiments is neither stated in the results nor figure legend. 9. “Selective” inhibitors of glucosylceramide synthase (UGCG) were applied to test a targetable metabolic vulnerability. The literature on the specificity of UGCG inhibitors is not straightforward, especially in regards to the two inhibitors used. However, it is quite evident that inhibition of UGCG does not reduce all glucosylceramide-related glycosphingolipids. Also, studies focused on UGCG are quite controversial. The authors found cytotoxicity with accompanied alteration of metabolites in 2 cell lines. Again, other non-glycosphingolipid related drugs might cause the same effect. Have other drugs been tested with focusing on identical readout? Are PPMP and PDPM working the same way as shown for eliglustat? Does Cav-1 overexpression (e.g. LNCaP) result in enhanced resistance to eliglustat? 10. It is not clear where “… and acute (6hr) accumulation of intracellular ceramides…” (statement row 282-285) is demonstrated. There are no data on detailed expression of metabolites as stated. 11. The colors for treatment (eligustat, PDMP, PDMP, vehicle) in figure 5B should be identical. 11. Different series of glycosphingolipids play various roles in cancer. The limited number of glycosphingolipids investigated does not allow drawing conclusions for glycosphingolipids in general taken the heterogeneity of this class of biomolecules. Thus, it would be of benefit to reflect findings reported with the literature on glycosphingolipids in the context of cancer in the discussion. Reviewer #3 (Remarks to the Author): In this study, Vykoukal et al identify a Cav-1 and sphingolipid/glycosphingolipids and sphingomyelin) that is prognostic of prostate cancer disease progression and present a potentially interesting mechanism whereby prostate cancer cells utilize Cav-1 and increased ceramide to glycosphingo conversion to mitigate mito-toxicity potentially via increased clearance of ceramides. In addition they replicate the plasma signature of patients in tumor bearing mice. I found the data to be quite interesting. The paper covers a lot of ground and proposes a fairly detailed mechanism. Various technical points need to be addressed, and the mechanism could be refined a bit more. Altogether this is a good mixture of clinical and basic studies. Comments are on the heavy side, but this has as much to do with our interest as it does with science. Major comments: 1. I am somewhat confused by their overarching mechanism: The authors demonstrate nicely that tumor cells upregulate ceramide synthesis from SM, but cells must then deal with (ceramide induced) mitochondrial toxicity by shuttling out glycosphingolipids. The mechanism (as written) doesn’t make sense to me, as it is a futile cycle with large bioenergetic/biosynthetic costs. If the ceramide toxicity point is important to the authors, they should test this with rescue studies (perhaps with inhibitors?). There are various other reasons why mitochondria are damaged, so I think they can work around it. Along these lines, are mitophagy markers generally upregulated in Cav1 KD cells? 2. Figure 2G must include the abundance of total (labeled and unlabeled) species to make any claim about fluxes. The LnCAPs could have a much larger pool size (though I think results will hold). The 10-fold abundance claim about d9-ceramide is misleading without this information. 3. This important and elegant experiment would be better tested in isogenic Cav1 +/- cells (as in Fig 2C/D). As currently shown, the Cav1 expression is correlative with their flux findings. These 3 cell lines are as different as 3 PCa patients – different genetic backgrounds, etc. 4. The impact of Cav1 modulation on cell growth should be shown. This also puts the flux and metabolomics data in better context. Is this dependent on lipids in the medium? It would provide the reader with more confidence in their mechanism and data in Figure 3. I like the mitochondria efflux concept, but this is not supported enough by the data provided. 5. Are significant amounts of mitochondrial DNA or protein present in EVs? Not +/- Cav1 (this will certainly be the case), instead the authors should compare vesicular mitochondrial pools to total to determine whether this flux is appreciable. 6. If Cav-1 protects against mito toxicity, is overall mito function altered with Cav-1 KD such as oxygen consumption etc.? 7. The clinical data is impactful, but none of the metabolites in question were statistically significant in their initial analysis (as the authors note). I am fine with the subsequent grouping/signature that focuses on ceramide metabolism, but these trends should be considered in the context of other clinical parameters. Could obesity or BMI be a driver of the dyslipidemia? Most plasma lipids are on liver-derived vLDL particles and thus reflects hepatic metabolism more than the tumor. On that front, it would be helpful to quantify the amount of sphingolipids in patient EVs vs vLDLs to know what they are observing in the clinical samples. Minor points: 1. It’s not my area (making it all the more important), but the FL image results weren’t clear to me initially. I understand the point from the text and see the image, but it could be presented better. Some quantitation of FL imaging would allow the authors to show the highlight the difference to the reader more clearly. 2. How was the lipid abundance in figure s3, 2c-d, 2g normalized, was this to cell number or protein? 3. Authors note “Cav-1-mediated clearance of ‘damaged’ mitochondria”, then would you expect less LC3BII with eliglustat? 4. Please add a scale bar to 6f images 5. Please plot individual values on 6 b-c so that the biological variation is clear.	Reviewers' comments: Reviewer #1 (Remarks to the Author): This manuscript reports the association between plasma sphingolipid profile and disease progression in low risk localized prostate cancer, and attempts to link this observation to a targetable biology in the cancer cells involving scavenging and metabolism of extracellular sphingolipids. The manuscript features some interesting lipidomic profiling and sphingolipid biochemistry, but is ultimately limited by several experimental concerns and the use of laboratory models that poorly recapitulate clinical disease, particularly for early stage localized prostate cancer which is the relevant patient group in active surveillance cohorts. The cell line models used in this manuscript are not typical of active surveillance patients or indeed most clinical prostate tumors. The only lines that express CAV1 are PC-3, RM9 and DU145, all of which are androgen receptor negative and reflect a very small proportion of very advanced or neuroendocrine prostate tumors. Virtually all early stage and low risk primary tumors express the androgen receptor and all of the AR-positive cell lines here (and seemingly most clinical tumor epithelium on Protein atlas) appear to express little to no CAV1, which casts some doubt on the mechanistic links being drawn by the authors between the plasma and cancer cell data. Major comments: 1. Lipidomics data is not quantitative in this manuscript but rather expressed as HRs relative to historical plasma values- this is not explained. Can the authors please provide further information about how specific metabolites/lipids were identified and which standards were used? Was any interference from the presence of EDTA and/or age of the plasma samples evident in the samples detected for metabolomics? Were the plasma samples from fasted individuals? 2. I have a concern over some of the clinical terminology used in the manuscript- when considering upgrading in an active surveillance cohort, this should not be described as aggressive versus indolent prostate cancer. There is a wealth of evidence that the vast majority of even those men experiencing upgrading or other measures of disease progression will not ultimately develop clinical metastases or die from prostate cancer. Even biochemical relapse following prostatectomy has been shown to be a poor surrogate for overall survival. This aspect of the manuscript and its title needs to be described more accurately, and the results interpreted with this in mind- the signatures found are really prognostic for disease progression at best, not aggressive disease. 3. Please justify the 25%:75% cutpoint for the Kaplan-Meier survival curve in Figure 1B. Importantly, when considering an active surveillance cohort, “Survival probability” in this context (Y-axis) is misleading and should more clearly relate to disease progression or upgrading. Figure 1C is not well explained in the text and it is not evident why pathways involved in sphingolipid metabolism were not included considering the focus of the rest of Figure 1. Can the authors please also include a figure presenting the absolute expression data for CAV1 in the TCGA samples rather than relative levels? Alternatively IHC staining data from their own cohorts. This is critical to support the link between the plasma data and the mechanistic data in the rest of the manuscript. My quick impression is that the abundance is rather low in most localized tumor samples. 4. For all of the CAV1 overexpression (LNCaP) and knockdown (PC-3) data in the manuscript, the degree of overexpression or knockdown is not shown anywhere in the manuscript, figures, legends or Supplemental data. This is unacceptable and needs to be included for all major experiments. 5. Figure 2E lacks clarity and could be annotated to emphasize the main points regarding the biochemical pathways being highlighted by the authors. 6. Figure 3: CAV1 staining should be included in all of these experiments to show the extent and efficiency of knockdown in the cell populations shown- particularly important when highlighting individual cells. This may also underpin the fact that there appear to be differences in the appearance of the mitochondria in the PC-3M cells in panel A versus Panel C. 7. Figure 4: How does the total number of EVs change in response to altering CAV1 levels? 8. Figure 6: There is no information in the manuscript or figure legends regarding the numbers of mice used in the experiments. The IHC data on single tumors is inadequate in the absence of any quantitative data on the remainder of the mice in the study. It is not evident whether the effects of eliglustat on tumor growth are due to systemic effects versus a targeted effect on the tumors- was any toxicity noted? A parallel experiment to confirm this using stable knockdown of CAV1 in the RM9 cells would help distinguish between these two mechanisms. Minor comments: 1. Certain figures within the manuscript lack clarity or are illegible. Please consider increasing font sizes and image quality for Fig 1B, 2E, 5B at a minimum. Reviewer #2 (Remarks to the Author): The authors of the manuscript entitled “Cav-1-Mediated Sphingolipid Oncometabolism Underlies a Metabolic Vulnerability of Aggressive Prostate Cancer” report on caveolin-1-dependent effects on the metabolome in prostate cancer by plasma from two independent cohorts of impressive size. Moreover, CCLE and TCGA transcriptomic data are incorporated throughout the study partially supporting findings. Mechanistically, three standard prostate cancer cell lines are applied throughout the study using transient silencing and overexpression of Cav-1 as well as inhibitors/ Cav- 1 specific antibody. The investigation of metabolic rewiring in prostate cancer seems to be novel. The manuscript is well written, however, there are quite a few major points (especially in regards to investigations in potential mechanisms) which require a more detailed determination for current conclusions made by the authors. 1. In regards to the volcano plot shown in Figure 1B, there seems to be a general trend towards higher HR in “early disease progression” indicating that there is no single metabolite detected suggesting a smaller risk. Is this the correct interpretation and if yes, does it mean there is a generally higher abundance of detected metabolites in the DP group? 2. The authors utilize a “sphingolipid signature” to determine the prediction power for progression- free survival separating the cohort into top 25th percentile versus bottom 75th percentile, unfortunately, without providing detailed information how this arbitrary threshold was selected. Beside further information needed and without knowing the rational for an optimal threshold, it is suggested to split the entire cohort into quartiles which might reflect prediction power closer to what has been shown in the manuscript. How does a quartile split look like? 3. The authors have accessed the TCGA prostate cancer transcriptomic data to provide further evidence of CAV-1 in prostate cancer and to study the relationship to lipid metabolism. Here, clinicopathological data accompanied with the data sets should be thoroughly investigated in regards to CAV-1/ gene sets. Has this been done? Neither the results nor the figure legends provide details on the generation of “iCluster-1 to 3”. An appropriate dendrogram on the patients and possible prostate cancer subgrouping (if there are any) might be of benefit. 4. The Western blot Figure 2B (bottom of the images) has to be improved. Here, at least a repetitive Western blot with reference protein should be shown for all three cell lines together. In addition, an appropriate quantification should be done on the images. Stating “…exhibited substantially higher fluorescence accumulation” is insufficient. In addition, the scale bar in cell culture images is missing. 5. There is no clear evidence that neither overexpression or silencing of Cav-1 in selected cell lines is functional. The Cav-1 antibody seems to work well in Western blot (Figure 2A). Thus, the overexpression and silencing could be easily demonstrated in all cell lines used throughout this study. 6. CAV1-dependent investigation of the CCLE and TCGA in Figure 2 has been done in order to study the possible association of CAV1 and metabolite enzyme-encoding gene expression. Here, the description in the text (row 189 to 196) is not obvious in the related figures. It might be true for glycosphingolipids, but not for ceramides. A more sophisticated analysis should be done. Moreover, the stratification into high or low CAV1 is insufficiently explained. Why were only the highest and lowest quartiles chosen? What are the remaining patient numbers in both groups? Is this statistically tested? 7. The authors utilize LC-MS for tracing uptake of extracellular sphingomyelins and subsequent synthesis of ceramides and glycosphingolipids in the three prostate cancer cell lines. Drawing conclusions from the cell lines based on their endogenous Cav-1 expression doesn’t seem to be appropriate. Thus, the authors should show in at least 2 cell lines that their results are Cav-1 (in- )dependent by either silencing or overexpressing Cav-1. In general, comparing cell lines based on their Cav-1 expression only, does not adequately allow to address the function of Cav-1 as differences might be unrelated to the protein. Here, additional functional tests as already performed at the beginning should be consequently applied in almost all the experiments. 8. For entire figure 3, scale bar is missing and statistical evaluation is needed. The number of experiments is neither stated in the results nor figure legend. 9. “Selective” inhibitors of glucosylceramide synthase (UGCG) were applied to test a targetable metabolic vulnerability. The literature on the specificity of UGCG inhibitors is not straightforward, especially in regards to the two inhibitors used. However, it is quite evident that inhibition of UGCG does not reduce all glucosylceramide-related glycosphingolipids. Also, studies focused on UGCG are quite controversial. The authors found cytotoxicity with accompanied alteration of metabolites in 2 cell lines. Again, other non-glycosphingolipid related drugs might cause the same effect. Have other drugs been tested with focusing on identical readout? Are PPMP and PDPM working the same way as shown for eliglustat? Does Cav-1 overexpression (e.g. LNCaP) result in enhanced resistance to eliglustat? 10. It is not clear where “… and acute (6hr) accumulation of intracellular ceramides…” (statement row 282-285) is demonstrated. There are no data on detailed expression of metabolites as stated. 11. The colors for treatment (eligustat, PDMP, PDMP, vehicle) in figure 5B should be identical. 11. Different series of glycosphingolipids play various roles in cancer. The limited number of glycosphingolipids investigated does not allow drawing conclusions for glycosphingolipids in general taken the heterogeneity of this class of biomolecules. Thus, it would be of benefit to reflect findings reported with the literature on glycosphingolipids in the context of cancer in the discussion. Reviewer #3 (Remarks to the Author): In this study, Vykoukal et al identify a Cav-1 and sphingolipid/glycosphingolipids and sphingomyelin) that is prognostic of prostate cancer disease progression and present a potentially interesting mechanism whereby prostate cancer cells utilize Cav-1 and increased ceramide to glycosphingo conversion to mitigate mito-toxicity potentially via increased clearance of ceramides. In addition they replicate the plasma signature of patients in tumor bearing mice. I found the data to be quite interesting. The paper covers a lot of ground and proposes a fairly detailed mechanism. Various technical points need to be addressed, and the mechanism could be refined a bit more. Altogether this is a good mixture of clinical and basic studies. Comments are on the heavy side, but this has as much to do with our interest as it does with science. Major comments: 1. I am somewhat confused by their overarching mechanism: The authors demonstrate nicely that tumor cells upregulate ceramide synthesis from SM, but cells must then deal with (ceramide induced) mitochondrial toxicity by shuttling out glycosphingolipids. The mechanism (as written) doesn’t make sense to me, as it is a futile cycle with large bioenergetic/biosynthetic costs. If the ceramide toxicity point is important to the authors, they should test this with rescue studies (perhaps with inhibitors?). There are various other reasons why mitochondria are damaged, so I think they can work around it. Along these lines, are mitophagy markers generally upregulated in Cav1 KD cells? 2. Figure 2G must include the abundance of total (labeled and unlabeled) species to make any claim about fluxes. The LnCAPs could have a much larger pool size (though I think results will hold). The 10-fold abundance claim about d9-ceramide is misleading without this information. 3. This important and elegant experiment would be better tested in isogenic Cav1 +/- cells (as in Fig 2C/D). As currently shown, the Cav1 expression is correlative with their flux findings. These 3 cell lines are as different as 3 PCa patients – different genetic backgrounds, etc. 4. The impact of Cav1 modulation on cell growth should be shown. This also puts the flux and metabolomics data in better context. Is this dependent on lipids in the medium? It would provide the reader with more confidence in their mechanism and data in Figure 3. I like the mitochondria efflux concept, but this is not supported enough by the data provided. 5. Are significant amounts of mitochondrial DNA or protein present in EVs? Not +/- Cav1 (this will certainly be the case), instead the authors should compare vesicular mitochondrial pools to total to determine whether this flux is appreciable. 6. If Cav-1 protects against mito toxicity, is overall mito function altered with Cav-1 KD such as oxygen consumption etc.? 7. The clinical data is impactful, but none of the metabolites in question were statistically significant in their initial analysis (as the authors note). I am fine with the subsequent grouping/signature that focuses on ceramide metabolism, but these trends should be considered in the context of other clinical parameters. Could obesity or BMI be a driver of the dyslipidemia? Most plasma lipids are on liver-derived vLDL particles and thus reflects hepatic metabolism more than the tumor. On that front, it would be helpful to quantify the amount of sphingolipids in patient EVs vs vLDLs to know what they are observing in the clinical samples. Minor points: 1. It’s not my area (making it all the more important), but the FL image results weren’t clear to me initially. I understand the point from the text and see the image, but it could be presented better. Some quantitation of FL imaging would allow the authors to show the highlight the difference to the reader more clearly. 2. How was the lipid abundance in figure s3, 2c-d, 2g normalized, was this to cell number or protein? 3. Authors note “Cav-1-mediated clearance of ‘damaged’ mitochondria”, then would you expect less LC3BII with eliglustat? 4. Please add a scale bar to 6f images 5. Please plot individual values on 6 b-c so that the biological variation is clear.	TPR	NC_BS_BIOCH_009_RR1.txt	{ "criticism": 32, "example": 16, "importance_and_relevance": 9, "materials_and_methods": 76, "praise": 9, "presentation_and_reporting": 28, "results_and_discussion": 17, "suggestion_and_solution": 37, "total": 155 }	{ "criticism": 0.2064516129032258, "example": 0.1032258064516129, "importance_and_relevance": 0.05806451612903226, "materials_and_methods": 0.49032258064516127, "praise": 0.05806451612903226, "presentation_and_reporting": 0.18064516129032257, "results_and_discussion": 0.10967741935483871, "suggestion_and_solution": 0.23870967741935484 }	1.445161	[ { "content": "Please review this paper.", "role": "user" }, { "content": "Reviewers' comments: \n\n \n\nReviewer #1 (Remarks to the Author): \n\n \n\nThis manuscript reports the association between plasma sphingolipid profile and disease progression \nin low risk localized prostate cancer, and attempts to link this observation to a targetable biology in \nthe cancer cells involving scavenging and metabolism of extracellular sphingolipids. The manuscript \nfeatures some interesting lipidomic profiling and sphingolipid biochemistry, but is ultimately limited \nby several experimental concerns and the use of laboratory models that poorly recapitulate clinical \ndisease, particularly for early stage localized prostate cancer which is the relevant patient group in \nactive surveillance cohorts. \n\n \n\nThe cell line models used in this manuscript are not typical of active surveillance patients or indeed \nmost clinical prostate tumors. The only lines that express CAV1 are PC-3, RM9 and DU145, all of \nwhich are androgen receptor negative and reflect a very small proportion of very advanced or \nneuroendocrine prostate tumors. Virtually all early stage and low risk primary tumors express the \nandrogen receptor and all of the AR-positive cell lines here (and seemingly most clinical tumor \nepithelium on Protein atlas) appear to express little to no CAV1, which casts some doubt on the \nmechanistic links being drawn by the authors between the plasma and cancer cell data. \n\n \n\nMajor comments: \n\n \n\n1. Lipidomics data is not quantitative in this manuscript but rather expressed as HRs relative to \nhistorical plasma values- this is not explained. Can the authors please provide further information \nabout how specific metabolites/lipids were identified and which standards were used? Was any \ninterference from the presence of EDTA and/or age of the plasma samples evident in the samples \ndetected for metabolomics? Were the plasma samples from fasted individuals? \n\n \n\n2. I have a concern over some of the clinical terminology used in the manuscript- when considering \nupgrading in an active surveillance cohort, this should not be described as aggressive versus indolent \nprostate cancer. There is a wealth of evidence that the vast majority of even those men experiencing \nupgrading or other measures of disease progression will not ultimately develop clinical metastases \nor die from prostate cancer. Even biochemical relapse following prostatectomy has been shown to \nbe a poor surrogate for overall survival. This aspect of the manuscript and its title needs to be \ndescribed more accurately, and the results interpreted with this in mind- the signatures found are \nreally prognostic for disease progression at best, not aggressive disease. \n\n\f \n\n3. Please justify the 25%:75% cutpoint for the Kaplan-Meier survival curve in Figure 1B. Importantly, \nwhen considering an active surveillance cohort, “Survival probability” in this context (Y-axis) is \nmisleading and should more clearly relate to disease progression or upgrading. Figure 1C is not well \nexplained in the text and it is not evident why pathways involved in sphingolipid metabolism were \nnot included considering the focus of the rest of Figure 1. Can the authors please also include a \nfigure presenting the absolute expression data for CAV1 in the TCGA samples rather than relative \nlevels? Alternatively IHC staining data from their own cohorts. This is critical to support the link \nbetween the plasma data and the mechanistic data in the rest of the manuscript. My quick \nimpression is that the abundance is rather low in most localized tumor samples. \n\n \n\n4. For all of the CAV1 overexpression (LNCaP) and knockdown (PC-3) data in the manuscript, the \ndegree of overexpression or knockdown is not shown anywhere in the manuscript, figures, legends \nor Supplemental data. This is unacceptable and needs to be included for all major experiments. \n\n \n\n5. Figure 2E lacks clarity and could be annotated to emphasize the main points regarding the \nbiochemical pathways being highlighted by the authors. \n\n \n\n6. Figure 3: CAV1 staining should be included in all of these experiments to show the extent and \nefficiency of knockdown in the cell populations shown- particularly important when highlighting \nindividual cells. This may also underpin the fact that there appear to be differences in the \nappearance of the mitochondria in the PC-3M cells in panel A versus Panel C. \n\n \n\n7. Figure 4: How does the total number of EVs change in response to altering CAV1 levels? \n\n \n\n8. Figure 6: There is no information in the manuscript or figure legends regarding the numbers of \nmice used in the experiments. The IHC data on single tumors is inadequate in the absence of any \nquantitative data on the remainder of the mice in the study. It is not evident whether the effects of \neliglustat on tumor growth are due to systemic effects versus a targeted effect on the tumors- was \nany toxicity noted? A parallel experiment to confirm this using stable knockdown of CAV1 in the RM9 \ncells would help distinguish between these two mechanisms. \n\n \n\nMinor comments: \n\n \n\n\f1. Certain figures within the manuscript lack clarity or are illegible. Please consider increasing font \nsizes and image quality for Fig 1B, 2E, 5B at a minimum. \n\n \n\n \n\n \n\nReviewer #2 (Remarks to the Author): \n\n \n\nThe authors of the manuscript entitled “Cav-1-Mediated Sphingolipid Oncometabolism Underlies a \nMetabolic Vulnerability of Aggressive Prostate Cancer” report on caveolin-1-dependent effects on \nthe metabolome in prostate cancer by plasma from two independent cohorts of impressive size. \nMoreover, CCLE and TCGA transcriptomic data are incorporated throughout the study partially \nsupporting findings. Mechanistically, three standard prostate cancer cell lines are applied \nthroughout the study using transient silencing and overexpression of Cav-1 as well as inhibitors/ Cav-\n1 specific antibody. The investigation of metabolic rewiring in prostate cancer seems to be novel. \nThe manuscript is well written, however, there are quite a few major points (especially in regards to \ninvestigations in potential mechanisms) which require a more detailed determination for current \nconclusions made by the authors. \n\n \n\n1. In regards to the volcano plot shown in Figure 1B, there seems to be a general trend towards \nhigher HR in “early disease progression” indicating that there is no single metabolite detected \nsuggesting a smaller risk. Is this the correct interpretation and if yes, does it mean there is a \ngenerally higher abundance of detected metabolites in the DP group? \n\n \n\n2. The authors utilize a “sphingolipid signature” to determine the prediction power for progression-\nfree survival separating the cohort into top 25th percentile versus bottom 75th percentile, \nunfortunately, without providing detailed information how this arbitrary threshold was selected. \nBeside further information needed and without knowing the rational for an optimal threshold, it is \nsuggested to split the entire cohort into quartiles which might reflect prediction power closer to \nwhat has been shown in the manuscript. How does a quartile split look like? \n\n \n\n3. The authors have accessed the TCGA prostate cancer transcriptomic data to provide further \nevidence of CAV-1 in prostate cancer and to study the relationship to lipid metabolism. Here, \nclinicopathological data accompanied with the data sets should be thoroughly investigated in \nregards to CAV-1/ gene sets. Has this been done? Neither the results nor the figure legends provide \ndetails on the generation of “iCluster-1 to 3”. An appropriate dendrogram on the patients and \npossible prostate cancer subgrouping (if there are any) might be of benefit. \n\n\f \n\n4. The Western blot Figure 2B (bottom of the images) has to be improved. Here, at least a repetitive \nWestern blot with reference protein should be shown for all three cell lines together. In addition, an \nappropriate quantification should be done on the images. Stating “…exhibited substantially higher \nfluorescence accumulation” is insufficient. In addition, the scale bar in cell culture images is missing. \n\n \n\n5. There is no clear evidence that neither overexpression or silencing of Cav-1 in selected cell lines is \nfunctional. The Cav-1 antibody seems to work well in Western blot (Figure 2A). Thus, the \noverexpression and silencing could be easily demonstrated in all cell lines used throughout this \nstudy. \n\n \n\n6. CAV1-dependent investigation of the CCLE and TCGA in Figure 2 has been done in order to study \nthe possible association of CAV1 and metabolite enzyme-encoding gene expression. Here, the \ndescription in the text (row 189 to 196) is not obvious in the related figures. It might be true for \nglycosphingolipids, but not for ceramides. A more sophisticated analysis should be done. Moreover, \nthe stratification into high or low CAV1 is insufficiently explained. Why were only the highest and \nlowest quartiles chosen? What are the remaining patient numbers in both groups? Is this statistically \ntested? \n\n \n\n7. The authors utilize LC-MS for tracing uptake of extracellular sphingomyelins and subsequent \nsynthesis of ceramides and glycosphingolipids in the three prostate cancer cell lines. Drawing \nconclusions from the cell lines based on their endogenous Cav-1 expression doesn’t seem to be \nappropriate. Thus, the authors should show in at least 2 cell lines that their results are Cav-1 (in-\n)dependent by either silencing or overexpressing Cav-1. In general, comparing cell lines based on \ntheir Cav-1 expression only, does not adequately allow to address the function of Cav-1 as \ndifferences might be unrelated to the protein. Here, additional functional tests as already performed \nat the beginning should be consequently applied in almost all the experiments. \n\n \n\n8. For entire figure 3, scale bar is missing and statistical evaluation is needed. The number of \nexperiments is neither stated in the results nor figure legend. \n\n \n\n9. “Selective” inhibitors of glucosylceramide synthase (UGCG) were applied to test a targetable \nmetabolic vulnerability. The literature on the specificity of UGCG inhibitors is not straightforward, \nespecially in regards to the two inhibitors used. However, it is quite evident that inhibition of UGCG \ndoes not reduce all glucosylceramide-related glycosphingolipids. Also, studies focused on UGCG are \nquite controversial. The authors found cytotoxicity with accompanied alteration of metabolites in 2 \ncell lines. Again, other non-glycosphingolipid related drugs might cause the same effect. Have other \ndrugs been tested with focusing on identical readout? Are PPMP and PDPM working the same way \n\n\fas shown for eliglustat? Does Cav-1 overexpression (e.g. LNCaP) result in enhanced resistance to \neliglustat? \n\n \n\n10. It is not clear where “… and acute (6hr) accumulation of intracellular ceramides…” (statement \nrow 282-285) is demonstrated. There are no data on detailed expression of metabolites as stated. \n\n \n\n11. The colors for treatment (eligustat, PDMP, PDMP, vehicle) in figure 5B should be identical. \n\n \n\n11. Different series of glycosphingolipids play various roles in cancer. The limited number of \nglycosphingolipids investigated does not allow drawing conclusions for glycosphingolipids in general \ntaken the heterogeneity of this class of biomolecules. Thus, it would be of benefit to reflect findings \nreported with the literature on glycosphingolipids in the context of cancer in the discussion. \n\n \n\n \n\nReviewer #3 (Remarks to the Author): \n\n \n\nIn this study, Vykoukal et al identify a Cav-1 and sphingolipid/glycosphingolipids and sphingomyelin) \nthat is prognostic of prostate cancer disease progression and present a potentially interesting \nmechanism whereby prostate cancer cells utilize Cav-1 and increased ceramide to glycosphingo \nconversion to mitigate mito-toxicity potentially via increased clearance of ceramides. In addition \nthey replicate the plasma signature of patients in tumor bearing mice. \n\n \n\nI found the data to be quite interesting. The paper covers a lot of ground and proposes a fairly \ndetailed mechanism. Various technical points need to be addressed, and the mechanism could be \nrefined a bit more. Altogether this is a good mixture of clinical and basic studies. Comments are on \nthe heavy side, but this has as much to do with our interest as it does with science. \n\n \n\nMajor comments: \n\n1. I am somewhat confused by their overarching mechanism: The authors demonstrate nicely that \ntumor cells upregulate ceramide synthesis from SM, but cells must then deal with (ceramide \ninduced) mitochondrial toxicity by shuttling out glycosphingolipids. The mechanism (as written) \ndoesn’t make sense to me, as it is a futile cycle with large bioenergetic/biosynthetic costs. If the \nceramide toxicity point is important to the authors, they should test this with rescue studies \n(perhaps with inhibitors?). There are various other reasons why mitochondria are damaged, so I \n\n\fthink they can work around it. Along these lines, are mitophagy markers generally upregulated in \nCav1 KD cells? \n\n \n\n2. Figure 2G must include the abundance of total (labeled and unlabeled) species to make any claim \nabout fluxes. The LnCAPs could have a much larger pool size (though I think results will hold). The \n10-fold abundance claim about d9-ceramide is misleading without this information. \n\n \n\n3. This important and elegant experiment would be better tested in isogenic Cav1 +/- cells (as in Fig \n2C/D). As currently shown, the Cav1 expression is correlative with their flux findings. These 3 cell \nlines are as different as 3 PCa patients – different genetic backgrounds, etc. \n\n \n\n4. The impact of Cav1 modulation on cell growth should be shown. This also puts the flux and \nmetabolomics data in better context. Is this dependent on lipids in the medium? It would provide \nthe reader with more confidence in their mechanism and data in Figure 3. I like the mitochondria \nefflux concept, but this is not supported enough by the data provided. \n\n \n\n5. Are significant amounts of mitochondrial DNA or protein present in EVs? Not +/- Cav1 (this will \ncertainly be the case), instead the authors should compare vesicular mitochondrial pools to total to \ndetermine whether this flux is appreciable. \n\n \n\n6. If Cav-1 protects against mito toxicity, is overall mito function altered with Cav-1 KD such as \noxygen consumption etc.? \n\n \n\n7. The clinical data is impactful, but none of the metabolites in question were statistically significant \nin their initial analysis (as the authors note). I am fine with the subsequent grouping/signature that \nfocuses on ceramide metabolism, but these trends should be considered in the context of other \nclinical parameters. Could obesity or BMI be a driver of the dyslipidemia? Most plasma lipids are on \nliver-derived vLDL particles and thus reflects hepatic metabolism more than the tumor. On that \nfront, it would be helpful to quantify the amount of sphingolipids in patient EVs vs vLDLs to know \nwhat they are observing in the clinical samples. \n\n \n\n \n\nMinor points: \n\n\f1. It’s not my area (making it all the more important), but the FL image results weren’t clear to me \ninitially. I understand the point from the text and see the image, but it could be presented better. \nSome quantitation of FL imaging would allow the authors to show the highlight the difference to the \nreader more clearly. \n\n \n\n2. How was the lipid abundance in figure s3, 2c-d, 2g normalized, was this to cell number or protein? \n\n \n\n3. Authors note “Cav-1-mediated clearance of ‘damaged’ mitochondria”, then would you expect less \nLC3BII with eliglustat? \n\n \n\n4. Please add a scale bar to 6f images \n\n \n\n5. Please plot individual values on 6 b-c so that the biological variation is clear. \n\n\f\n", "role": "assistant" } ]	2,020
Caveolin-1-mediated sphingolipid oncometabolism underlies a metabolic vulnerability of	REVIEWERS' COMMENTS: Reviewer#1: The authors have thoughtfully addressed the critiques. I recommend this article for publication.	REVIEWERS' COMMENTS: Reviewer#1: The authors have thoughtfully addressed the critiques. I recommend this article for publication.	TPR	NC_BS_BIOCH_009_RR3.txt	{ "criticism": 0, "example": 0, "importance_and_relevance": 0, "materials_and_methods": 0, "praise": 2, "presentation_and_reporting": 0, "results_and_discussion": 0, "suggestion_and_solution": 1, "total": 2 }	{ "criticism": 0, "example": 0, "importance_and_relevance": 0, "materials_and_methods": 0, "praise": 1, "presentation_and_reporting": 0, "results_and_discussion": 0, "suggestion_and_solution": 0.5 }	1.5	[ { "content": "Please review this paper.", "role": "user" }, { "content": "\n\fREVIEWERS' COMMENTS: \n\n \n\nReviewer#1: \n\n \n\nThe authors have thoughtfully addressed the critiques. I recommend this article for publication. \n\n\f\n", "role": "assistant" } ]	2,020
Caveolin-1-mediated sphingolipid oncometabolism underlies a metabolic vulnerability of	Reviewers' comments: Reviewer #1 (Remarks to the Author): The revisions to the manuscript have enhanced the clarity and quality of the manuscript text and key figures. The critical amendments I requested to the clinical descriptors are now accurate and there is greater detail about the key experimental and analysis details that were previously missing or incomplete. The inclusion of quantitative data for the animal studies is appreciated and markedly strengthens this aspect, but a major issue raised by myself and another reviewer remains unaddressed- the quantification of data and confirmation of CAV1 modulation in the imunofluorescent studies in Figure 4. With relatively few cells being shown in the images, it is particularly important to include quantitative data with larger numbers of cells- this is now relatively routine for immunofluorescence work thus it is surprising it has not been done to support a major conclusion of the manuscript. Regarding CAV-1 detection, despite the western blots now included for an independent experiment, given the known heterogeneity of expression on an individual cell level in even stably transfected lines, CAV-1 levels in the images for Figure 4 remain an important experiment to include to ensure the data and changes are interpreted appropriately. For that purpose live cell imaging is not necessary and standard immunofluorescent protocols could be used to visualize CAV1 and the other markers included. Other Comments: 1. The authors have answered the queries regarding the metabolomics analysis of the clinical samples used in the study. Some of these issues (patient fasting status, freeze-thawing etc) are intrinsic to large clinical cohorts of active surveillance and understandably cannot be revisited by the investigators. However for benefit of the readers and interpretation of results compared to other cohorts, it would be important to include these details in the materials and methods section of the manuscript, not only the rebuttal. 2. Figure 1C is now much clearer. The data for the normalized gene expression should be included as supplementary, and the authors could consider expressing the heatmaps data as median rather than average values, as the average will be much more strongly influenced by outlier or high abundance genes (or both). Minor comments: 1. For Figure 5B (now Figure 6B), despite some improved labeling, the figure is proportionally far too small compared to the other figure panels and is again impossible to read and interpret. 2. Line 350: Figure 5E should be Figure 6E. Reviewer #2 (Remarks to the Author): The authors of the revised manuscript entitled “Cav-1-Mediated Sphingolipid Oncometabolism Underlies a Metabolic Vulnerability of Prostate Cancer” have appropriately addressed all questions raised by the reviewer. Reviewer #3 (Remarks to the Author): The authors have provided a fairly thorough and verbose (longer is not always better) response to our critiques, arguing away most points. See some minor issues below. I would argue that “mitochondria” is a central part of their mechanism (I see the term repeated lots in Figure 7), why is a simple validation of actual mitochondrial components in their EVs beyond the scope of this paper? They propose an interesting and complex mechanism. The link to mitochondria isn’t so strong and probably not critical here, so I would suggest backing off some of those aspects. It’s not just mitochondria that need turnover. While I think questions remain about some specific details, in my view the paper and concept should be published, but their claims need to be kept in check. I am specifically referring to the abstract sentence: “Our findings demonstrate involvement of this metabolic program in support of vesicle formation that facilitates trafficking of ‘damaged’ mitochondrial components from the cell.” This has not been shown in my view, although they hope this is the mechanism. I would cut or show me some “damaged” mitochondrial components and their functional impact. Some minor corrections: There is a typo (line 350), where we think Figure 5E should be 6E For figures 2D/F, using the term “relative abundance” for the axis title would be much better, as “area units” is vague. Please explain how these were normalized in the legend for the reader.	Reviewers' comments: Reviewer #1 (Remarks to the Author): The revisions to the manuscript have enhanced the clarity and quality of the manuscript text and key figures. The critical amendments I requested to the clinical descriptors are now accurate and there is greater detail about the key experimental and analysis details that were previously missing or incomplete. The inclusion of quantitative data for the animal studies is appreciated and markedly strengthens this aspect, but a major issue raised by myself and another reviewer remains unaddressed- the quantification of data and confirmation of CAV1 modulation in the imunofluorescent studies in Figure 4. With relatively few cells being shown in the images, it is particularly important to include quantitative data with larger numbers of cells- this is now relatively routine for immunofluorescence work thus it is surprising it has not been done to support a major conclusion of the manuscript. Regarding CAV-1 detection, despite the western blots now included for an independent experiment, given the known heterogeneity of expression on an individual cell level in even stably transfected lines, CAV-1 levels in the images for Figure 4 remain an important experiment to include to ensure the data and changes are interpreted appropriately. For that purpose live cell imaging is not necessary and standard immunofluorescent protocols could be used to visualize CAV1 and the other markers included. Other Comments: 1. The authors have answered the queries regarding the metabolomics analysis of the clinical samples used in the study. Some of these issues (patient fasting status, freeze-thawing etc) are intrinsic to large clinical cohorts of active surveillance and understandably cannot be revisited by the investigators. However for benefit of the readers and interpretation of results compared to other cohorts, it would be important to include these details in the materials and methods section of the manuscript, not only the rebuttal. 2. Figure 1C is now much clearer. The data for the normalized gene expression should be included as supplementary, and the authors could consider expressing the heatmaps data as median rather than average values, as the average will be much more strongly influenced by outlier or high abundance genes (or both). Minor comments: 1. For Figure 5B (now Figure 6B), despite some improved labeling, the figure is proportionally far too small compared to the other figure panels and is again impossible to read and interpret. 2. Line 350: Figure 5E should be Figure 6E. Reviewer #2 (Remarks to the Author): The authors of the revised manuscript entitled “Cav-1-Mediated Sphingolipid Oncometabolism Underlies a Metabolic Vulnerability of Prostate Cancer” have appropriately addressed all questions raised by the reviewer. Reviewer #3 (Remarks to the Author): The authors have provided a fairly thorough and verbose (longer is not always better) response to our critiques, arguing away most points. See some minor issues below. I would argue that “mitochondria” is a central part of their mechanism (I see the term repeated lots in Figure 7), why is a simple validation of actual mitochondrial components in their EVs beyond the scope of this paper? They propose an interesting and complex mechanism. The link to mitochondria isn’t so strong and probably not critical here, so I would suggest backing off some of those aspects. It’s not just mitochondria that need turnover. While I think questions remain about some specific details, in my view the paper and concept should be published, but their claims need to be kept in check. I am specifically referring to the abstract sentence: “Our findings demonstrate involvement of this metabolic program in support of vesicle formation that facilitates trafficking of ‘damaged’ mitochondrial components from the cell.” This has not been shown in my view, although they hope this is the mechanism. I would cut or show me some “damaged” mitochondrial components and their functional impact. Some minor corrections: There is a typo (line 350), where we think Figure 5E should be 6E For figures 2D/F, using the term “relative abundance” for the axis title would be much better, as “area units” is vague. Please explain how these were normalized in the legend for the reader.	TPR	NC_BS_BIOCH_009_RR2.txt	{ "criticism": 6, "example": 6, "importance_and_relevance": 2, "materials_and_methods": 12, "praise": 5, "presentation_and_reporting": 9, "results_and_discussion": 4, "suggestion_and_solution": 9, "total": 29 }	{ "criticism": 0.20689655172413793, "example": 0.20689655172413793, "importance_and_relevance": 0.06896551724137931, "materials_and_methods": 0.41379310344827586, "praise": 0.1724137931034483, "presentation_and_reporting": 0.3103448275862069, "results_and_discussion": 0.13793103448275862, "suggestion_and_solution": 0.3103448275862069 }	1.827586	[ { "content": "Please review this paper.", "role": "user" }, { "content": "\n \n \n\n\fReviewers' comments: \n\n \n\nReviewer #1 (Remarks to the Author): \n\n \n\nThe revisions to the manuscript have enhanced the clarity and quality of the manuscript text and key \nfigures. The critical amendments I requested to the clinical descriptors are now accurate and there is \ngreater detail about the key experimental and analysis details that were previously missing or \nincomplete. The inclusion of quantitative data for the animal studies is appreciated and markedly \nstrengthens this aspect, but a major issue raised by myself and another reviewer remains \nunaddressed- the quantification of data and confirmation of CAV1 modulation in the \nimunofluorescent studies in Figure 4. With relatively few cells being shown in the images, it is \nparticularly important to include quantitative data with larger numbers of cells- this is now relatively \nroutine for immunofluorescence work thus it is surprising it has not been done to support a major \nconclusion of the manuscript. Regarding CAV-1 detection, despite the western blots now included \nfor an independent experiment, given the known heterogeneity of expression on an individual cell \nlevel in even stably transfected lines, CAV-1 levels in the images for Figure 4 remain an important \nexperiment to include to ensure the data and changes are interpreted appropriately. For that \npurpose live cell imaging is not necessary and standard immunofluorescent protocols could be used \nto visualize CAV1 and the other markers included. \n\n \n\nOther Comments: \n\n \n\n1. The authors have answered the queries regarding the metabolomics analysis of the clinical \nsamples used in the study. Some of these issues (patient fasting status, freeze-thawing etc) are \nintrinsic to large clinical cohorts of active surveillance and understandably cannot be revisited by the \ninvestigators. However for benefit of the readers and interpretation of results compared to other \ncohorts, it would be important to include these details in the materials and methods section of the \nmanuscript, not only the rebuttal. \n\n \n\n2. Figure 1C is now much clearer. The data for the normalized gene expression should be included as \nsupplementary, and the authors could consider expressing the heatmaps data as median rather than \naverage values, as the average will be much more strongly influenced by outlier or high abundance \ngenes (or both). \n\n \n\nMinor comments: \n\n \n\n\f1. For Figure 5B (now Figure 6B), despite some improved labeling, the figure is proportionally far too \nsmall compared to the other figure panels and is again impossible to read and interpret. \n\n \n\n2. Line 350: Figure 5E should be Figure 6E. \n\n \n\n \n\n \n\nReviewer #2 (Remarks to the Author): \n\n \n\nThe authors of the revised manuscript entitled “Cav-1-Mediated Sphingolipid Oncometabolism \nUnderlies a Metabolic Vulnerability of Prostate Cancer” have appropriately addressed all questions \nraised by the reviewer. \n\n \n\n \n\nReviewer #3 (Remarks to the Author): \n\n \n\nThe authors have provided a fairly thorough and verbose (longer is not always better) response to \nour critiques, arguing away most points. See some minor issues below. I would argue that \n“mitochondria” is a central part of their mechanism (I see the term repeated lots in Figure 7), why is \na simple validation of actual mitochondrial components in their EVs beyond the scope of this paper? \n\n \n\nThey propose an interesting and complex mechanism. The link to mitochondria isn’t so strong and \nprobably not critical here, so I would suggest backing off some of those aspects. It’s not just \nmitochondria that need turnover. While I think questions remain about some specific details, in my \nview the paper and concept should be published, but their claims need to be kept in check. I am \nspecifically referring to the abstract sentence: \n\n“Our findings demonstrate involvement of this metabolic program in support of vesicle formation \nthat facilitates trafficking of ‘damaged’ mitochondrial components from the cell.” \n\n \n\nThis has not been shown in my view, although they hope this is the mechanism. I would cut or show \nme some “damaged” mitochondrial components and their functional impact. \n\nSome minor corrections: \n\n\fThere is a typo (line 350), where we think Figure 5E should be 6E \n\nFor figures 2D/F, using the term “relative abundance” for the axis title would be much better, as \n“area units” is vague. Please explain how these were normalized in the legend for the reader. \n\n\f", "role": "assistant" } ]	2,020
Structural mechanism for replication origin binding and remodeling by a metazoan origin	REVIEWERS' COMMENTS: Reviewer #1 (Remarks to the Author): JM Schmidt and F Bleichert report a structural mechanism for replication origin binding and remodeling by the Drosophyla melanogaster’s Origin Recognition Complex (ORC) and its co-loader Cdc6. While ORC recognizes consensus DNA sequences in S. cerevisiae, metazoan ORC (such as D. melanogaster) associates with chromosomal DNA independently of specific sequence elements. This work aims to decipher the structural basis for DNA engagement by ORC in the absence of specific consensus DNA sequences. The authors report near atomic resolution cryo-EM structures of D. melanogaster ORC bound to different DNA substrates in the absence and presence of its co-loader Cdc6. These structures show how the ATPase modules of subunits Orc1 and Orc4 constitute the primary DNA binding site in the center of the ORC ring and cooperate with the winged-helix domains to stabilize DNA bending. Correlating structural observations with biochemical data further shows that DNA sequence modulates DNA binding and remodeling, and that DNA bending in turn promotes Mcm2-7 loading. To explain the distinct DNA sequence-dependencies of S. cerevisiae and D. melanogaster (metazoan), the DNA contacts between S. cerevisiae and metazoan ORC are extensively compared. These findings support a model in which DNA geometry and DNA bendability contribute to the selection of chromosomal Mcm2- 7 helicase loading sites in metazoans. This work answers fundamental questions about the initiation of DNA replication in eukaryotes. The manuscript is well-written, the figures are nice and informative. The conclusions are sound and based on an impressive amount of functional and structural data. Please find below a few points that the authors should consider when preparing the revised version of their manuscript. Major points: The DNA contacts are extensively described. My main concern is that non-specialist readers may find it difficult to capture some of the article's important conclusions. - Figure 7 summaries the main conclusions of the article. The figure is very nice and informative. However, the figure may be extended by detailing and numbering the different conformational transitions in metazoan ORC and DNA remodeling. - A movie may be useful to compare the DNA contacts between S. cerevisiae and metazoan ORC, and explain the distinct DNA sequence-dependencies of S. cerevisiae and D. melanogaster. Minor points: - P.2: … the eukaryotic kingdom… should be … the eukaryotic domain of life … (The taxonomic rank of Eukarya is a domain not a kingdom). - P.9: “backbone phosphates” may be “phosphate backbones”. - P.11: You may consider splitting the following sentence: “Two notable exceptions include the WH domain of Orc2, which is detached from the AAA+ layer and flexible in the absence of Cdc6 (Extended Data Figures 6E and 6G), and the arginine finger helix in Orc1, a region that interacts with the Cdc6 AAA+ module in the ternary complex and is less ordered in the DNA-free DmORC and the DmORC·DNA assembly, suggesting it undergoes a disorder-to-order transition upon Cdc6 binding.” - “van der Waals contacts” or “van der Waals interactions” is used in many occasions in the manuscript to describe the interactions within the different protein complexes. Hydrophobic interactions may be more appropriate ? - The interactions within the different protein complexes are extensively described. However several sentences may be confusing for the readers. The following sentences may be revised: P.7: In agreement with this evolutionary kinship, Orc1 and Cdc6 are predicated on the same protein folds and adopt similar 3D structures in the ternary initiator·co-loader complex. P.9: The ISMs of Orc1, Orc4, and Cdc6 in particular form several hydrogen bonds and van der Waals contacts with the sugar-phosphate backbone of one of the DNA strands that together extend for half a turn of the duplex - P31-32 (mat & meth): imidazole p31 … or … Imidazole p32 : please homogenize. Reviewer #2 (Remarks to the Author): Schmidt and Bleichert report the reconstitution of the MCM helicase loading reaction in vitro, using purified Drosophila proteins, and the cryo-EM structure of the Drosophila ORC-DNA and ORC-DNA- Cdc6 complex. They find that, in both complexes, DNA is bent as it traverses the central pore of the ORC ring. The ORC-Cdc6 ring forms a flat ring configuration, structurally distinct from previous helicase loading intermediates observed in yeast. The ORC-Cdc6-DNA complex also contains a conformation distinct from the auto-inhibited state previously observed for the isolated ORC complex. Using structural and biochemical tools, the authors describe a new B-loop element in Orc1 that is responsible for the high-affinity interaction of Drosophila ORC with DNA. The authors also describe the cryo-EM structures of DmORC-Cdc6 bound to DNA of different sequence compositions. Interestingly, they establish, that duplex DNA segments differing in sequence can be bent to different degrees, and that DNA bendability affects MCM loading efficiency. This work is simply outstanding. The in vitro reconstitution of the MCM loading with metazoan proteins alone would grant in my opinion publication in a high impact journal. In addition to this however, the authors report an extremely high quality structural and biochemical study, describing important physical aspects of helicase loading by ORC. I only have a few minor comments: 1. The manuscript in general is very clearly written, although I must say in certain passages the language appears unnecessarily complicated. For example, in the abstract, “a nucleotidyl hydrolase that uses adenosine triphosphate (ATP) binding and turnover to control its association with chromosomal regions” could be summarized with “a DNA binding ATPase”. I would suggest that the authors critically read their manuscript to eliminate such passages, to improve an otherwise exquisitely written manuscript. 2. I would also recommend that the authors put a bigger emphasis on their achievement of the reconstitution of MCM loading using purified Drosophila proteins. Again, this is probably the most significant aspect of the work presented, and at the moment it feels like something appended to the study, rather than its central piece. 3. Negative stain EM micrographs should be added as supplementary material for the MCM loading. 4. The authors describe an ORC-Cdc6 complex that exists in a flat configuration that is different from the spiral form described for the yeast OCCM, and they speculate that MCM engagement could reconfigure the ORC-Cdc6. Could the authors clarify to what extent the ATPase sites in the Orc-Cdc6 complex would be altered?	REVIEWERS' COMMENTS: Reviewer #1 (Remarks to the Author): JM Schmidt and F Bleichert report a structural mechanism for replication origin binding and remodeling by the Drosophyla melanogaster’s Origin Recognition Complex (ORC) and its co-loader Cdc6. While ORC recognizes consensus DNA sequences in S. cerevisiae, metazoan ORC (such as D. melanogaster) associates with chromosomal DNA independently of specific sequence elements. This work aims to decipher the structural basis for DNA engagement by ORC in the absence of specific consensus DNA sequences. The authors report near atomic resolution cryo-EM structures of D. melanogaster ORC bound to different DNA substrates in the absence and presence of its co-loader Cdc6. These structures show how the ATPase modules of subunits Orc1 and Orc4 constitute the primary DNA binding site in the center of the ORC ring and cooperate with the winged-helix domains to stabilize DNA bending. Correlating structural observations with biochemical data further shows that DNA sequence modulates DNA binding and remodeling, and that DNA bending in turn promotes Mcm2-7 loading. To explain the distinct DNA sequence-dependencies of S. cerevisiae and D. melanogaster (metazoan), the DNA contacts between S. cerevisiae and metazoan ORC are extensively compared. These findings support a model in which DNA geometry and DNA bendability contribute to the selection of chromosomal Mcm2- 7 helicase loading sites in metazoans. This work answers fundamental questions about the initiation of DNA replication in eukaryotes. The manuscript is well-written, the figures are nice and informative. The conclusions are sound and based on an impressive amount of functional and structural data. Please find below a few points that the authors should consider when preparing the revised version of their manuscript. Major points: The DNA contacts are extensively described. My main concern is that non-specialist readers may find it difficult to capture some of the article's important conclusions. - Figure 7 summaries the main conclusions of the article. The figure is very nice and informative. However, the figure may be extended by detailing and numbering the different conformational transitions in metazoan ORC and DNA remodeling. - A movie may be useful to compare the DNA contacts between S. cerevisiae and metazoan ORC, and explain the distinct DNA sequence-dependencies of S. cerevisiae and D. melanogaster. Minor points: - P.2: … the eukaryotic kingdom… should be … the eukaryotic domain of life … (The taxonomic rank of Eukarya is a domain not a kingdom). - P.9: “backbone phosphates” may be “phosphate backbones”. - P.11: You may consider splitting the following sentence: “Two notable exceptions include the WH domain of Orc2, which is detached from the AAA+ layer and flexible in the absence of Cdc6 (Extended Data Figures 6E and 6G), and the arginine finger helix in Orc1, a region that interacts with the Cdc6 AAA+ module in the ternary complex and is less ordered in the DNA-free DmORC and the DmORC·DNA assembly, suggesting it undergoes a disorder-to-order transition upon Cdc6 binding.” - “van der Waals contacts” or “van der Waals interactions” is used in many occasions in the manuscript to describe the interactions within the different protein complexes. Hydrophobic interactions may be more appropriate ? - The interactions within the different protein complexes are extensively described. However several sentences may be confusing for the readers. The following sentences may be revised: P.7: In agreement with this evolutionary kinship, Orc1 and Cdc6 are predicated on the same protein folds and adopt similar 3D structures in the ternary initiator·co-loader complex. P.9: The ISMs of Orc1, Orc4, and Cdc6 in particular form several hydrogen bonds and van der Waals contacts with the sugar-phosphate backbone of one of the DNA strands that together extend for half a turn of the duplex - P31-32 (mat & meth): imidazole p31 … or … Imidazole p32 : please homogenize. Reviewer #2 (Remarks to the Author): Schmidt and Bleichert report the reconstitution of the MCM helicase loading reaction in vitro, using purified Drosophila proteins, and the cryo-EM structure of the Drosophila ORC-DNA and ORC-DNA- Cdc6 complex. They find that, in both complexes, DNA is bent as it traverses the central pore of the ORC ring. The ORC-Cdc6 ring forms a flat ring configuration, structurally distinct from previous helicase loading intermediates observed in yeast. The ORC-Cdc6-DNA complex also contains a conformation distinct from the auto-inhibited state previously observed for the isolated ORC complex. Using structural and biochemical tools, the authors describe a new B-loop element in Orc1 that is responsible for the high-affinity interaction of Drosophila ORC with DNA. The authors also describe the cryo-EM structures of DmORC-Cdc6 bound to DNA of different sequence compositions. Interestingly, they establish, that duplex DNA segments differing in sequence can be bent to different degrees, and that DNA bendability affects MCM loading efficiency. This work is simply outstanding. The in vitro reconstitution of the MCM loading with metazoan proteins alone would grant in my opinion publication in a high impact journal. In addition to this however, the authors report an extremely high quality structural and biochemical study, describing important physical aspects of helicase loading by ORC. I only have a few minor comments: 1. The manuscript in general is very clearly written, although I must say in certain passages the language appears unnecessarily complicated. For example, in the abstract, “a nucleotidyl hydrolase that uses adenosine triphosphate (ATP) binding and turnover to control its association with chromosomal regions” could be summarized with “a DNA binding ATPase”. I would suggest that the authors critically read their manuscript to eliminate such passages, to improve an otherwise exquisitely written manuscript. 2. I would also recommend that the authors put a bigger emphasis on their achievement of the reconstitution of MCM loading using purified Drosophila proteins. Again, this is probably the most significant aspect of the work presented, and at the moment it feels like something appended to the study, rather than its central piece. 3. Negative stain EM micrographs should be added as supplementary material for the MCM loading. 4. The authors describe an ORC-Cdc6 complex that exists in a flat configuration that is different from the spiral form described for the yeast OCCM, and they speculate that MCM engagement could reconfigure the ORC-Cdc6. Could the authors clarify to what extent the ATPase sites in the Orc-Cdc6 complex would be altered?	TPR	NC_BS_BIOCH_010_RR1.txt	{ "criticism": 3, "example": 6, "importance_and_relevance": 6, "materials_and_methods": 18, "praise": 6, "presentation_and_reporting": 13, "results_and_discussion": 4, "suggestion_and_solution": 14, "total": 48 }	{ "criticism": 0.0625, "example": 0.125, "importance_and_relevance": 0.125, "materials_and_methods": 0.375, "praise": 0.125, "presentation_and_reporting": 0.2708333333333333, "results_and_discussion": 0.08333333333333333, "suggestion_and_solution": 0.2916666666666667 }	1.458333	[ { "content": "Please review this paper.", "role": "user" }, { "content": "REVIEWERS' COMMENTS: \n\nReviewer #1 (Remarks to the Author): \n\nJM Schmidt and F Bleichert report a structural mechanism for replication origin binding and remodeling \nby the Drosophyla melanogaster’s Origin Recognition Complex (ORC) and its co-loader Cdc6. While \nORC recognizes consensus DNA sequences in S. cerevisiae, metazoan ORC (such as D. melanogaster) \nassociates with chromosomal DNA independently of specific sequence elements. This work aims to \ndecipher the structural basis for DNA engagement by ORC in the absence of specific consensus DNA \nsequences. \nThe authors report near atomic resolution cryo-EM structures of D. melanogaster ORC bound to \ndifferent DNA substrates in the absence and presence of its co-loader Cdc6. These structures show \nhow the ATPase modules of subunits Orc1 and Orc4 constitute the primary DNA binding site in the \ncenter of the ORC ring and cooperate with the winged-helix domains to stabilize DNA bending. \nCorrelating structural observations with biochemical data further shows that DNA sequence modulates \nDNA binding and remodeling, and that DNA bending in turn promotes Mcm2-7 loading. To explain the \ndistinct DNA sequence-dependencies of S. cerevisiae and D. melanogaster (metazoan), the DNA \ncontacts between S. cerevisiae and metazoan ORC are extensively compared. These findings support a \nmodel in which DNA geometry and DNA bendability contribute to the selection of chromosomal Mcm2-\n7 helicase loading sites in metazoans. \nThis work answers fundamental questions about the initiation of DNA replication in eukaryotes. The \nmanuscript is well-written, the figures are nice and informative. The conclusions are sound and based \non an impressive amount of functional and structural data. Please find below a few points that the \nauthors should consider when preparing the revised version of their manuscript. \n\nMajor points: \n\nThe DNA contacts are extensively described. My main concern is that non-specialist readers may find \nit difficult to capture some of the article's important conclusions. \n- Figure 7 summaries the main conclusions of the article. The figure is very nice and informative. \nHowever, the figure may be extended by detailing and numbering the different conformational \ntransitions in metazoan ORC and DNA remodeling. \n- A movie may be useful to compare the DNA contacts between S. cerevisiae and metazoan ORC, and \nexplain the distinct DNA sequence-dependencies of S. cerevisiae and D. melanogaster. \n\nMinor points: \n- P.2: … the eukaryotic kingdom… should be … the eukaryotic domain of life … (The taxonomic rank of \nEukarya is a domain not a kingdom). \n- P.9: “backbone phosphates” may be “phosphate backbones”. \n- P.11: You may consider splitting the following sentence: “Two notable exceptions include the WH \ndomain of Orc2, which is detached from the AAA+ layer and flexible in the absence of Cdc6 (Extended \nData Figures 6E and 6G), and the arginine finger helix in Orc1, a region that interacts with the Cdc6 \nAAA+ module in the ternary complex and is less ordered in the DNA-free DmORC and the \nDmORC·DNA assembly, suggesting it undergoes a disorder-to-order transition upon Cdc6 binding.” \n- “van der Waals contacts” or “van der Waals interactions” is used in many occasions in the \nmanuscript to describe the interactions within the different protein complexes. Hydrophobic \ninteractions may be more appropriate ? \n- The interactions within the different protein complexes are extensively described. However several \nsentences may be confusing for the readers. The following sentences may be revised: \nP.7: In agreement with this evolutionary kinship, Orc1 and Cdc6 are predicated on the same protein \nfolds and adopt similar 3D structures in the ternary initiator·co-loader complex. \nP.9: The ISMs of Orc1, Orc4, and Cdc6 in particular form several hydrogen bonds and van der Waals \ncontacts with the sugar-phosphate backbone of one of the DNA strands that together extend for half a \n\n\fturn of the duplex \n- P31-32 (mat & meth): imidazole p31 … or … Imidazole p32 : please homogenize. \n\nReviewer #2 (Remarks to the Author): \n\nSchmidt and Bleichert report the reconstitution of the MCM helicase loading reaction in vitro, using \npurified Drosophila proteins, and the cryo-EM structure of the Drosophila ORC-DNA and ORC-DNA-\nCdc6 complex. They find that, in both complexes, DNA is bent as it traverses the central pore of the \nORC ring. The ORC-Cdc6 ring forms a flat ring configuration, structurally distinct from previous \nhelicase loading intermediates observed in yeast. The ORC-Cdc6-DNA complex also contains a \nconformation distinct from the auto-inhibited state previously observed for the isolated ORC complex. \nUsing structural and biochemical tools, the authors describe a new B-loop element in Orc1 that is \nresponsible for the high-affinity interaction of Drosophila ORC with DNA. The authors also describe the \ncryo-EM structures of DmORC-Cdc6 bound to DNA of different sequence compositions. Interestingly, \nthey establish, that duplex DNA segments differing in sequence can be bent to different degrees, and \nthat DNA bendability affects MCM loading efficiency. \n\nThis work is simply outstanding. The in vitro reconstitution of the MCM loading with metazoan proteins \nalone would grant in my opinion publication in a high impact journal. In addition to this however, the \nauthors report an extremely high quality structural and biochemical study, describing important \nphysical aspects of helicase loading by ORC. \n\nI only have a few minor comments: \n\n1. The manuscript in general is very clearly written, although I must say in certain passages the \nlanguage appears unnecessarily complicated. For example, in the abstract, “a nucleotidyl hydrolase \nthat uses adenosine triphosphate (ATP) binding and turnover to control its association with \nchromosomal regions” could be summarized with “a DNA binding ATPase”. I would suggest that the \nauthors critically read their manuscript to eliminate such passages, to improve an otherwise \nexquisitely written manuscript. \n\n2. I would also recommend that the authors put a bigger emphasis on their achievement of the \nreconstitution of MCM loading using purified Drosophila proteins. Again, this is probably the most \nsignificant aspect of the work presented, and at the moment it feels like something appended to the \nstudy, rather than its central piece. \n\n3. Negative stain EM micrographs should be added as supplementary material for the MCM loading. \n\n4. The authors describe an ORC-Cdc6 complex that exists in a flat configuration that is different from \nthe spiral form described for the yeast OCCM, and they speculate that MCM engagement could \nreconfigure the ORC-Cdc6. Could the authors clarify to what extent the ATPase sites in the Orc-Cdc6 \ncomplex would be altered? \n\n\f\n", "role": "assistant" } ]	2,020
Enhancing intracellular accumulation and target engagement of PROTACs with reversible	REVIEWERS' COMMENTS: Reviewer #1 (Remarks to the Author): The authors have adequately dealt with the previous comments that I raised. Reviewer #2 (Remarks to the Author): The authors have done a nice job to carefully address questions raised by all reviewers by performing additional experiments and making new compounds. I found the manuscript now acceptable for publication in Nature Communications.	REVIEWERS' COMMENTS: Reviewer #1 (Remarks to the Author): The authors have adequately dealt with the previous comments that I raised. Reviewer #2 (Remarks to the Author): The authors have done a nice job to carefully address questions raised by all reviewers by performing additional experiments and making new compounds. I found the manuscript now acceptable for publication in Nature Communications.	TPR	NC_BS_BIOCH_011_RR2.txt	{ "criticism": 0, "example": 0, "importance_and_relevance": 0, "materials_and_methods": 1, "praise": 3, "presentation_and_reporting": 0, "results_and_discussion": 0, "suggestion_and_solution": 0, "total": 3 }	{ "criticism": 0, "example": 0, "importance_and_relevance": 0, "materials_and_methods": 0.3333333333333333, "praise": 1, "presentation_and_reporting": 0, "results_and_discussion": 0, "suggestion_and_solution": 0 }	1.333333	[ { "content": "Please review this paper.", "role": "user" }, { "content": "\n \n\n \n\n \n \n \n \n \n\n\fREVIEWERS' COMMENTS: \n\n \n\nReviewer #1 (Remarks to the Author): \n\n \n\nThe authors have adequately dealt with the previous comments that I raised. \n\n \n\n \n\nReviewer #2 (Remarks to the Author): \n\n \n\nThe authors have done a nice job to carefully address questions raised by all reviewers by \nperforming additional experiments and making new compounds. I found the manuscript now \nacceptable for publication in Nature Communications. \n\n \n\n\f\n", "role": "assistant" } ]	2,019
Enhancing intracellular accumulation and target engagement of PROTACs with reversible	Reviewers' comments: Reviewer #1 (Remarks to the Author): This is an interesting study of the effect of reversible covalent Protac ligands on protein degradation. The authors provide compelling data for the effectiveness of their RC-1 ligand in producing BTK degradation. I cannot comment on the underlying chemistry but the biological experiments have been carefully performed and generally well analysed (but see below). I do, however, have a few issues with some of the target engagement calculations and the determination of IC50 values using the NanoBRET ligand binding approach. According to Figure 4 the IC50 values cited in the Tables have been calculated from the concentration that reduces the NanoBRET signal by 50% - which is not correct. The IC50 value should be calculated from the specific component of binding after subtraction of non-specific binding. In Fig 4b it is clear that the inhibition curve obtained with RC-1 levels off significantly above zero binding. This non-displaceable binding is likely to represent non-specific binding. The IC50 will therefore be at lower concentrations than currently calculated. The same is true for Fig 4c and Fig 4d. Some information should be provided on the occupancy of the fixed concentration of fluorescent ligand used in the assay. Target Engagement IC50 is not “the compound concentration required to achieve 50% BTK target engagement measured using a NanoBRET BTK in-cell target engagement assay” as described in the Table legends. It is the concentration of compound required to inhibit binding of the fixed concentration of fluorescent probe. The IC50 will very much depend on the concentration of probe used in the experiments. A much clear explanation is required on this. I also do not fully understand what is meant by ‘normalised target engagement’ in Table 1 and 2. Occupancy (target engagement) is normally represented by [ligand]/([Ligand] + Kd). What do these normalised target engagement values represent in the Tables? I struggled to understand the fluorophore phase separation-based assay for imaging ternary complex formation. It looked to me to be a general measure of protein aggregation with the complexes being quite large (as judged by their size with respect to the scale bar). I failed to understand how this was a measure of ternary complex formation. Surely the spatial resolution is too low? Reviewer #2 (Remarks to the Author): In this manuscript, Guo et al. developed RC-1 as the first reversible covalent BTK PROTAC, which has high target occupancy and is effective both as an inhibitor and a degrader. They provided biochemical and cellular data to support their conclusion that RC-1 induces mechanism-based BTK degradation against both wild-type BTK and C481S mutant form. The authors compared RC-1 with reversible noncovalent degrader (RNC-1) and irreversible covalent degrader (IRC-1) and found RC-1 shows better activity in cells and suggested that reversible covalent degrader could be a good strategy for the development of PORTACS. In addition, they demonstrated that the enhanced cellular permeability is responsible for better cellular potencies for RC-1 than RNC-1 and ICR-1. Their study suggests that “introducing a cyano-acrylamide moiety can be a general strategy to enhance cellular permeability for PROTACs.” To test this idea further, they have designed FLT3 degraders and obtained similar results. Finally, they tested RC-1 in vivo, shows BTK degradation in spleen, which suggests reversible covalent BTK degraders may be useful for in vivo studies. Overall, this study was well done and is significant for the field of PROTACs. It is recommended for publication in Nature Communications with the following revisions. 1. Since the most attractive point of the manuscript is “introducing a cyano-acrylamide moiety can be a general strategy to enhance cellular permeability for PROTACs”, it is more convincing to provide direct evidence about the improved cellular permeability for RC-1 over other compounds using the Caco-2 cell-permeability assay or other similar assas. 2. It will be useful to make another reversible noncovalent compound as reference, which is directly reduce the double bond of RC-1. Compare its cell permeability with RC-1 may give us more information, such as whether the enhanced permeability is caused by adding lipophilic groups (cyano and di-methyl) or because of the reversible covalent ability. 3. Have the authors compared the BTK degradation between Crews’s compound (such as MT802) and RC-1 in Moml-14 cell line or other cell lines? It will be useful to compare these two compounds, since MT802 is an optimal reversible noncovalent degrader. 4. In the in vivo PD study, the drug was dosed for 7 days. The PD effect is not as good what has been observed in vitro. Since degradation of BTK is already profound within 24 hours in cells, what was the reason for the 7-day dosing and for the relative less profound degradation of BTK? Reviewer #3 (Remarks to the Author): This is a well-done and timely study on the effect of semi-reversible BTK PROTACs and showing that these molecules are able to degrade BTK in cells. They perform followup proteomic studies and also compare their PROTACs to other BTK PROTACs that have been previously synthesized. This study complements recent cysteine-targeting semi-reversible BTK PROTAC papers from Nir London's lab and Pfizer that have been recently posted on ChemRxiv and BioRxiv. This study should be published as is.	Reviewers' comments: Reviewer #1 (Remarks to the Author): This is an interesting study of the effect of reversible covalent Protac ligands on protein degradation. The authors provide compelling data for the effectiveness of their RC-1 ligand in producing BTK degradation. I cannot comment on the underlying chemistry but the biological experiments have been carefully performed and generally well analysed (but see below). I do, however, have a few issues with some of the target engagement calculations and the determination of IC50 values using the NanoBRET ligand binding approach. According to Figure 4 the IC50 values cited in the Tables have been calculated from the concentration that reduces the NanoBRET signal by 50% - which is not correct. The IC50 value should be calculated from the specific component of binding after subtraction of non-specific binding. In Fig 4b it is clear that the inhibition curve obtained with RC-1 levels off significantly above zero binding. This non-displaceable binding is likely to represent non-specific binding. The IC50 will therefore be at lower concentrations than currently calculated. The same is true for Fig 4c and Fig 4d. Some information should be provided on the occupancy of the fixed concentration of fluorescent ligand used in the assay. Target Engagement IC50 is not “the compound concentration required to achieve 50% BTK target engagement measured using a NanoBRET BTK in-cell target engagement assay” as described in the Table legends. It is the concentration of compound required to inhibit binding of the fixed concentration of fluorescent probe. The IC50 will very much depend on the concentration of probe used in the experiments. A much clear explanation is required on this. I also do not fully understand what is meant by ‘normalised target engagement’ in Table 1 and 2. Occupancy (target engagement) is normally represented by [ligand]/([Ligand] + Kd). What do these normalised target engagement values represent in the Tables? I struggled to understand the fluorophore phase separation-based assay for imaging ternary complex formation. It looked to me to be a general measure of protein aggregation with the complexes being quite large (as judged by their size with respect to the scale bar). I failed to understand how this was a measure of ternary complex formation. Surely the spatial resolution is too low? Reviewer #2 (Remarks to the Author): In this manuscript, Guo et al. developed RC-1 as the first reversible covalent BTK PROTAC, which has high target occupancy and is effective both as an inhibitor and a degrader. They provided biochemical and cellular data to support their conclusion that RC-1 induces mechanism-based BTK degradation against both wild-type BTK and C481S mutant form. The authors compared RC-1 with reversible noncovalent degrader (RNC-1) and irreversible covalent degrader (IRC-1) and found RC-1 shows better activity in cells and suggested that reversible covalent degrader could be a good strategy for the development of PORTACS. In addition, they demonstrated that the enhanced cellular permeability is responsible for better cellular potencies for RC-1 than RNC-1 and ICR-1. Their study suggests that “introducing a cyano-acrylamide moiety can be a general strategy to enhance cellular permeability for PROTACs.” To test this idea further, they have designed FLT3 degraders and obtained similar results. Finally, they tested RC-1 in vivo, shows BTK degradation in spleen, which suggests reversible covalent BTK degraders may be useful for in vivo studies. Overall, this study was well done and is significant for the field of PROTACs. It is recommended for publication in Nature Communications with the following revisions. 1. Since the most attractive point of the manuscript is “introducing a cyano-acrylamide moiety can be a general strategy to enhance cellular permeability for PROTACs”, it is more convincing to provide direct evidence about the improved cellular permeability for RC-1 over other compounds using the Caco-2 cell-permeability assay or other similar assas. 2. It will be useful to make another reversible noncovalent compound as reference, which is directly reduce the double bond of RC-1. Compare its cell permeability with RC-1 may give us more information, such as whether the enhanced permeability is caused by adding lipophilic groups (cyano and di-methyl) or because of the reversible covalent ability. 3. Have the authors compared the BTK degradation between Crews’s compound (such as MT802) and RC-1 in Moml-14 cell line or other cell lines? It will be useful to compare these two compounds, since MT802 is an optimal reversible noncovalent degrader. 4. In the in vivo PD study, the drug was dosed for 7 days. The PD effect is not as good what has been observed in vitro. Since degradation of BTK is already profound within 24 hours in cells, what was the reason for the 7-day dosing and for the relative less profound degradation of BTK? Reviewer #3 (Remarks to the Author): This is a well-done and timely study on the effect of semi-reversible BTK PROTACs and showing that these molecules are able to degrade BTK in cells. They perform followup proteomic studies and also compare their PROTACs to other BTK PROTACs that have been previously synthesized. This study complements recent cysteine-targeting semi-reversible BTK PROTAC papers from Nir London's lab and Pfizer that have been recently posted on ChemRxiv and BioRxiv. This study should be published as is.	TPR	NC_BS_BIOCH_011_RR1.txt	{ "criticism": 6, "example": 3, "importance_and_relevance": 4, "materials_and_methods": 32, "praise": 7, "presentation_and_reporting": 5, "results_and_discussion": 4, "suggestion_and_solution": 8, "total": 47 }	{ "criticism": 0.1276595744680851, "example": 0.06382978723404255, "importance_and_relevance": 0.0851063829787234, "materials_and_methods": 0.6808510638297872, "praise": 0.14893617021276595, "presentation_and_reporting": 0.10638297872340426, "results_and_discussion": 0.0851063829787234, "suggestion_and_solution": 0.1702127659574468 }	1.468085	[ { "content": "Please review this paper.", "role": "user" }, { "content": "Reviewers' comments: \n\n \n\nReviewer #1 (Remarks to the Author): \n\n \n\nThis is an interesting study of the effect of reversible covalent Protac ligands on protein degradation. \nThe authors provide compelling data for the effectiveness of their RC-1 ligand in producing BTK \ndegradation. I cannot comment on the underlying chemistry but the biological experiments have \nbeen carefully performed and generally well analysed (but see below). \n\n \n\nI do, however, have a few issues with some of the target engagement calculations and the \ndetermination of IC50 values using the NanoBRET ligand binding approach. According to Figure 4 the \nIC50 values cited in the Tables have been calculated from the concentration that reduces the \nNanoBRET signal by 50% - which is not correct. The IC50 value should be calculated from the specific \ncomponent of binding after subtraction of non-specific binding. In Fig 4b it is clear that the inhibition \ncurve obtained with RC-1 levels off significantly above zero binding. This non-displaceable binding is \nlikely to represent non-specific binding. The IC50 will therefore be at lower concentrations than \ncurrently calculated. The same is true for Fig 4c and Fig 4d. Some information should be provided on \nthe occupancy of the fixed concentration of fluorescent ligand used in the assay. \n\n \n\nTarget Engagement IC50 is not “the compound concentration required to achieve 50% BTK target \nengagement measured using a NanoBRET BTK in-cell target engagement assay” as described in the \nTable legends. It is the concentration of compound required to inhibit binding of the fixed \nconcentration of fluorescent probe. The IC50 will very much depend on the concentration of probe \nused in the experiments. A much clear explanation is required on this. \n\n \n\nI also do not fully understand what is meant by ‘normalised target engagement’ in Table 1 and 2. \nOccupancy (target engagement) is normally represented by [ligand]/([Ligand] + Kd). What do these \nnormalised target engagement values represent in the Tables? \n\n \n\nI struggled to understand the fluorophore phase separation-based assay for imaging ternary complex \nformation. It looked to me to be a general measure of protein aggregation with the complexes being \nquite large (as judged by their size with respect to the scale bar). I failed to understand how this was \na measure of ternary complex formation. Surely the spatial resolution is too low? \n\n \n\n \n\n\f \n\nReviewer #2 (Remarks to the Author): \n\n \n\nIn this manuscript, Guo et al. developed RC-1 as the first reversible covalent BTK PROTAC, which has \nhigh target occupancy and is effective both as an inhibitor and a degrader. They provided \nbiochemical and cellular data to support their conclusion that RC-1 induces mechanism-based BTK \ndegradation against both wild-type BTK and C481S mutant form. \n\nThe authors compared RC-1 with reversible noncovalent degrader (RNC-1) and irreversible covalent \ndegrader (IRC-1) and found RC-1 shows better activity in cells and suggested that reversible covalent \ndegrader could be a good strategy for the development of PORTACS. In addition, they demonstrated \nthat the enhanced cellular permeability is responsible for better cellular potencies for RC-1 than \nRNC-1 and ICR-1. Their study suggests that “introducing a cyano-acrylamide moiety can be a general \nstrategy to enhance cellular permeability for PROTACs.” To test this idea further, they have designed \nFLT3 degraders and obtained similar results. Finally, they tested RC-1 in vivo, shows BTK degradation \nin spleen, which suggests reversible covalent BTK degraders may be useful for in vivo studies. \n\n \n\nOverall, this study was well done and is significant for the field of PROTACs. It is recommended for \npublication in Nature Communications with the following revisions. \n\n \n\n1. Since the most attractive point of the manuscript is “introducing a cyano-acrylamide moiety can \nbe a general strategy to enhance cellular permeability for PROTACs”, it is more convincing to provide \ndirect evidence about the improved cellular permeability for RC-1 over other compounds using the \nCaco-2 cell-permeability assay or other similar assas. \n\n \n\n2. It will be useful to make another reversible noncovalent compound as reference, which is directly \nreduce the double bond of RC-1. Compare its cell permeability with RC-1 may give us more \ninformation, such as whether the enhanced permeability is caused by adding lipophilic groups \n(cyano and di-methyl) or because of the reversible covalent ability. \n\n \n\n3. Have the authors compared the BTK degradation between Crews’s compound (such as MT802) \nand RC-1 in Moml-14 cell line or other cell lines? It will be useful to compare these two compounds, \nsince MT802 is an optimal reversible noncovalent degrader. \n\n \n\n4. In the in vivo PD study, the drug was dosed for 7 days. The PD effect is not as good what has been \nobserved in vitro. Since degradation of BTK is already profound within 24 hours in cells, what was \nthe reason for the 7-day dosing and for the relative less profound degradation of BTK? \n\n\f \n\n \n\n \n\nReviewer #3 (Remarks to the Author): \n\n \n\nThis is a well-done and timely study on the effect of semi-reversible BTK PROTACs and showing that \nthese molecules are able to degrade BTK in cells. They perform followup proteomic studies and also \ncompare their PROTACs to other BTK PROTACs that have been previously synthesized. This study \ncomplements recent cysteine-targeting semi-reversible BTK PROTAC papers from Nir London's lab \nand Pfizer that have been recently posted on ChemRxiv and BioRxiv. This study should be published \nas is. \n\n\f\f\n", "role": "assistant" } ]	2,019
Orthophosphate increases the efficiency of slow muscle-myosin isoform in the presence of	Reviewers' comments: Reviewer #1 (Remarks to the Author): The authors have resubmitted a greatly improved version of their manuscript. I only have a few additional comments on the rebuttal responses and associated manuscript updates. Major comments: 1. The authors justify their choice to use permeabilized rabbit soleus muscle preparations over a permeabilized cardiac muscle system with (i) the lab’s previous experiences with skinned rabbit soleus and (ii) the want to have a homogeneous ensemble of beta-MHC. In response to the first point, what prevents the authors from using skinned rat trabeculae? There are many examples of other studies successfully using skinned rat trabeculae to study OM. The reasoning behind the second argument is understandable as human cardiac MHC, as the authors mention, is predominantly of the beta isoform. However, it seems that it would still be preferable to study the joint effects of [OM] and [Pi] in the rat cardiac system. This data would be more readily compared to previous OM studies and more translatable to other cardiac systems despite the differences in MHC isoform expression. Additionally, the work in (Pinzauti et al. 2018) shows that a similar model of sarcomere compliance and strain can be applied to rat trabeculae; comparing table 2 in (Pinzauti et al. 2018 ) to table 2 in (Percario et al. 2018) further reveals that the necessary parameters for skinned rat trabeculae could be determined. Were there other more substantial limitations to using skinned rat cardiac trabeculae that were not mentioned in the manuscript and rebuttal? 2. The addition of Table 3 greatly improved the clarity and presentation of the proposed kinetic model. It would further help if the authors would add uncertainties to the parameters that were “adjusted” or “set” to indicate how sensitive the model fit is. 3. The authors should include some goodness-of-fit metric to quantify how well the model fits the experimental data. Minor Comments: 1. In the system of equations variables for [OM], [ATP], and [ADP] lack the square brackets. This makes them difficult to distinguish from the other variables in the equations. References: Pinzauti, F., I. Pertici, M. Reconditi, T. Narayanan, G.J.M. Stienen, G. Piazzesi, V. Lombardi, M. Linari, and M. Caremani. 2018. The force and stiffness of myosin motors in the isometric twitch of a cardiac trabecula and the effect of the extracellular calcium concentration. J. Physiol. 596:2581–2596. doi:10.1113/JP275579. Percario, V., S. Boncompagni, F. Protasi, I. Pertici, F. Pinzauti, and M. Caremani. 2018. Mechanical parameters of the molecular motor myosin II determined in permeabilised fibres from slow and fast skeletal muscles of the rabbit. J. Physiol. 596:1243–1257. doi:10.1113/JP275404. Reviewer #2 (Remarks to the Author): The authors have addressed my comments in their revisions and I thank them for their consideration of these comments. The paper is strengthened by their revisions. I have only the following minor suggestions for the Introduction: 1) After the first sentence in the introduction, I would include a sentence about heart failure with reduced ejection fraction. While this is an acquired cardiomyopathy, increasing cardiac contractility may also play an important role in treatment. 2) In the third sentence of the introduction - omecamtiv should not be capitalized. Reviewer #3 (Remarks to the Author): The authors have provided a satisfactory answer to my questions. The paper is adequate for publication.	Reviewers' comments: Reviewer #1 (Remarks to the Author): The authors have resubmitted a greatly improved version of their manuscript. I only have a few additional comments on the rebuttal responses and associated manuscript updates. Major comments: 1. The authors justify their choice to use permeabilized rabbit soleus muscle preparations over a permeabilized cardiac muscle system with (i) the lab’s previous experiences with skinned rabbit soleus and (ii) the want to have a homogeneous ensemble of beta-MHC. In response to the first point, what prevents the authors from using skinned rat trabeculae? There are many examples of other studies successfully using skinned rat trabeculae to study OM. The reasoning behind the second argument is understandable as human cardiac MHC, as the authors mention, is predominantly of the beta isoform. However, it seems that it would still be preferable to study the joint effects of [OM] and [Pi] in the rat cardiac system. This data would be more readily compared to previous OM studies and more translatable to other cardiac systems despite the differences in MHC isoform expression. Additionally, the work in (Pinzauti et al. 2018) shows that a similar model of sarcomere compliance and strain can be applied to rat trabeculae; comparing table 2 in (Pinzauti et al. 2018 ) to table 2 in (Percario et al. 2018) further reveals that the necessary parameters for skinned rat trabeculae could be determined. Were there other more substantial limitations to using skinned rat cardiac trabeculae that were not mentioned in the manuscript and rebuttal? 2. The addition of Table 3 greatly improved the clarity and presentation of the proposed kinetic model. It would further help if the authors would add uncertainties to the parameters that were “adjusted” or “set” to indicate how sensitive the model fit is. 3. The authors should include some goodness-of-fit metric to quantify how well the model fits the experimental data. Minor Comments: 1. In the system of equations variables for [OM], [ATP], and [ADP] lack the square brackets. This makes them difficult to distinguish from the other variables in the equations. References: Pinzauti, F., I. Pertici, M. Reconditi, T. Narayanan, G.J.M. Stienen, G. Piazzesi, V. Lombardi, M. Linari, and M. Caremani. 2018. The force and stiffness of myosin motors in the isometric twitch of a cardiac trabecula and the effect of the extracellular calcium concentration. J. Physiol. 596:2581–2596. doi:10.1113/JP275579. Percario, V., S. Boncompagni, F. Protasi, I. Pertici, F. Pinzauti, and M. Caremani. 2018. Mechanical parameters of the molecular motor myosin II determined in permeabilised fibres from slow and fast skeletal muscles of the rabbit. J. Physiol. 596:1243–1257. doi:10.1113/JP275404. Reviewer #2 (Remarks to the Author): The authors have addressed my comments in their revisions and I thank them for their consideration of these comments. The paper is strengthened by their revisions. I have only the following minor suggestions for the Introduction: 1) After the first sentence in the introduction, I would include a sentence about heart failure with reduced ejection fraction. While this is an acquired cardiomyopathy, increasing cardiac contractility may also play an important role in treatment. 2) In the third sentence of the introduction - omecamtiv should not be capitalized. Reviewer #3 (Remarks to the Author): The authors have provided a satisfactory answer to my questions. The paper is adequate for publication.	TPR	NC_BS_BIOPH_002_RR2.txt	{ "criticism": 2, "example": 2, "importance_and_relevance": 1, "materials_and_methods": 11, "praise": 5, "presentation_and_reporting": 4, "results_and_discussion": 0, "suggestion_and_solution": 7, "total": 42 }	{ "criticism": 0.047619047619047616, "example": 0.047619047619047616, "importance_and_relevance": 0.023809523809523808, "materials_and_methods": 0.2619047619047619, "praise": 0.11904761904761904, "presentation_and_reporting": 0.09523809523809523, "results_and_discussion": 0, "suggestion_and_solution": 0.16666666666666666 }	0.761905	[ { "content": "Please review this paper.", "role": "user" }, { "content": "Reviewers' comments:\nReviewer #1 (Remarks to the Author):\nThe authors have resubmitted a greatly improved version of their manuscript. I only have a few additional comments on the rebuttal responses and associated manuscript updates.\nMajor comments:\n1. The authors justify their choice to use permeabilized rabbit soleus muscle preparations over a permeabilized cardiac muscle system with (i) the lab’s previous experiences with skinned rabbit soleus and (ii) the want to have a homogeneous ensemble of beta-MHC. In response to the first point, what prevents the authors from using skinned rat trabeculae? There are many examples of other studies successfully using skinned rat trabeculae to study OM.\nThe reasoning behind the second argument is understandable as human cardiac MHC, as the authors mention, is predominantly of the beta isoform. However, it seems that it would still be preferable to study the joint effects of [OM] and [Pi] in the rat cardiac system. This data would be more readily compared to previous OM studies and more translatable to other cardiac systems despite the differences in MHC isoform expression. Additionally, the work in (Pinzauti et al. 2018) shows that a similar model of sarcomere compliance and strain can be applied to rat trabeculae; comparing table 2 in (Pinzauti et al. 2018 ) to table 2 in (Percario et al. 2018) further reveals that the necessary parameters for skinned rat trabeculae could be determined. Were there other more substantial limitations to using skinned rat cardiac trabeculae that were not mentioned in the manuscript and rebuttal?\n2. The addition of Table 3 greatly improved the clarity and presentation of the proposed kinetic model. It would further help if the authors would add uncertainties to the parameters that were “adjusted” or “set” to indicate how sensitive the model fit is.\n3. The authors should include some goodness-of-fit metric to quantify how well the model fits the experimental data.\nMinor Comments:\n1. In the system of equations variables for [OM], [ATP], and [ADP] lack the square brackets. This makes them difficult to distinguish from the other variables in the equations.\nReferences:\nPinzauti, F., I. Pertici, M. Reconditi, T. Narayanan, G.J.M. Stienen, G. Piazzesi, V. Lombardi, M. Linari, and M. Caremani. 2018. The force and stiffness of myosin motors in the isometric twitch of a cardiac trabecula and the effect of the extracellular calcium concentration. J. Physiol. 596:2581–2596. doi:10.1113/JP275579.\nPercario, V., S. Boncompagni, F. Protasi, I. Pertici, F. Pinzauti, and M. Caremani. 2018. Mechanical parameters of the molecular motor myosin II determined in permeabilised fibres from slow and fast skeletal muscles of the rabbit. J. Physiol. 596:1243–1257. doi:10.1113/JP275404.\nReviewer #2 (Remarks to the Author):\nThe authors have addressed my comments in their revisions and I thank them for their consideration of these comments. The paper is strengthened by their revisions.\nI have only the following minor suggestions for the Introduction:\n1) After the first sentence in the introduction, I would include a sentence about heart failure with reduced ejection fraction. While this is an acquired cardiomyopathy, increasing cardiac contractility may also play an important role in treatment.\n2) In the third sentence of the introduction - omecamtiv should not be capitalized.\nReviewer #3 (Remarks to the Author):\nThe authors have provided a satisfactory answer to my questions. The paper is adequate for publication.\n", "role": "assistant" } ]	2,020
Orthophosphate increases the efficiency of slow muscle-myosin isoform in the presence of	Reviewers' comments: Reviewer #1 (Remarks to the Author): Summary: The authors of the submitted manuscript present data investigating the mechanism by which the systolic heart failure (HF) drug, Omecamtiv Mecarbil (OM), affects the force output and cycling kinetic of rabbit soleus muscle in situ. Mechanical experiments confirmed previously published results that OM increases force at submaximal [Ca¬2+], decreases force near maximal [Ca2+], increases calcium sensitivity, and decreases cooperativity. Using a simplified spring-model of the sarcomere, the authors show that the OM-induced reduction in force at max [Ca2+] can be attributed to the reduction in the average strain experienced by each attached motor, as opposed to a reduction in the number of bound motors. At submaximal [Ca2+], OM-induced force increases were attributed to an increase in the number of bound motors, which supports previous findings. Interestingly, the addition of physiological concentrations of inorganic phosphate (Pi) significantly mitigates the OM-induced force reduction at maximal [Ca2+]. The addition of Pi also had significant dose dependent effects on the rate of force redevelopment and the average motor strain. As a second source of kinetic insight, the authors measured the overall ATPase rate (kcat) in isometrically contracting muscle to find that OM did not significantly impact the Pi dependence of kcat but did reduce the muscle’s tension cost. As the authors mention, this finding is specific to the context of high load contraction. To explain the observed effects of OM and Pi, the authors extend a previously published kinetic model of muscle myosin. The main feature of the model is the addition of an OM pathway that inhibits the power stroke and introduces an isomerization step following P¬¬i where the new state has an increased Pi affinity. The new state allows for competitive interactions between OM and Pi¬ in which P¬i can allosterically displace OM, thus allowing the motor to reenter the usual force producing ATPase cycle. Finally, the authors suggest that their experimental findings and model predictions may identify an energy-efficient response to OM during exercise when intercellular Pi increases within the heart. Major Comments: 1. While it is understandable from the experimental perspective why OM’s effects were studied in soleus skeletal muscle instead of cardiac muscle, the authors translate results between the two systems without much discussion of potential limitations and caveats. Such translation may be appropriate for single molecule results as the dominant myosin isoform in soleus and cardiac muscle are so similar (ref 1) but this is harder to justify at the fiber level where other important sarcomeric proteins differ in isoform and functional role. Furthermore, considering that the vast majority of fiber level studies on OM are in cardiac muscle and that little has been previously published on OM’s effect on soleus skeletal muscle, a detailed limitations section and/or additional experimental evidence is needed to contextualize/justify the translation from soleus to cardiac muscle. 2. In the introduction (manuscript lines 63-69) the authors use the fact that intracellular [Ca2+] remains below saturating levels during left ventricular systole in their proposed mechanism explaining how OM and Pi jointly impact muscle function. Submaximal [Ca2+] data is reported for OM in the absence of presence of Pi in figure 3, but data for the joint effect of OM and Pi is only reported for maximal [Ca2+]. It would greatly strengthen the proposed mechanism if the authors would repeat the Pi experiments reported in figures 4 and 5 at submaximal [Ca2+]. This would ensure that cooperative mechanisms prevalent at submaximal [Ca2+]2, whose impacts may be difficult to predict in the context of a drug, do not significantly impact the results. 3. There are important details missing about the development and implementation of the kinetic model shown in figure 6. First, each rate in table 3 should be accompanied with a reference for which the value was obtained from. If the value had to be inferred, then a similar indication should be made in the table with a written or mathematical justification for why or how the parameter was assigned a specific value. Second, the force generating states used to predict the OM/Pi dependencies of the mechanical and kinetic parameters in figures 2, 4, 5 need to be clearly marked in figure 6. Third, at a minimum, the general form of the differential equations underlying the kinetic scheme should be stated in the supplemental information. Ideally, however, the whole system of equations would be written down. Finally, what numerical methods were used to solve the system or what software was used? The inclusion of each of these would significantly add credibility and transparency to the model and following conclusions. 4. As the manuscript mentions (manuscript lines 41-42), it is well known that OM increases the rate of phosphate release (refs 3, 4). However, the kinetic rates found in table 3 do not reflect this; the rate of phosphate release normal XB cycle (k+4) is given a value of 25s-1, while the rate of phosphate release in the OM affected cycle (k+10) is only 7s-1. Please comment on why this is. 5. While it is true that this is the first study to examine OM’s effects in situ by combining fast sarcomere-level mechanics and in situ enzyme kinetics, the individual methodologies are well established (refs 5-7). Furthermore, previous studies using different fiber-level techniques have arrived at very similar conclusions as the ones in this study. For instance, OM’s mechanism of force enhancement and its potential effect on the muscle’s tension cost had both been previously predicted (ref 8). Finally, the manuscript’s primary novel contribution, the joint effect of Pi and OM on contractile performance, was not obtained in the most relevant muscle system (cardiac muscle), which limits its usefulness in understanding OM’s mechanism. The authors would well be served by making a stronger case for the unique contribution of the data within the context of the substantial literature on OM that already exists. Minor Comments: 1. In figures 1D and the solid trace in 4B that display the rates of force redevelopment (rTD) as functions of [OM] and [Pi], the points closest to the y-axis are presumably measured under the same experimental conditions: [OM]=0mM and [Pi]=1mM. However the furthest left value in 1D is (3.7±2)s-1 while the furthest left value in figure 4B is significantly lower, approximately (2.7±2) s-1, but increases with [Pi] to a value close to 3.7s-1. Should this be the case? 2. The methods section explains the experimental apparatus used at the Amsterdam lab to measure ATPase rate. However, it is not clear how the force was measured and recorded during the ATPase experiments. Please clarify this process in the methods section and include any relevant instrument specifications. 3. Please report the temperature at which ATPase rates were measured at the Amsterdam lab. References 1. Schiaffino, S. & Reggiani, C. Molecular diversity of myofibrillar proteins: gene regulation and functional significance. Physiol. Rev. 76, 371–423 (1996). 2. Gordon, A. M., Homsher, E. & Regnier, M. Regulation of contraction in striated muscle. Physiol. Rev. 80, 853–924 (2000). 3. Malik, F. I. et al. Cardiac myosin activation: A potential therapeutic approach for systolic heart failure. Science (80-. ). 331, 1439–1443 (2011). 4. Woody, M. S. et al. Positive cardiac inotrope omecamtiv mecarbil activates muscle despite suppressing the myosin working stroke. Nat. Commun. 9, 1–11 (2018). 5. Goldman, Y. Kinetics Of The Actomyosin ATPase In Muscle Fibers. Annu. Rev. Physiol. 49, 637–654 (1987). 6. Caremani, M., Melli, L., Dolfi, M., Lombardi, V. & Linari, M. Force and number of myosin motors during muscle shortening and the coupling with the release of the ATP hydrolysis products. J. Physiol. 593, 3313–3332 (2015). 7. Percario, V. et al. Mechanical parameters of the molecular motor myosin II determined in permeabilised fibres from slow and fast skeletal muscles of the rabbit. J. Physiol. 596, 1243–1257 (2018). 8. Kieu, T. T., Awinda, P. O. & Tanner, B. C. W. Omecamtiv Mecarbil Slows Myosin Kinetics in Skinned Rat Myocardium at Physiological Temperature. Biophys. J. 116, 2149–2160 (2019). Reviewer #2 (Remarks to the Author): Major Points The authors explore the effect of phosphate on the action of omecamtiv mecarbil – which surprisingly has a significant effect on muscle mechanics. Overall, this is a thoughtful set of experiments that grow the understanding of how omecamtiv mecarbil works to increase muscle contractility and the energetics of its action. Many will be interested in the conclusions of the paper. However, overall the paper is complex for the reader not already an expert in muscle mechanics. I would suggest the authors simplify the text substantially. The key findings of the paper reviewed in the discussion are lost because of the complexity of the discussion. Cardiac contraction is more of a twitch (the time of systole ~300 msec is approximately the same as a single turnover of ATP by myosin). Do the results for an isometric contraction differ from what might happen during a short isokinetic contraction? Is there an experiment the authors can do to understand if there is a difference? In regard to the following sentence and central to the conclusions of the paper: “The inhibitory effect of OM on force generation is reversed by Pi, which acts as an allosteric competitor that allows the OM-bound no-force-generating motors to release OM and re-enter the force-generating cycle without any change in the rate of ATP hydrolysis.” How is the alternate model - that of OM stabilizing the lever arm in the pre-powerstroke position, engaging the thin filament, releasing Pi and OM together, and then myosin proceeding to generate force – excluded? Or should it be considered as an alternate model? I would suggest a table in the text of the paper that describes the various parameters as an easy reference for the reader working through the paper. Minor Comments Abstract The statement below seems to also apply to resting conditions – why the focus on exercise? Suggest you end sentence after the word “patients”. “This mechanism could underpin an energetically efficient reduction of systolic tension cost in OM- treated patients under exercise, when [Pi] increases with heart-beat frequency.” Introduction Misspellings like disfunction (dysfunction) and some other typos and grammatical errors – authors please correct. Omecamtiv mecarbil not Omecamtiv Mecarbil This statement - “while the velocity of actin sliding in the in vitro motility assay (IVMA) is reduced15- 17”- is not needed and distracts from the main relevant description of the mechanism of action. Long sentence – please revise. “Here we determine the effect of OM on the number and force of the attached myosin motors and on the rate of ATP hydrolysis and their modulation by physiological concentrations of inorganic phosphate (Pi), by applying fast sarcomere level mechanics to Ca2+-activated demembranated fibres from rabbit soleus muscle, which contains the slow myosin isoform.” Suggest use of the work “contractile” rather than “inotropic” in the sentence below. Inotropic generally references an increase in the rate and extent of contractility. OM doesn’t appear to increase the rate of cardiac contractility. “This effect of Pi on the OM-bound motors represents an energetically efficient inotropic response in patients under OM treatment during exercise and increased workload, when [Pi] increases23,24 with the increase of heart beat frequency.” Results The authors are applauded for examining the concentration effect relationship of omecamtiv mecarbil in Figure 1. They do focus on the changes at 1 uM. The target range of concentration of omecamtiv mecarbil in humans is 300-750 ng/mL and the free fraction of OM is only 20%. So a plasma concentration of 400 ng/mL translates to a free concentration of 0.2 uM, 5x lower than 1 uM, where the effects are more modest and probably more clinically relevant. The authors should include some text in the results section about the magnitude of the effects in the clinically relevant range of omecamtiv mecarbil concentrations and not just focus on 1 uM. Unusual to reference to Amsterdam or Florence experiments. The paper’s results usually are described without reference to location. Generally, the authors should try to simplify the text in the results section if possible. It is difficult to follow – a table of measured parameters would be helpful as mentioned above so that the reader can easily figure out the meaning of the many different variables used. Discussion The paper discussion should focus on how these findings contribute to our understanding how OM works in conditions that mimic living, beating cardiac muscle. The relevance of a mechanism of action for omecamtiv mecarbil described previously in the absence of phosphate should be questioned since there is always mM levels of phosphate in working muscle. The authors should compare/contrast their findings with prior findings that did not examine the effects of physiological phosphate and focus on what are the implications to the energetics of cardiac contraction in the presence of omecamtiv mecarbil. Reviewer #3 (Remarks to the Author): This contribution from the Lombardi group provides critical insights into the mechanism of Omecamtiv mercarbil (OM), an activator of cardiac myosin that is in Phase 3 clinical trials for heart failure. The data from this laboratory is as always of excellent quality - it combines sarcomere-level mechanics and ATPase measurements in single slow demembranated fibres from rabbit soleus. A major finding from this paper is the impact of Pi ions on OM-bound myosin heads. In the absence of Pi, OM leads to an increase in ATP consumption due to the inability of OM-bound heads to complete the force-generating power stroke, and thus detaching without completing the powerstoke but having consumed ATP. The addition of physiological levels of ATP reverses the excess ATP consumption and brings it back to the consumption in the absence of drug. This is of great importance since it shows a previously unappreciated effect of OM – its ability to reduce the energetic cost of systolic tension when Pi concentrations increases with heart-beat frequency. Without knowledge this Pi effect, it has been difficult to explain the impact of the drug on the working heart from the experimental data published to date. While the data is well presented and of very high quality, I have reservations concerning the modeling and its justification. The authors should at least indicate that the model proposed is only one possible model. There is for example no evidence that rebinding of Pi would lead to release of OM as indicated in step 14. In fact there is no basis for suggesting the transition (11) to a state that has a higher affinity for Pi exists. If it is based on other data in the literature, this should be cited. However, it seems simpler to propose that OM increases the occupancy of the state formed by Pi release, allowing exogenous Pi to increase the population of the same state that detaches in the model (green) in the absence of OM. But that OM inhibits its detachment and futile loss of ADP and Pi. In any event, it is important that the authors acknowledge that they have chosen one of the possible models to account for the data since controversy exists as to the states sensitive to Pi rebinding to the myosin head during the powerstroke. Overall, this is a very strong and well written paper that provides a critical new insight into how a cardiac/slow myosin stimulator can achieve positive results in the clinic. The data is state of the art and the findings will be of wide interest.	Reviewers' comments: Reviewer #1 (Remarks to the Author): Summary: The authors of the submitted manuscript present data investigating the mechanism by which the systolic heart failure (HF) drug, Omecamtiv Mecarbil (OM), affects the force output and cycling kinetic of rabbit soleus muscle in situ. Mechanical experiments confirmed previously published results that OM increases force at submaximal [Ca¬2+], decreases force near maximal [Ca2+], increases calcium sensitivity, and decreases cooperativity. Using a simplified spring-model of the sarcomere, the authors show that the OM-induced reduction in force at max [Ca2+] can be attributed to the reduction in the average strain experienced by each attached motor, as opposed to a reduction in the number of bound motors. At submaximal [Ca2+], OM-induced force increases were attributed to an increase in the number of bound motors, which supports previous findings. Interestingly, the addition of physiological concentrations of inorganic phosphate (Pi) significantly mitigates the OM-induced force reduction at maximal [Ca2+]. The addition of Pi also had significant dose dependent effects on the rate of force redevelopment and the average motor strain. As a second source of kinetic insight, the authors measured the overall ATPase rate (kcat) in isometrically contracting muscle to find that OM did not significantly impact the Pi dependence of kcat but did reduce the muscle’s tension cost. As the authors mention, this finding is specific to the context of high load contraction. To explain the observed effects of OM and Pi, the authors extend a previously published kinetic model of muscle myosin. The main feature of the model is the addition of an OM pathway that inhibits the power stroke and introduces an isomerization step following P¬¬i where the new state has an increased Pi affinity. The new state allows for competitive interactions between OM and Pi¬ in which P¬i can allosterically displace OM, thus allowing the motor to reenter the usual force producing ATPase cycle. Finally, the authors suggest that their experimental findings and model predictions may identify an energy-efficient response to OM during exercise when intercellular Pi increases within the heart. Major Comments: 1. While it is understandable from the experimental perspective why OM’s effects were studied in soleus skeletal muscle instead of cardiac muscle, the authors translate results between the two systems without much discussion of potential limitations and caveats. Such translation may be appropriate for single molecule results as the dominant myosin isoform in soleus and cardiac muscle are so similar (ref 1) but this is harder to justify at the fiber level where other important sarcomeric proteins differ in isoform and functional role. Furthermore, considering that the vast majority of fiber level studies on OM are in cardiac muscle and that little has been previously published on OM’s effect on soleus skeletal muscle, a detailed limitations section and/or additional experimental evidence is needed to contextualize/justify the translation from soleus to cardiac muscle. 2. In the introduction (manuscript lines 63-69) the authors use the fact that intracellular [Ca2+] remains below saturating levels during left ventricular systole in their proposed mechanism explaining how OM and Pi jointly impact muscle function. Submaximal [Ca2+] data is reported for OM in the absence of presence of Pi in figure 3, but data for the joint effect of OM and Pi is only reported for maximal [Ca2+]. It would greatly strengthen the proposed mechanism if the authors would repeat the Pi experiments reported in figures 4 and 5 at submaximal [Ca2+]. This would ensure that cooperative mechanisms prevalent at submaximal [Ca2+]2, whose impacts may be difficult to predict in the context of a drug, do not significantly impact the results. 3. There are important details missing about the development and implementation of the kinetic model shown in figure 6. First, each rate in table 3 should be accompanied with a reference for which the value was obtained from. If the value had to be inferred, then a similar indication should be made in the table with a written or mathematical justification for why or how the parameter was assigned a specific value. Second, the force generating states used to predict the OM/Pi dependencies of the mechanical and kinetic parameters in figures 2, 4, 5 need to be clearly marked in figure 6. Third, at a minimum, the general form of the differential equations underlying the kinetic scheme should be stated in the supplemental information. Ideally, however, the whole system of equations would be written down. Finally, what numerical methods were used to solve the system or what software was used? The inclusion of each of these would significantly add credibility and transparency to the model and following conclusions. 4. As the manuscript mentions (manuscript lines 41-42), it is well known that OM increases the rate of phosphate release (refs 3, 4). However, the kinetic rates found in table 3 do not reflect this; the rate of phosphate release normal XB cycle (k+4) is given a value of 25s-1, while the rate of phosphate release in the OM affected cycle (k+10) is only 7s-1. Please comment on why this is. 5. While it is true that this is the first study to examine OM’s effects in situ by combining fast sarcomere-level mechanics and in situ enzyme kinetics, the individual methodologies are well established (refs 5-7). Furthermore, previous studies using different fiber-level techniques have arrived at very similar conclusions as the ones in this study. For instance, OM’s mechanism of force enhancement and its potential effect on the muscle’s tension cost had both been previously predicted (ref 8). Finally, the manuscript’s primary novel contribution, the joint effect of Pi and OM on contractile performance, was not obtained in the most relevant muscle system (cardiac muscle), which limits its usefulness in understanding OM’s mechanism. The authors would well be served by making a stronger case for the unique contribution of the data within the context of the substantial literature on OM that already exists. Minor Comments: 1. In figures 1D and the solid trace in 4B that display the rates of force redevelopment (rTD) as functions of [OM] and [Pi], the points closest to the y-axis are presumably measured under the same experimental conditions: [OM]=0mM and [Pi]=1mM. However the furthest left value in 1D is (3.7±2)s-1 while the furthest left value in figure 4B is significantly lower, approximately (2.7±2) s-1, but increases with [Pi] to a value close to 3.7s-1. Should this be the case? 2. The methods section explains the experimental apparatus used at the Amsterdam lab to measure ATPase rate. However, it is not clear how the force was measured and recorded during the ATPase experiments. Please clarify this process in the methods section and include any relevant instrument specifications. 3. Please report the temperature at which ATPase rates were measured at the Amsterdam lab. References 1. Schiaffino, S. & Reggiani, C. Molecular diversity of myofibrillar proteins: gene regulation and functional significance. Physiol. Rev. 76, 371–423 (1996). 2. Gordon, A. M., Homsher, E. & Regnier, M. Regulation of contraction in striated muscle. Physiol. Rev. 80, 853–924 (2000). 3. Malik, F. I. et al. Cardiac myosin activation: A potential therapeutic approach for systolic heart failure. Science (80-. ). 331, 1439–1443 (2011). 4. Woody, M. S. et al. Positive cardiac inotrope omecamtiv mecarbil activates muscle despite suppressing the myosin working stroke. Nat. Commun. 9, 1–11 (2018). 5. Goldman, Y. Kinetics Of The Actomyosin ATPase In Muscle Fibers. Annu. Rev. Physiol. 49, 637–654 (1987). 6. Caremani, M., Melli, L., Dolfi, M., Lombardi, V. & Linari, M. Force and number of myosin motors during muscle shortening and the coupling with the release of the ATP hydrolysis products. J. Physiol. 593, 3313–3332 (2015). 7. Percario, V. et al. Mechanical parameters of the molecular motor myosin II determined in permeabilised fibres from slow and fast skeletal muscles of the rabbit. J. Physiol. 596, 1243–1257 (2018). 8. Kieu, T. T., Awinda, P. O. & Tanner, B. C. W. Omecamtiv Mecarbil Slows Myosin Kinetics in Skinned Rat Myocardium at Physiological Temperature. Biophys. J. 116, 2149–2160 (2019). Reviewer #2 (Remarks to the Author): Major Points The authors explore the effect of phosphate on the action of omecamtiv mecarbil – which surprisingly has a significant effect on muscle mechanics. Overall, this is a thoughtful set of experiments that grow the understanding of how omecamtiv mecarbil works to increase muscle contractility and the energetics of its action. Many will be interested in the conclusions of the paper. However, overall the paper is complex for the reader not already an expert in muscle mechanics. I would suggest the authors simplify the text substantially. The key findings of the paper reviewed in the discussion are lost because of the complexity of the discussion. Cardiac contraction is more of a twitch (the time of systole ~300 msec is approximately the same as a single turnover of ATP by myosin). Do the results for an isometric contraction differ from what might happen during a short isokinetic contraction? Is there an experiment the authors can do to understand if there is a difference? In regard to the following sentence and central to the conclusions of the paper: “The inhibitory effect of OM on force generation is reversed by Pi, which acts as an allosteric competitor that allows the OM-bound no-force-generating motors to release OM and re-enter the force-generating cycle without any change in the rate of ATP hydrolysis.” How is the alternate model - that of OM stabilizing the lever arm in the pre-powerstroke position, engaging the thin filament, releasing Pi and OM together, and then myosin proceeding to generate force – excluded? Or should it be considered as an alternate model? I would suggest a table in the text of the paper that describes the various parameters as an easy reference for the reader working through the paper. Minor Comments Abstract The statement below seems to also apply to resting conditions – why the focus on exercise? Suggest you end sentence after the word “patients”. “This mechanism could underpin an energetically efficient reduction of systolic tension cost in OM- treated patients under exercise, when [Pi] increases with heart-beat frequency.” Introduction Misspellings like disfunction (dysfunction) and some other typos and grammatical errors – authors please correct. Omecamtiv mecarbil not Omecamtiv Mecarbil This statement - “while the velocity of actin sliding in the in vitro motility assay (IVMA) is reduced15- 17”- is not needed and distracts from the main relevant description of the mechanism of action. Long sentence – please revise. “Here we determine the effect of OM on the number and force of the attached myosin motors and on the rate of ATP hydrolysis and their modulation by physiological concentrations of inorganic phosphate (Pi), by applying fast sarcomere level mechanics to Ca2+-activated demembranated fibres from rabbit soleus muscle, which contains the slow myosin isoform.” Suggest use of the work “contractile” rather than “inotropic” in the sentence below. Inotropic generally references an increase in the rate and extent of contractility. OM doesn’t appear to increase the rate of cardiac contractility. “This effect of Pi on the OM-bound motors represents an energetically efficient inotropic response in patients under OM treatment during exercise and increased workload, when [Pi] increases23,24 with the increase of heart beat frequency.” Results The authors are applauded for examining the concentration effect relationship of omecamtiv mecarbil in Figure 1. They do focus on the changes at 1 uM. The target range of concentration of omecamtiv mecarbil in humans is 300-750 ng/mL and the free fraction of OM is only 20%. So a plasma concentration of 400 ng/mL translates to a free concentration of 0.2 uM, 5x lower than 1 uM, where the effects are more modest and probably more clinically relevant. The authors should include some text in the results section about the magnitude of the effects in the clinically relevant range of omecamtiv mecarbil concentrations and not just focus on 1 uM. Unusual to reference to Amsterdam or Florence experiments. The paper’s results usually are described without reference to location. Generally, the authors should try to simplify the text in the results section if possible. It is difficult to follow – a table of measured parameters would be helpful as mentioned above so that the reader can easily figure out the meaning of the many different variables used. Discussion The paper discussion should focus on how these findings contribute to our understanding how OM works in conditions that mimic living, beating cardiac muscle. The relevance of a mechanism of action for omecamtiv mecarbil described previously in the absence of phosphate should be questioned since there is always mM levels of phosphate in working muscle. The authors should compare/contrast their findings with prior findings that did not examine the effects of physiological phosphate and focus on what are the implications to the energetics of cardiac contraction in the presence of omecamtiv mecarbil. Reviewer #3 (Remarks to the Author): This contribution from the Lombardi group provides critical insights into the mechanism of Omecamtiv mercarbil (OM), an activator of cardiac myosin that is in Phase 3 clinical trials for heart failure. The data from this laboratory is as always of excellent quality - it combines sarcomere-level mechanics and ATPase measurements in single slow demembranated fibres from rabbit soleus. A major finding from this paper is the impact of Pi ions on OM-bound myosin heads. In the absence of Pi, OM leads to an increase in ATP consumption due to the inability of OM-bound heads to complete the force-generating power stroke, and thus detaching without completing the powerstoke but having consumed ATP. The addition of physiological levels of ATP reverses the excess ATP consumption and brings it back to the consumption in the absence of drug. This is of great importance since it shows a previously unappreciated effect of OM – its ability to reduce the energetic cost of systolic tension when Pi concentrations increases with heart-beat frequency. Without knowledge this Pi effect, it has been difficult to explain the impact of the drug on the working heart from the experimental data published to date. While the data is well presented and of very high quality, I have reservations concerning the modeling and its justification. The authors should at least indicate that the model proposed is only one possible model. There is for example no evidence that rebinding of Pi would lead to release of OM as indicated in step 14. In fact there is no basis for suggesting the transition (11) to a state that has a higher affinity for Pi exists. If it is based on other data in the literature, this should be cited. However, it seems simpler to propose that OM increases the occupancy of the state formed by Pi release, allowing exogenous Pi to increase the population of the same state that detaches in the model (green) in the absence of OM. But that OM inhibits its detachment and futile loss of ADP and Pi. In any event, it is important that the authors acknowledge that they have chosen one of the possible models to account for the data since controversy exists as to the states sensitive to Pi rebinding to the myosin head during the powerstroke. Overall, this is a very strong and well written paper that provides a critical new insight into how a cardiac/slow myosin stimulator can achieve positive results in the clinic. The data is state of the art and the findings will be of wide interest.	TPR	NC_BS_BIOPH_002_RR1.txt	{ "criticism": 13, "example": 9, "importance_and_relevance": 10, "materials_and_methods": 50, "praise": 9, "presentation_and_reporting": 23, "results_and_discussion": 28, "suggestion_and_solution": 30, "total": 142 }	{ "criticism": 0.09154929577464789, "example": 0.06338028169014084, "importance_and_relevance": 0.07042253521126761, "materials_and_methods": 0.352112676056338, "praise": 0.06338028169014084, "presentation_and_reporting": 0.1619718309859155, "results_and_discussion": 0.19718309859154928, "suggestion_and_solution": 0.2112676056338028 }	1.211268	[ { "content": "Please review this paper.", "role": "user" }, { "content": "Reviewers' comments:\nReviewer #1 (Remarks to the Author):\nSummary: The authors of the submitted manuscript present data investigating the mechanism by which the systolic heart failure (HF) drug, Omecamtiv Mecarbil (OM), affects the force output and cycling kinetic of rabbit soleus muscle in situ. Mechanical experiments confirmed previously published results that OM increases force at submaximal [Ca¬2+], decreases force near maximal [Ca2+], increases calcium sensitivity, and decreases cooperativity. Using a simplified spring-model of the sarcomere, the authors show that the OM-induced reduction in force at max [Ca2+] can be attributed to the reduction in the average strain experienced by each attached motor, as opposed to a reduction in the number of bound motors. At submaximal [Ca2+], OM-induced force increases were attributed to an increase in the number of bound motors, which supports previous findings. Interestingly, the addition of physiological concentrations of inorganic phosphate (Pi) significantly mitigates the OM-induced force reduction at maximal [Ca2+]. The addition of Pi also had significant dose dependent effects on the rate of force redevelopment and the average motor strain. As a second source of kinetic insight, the authors measured the overall ATPase rate (kcat) in isometrically contracting muscle to find that OM did not significantly impact the Pi dependence of kcat but did reduce the muscle’s tension cost. As the authors mention, this finding is specific to the context of high load contraction. To explain the observed effects of OM and Pi, the authors extend a previously published kinetic model of muscle myosin. The main feature of the model is the addition of an OM pathway that inhibits the power stroke and introduces an isomerization step following P¬¬i where the new state has an increased Pi affinity. The new state allows for competitive interactions between OM and Pi¬ in which P¬i can allosterically displace OM, thus allowing the motor to reenter the usual force producing ATPase cycle. Finally, the authors suggest that their experimental findings and model predictions may identify an energy-efficient response to OM during exercise when intercellular Pi increases within the heart.\nMajor Comments:\n1. While it is understandable from the experimental perspective why OM’s effects were studied in soleus skeletal muscle instead of cardiac muscle, the authors translate results between the two systems without much discussion of potential limitations and caveats. Such translation may be appropriate for single molecule results as the dominant myosin isoform in soleus and cardiac muscle are so similar (ref 1) but this is harder to justify at the fiber level where other important sarcomeric proteins differ in isoform and functional role. Furthermore, considering that the vast majority of fiber level studies on OM are in cardiac muscle and that little has been previously published on OM’s effect on soleus skeletal muscle, a detailed limitations section and/or additional experimental evidence is needed to contextualize/justify the translation from soleus to cardiac muscle.\n2. In the introduction (manuscript lines 63-69) the authors use the fact that intracellular [Ca2+] remains below saturating levels during left ventricular systole in their proposed mechanism explaining how OM and Pi jointly impact muscle function. Submaximal [Ca2+] data is reported for OM in the absence of presence of Pi in figure 3, but data for the joint effect of OM and Pi is only reported for maximal [Ca2+]. It would greatly strengthen the proposed mechanism if the authors would repeat the Pi experiments reported in figures 4 and 5 at submaximal [Ca2+]. This would ensure that cooperative mechanisms prevalent at submaximal [Ca2+]2, whose impacts may be difficult to predict in the context of a drug, do not significantly impact the results.\n3. There are important details missing about the development and implementation of the kinetic model shown in figure 6. First, each rate in table 3 should be accompanied with a reference for which the value was obtained from. If the value had to be inferred, then a similar indication should be made in the table with a written or mathematical justification for why or how the parameter was assigned a specific value. Second, the force generating states used to predict the OM/Pi dependencies of the mechanical and kinetic parameters in figures 2, 4, 5 need to be clearly marked in figure 6. Third, at a minimum, the general form of the differential equations underlying the kinetic scheme should be stated in the supplemental information. Ideally, however, the whole system of equations would be written down. Finally, what numerical methods were used to solve the system or what software was used? The inclusion of each of these would significantly add credibility and transparency to the model and following conclusions.\n4. As the manuscript mentions (manuscript lines 41-42), it is well known that OM increases the rate of phosphate release (refs 3, 4). However, the kinetic rates found in table 3 do not reflect this; the rate of phosphate release normal XB cycle (k+4) is given a value of 25s-1, while the rate of phosphate release in the OM affected cycle (k+10) is only 7s-1. Please comment on why this is.\n5. While it is true that this is the first study to examine OM’s effects in situ by combining fast sarcomere-level mechanics and in situ enzyme kinetics, the individual methodologies are well established (refs 5-7). Furthermore, previous studies using different fiber-level techniques have arrived at very similar conclusions as the ones in this study. For instance, OM’s mechanism of force enhancement and its potential effect on the muscle’s tension cost had both been previously predicted (ref 8). Finally, the manuscript’s primary novel contribution, the joint effect of Pi and OM on contractile performance, was not obtained in the most relevant muscle system (cardiac muscle), which limits its usefulness in understanding OM’s mechanism. The authors would well be served by making a stronger case for the unique contribution of the data within the context of the substantial literature on OM that already exists.\nMinor Comments:\n1. In figures 1D and the solid trace in 4B that display the rates of force redevelopment (rTD) as functions of [OM] and [Pi], the points closest to the y-axis are presumably measured under the same experimental conditions: [OM]=0mM and [Pi]=1mM. However the furthest left value in 1D is (3.7±2)s-1 while the furthest left value in figure 4B is significantly lower, approximately (2.7±2) s-1, but increases with [Pi] to a value close to 3.7s-1. Should this be the case?\n2. The methods section explains the experimental apparatus used at the Amsterdam lab to measure ATPase rate. However, it is not clear how the force was measured and recorded during the ATPase experiments. Please clarify this process in the methods section and include any relevant instrument specifications.\n3. Please report the temperature at which ATPase rates were measured at the Amsterdam lab.\nReferences\n1. Schiaffino, S. & Reggiani, C. Molecular diversity of myofibrillar proteins: gene regulation and functional significance. Physiol. Rev. 76, 371–423 (1996).\n2. Gordon, A. M., Homsher, E. & Regnier, M. Regulation of contraction in striated muscle. Physiol. Rev. 80, 853–924 (2000).\n3. Malik, F. I. et al. Cardiac myosin activation: A potential therapeutic approach for systolic heart failure. Science (80-. ). 331, 1439–1443 (2011).\n4. Woody, M. S. et al. Positive cardiac inotrope omecamtiv mecarbil activates muscle despite suppressing the myosin working stroke. Nat. Commun. 9, 1–11 (2018).\n5. Goldman, Y. Kinetics Of The Actomyosin ATPase In Muscle Fibers. Annu. Rev. Physiol. 49, 637–654 (1987).\n6. Caremani, M., Melli, L., Dolfi, M., Lombardi, V. & Linari, M. Force and number of myosin motors during muscle shortening and the coupling with the release of the ATP hydrolysis products. J. Physiol. 593, 3313–3332 (2015).\n7. Percario, V. et al. Mechanical parameters of the molecular motor myosin II determined in permeabilised fibres from slow and fast skeletal muscles of the rabbit. J. Physiol. 596, 1243–1257 (2018).\n8. Kieu, T. T., Awinda, P. O. & Tanner, B. C. W. Omecamtiv Mecarbil Slows Myosin Kinetics in Skinned Rat Myocardium at Physiological Temperature. Biophys. J. 116, 2149–2160 (2019).\nReviewer #2 (Remarks to the Author):\nMajor Points\nThe authors explore the effect of phosphate on the action of omecamtiv mecarbil – which surprisingly has a significant effect on muscle mechanics. Overall, this is a thoughtful set of experiments that grow the understanding of how omecamtiv mecarbil works to increase muscle contractility and the energetics of its action. Many will be interested in the conclusions of the paper. However, overall the paper is complex for the reader not already an expert in muscle mechanics. I would suggest the authors simplify the text substantially. The key findings of the paper reviewed in the discussion are lost because of the complexity of the discussion.\nCardiac contraction is more of a twitch (the time of systole ~300 msec is approximately the same as a single turnover of ATP by myosin). Do the results for an isometric contraction differ from what might happen during a short isokinetic contraction? Is there an experiment the authors can do to understand if there is a difference?\nIn regard to the following sentence and central to the conclusions of the paper:\n“The inhibitory effect of OM on force generation is reversed by Pi, which acts as an allosteric competitor that allows the OM-bound no-force-generating motors to release OM and re-enter the force-generating cycle without any change in the rate of ATP hydrolysis.”\nHow is the alternate model - that of OM stabilizing the lever arm in the pre-powerstroke position, engaging the thin filament, releasing Pi and OM together, and then myosin proceeding to generate force – excluded? Or should it be considered as an alternate model?\nI would suggest a table in the text of the paper that describes the various parameters as an easy reference for the reader working through the paper.\nMinor Comments\nAbstract\nThe statement below seems to also apply to resting conditions – why the focus on exercise? Suggest you end sentence after the word “patients”.\n“This mechanism could underpin an energetically efficient reduction of systolic tension cost in OM- treated patients under exercise, when [Pi] increases with heart-beat frequency.”\nIntroduction\nMisspellings like disfunction (dysfunction) and some other typos and grammatical errors – authors please correct.\nOmecamtiv mecarbil not Omecamtiv Mecarbil\nThis statement - “while the velocity of actin sliding in the in vitro motility assay (IVMA) is reduced15- 17”- is not needed and distracts from the main relevant description of the mechanism of action.\nLong sentence – please revise.\n“Here we determine the effect of OM on the number and force of the attached myosin motors and on the rate of ATP hydrolysis and their modulation by physiological concentrations of inorganic phosphate (Pi), by applying fast sarcomere level mechanics to Ca2+-activated demembranated fibres from rabbit soleus muscle, which contains the slow myosin isoform.”\nSuggest use of the work “contractile” rather than “inotropic” in the sentence below. Inotropic generally references an increase in the rate and extent of contractility. OM doesn’t appear to increase the rate of cardiac contractility.\n“This effect of Pi on the OM-bound motors represents an energetically efficient inotropic response in patients under OM treatment during exercise and increased workload, when [Pi] increases23,24 with the increase of heart beat frequency.”\nResults\nThe authors are applauded for examining the concentration effect relationship of omecamtiv mecarbil in Figure 1. They do focus on the changes at 1 uM. The target range of concentration of omecamtiv mecarbil in humans is 300-750 ng/mL and the free fraction of OM is only 20%. So a plasma concentration of 400 ng/mL translates to a free concentration of 0.2 uM, 5x lower than 1 uM, where the effects are more modest and probably more clinically relevant. The authors should include some text in the results section about the magnitude of the effects in the clinically relevant range of omecamtiv mecarbil concentrations and not just focus on 1 uM.\nUnusual to reference to Amsterdam or Florence experiments. The paper’s results usually are described without reference to location.\nGenerally, the authors should try to simplify the text in the results section if possible. It is difficult to follow – a table of measured parameters would be helpful as mentioned above so that the reader can easily figure out the meaning of the many different variables used.\nDiscussion\nThe paper discussion should focus on how these findings contribute to our understanding how OM works in conditions that mimic living, beating cardiac muscle. The relevance of a mechanism of action for omecamtiv mecarbil described previously in the absence of phosphate should be questioned since there is always mM levels of phosphate in working muscle. The authors should compare/contrast their findings with prior findings that did not examine the effects of physiological phosphate and focus on what are the implications to the energetics of cardiac contraction in the presence of omecamtiv mecarbil.\nReviewer #3 (Remarks to the Author):\nThis contribution from the Lombardi group provides critical insights into the mechanism of Omecamtiv mercarbil (OM), an activator of cardiac myosin that is in Phase 3 clinical trials for heart failure. The data from this laboratory is as always of excellent quality - it combines sarcomere-level mechanics and ATPase measurements in single slow demembranated fibres from rabbit soleus.\nA major finding from this paper is the impact of Pi ions on OM-bound myosin heads. In the absence of Pi, OM leads to an increase in ATP consumption due to the inability of OM-bound heads to complete the force-generating power stroke, and thus detaching without completing the powerstoke but having consumed ATP. The addition of physiological levels of ATP reverses the excess ATP consumption and brings it back to the consumption in the absence of drug. This is of great importance since it shows a previously unappreciated effect of OM – its ability to reduce the energetic cost of systolic tension when Pi concentrations increases with heart-beat frequency. Without knowledge this Pi effect, it has been difficult to explain the impact of the drug on the working heart from the experimental data published to date.\nWhile the data is well presented and of very high quality, I have reservations concerning the modeling and its justification. The authors should at least indicate that the model proposed is only one possible model. There is for example no evidence that rebinding of Pi would lead to release of OM as indicated in step 14. In fact there is no basis for suggesting the transition (11) to a state that has a higher affinity for Pi exists. If it is based on other data in the literature, this should be cited. However, it seems simpler to propose that OM increases the occupancy of the state formed by Pi release, allowing exogenous Pi to increase the population of the same state that detaches in the model (green) in the absence of OM. But that OM inhibits its detachment and futile loss of ADP and Pi. In any event, it is important that the authors acknowledge that they have chosen one of the possible models to account for the data since controversy exists as to the states sensitive to Pi rebinding to the myosin head during the powerstroke.\nOverall, this is a very strong and well written paper that provides a critical new insight into how a cardiac/slow myosin stimulator can achieve positive results in the clinic. The data is state of the art and the findings will be of wide interest.\n", "role": "assistant" } ]	2,020
TrkA undergoes a tetramer-to-dimer conversion to open TrkH which enables changes in mem	REVIEWERS' COMMENTS: Reviewer #1 (Remarks to the Author): Zhang et al. provide a revised manuscript, which improved in readability and clarity (also about the limitations of their study). I am satisfied with most of their replies. The only remaining problem for me is the link to physiology. First, it is in no way new that potassium transporters and channels in general contribute to the establishment of the membrane potential. However, previous studies (general and on TrkHA/KtrBA) clearly showed the link between potassium accumulation and membrane potential adjustment (e.g. Kashket & Barker, 1977, Tokuda et al, 1981; Ochromel et al, 2011; Castaneda-Garcia et al, 2011), which Zhang et al. appear to question. But, their results in the changed membrane potential could as well reflect the disturbed potassium homeostasis upon abolishing the effective closure of TrkH. This said, I find the presented data with the given discussion, not further detailing what in their opinion happens other than K+ translocation, rather meaningless. The authors should either provide further data determining intracellular K+ and Na+ as well as the pH or remove the data. At least they should reference former papers and relate their suggestion to those publications. Also, I am not in favor of the changed title because ‘TrkA alters the membrane potential’ in my opinion is not even describing what the authors claim. I understand their argumentation rather as saying that the functional TrkHA system is involved in membrane potential adjustment. Additionally, there are several sentences that lack a word or have one too much as a result of changes to the manuscript (e.g. line 54., line 96, line 123, line 178). The authors should carefully revise their manuscript. Reviewer #2 (Remarks to the Author): I am satisfied with the changes made to the manuscript. I only suggest that the authors revise carefully the text and figures as I found a few obvious mistakes. For example: Supplementary Figure 5a: The two different interfaces are both labeled as HN1. I believe that the one on the right should be labeled HN2. Supplementary Figure 5b: One of the residues is labeled APS44, it should read ASP44. Legend of Supplementary Figure 11. The authors have chosen to use KtrBA instead of the established KtrAB. Although it is understandable to try to harmonize the nomenclature across the field (TrkHA and KtrBA), this may cause a confusion among the readers if they try to follow the literature. I suggest that either the authors use the established KtrAB or stick to the term KtrB- KtrA used in the main text.	REVIEWERS' COMMENTS: Reviewer #1 (Remarks to the Author): Zhang et al. provide a revised manuscript, which improved in readability and clarity (also about the limitations of their study). I am satisfied with most of their replies. The only remaining problem for me is the link to physiology. First, it is in no way new that potassium transporters and channels in general contribute to the establishment of the membrane potential. However, previous studies (general and on TrkHA/KtrBA) clearly showed the link between potassium accumulation and membrane potential adjustment (e.g. Kashket & Barker, 1977, Tokuda et al, 1981; Ochromel et al, 2011; Castaneda-Garcia et al, 2011), which Zhang et al. appear to question. But, their results in the changed membrane potential could as well reflect the disturbed potassium homeostasis upon abolishing the effective closure of TrkH. This said, I find the presented data with the given discussion, not further detailing what in their opinion happens other than K+ translocation, rather meaningless. The authors should either provide further data determining intracellular K+ and Na+ as well as the pH or remove the data. At least they should reference former papers and relate their suggestion to those publications. Also, I am not in favor of the changed title because ‘TrkA alters the membrane potential’ in my opinion is not even describing what the authors claim. I understand their argumentation rather as saying that the functional TrkHA system is involved in membrane potential adjustment. Additionally, there are several sentences that lack a word or have one too much as a result of changes to the manuscript (e.g. line 54., line 96, line 123, line 178). The authors should carefully revise their manuscript. Reviewer #2 (Remarks to the Author): I am satisfied with the changes made to the manuscript. I only suggest that the authors revise carefully the text and figures as I found a few obvious mistakes. For example: Supplementary Figure 5a: The two different interfaces are both labeled as HN1. I believe that the one on the right should be labeled HN2. Supplementary Figure 5b: One of the residues is labeled APS44, it should read ASP44. Legend of Supplementary Figure 11. The authors have chosen to use KtrBA instead of the established KtrAB. Although it is understandable to try to harmonize the nomenclature across the field (TrkHA and KtrBA), this may cause a confusion among the readers if they try to follow the literature. I suggest that either the authors use the established KtrAB or stick to the term KtrB- KtrA used in the main text.	TPR	NC_BS_BIOPH_003_RR2.txt	{ "criticism": 4, "example": 4, "importance_and_relevance": 0, "materials_and_methods": 4, "praise": 3, "presentation_and_reporting": 10, "results_and_discussion": 2, "suggestion_and_solution": 9, "total": 26 }	{ "criticism": 0.15384615384615385, "example": 0.15384615384615385, "importance_and_relevance": 0, "materials_and_methods": 0.15384615384615385, "praise": 0.11538461538461539, "presentation_and_reporting": 0.38461538461538464, "results_and_discussion": 0.07692307692307693, "suggestion_and_solution": 0.34615384615384615 }	1.384615	[ { "content": "Please review this paper.", "role": "user" }, { "content": "\n \n \n\n\fREVIEWERS' COMMENTS: \n \nReviewer #1 (Remarks to the Author): \n \nZhang et al. provide a revised manuscript, which improved in readability and clarity (also about \nthe limitations of their study). I am satisfied with most of their replies. The only remaining \nproblem for me is the link to physiology. First, it is in no way new that potassium transporters and \nchannels in general contribute to the establishment of the membrane potential. However, previous \nstudies (general and on TrkHA/KtrBA) clearly showed the link between potassium accumulation \nand membrane potential adjustment (e.g. Kashket & Barker, 1977, Tokuda et al, 1981; Ochromel \net al, 2011; Castaneda-Garcia et al, 2011), which Zhang et al. appear to question. But, their \nresults in the changed membrane potential could as well reflect the disturbed potassium \nhomeostasis upon abolishing the effective closure of TrkH. This said, I find the presented data with \nthe given discussion, not further detailing what in their opinion happens other than K+ \ntranslocation, rather meaningless. The authors should either provide further data determining \nintracellular K+ and Na+ as well as the pH or remove the data. At least they should reference \nformer papers and relate their suggestion to those publications. Also, I am not in favor of the \nchanged title because ‘TrkA alters the membrane potential’ in my opinion is not even describing \nwhat the authors claim. I understand their argumentation rather as saying that the functional \nTrkHA system is involved in membrane potential adjustment. \nAdditionally, there are several sentences that lack a word or have one too much as a result of \nchanges to the manuscript (e.g. line 54., line 96, line 123, line 178). The authors should carefully \nrevise their manuscript. \n \n \n \nReviewer #2 (Remarks to the Author): \n \nI am satisfied with the changes made to the manuscript. \n \nI only suggest that the authors revise carefully the text and figures as I found a few obvious \nmistakes. \nFor example: \nSupplementary Figure 5a: The two different interfaces are both labeled as HN1. I believe that the \none on the right should be labeled HN2. \n \nSupplementary Figure 5b: One of the residues is labeled APS44, it should read ASP44. \n \nLegend of Supplementary Figure 11. The authors have chosen to use KtrBA instead of the \nestablished KtrAB. Although it is understandable to try to harmonize the nomenclature across the \nfield (TrkHA and KtrBA), this may cause a confusion among the readers if they try to follow the \nliterature. I suggest that either the authors use the established KtrAB or stick to the term KtrB-\nKtrA used in the main text. \n \n \n\n\f", "role": "assistant" } ]	2,020
TrkA undergoes a tetramer-to-dimer conversion to open TrkH which enables changes in mem	Reviewers' comments: Reviewer #1 (Remarks to the Author): Based on three new structures of the TrkAH complex in the presence of ADP, ATP and ATPgammaS, respectively, Zhang et al suggest a novel mechanism of nucleotide-dependent gating of an RCK protein-regulated K+ channel. This is particularly interesting as for many of those channels a mechanism of gating has been suggested, but sufficient evidence from structural and functional experiments is missing. Here, for TrkAH Zhang et al. propose that ATP binding leads to the dissociation of the tetrameric TrkA ring into two TrkA dimers, which consequently opens the pore subunits by widening the pore-lining helices and displacing the intramembrane loop. In contrast, two interaction sites of TrkA and TrkH in the ADP-bound conformation stabilize the channel in a closed conformation. This new hypothesis will be of high interest for the K+ channel community and the membrane transport field in general. However, to me the manuscript needs a significant improvement in data representation to justify the drawn conclusions and additional functional experiments are required to prove the raised hypothesis. Most important are the following aspects: 1. For all relevant areas (HN1 interface, HN2 interface, intramembrane loop, selectivity filter, ligands) the density maps should be represented along with the models to visualize the reliability of the models. In particular, the authors mention that residues 155 – 173 of TrkH and 34 – 39 of TrkA in the ADP-bound state were not resolved (although at least residues 37 – 39 of TrkA are shown in figure 4 and – strangely – are helical and should be better resolved than the successive loop). Hence I wonder how well the hydrogen bond between Thr175 from TrkH and Gln40 from TrkA and the salt bridge between Arg177 from TrkH and Asp44 from TrkA are resolved. A clear HN1/HN2 figure at all three conditions (similar to Fig. 4a) with density maps is required in the main manuscript. Instead, I suggest to remove Fig. 4b as it is repetition of Fig. 1a. The distances as indicated in Fig. 4b could be indicated there. 2. In support of the hypothesized role of the HN1 interface the authors should provide a comparison with the available TrkH structure. How does the TrkH structure compare to the ATP- bound TrkAH structure? One would assume that the pore-lining helices adopt similar conformations because both are not restricted. The intramembrane loop on the other hand could adopt a different conformation because the “pulling element” is missing. Or is the assembled ring rather pushing on the intramembrane loop (cf. line 157)? How do the patch clamp measurements of the T175A mutant compare to TrkH in the absence of TrkA? 3. The authors suggest that N2 binds ADP, ATP and ATPgammaS, while N1 only binds ATP and ATPgammaS (lines 190ff). If so, this finding would finally provide an explanation on how TrkAH selectively binds ATP over ADP and why under physiological conditions activation could actually happen. However, the low resolution of N1 in the ATP-bound conformation did not allow resolving the bound ligand. Further, the binding affinities of the N1 and N2 sites for the different nucleotides are not known. The manuscript lacks an explanation on why ADP would not bind to N1. I suggest the performing of ITC measurements to determine N values and binding affinities for the different nucleotides. The R100A mutant likely would even allow the determination of affinities for the N2 site only, while a similar mutation in the N2 site would allow the determination of N1 binding affinities. In fact, those experiments could be performed on the TrkA subunit only to further simplify the experiments. 4. Further, one should ask which role nucleotide binding to the N2 site has. Here, a mutation similar to the R100A mutation in N1 should be inserted to hinder ATP binding to the N2 site. 5. There is no evidence from any functional data for the suggested role of TrkAH (lines 297ff.). Further, to me it does not sound very plausible. Why would it be useful for the cell to pump even more protons, if the ATP concentration is already high? This in consequence would actually lead to a drastic alkalization of the cell, which most likely would be problematic. In contrast, TrkAH and KtrAB have been suggested to be involved in pH regulation. I suggest to remove this far-fetched hypothesis on the physiological role of TrkAH or to develop it better. In addition to these five main points I have several smaller comments/questions: 6. Line 10: Explain RCK 7. Lines 116f.: It is known that at least in Gram-positive bacteria c-di-AMP binding to RCK-C closes TrkH. Thus the speculative sentence about C1 and C2 should be removed. 8. The order of the figures in the text (1,5,4,3,2,6,7) is very confusing 9. Figure S3, panel d: The resolution of the map is impossible to judge. Show a slice through the map. Panel i: Show densities of important areas (intramembrane loop, selectivity filter, AN1, AN2, ligands) and not of any random area. 10. I wonder whether you observed the dissociation of the tetrameic TrkA when adding ATP to isolated TrkA. This could be easily analysed by negative-staining EM. 11. How confident are the authors that the K+ is not yet rehydrated when passing the intramembrane loop (lines 147ff)? Is the opening of the pore seen really sufficient to allow K+ translocation 12. Which effect could the sandwich conformation have on the observed conformational changes? Might it hinder a complete dissociation of TrkA and TrkH? 13. Line 177: Specify the term “intermediate” 14. Lines 202f: “feeble HN1 interface”. Show interaction with density map! Figure S6e only shows the TrkH subunit, which doesn’t help the interpretation. 15. HN1 interface: What happens if you enforce the interface by a disulfide bond? Is it ATP insensitive? 16. What is the role of the relative rotation of the TrkH protomers? 17. The cartoon in Fig. 7 is very difficult to understand. Breaking of tetramer into dimers is not clear, HN1 and HN2 sites should be indicated. 18. Several panels particularly in the supplements are invisible/unreadable (e.g. I220C in S5, resolution map in S3d) Reviewer #2 (Remarks to the Author): The manuscript entitled “Opening of the TrkH channel by a tetramer-to-dimer conversion of the associated TrkA” describes structures of the TrkHA complex. This is a bacterial cation channel which belongs to the SKT superfamily of channels and transporters with important roles in K+ homeostasis of bacteria, archaea, fungi and plants. The manuscript presents three different conformational states of the TrkHA complex, providing a unique view of its mechanism of activation. Crucially, it provides the first description of the open state of a SKT channel. This manuscript will be of interest for all of those that are interested in ion channel structure and function and also for those interested in bacterial physiology. Major issues: 1- A conclusion from this paper relates to the impact of ligand binding to the different sites (RCK1 and RCK2). In particular, ADP binds just to RCK2 while ATPγS binds to both RCK1 and RCK2. The authors reasonably conclude that ATP will likely bind to RCK1 and RCK2. The importance of this conclusion must rely on solid data and it is surprising that the conformation of ATPγS in RCK2 in the new structure at 3.8Å is different from the conformation of the same ligand in a previous structure at 3.0Å. These differences occur at the level of the position by the base (the base in the new structure is flipped relative to the previous structure) and at the level of the interactions established by the phosphates (see description in lines 184 to 190 and confront with previous structure, PDB code: 4J9V). Are the authors sure that the low resolution of the new structure allows them to make these changes in the ligand? In addition, I would strongly suggest that the panels in Figure 5 show the protein structures in identical views to facilitate a comparison between the different ligands and binding sites. 2- I would like for the authors to consider that there are other structural regions that have a role in the closure/opening of TrkH besides the HN1 interface. After all their functional data shows that mutations at the interface still allow gating since there is still a large reduction in open probability (from 80% open with ATP to ~13% open with ADP versus in the WT where it changed from 87% to 2%). In addition, a) the authors describe that changes in the intramembrane loop are connected with an increase in the distance separating the selectivity filter from the HN2 interface (Lines 140 to 145). By the way, this change is not explained; and b) that the helix D3M2b helix, which is directly connected to the intramembrane loop, has moved in the ATP versus the ADP structure (Figure 2d). Without minimizing the importance of the HN1 contact for the stability of the closed state, why are the authors ignoring all these other changes as part of the overall mechanism of moving the intramembrane loop out of the way? 3- I was surprised not to see more information about the ion pore of the "partially" open TrkHA- ATP structure. In particular, the authors have decided not to show any detailed information about the intramembrane loop (both the structure and the experimental map). It seems to me that the functional importance of this region and good resolution of the TrkHA-ATP structure merits at least a figure showing the intramembrane loop structure and the experimental map. 4- One of the major issues I have found with the manuscript is its organization. The main text does not follow the organization of the figures and this means that the data is spread-out across different figures, both main and supplementary, making reading the text and looking/thinking about the data a fairly cumbersome process. Without being to bullish with this request, I would like to suggest that the authors consider a reorganization of the main text so that the manuscript starts with a description of the major differences between the ADP and ATP structures, followed by a description of the details that underlie those major differences and that support their mechanism proposal, including the presentation of the ATPγS structure. I am convinced that this will increase the readability of the manuscript and its impact. Not sure as well that the comparison of the new structures with the previous TrkHA-NADH structure should take such a prominent place in the early sections of the manuscript. It takes a whole paragraph in the section describing the TrkHA-ADP structure and also in the final paragraph of the description of TrkHA-ATP structure and becomes a distraction to the understanding of the new data. I suggest that you discuss the differences between the new structures and the TrkHA- NADH structure in a single paragraph later in the manuscript. Minor issues: 1- The size of some of the Figures is too small. For example, all the panels in Supp. Figure 3 and Supp. Figures 4g and 4 h are virtually impossible to analyze even after extensive zooming. In addition, labels in Figures 4a, and Figure 5 are also too small. 2- Resolution of left-hand side panels in Figure 5 for ADP and ATPγS is poor. 3- Supplementary Figure 1c serves no apparent purpose and is very difficult to analyze due to lack of contrast between ligand and protein. 4- There is something wrong with Stereo figures in Supp. Figure 4 g and h. It appears that the panels are inverted. 5- In line 114, the differences in the N2-N2 interface are mentioned when discussing changes between the TrkHA-ADP and TrkHA-NADH structures and Supp. Figure 2F is referenced. However, the figure only shows the N2-N2 interface of the TrkHA-NADH structure and so these differences are not easy to perceive. 6- Not clear what is being presented in Supp. Figures 4a and 4b, they appear to represent the two N2-N2 interfaces from the isolated TrkA-ATPγS structure, but why do the interfaces look different? 7- The PDB code indicated in panel Supp. Figure 4e is wrong, the TrkA-ATPγS PDB code is not 4J9U, it is 4J9V. 8- The Supp. Figure 4F panels are labeled as interfaces when they should be labeled “binding sites”. 9- In Supp. Figure 6 the figures are labeled “Supperimposed” when it should read ”Superimposed”. 10- In Suppl. Figure 6e, the contrast between D1D2 loop residues and the rest of structure is poor, making analysis of the structure difficult. 11- Supplemental Table 1 should show Ramachandran plot distribution of residues as Favored, Allowed and Disallowed as this is a refinement independent parameter. Reviewer #3 (Remarks to the Author): The authors studied the structures of TrkH-TrkA complex in the presence of ADP or ATP. The structures of TrkHA-ADP and TrkHA-ATPS were solved by X-ray crystallography. TrkH in the TrkHA-ADP is in a closed state. The structure of TrkHA-ATP was solved by cryo-EM. TrkH in the TrkHA-ATP is in an open state. However, the N1 domain is partially resolved and the C1 and C2 domains of TrkA are not resolved at all. The major conformational change is the conversion of a tetramer to a dimer of TrkA and a loss of constraints on TrkH. Although quite a few local structures are presented, the conversion is not clear. It is better to provide a few top views of the TM region and the intracellular region separately, and label some of the dimensions. In that way, the tetramer to dimer conversion can be seen easily. Also, comparison with the published NADH bound TrkA in top view will be helpful. For local structures, it is better to label them in the top views of corresponding regions. It might be better to re-organize the figures to focus on the conversion from tetramer to dimer. Here are some detailed comments. • In the introduction, it is better to give more information about the difference between TrkH and KtrB, and TrkA and KtrA. • Fig 1: Are A and B in the same orientation? • Fig 2C: It is better to replace yellow with another color. It is easy to confuse it with the yellow color in A. • Fig S2A: Is there any reason to show D2, D4, D1, and D3 instead of D1-4? • Page 6: "Second, the two TrkH protomers have a relative rotation of 12.5 (Supplementary Figures 6a and b)." Are the two protomers tilted in the plane perpendicular to the membrane or rotated in the plane parallel to the membrane? • Page 6: "The N1-N1 interface remains unchanged while the N2-N2 interface adjusts to accommodate the changes (Supplementary Figure 2f)." Fig s2f doesn't show the difference or similarity at all. • Page 7: "In the presence of ATP, TrkAs become two dimers and no longer form a tetrameric ring (Figure 3a and b).". It is really hard to see the dimers. A top view of the TrkAs will be helpful. A comparison of the whole intracellular dimers/rings will be needed. • Page 25: "TrkH-TrkA complex was assembled by mixing TrkH and TrkA in 1.2 to 1 molar ratio." Based on the structure, the molar ratio is 1 TrkH to 2 TrkA. Why was 1.2 to 1 molar ratio used? • Page 27: ". Each micrograph was collected in 5 frames/s for 10s, with a dose of 8e-/Å/s". Should it be 8e-/pixel/s? • Page 27: "These particles were reconstituted into a 6 Å map through RELION 3D refinement". Shouldn't it be reconstructed instead of reconstituted? • Page 27: Was any symmetry used in 3D reconstruction?	Reviewers' comments: Reviewer #1 (Remarks to the Author): Based on three new structures of the TrkAH complex in the presence of ADP, ATP and ATPgammaS, respectively, Zhang et al suggest a novel mechanism of nucleotide-dependent gating of an RCK protein-regulated K+ channel. This is particularly interesting as for many of those channels a mechanism of gating has been suggested, but sufficient evidence from structural and functional experiments is missing. Here, for TrkAH Zhang et al. propose that ATP binding leads to the dissociation of the tetrameric TrkA ring into two TrkA dimers, which consequently opens the pore subunits by widening the pore-lining helices and displacing the intramembrane loop. In contrast, two interaction sites of TrkA and TrkH in the ADP-bound conformation stabilize the channel in a closed conformation. This new hypothesis will be of high interest for the K+ channel community and the membrane transport field in general. However, to me the manuscript needs a significant improvement in data representation to justify the drawn conclusions and additional functional experiments are required to prove the raised hypothesis. Most important are the following aspects: 1. For all relevant areas (HN1 interface, HN2 interface, intramembrane loop, selectivity filter, ligands) the density maps should be represented along with the models to visualize the reliability of the models. In particular, the authors mention that residues 155 – 173 of TrkH and 34 – 39 of TrkA in the ADP-bound state were not resolved (although at least residues 37 – 39 of TrkA are shown in figure 4 and – strangely – are helical and should be better resolved than the successive loop). Hence I wonder how well the hydrogen bond between Thr175 from TrkH and Gln40 from TrkA and the salt bridge between Arg177 from TrkH and Asp44 from TrkA are resolved. A clear HN1/HN2 figure at all three conditions (similar to Fig. 4a) with density maps is required in the main manuscript. Instead, I suggest to remove Fig. 4b as it is repetition of Fig. 1a. The distances as indicated in Fig. 4b could be indicated there. 2. In support of the hypothesized role of the HN1 interface the authors should provide a comparison with the available TrkH structure. How does the TrkH structure compare to the ATP- bound TrkAH structure? One would assume that the pore-lining helices adopt similar conformations because both are not restricted. The intramembrane loop on the other hand could adopt a different conformation because the “pulling element” is missing. Or is the assembled ring rather pushing on the intramembrane loop (cf. line 157)? How do the patch clamp measurements of the T175A mutant compare to TrkH in the absence of TrkA? 3. The authors suggest that N2 binds ADP, ATP and ATPgammaS, while N1 only binds ATP and ATPgammaS (lines 190ff). If so, this finding would finally provide an explanation on how TrkAH selectively binds ATP over ADP and why under physiological conditions activation could actually happen. However, the low resolution of N1 in the ATP-bound conformation did not allow resolving the bound ligand. Further, the binding affinities of the N1 and N2 sites for the different nucleotides are not known. The manuscript lacks an explanation on why ADP would not bind to N1. I suggest the performing of ITC measurements to determine N values and binding affinities for the different nucleotides. The R100A mutant likely would even allow the determination of affinities for the N2 site only, while a similar mutation in the N2 site would allow the determination of N1 binding affinities. In fact, those experiments could be performed on the TrkA subunit only to further simplify the experiments. 4. Further, one should ask which role nucleotide binding to the N2 site has. Here, a mutation similar to the R100A mutation in N1 should be inserted to hinder ATP binding to the N2 site. 5. There is no evidence from any functional data for the suggested role of TrkAH (lines 297ff.). Further, to me it does not sound very plausible. Why would it be useful for the cell to pump even more protons, if the ATP concentration is already high? This in consequence would actually lead to a drastic alkalization of the cell, which most likely would be problematic. In contrast, TrkAH and KtrAB have been suggested to be involved in pH regulation. I suggest to remove this far-fetched hypothesis on the physiological role of TrkAH or to develop it better. In addition to these five main points I have several smaller comments/questions: 6. Line 10: Explain RCK 7. Lines 116f.: It is known that at least in Gram-positive bacteria c-di-AMP binding to RCK-C closes TrkH. Thus the speculative sentence about C1 and C2 should be removed. 8. The order of the figures in the text (1,5,4,3,2,6,7) is very confusing 9. Figure S3, panel d: The resolution of the map is impossible to judge. Show a slice through the map. Panel i: Show densities of important areas (intramembrane loop, selectivity filter, AN1, AN2, ligands) and not of any random area. 10. I wonder whether you observed the dissociation of the tetrameic TrkA when adding ATP to isolated TrkA. This could be easily analysed by negative-staining EM. 11. How confident are the authors that the K+ is not yet rehydrated when passing the intramembrane loop (lines 147ff)? Is the opening of the pore seen really sufficient to allow K+ translocation 12. Which effect could the sandwich conformation have on the observed conformational changes? Might it hinder a complete dissociation of TrkA and TrkH? 13. Line 177: Specify the term “intermediate” 14. Lines 202f: “feeble HN1 interface”. Show interaction with density map! Figure S6e only shows the TrkH subunit, which doesn’t help the interpretation. 15. HN1 interface: What happens if you enforce the interface by a disulfide bond? Is it ATP insensitive? 16. What is the role of the relative rotation of the TrkH protomers? 17. The cartoon in Fig. 7 is very difficult to understand. Breaking of tetramer into dimers is not clear, HN1 and HN2 sites should be indicated. 18. Several panels particularly in the supplements are invisible/unreadable (e.g. I220C in S5, resolution map in S3d) Reviewer #2 (Remarks to the Author): The manuscript entitled “Opening of the TrkH channel by a tetramer-to-dimer conversion of the associated TrkA” describes structures of the TrkHA complex. This is a bacterial cation channel which belongs to the SKT superfamily of channels and transporters with important roles in K+ homeostasis of bacteria, archaea, fungi and plants. The manuscript presents three different conformational states of the TrkHA complex, providing a unique view of its mechanism of activation. Crucially, it provides the first description of the open state of a SKT channel. This manuscript will be of interest for all of those that are interested in ion channel structure and function and also for those interested in bacterial physiology. Major issues: 1- A conclusion from this paper relates to the impact of ligand binding to the different sites (RCK1 and RCK2). In particular, ADP binds just to RCK2 while ATPγS binds to both RCK1 and RCK2. The authors reasonably conclude that ATP will likely bind to RCK1 and RCK2. The importance of this conclusion must rely on solid data and it is surprising that the conformation of ATPγS in RCK2 in the new structure at 3.8Å is different from the conformation of the same ligand in a previous structure at 3.0Å. These differences occur at the level of the position by the base (the base in the new structure is flipped relative to the previous structure) and at the level of the interactions established by the phosphates (see description in lines 184 to 190 and confront with previous structure, PDB code: 4J9V). Are the authors sure that the low resolution of the new structure allows them to make these changes in the ligand? In addition, I would strongly suggest that the panels in Figure 5 show the protein structures in identical views to facilitate a comparison between the different ligands and binding sites. 2- I would like for the authors to consider that there are other structural regions that have a role in the closure/opening of TrkH besides the HN1 interface. After all their functional data shows that mutations at the interface still allow gating since there is still a large reduction in open probability (from 80% open with ATP to ~13% open with ADP versus in the WT where it changed from 87% to 2%). In addition, a) the authors describe that changes in the intramembrane loop are connected with an increase in the distance separating the selectivity filter from the HN2 interface (Lines 140 to 145). By the way, this change is not explained; and b) that the helix D3M2b helix, which is directly connected to the intramembrane loop, has moved in the ATP versus the ADP structure (Figure 2d). Without minimizing the importance of the HN1 contact for the stability of the closed state, why are the authors ignoring all these other changes as part of the overall mechanism of moving the intramembrane loop out of the way? 3- I was surprised not to see more information about the ion pore of the "partially" open TrkHA- ATP structure. In particular, the authors have decided not to show any detailed information about the intramembrane loop (both the structure and the experimental map). It seems to me that the functional importance of this region and good resolution of the TrkHA-ATP structure merits at least a figure showing the intramembrane loop structure and the experimental map. 4- One of the major issues I have found with the manuscript is its organization. The main text does not follow the organization of the figures and this means that the data is spread-out across different figures, both main and supplementary, making reading the text and looking/thinking about the data a fairly cumbersome process. Without being to bullish with this request, I would like to suggest that the authors consider a reorganization of the main text so that the manuscript starts with a description of the major differences between the ADP and ATP structures, followed by a description of the details that underlie those major differences and that support their mechanism proposal, including the presentation of the ATPγS structure. I am convinced that this will increase the readability of the manuscript and its impact. Not sure as well that the comparison of the new structures with the previous TrkHA-NADH structure should take such a prominent place in the early sections of the manuscript. It takes a whole paragraph in the section describing the TrkHA-ADP structure and also in the final paragraph of the description of TrkHA-ATP structure and becomes a distraction to the understanding of the new data. I suggest that you discuss the differences between the new structures and the TrkHA- NADH structure in a single paragraph later in the manuscript. Minor issues: 1- The size of some of the Figures is too small. For example, all the panels in Supp. Figure 3 and Supp. Figures 4g and 4 h are virtually impossible to analyze even after extensive zooming. In addition, labels in Figures 4a, and Figure 5 are also too small. 2- Resolution of left-hand side panels in Figure 5 for ADP and ATPγS is poor. 3- Supplementary Figure 1c serves no apparent purpose and is very difficult to analyze due to lack of contrast between ligand and protein. 4- There is something wrong with Stereo figures in Supp. Figure 4 g and h. It appears that the panels are inverted. 5- In line 114, the differences in the N2-N2 interface are mentioned when discussing changes between the TrkHA-ADP and TrkHA-NADH structures and Supp. Figure 2F is referenced. However, the figure only shows the N2-N2 interface of the TrkHA-NADH structure and so these differences are not easy to perceive. 6- Not clear what is being presented in Supp. Figures 4a and 4b, they appear to represent the two N2-N2 interfaces from the isolated TrkA-ATPγS structure, but why do the interfaces look different? 7- The PDB code indicated in panel Supp. Figure 4e is wrong, the TrkA-ATPγS PDB code is not 4J9U, it is 4J9V. 8- The Supp. Figure 4F panels are labeled as interfaces when they should be labeled “binding sites”. 9- In Supp. Figure 6 the figures are labeled “Supperimposed” when it should read ”Superimposed”. 10- In Suppl. Figure 6e, the contrast between D1D2 loop residues and the rest of structure is poor, making analysis of the structure difficult. 11- Supplemental Table 1 should show Ramachandran plot distribution of residues as Favored, Allowed and Disallowed as this is a refinement independent parameter. Reviewer #3 (Remarks to the Author): The authors studied the structures of TrkH-TrkA complex in the presence of ADP or ATP. The structures of TrkHA-ADP and TrkHA-ATPS were solved by X-ray crystallography. TrkH in the TrkHA-ADP is in a closed state. The structure of TrkHA-ATP was solved by cryo-EM. TrkH in the TrkHA-ATP is in an open state. However, the N1 domain is partially resolved and the C1 and C2 domains of TrkA are not resolved at all. The major conformational change is the conversion of a tetramer to a dimer of TrkA and a loss of constraints on TrkH. Although quite a few local structures are presented, the conversion is not clear. It is better to provide a few top views of the TM region and the intracellular region separately, and label some of the dimensions. In that way, the tetramer to dimer conversion can be seen easily. Also, comparison with the published NADH bound TrkA in top view will be helpful. For local structures, it is better to label them in the top views of corresponding regions. It might be better to re-organize the figures to focus on the conversion from tetramer to dimer. Here are some detailed comments. • In the introduction, it is better to give more information about the difference between TrkH and KtrB, and TrkA and KtrA. • Fig 1: Are A and B in the same orientation? • Fig 2C: It is better to replace yellow with another color. It is easy to confuse it with the yellow color in A. • Fig S2A: Is there any reason to show D2, D4, D1, and D3 instead of D1-4? • Page 6: "Second, the two TrkH protomers have a relative rotation of 12.5 (Supplementary Figures 6a and b)." Are the two protomers tilted in the plane perpendicular to the membrane or rotated in the plane parallel to the membrane? • Page 6: "The N1-N1 interface remains unchanged while the N2-N2 interface adjusts to accommodate the changes (Supplementary Figure 2f)." Fig s2f doesn't show the difference or similarity at all. • Page 7: "In the presence of ATP, TrkAs become two dimers and no longer form a tetrameric ring (Figure 3a and b).". It is really hard to see the dimers. A top view of the TrkAs will be helpful. A comparison of the whole intracellular dimers/rings will be needed. • Page 25: "TrkH-TrkA complex was assembled by mixing TrkH and TrkA in 1.2 to 1 molar ratio." Based on the structure, the molar ratio is 1 TrkH to 2 TrkA. Why was 1.2 to 1 molar ratio used? • Page 27: ". Each micrograph was collected in 5 frames/s for 10s, with a dose of 8e-/Å/s". Should it be 8e-/pixel/s? • Page 27: "These particles were reconstituted into a 6 Å map through RELION 3D refinement". Shouldn't it be reconstructed instead of reconstituted? • Page 27: Was any symmetry used in 3D reconstruction?	TPR	NC_BS_BIOPH_003_RR1.txt	{ "criticism": 33, "example": 43, "importance_and_relevance": 8, "materials_and_methods": 53, "praise": 3, "presentation_and_reporting": 46, "results_and_discussion": 8, "suggestion_and_solution": 37, "total": 165 }	{ "criticism": 0.2, "example": 0.2606060606060606, "importance_and_relevance": 0.048484848484848485, "materials_and_methods": 0.3212121212121212, "praise": 0.01818181818181818, "presentation_and_reporting": 0.2787878787878788, "results_and_discussion": 0.048484848484848485, "suggestion_and_solution": 0.22424242424242424 }	1.4	[ { "content": "Please review this paper.", "role": "user" }, { "content": "Reviewers' comments: \n \nReviewer #1 (Remarks to the Author): \n \nBased on three new structures of the TrkAH complex in the presence of ADP, ATP and \nATPgammaS, respectively, Zhang et al suggest a novel mechanism of nucleotide-dependent gating \nof an RCK protein-regulated K+ channel. This is particularly interesting as for many of those \nchannels a mechanism of gating has been suggested, but sufficient evidence from structural and \nfunctional experiments is missing. Here, for TrkAH Zhang et al. propose that ATP binding leads to \nthe dissociation of the tetrameric TrkA ring into two TrkA dimers, which consequently opens the \npore subunits by widening the pore-lining helices and displacing the intramembrane loop. In \ncontrast, two interaction sites of TrkA and TrkH in the ADP-bound conformation stabilize the \nchannel in a closed conformation. This new hypothesis will be of high interest for the K+ channel \ncommunity and the membrane transport field in general. However, to me the manuscript needs a \nsignificant improvement in data representation to justify the drawn conclusions and additional \nfunctional experiments are required to prove the raised hypothesis. Most important are the \nfollowing aspects: \n \n1. For all relevant areas (HN1 interface, HN2 interface, intramembrane loop, selectivity filter, \nligands) the density maps should be represented along with the models to visualize the reliability \nof the models. In particular, the authors mention that residues 155 – 173 of TrkH and 34 – 39 of \nTrkA in the ADP-bound state were not resolved (although at least residues 37 – 39 of TrkA are \nshown in figure 4 and – strangely – are helical and should be better resolved than the successive \nloop). Hence I wonder how well the hydrogen bond between Thr175 from TrkH and Gln40 from \nTrkA and the salt bridge between Arg177 from TrkH and Asp44 from TrkA are resolved. \nA clear HN1/HN2 figure at all three conditions (similar to Fig. 4a) with density maps is required in \nthe main manuscript. Instead, I suggest to remove Fig. 4b as it is repetition of Fig. 1a. The \ndistances as indicated in Fig. 4b could be indicated there. \n \n2. In support of the hypothesized role of the HN1 interface the authors should provide a \ncomparison with the available TrkH structure. How does the TrkH structure compare to the ATP-\nbound TrkAH structure? One would assume that the pore-lining helices adopt similar conformations \nbecause both are not restricted. The intramembrane loop on the other hand could adopt a different \nconformation because the “pulling element” is missing. Or is the assembled ring rather pushing on \nthe intramembrane loop (cf. line 157)? How do the patch clamp measurements of the T175A \nmutant compare to TrkH in the absence of TrkA? \n \n3. The authors suggest that N2 binds ADP, ATP and ATPgammaS, while N1 only binds ATP and \nATPgammaS (lines 190ff). If so, this finding would finally provide an explanation on how TrkAH \nselectively binds ATP over ADP and why under physiological conditions activation could actually \nhappen. However, the low resolution of N1 in the ATP-bound conformation did not allow resolving \nthe bound ligand. Further, the binding affinities of the N1 and N2 sites for the different nucleotides \nare not known. The manuscript lacks an explanation on why ADP would not bind to N1. I suggest \nthe performing of ITC measurements to determine N values and binding affinities for the different \nnucleotides. The R100A mutant likely would even allow the determination of affinities for the N2 \nsite only, while a similar mutation in the N2 site would allow the determination of N1 binding \naffinities. In fact, those experiments could be performed on the TrkA subunit only to further \nsimplify the experiments. \n \n4. Further, one should ask which role nucleotide binding to the N2 site has. Here, a mutation \nsimilar to the R100A mutation in N1 should be inserted to hinder ATP binding to the N2 site. \n \n5. There is no evidence from any functional data for the suggested role of TrkAH (lines 297ff.). \nFurther, to me it does not sound very plausible. Why would it be useful for the cell to pump even \nmore protons, if the ATP concentration is already high? This in consequence would actually lead to \na drastic alkalization of the cell, which most likely would be problematic. In contrast, TrkAH and \nKtrAB have been suggested to be involved in pH regulation. I suggest to remove this far-fetched \nhypothesis on the physiological role of TrkAH or to develop it better. \n \n\n\fIn addition to these five main points I have several smaller comments/questions: \n \n6. Line 10: Explain RCK \n \n7. Lines 116f.: It is known that at least in Gram-positive bacteria c-di-AMP binding to RCK-C closes \nTrkH. Thus the speculative sentence about C1 and C2 should be removed. \n \n8. The order of the figures in the text (1,5,4,3,2,6,7) is very confusing \n \n9. Figure S3, panel d: The resolution of the map is impossible to judge. Show a slice through the \nmap. Panel i: Show densities of important areas (intramembrane loop, selectivity filter, AN1, AN2, \nligands) and not of any random area. \n \n10. I wonder whether you observed the dissociation of the tetrameic TrkA when adding ATP to \nisolated TrkA. This could be easily analysed by negative-staining EM. \n \n11. How confident are the authors that the K+ is not yet rehydrated when passing the \nintramembrane loop (lines 147ff)? Is the opening of the pore seen really sufficient to allow K+ \ntranslocation \n \n12. Which effect could the sandwich conformation have on the observed conformational changes? \nMight it hinder a complete dissociation of TrkA and TrkH? \n \n13. Line 177: Specify the term “intermediate” \n \n14. Lines 202f: “feeble HN1 interface”. Show interaction with density map! Figure S6e only shows \nthe TrkH subunit, which doesn’t help the interpretation. \n \n15. HN1 interface: What happens if you enforce the interface by a disulfide bond? Is it ATP \ninsensitive? \n \n16. What is the role of the relative rotation of the TrkH protomers? \n \n17. The cartoon in Fig. 7 is very difficult to understand. Breaking of tetramer into dimers is not \nclear, HN1 and HN2 sites should be indicated. \n \n18. Several panels particularly in the supplements are invisible/unreadable (e.g. I220C in S5, \nresolution map in S3d) \n \n \n \nReviewer #2 (Remarks to the Author): \n \nThe manuscript entitled “Opening of the TrkH channel by a tetramer-to-dimer conversion of the \nassociated TrkA” describes structures of the TrkHA complex. This is a bacterial cation channel \nwhich belongs to the SKT superfamily of channels and transporters with important roles in K+ \nhomeostasis of bacteria, archaea, fungi and plants. The manuscript presents three different \nconformational states of the TrkHA complex, providing a unique view of its mechanism of \nactivation. Crucially, it provides the first description of the open state of a SKT channel. This \nmanuscript will be of interest for all of those that are interested in ion channel structure and \nfunction and also for those interested in bacterial physiology. \n \nMajor issues: \n1- A conclusion from this paper relates to the impact of ligand binding to the different sites (RCK1 \nand RCK2). In particular, ADP binds just to RCK2 while ATPγS binds to both RCK1 and RCK2. The \nauthors reasonably conclude that ATP will likely bind to RCK1 and RCK2. The importance of this \nconclusion must rely on solid data and it is surprising that the conformation of ATPγS in RCK2 in \nthe new structure at 3.8Å is different from the conformation of the same ligand in a previous \nstructure at 3.0Å. These differences occur at the level of the position by the base (the base in the \n\n\fnew structure is flipped relative to the previous structure) and at the level of the interactions \nestablished by the phosphates (see description in lines 184 to 190 and confront with previous \nstructure, PDB code: 4J9V). Are the authors sure that the low resolution of the new structure \nallows them to make these changes in the ligand? In addition, I would strongly suggest that the \npanels in Figure 5 show the protein structures in identical views to facilitate a comparison between \nthe different ligands and binding sites. \n2- I would like for the authors to consider that there are other structural regions that have a role \nin the closure/opening of TrkH besides the HN1 interface. After all their functional data shows that \nmutations at the interface still allow gating since there is still a large reduction in open probability \n(from 80% open with ATP to ~13% open with ADP versus in the WT where it changed from 87% \nto 2%). In addition, a) the authors describe that changes in the intramembrane loop are \nconnected with an increase in the distance separating the selectivity filter from the HN2 interface \n(Lines 140 to 145). By the way, this change is not explained; and b) that the helix D3M2b helix, \nwhich is directly connected to the intramembrane loop, has moved in the ATP versus the ADP \nstructure (Figure 2d). \nWithout minimizing the importance of the HN1 contact for the stability of the closed state, why are \nthe authors ignoring all these other changes as part of the overall mechanism of moving the \nintramembrane loop out of the way? \n3- I was surprised not to see more information about the ion pore of the \"partially\" open TrkHA-\nATP structure. In particular, the authors have decided not to show any detailed information about \nthe intramembrane loop (both the structure and the experimental map). It seems to me that the \nfunctional importance of this region and good resolution of the TrkHA-ATP structure merits at least \na figure showing the intramembrane loop structure and the experimental map. \n4- One of the major issues I have found with the manuscript is its organization. The main text \ndoes not follow the organization of the figures and this means that the data is spread-out across \ndifferent figures, both main and supplementary, making reading the text and looking/thinking \nabout the data a fairly cumbersome process. \nWithout being to bullish with this request, I would like to suggest that the authors consider a \nreorganization of the main text so that the manuscript starts with a description of the major \ndifferences between the ADP and ATP structures, followed by a description of the details that \nunderlie those major differences and that support their mechanism proposal, including the \npresentation of the ATPγS structure. I am convinced that this will increase the readability of the \nmanuscript and its impact. \nNot sure as well that the comparison of the new structures with the previous TrkHA-NADH \nstructure should take such a prominent place in the early sections of the manuscript. It takes a \nwhole paragraph in the section describing the TrkHA-ADP structure and also in the final paragraph \nof the description of TrkHA-ATP structure and becomes a distraction to the understanding of the \nnew data. I suggest that you discuss the differences between the new structures and the TrkHA-\nNADH structure in a single paragraph later in the manuscript. \n \nMinor issues: \n1- The size of some of the Figures is too small. For example, all the panels in Supp. Figure 3 and \nSupp. Figures 4g and 4 h are virtually impossible to analyze even after extensive zooming. In \naddition, labels in Figures 4a, and Figure 5 are also too small. \n2- Resolution of left-hand side panels in Figure 5 for ADP and ATPγS is poor. \n3- Supplementary Figure 1c serves no apparent purpose and is very difficult to analyze due to lack \nof contrast between ligand and protein. \n4- There is something wrong with Stereo figures in Supp. Figure 4 g and h. It appears that the \npanels are inverted. \n5- In line 114, the differences in the N2-N2 interface are mentioned when discussing changes \nbetween the TrkHA-ADP and TrkHA-NADH structures and Supp. Figure 2F is referenced. However, \nthe figure only shows the N2-N2 interface of the TrkHA-NADH structure and so these differences \nare not easy to perceive. \n6- Not clear what is being presented in Supp. Figures 4a and 4b, they appear to represent the two \nN2-N2 interfaces from the isolated TrkA-ATPγS structure, but why do the interfaces look different? \n7- The PDB code indicated in panel Supp. Figure 4e is wrong, the TrkA-ATPγS PDB code is not \n4J9U, it is 4J9V. \n8- The Supp. Figure 4F panels are labeled as interfaces when they should be labeled “binding \nsites”. \n\n\f9- In Supp. Figure 6 the figures are labeled “Supperimposed” when it should read ”Superimposed”. \n10- In Suppl. Figure 6e, the contrast between D1D2 loop residues and the rest of structure is poor, \nmaking analysis of the structure difficult. \n11- Supplemental Table 1 should show Ramachandran plot distribution of residues as Favored, \nAllowed and Disallowed as this is a refinement independent parameter. \n \n \n \n \nReviewer #3 (Remarks to the Author): \n \nThe authors studied the structures of TrkH-TrkA complex in the presence of ADP or ATP. The \nstructures of TrkHA-ADP and TrkHA-ATPS were solved by X-ray crystallography. TrkH in the \nTrkHA-ADP is in a closed state. The structure of TrkHA-ATP was solved by cryo-EM. TrkH in the \nTrkHA-ATP is in an open state. However, the N1 domain is partially resolved and the C1 and C2 \ndomains of TrkA are not resolved at all. The major conformational change is the conversion of a \ntetramer to a dimer of TrkA and a loss of constraints on TrkH. Although quite a few local structures \nare presented, the conversion is not clear. It is better to provide a few top views of the TM region \nand the intracellular region separately, and label some of the dimensions. In that way, the \ntetramer to dimer conversion can be seen easily. Also, comparison with the published NADH bound \nTrkA in top view will be helpful. For local structures, it is better to label them in the top views of \ncorresponding regions. It might be better to re-organize the figures to focus on the conversion \nfrom tetramer to dimer. \n \nHere are some detailed comments. \n• In the introduction, it is better to give more information about the difference between TrkH and \nKtrB, and TrkA and KtrA. \n• Fig 1: Are A and B in the same orientation? \n• Fig 2C: It is better to replace yellow with another color. It is easy to confuse it with the yellow \ncolor in A. \n• Fig S2A: Is there any reason to show D2, D4, D1, and D3 instead of D1-4? \n• Page 6: \"Second, the two TrkH protomers have a relative rotation of 12.5 (Supplementary \nFigures 6a and b).\" Are the two protomers tilted in the plane perpendicular to the membrane or \nrotated in the plane parallel to the membrane? \n• Page 6: \"The N1-N1 interface remains unchanged while the N2-N2 interface adjusts to \naccommodate the changes (Supplementary Figure 2f).\" Fig s2f doesn't show the difference or \nsimilarity at all. \n• Page 7: \"In the presence of ATP, TrkAs become two dimers and no longer form a tetrameric ring \n(Figure 3a and b).\". It is really hard to see the dimers. A top view of the TrkAs will be helpful. A \ncomparison of the whole intracellular dimers/rings will be needed. \n• Page 25: \"TrkH-TrkA complex was assembled by mixing TrkH and TrkA in 1.2 to 1 molar ratio.\" \nBased on the structure, the molar ratio is 1 TrkH to 2 TrkA. Why was 1.2 to 1 molar ratio used? \n• Page 27: \". Each micrograph was collected in 5 frames/s for 10s, with a dose of 8e-/Å/s\". Should \nit be 8e-/pixel/s? \n• Page 27: \"These particles were reconstituted into a 6 Å map through RELION 3D refinement\". \nShouldn't it be reconstructed instead of reconstituted? \n• Page 27: Was any symmetry used in 3D reconstruction? \n\n\f", "role": "assistant" } ]	2,020
Talin dissociates from RIAM and associates to vinculin sequentially in response to the	REVIEWERS' COMMENTS: Reviewer #2 (Remarks to the Author): The authors have satisfactorily addressed my comments.	REVIEWERS' COMMENTS: Reviewer #2 (Remarks to the Author): The authors have satisfactorily addressed my comments.	TPR	NC_BS_BIOPH_004_RR2.txt	{ "criticism": 0, "example": 0, "importance_and_relevance": 0, "materials_and_methods": 0, "praise": 1, "presentation_and_reporting": 0, "results_and_discussion": 0, "suggestion_and_solution": 0, "total": 1 }	{ "criticism": 0, "example": 0, "importance_and_relevance": 0, "materials_and_methods": 0, "praise": 1, "presentation_and_reporting": 0, "results_and_discussion": 0, "suggestion_and_solution": 0 }	1	[ { "content": "Please review this paper.", "role": "user" }, { "content": "REVIEWERS' COMMENTS:\nReviewer #2 (Remarks to the Author):\nThe authors have satisfactorily addressed my comments.", "role": "assistant" } ]	2,020
Talin dissociates from RIAM and associates to vinculin sequentially in response to the	Reviewers' comments: Reviewer #1 (Remarks to the Author): This technically very elegant study analyzes the force-dependent interaction of the two adhesion proteins RIAM and vinculin to talin, a crucial integrin activator and force transducer. By further developing a previously established in vitro reconstitution assay, the authors demonstrate the expected actomyosin-dependent binding of vinculin to talin’s R1–R3 domain but also provide direct evidence for a force-dependent interaction of vinculin with talin R11. The authors then show that actomyosin forces induce an unbinding of RIAM from talin R1–R3 and R11, while the interaction of RIAM with talin R7–R8 seems insensitive to actomyosin forces. Combining protein fragments from talin, vinculin and RIAM in one assay reveals a concomitant RIAM dissociation from and vinculin association with talin R1–R3 and talin R11 upon actomyosin contraction. Increasing concentrations of RIAM inhibit vinculin binding to R1–R3, but not to R11 suggesting a differential regulation of force-dependent vinculin binding to these domains. Overall, this is a well-written manuscript, the data are clear and statistically sound, and the results provide intriguing insights into the regulation of talin by mechanical force. Previous work already indicated that RIAM and vinculin undergo mutual exclusive interaction with R1–R3 (Goult et al. JBC, 2013), which is nicely confirmed in this paper. In addition, the manuscript shows how the additional RIAM/vinculin binding sites in R7–R8 and in R11 are regulated by actomyosin force. It would be nice to see whether increasing concentrations of vinculin (i.e. higher concentrations than used in Fig. 4E, G) can displace RIAM from talin, as this has been suggested before (Li e al, MBoC, 2013). However, the current data sets are convincing and do not require, from my perspective, any major adjustments. The only minor drawback of the study is the absence of any attempts to confirm the observations in an independent fashion. It would certainly increase the impact of the study if some of the predictions were being tested in cell culture experiments. I would not consider such cell culture experiments mandatory, because the data are conclusive and deserve to be published. Reviewer #2 (Remarks to the Author): In this manuscript, the authors use an elegant in vitro reconstitution system with time-lapse imaging to investigate mechanoresponses of talin in RIAM and vinculin bindings. They clearly show that RIAM dissociates from and vinculin binds to talin domains in response to assembly of the actomyosin cytoskeleton. They observe that different talin domains show distinct responses in actomyosin- dependent RIAM dissociation and vinculin association. With kinetics analyses, they show the sequential actions of talin in which actomyosin accumulation induces RIAM dissociation that is followed by vinculin binding. Furthermore, they reveal that RIAM inhibits actomyosin-dependent binding of vinculin to talin. This is an interesting and well-written paper that would contribute to our understanding of molecular basis for mechanoresponses of cell adhesion complexes. However, the reviewer has a few concerns that should be addressed before publication. 1) The authors assume that “force” is responsible for actomyosin-dependent RAIM dissociation from and vinculin association with talin. However, they applied actin and myosin II at a time and did not distinguish individual actions of these components, e.g., actin polymerization, F-actin bundling by myosin II, and force generation by actin-myosin II sliding. Therefore, it is currently unclear whether force is a true stimulus that induces the RAIM dissociation and the vinculin association in the system of this study. Several additional controls should be tested. For example, addition of 1) actin alone, and 2) actin together with myosin II that can crosslink F-actin but has no motor activity (e.g., N- ethylmaleimide-treated myosin II; Smith et al. 2007 Biophys J) would provide good controls. Furthermore, effects of inhibition of myosin II ATPase activity need to be tested. Even though conventional blebbistatin may not be compatible with time-lapse fluorescence imaging, a photostable and non-fluorescent derivative of blebbistatin is now commercially available. 2) Even in the presence of actomyosin, talin R7-R8 shows no apparent vinculin binding and only moderate RIAM dissociation. The authors discuss that the poor response of R7-R8 in vinculin binding and RIAM dissociation may arise from high mechanical stability of R7-R8. However, considering the fact that R7-R8 directly binds to F-actin (Atherton et al. 2015 Nat Commun), a simple possibility is that F-actin binding to R7-R8 may block RIAM dissociation from and vinculin binding to R7-R8. Alternatively, simultaneous pulling at R13 and at R7-R8 by actomyosin through F-actin binding to R13 and R7-R8 may result in no apparent tension development between R13 and R7-R8, leading to little RIAM dissociation/vinculin association. 3) In this in vitro study, the authors use the vinculin head domain as a vinculin model and show its actomyosin-dependent binding to talin domains. On the other hand, studies using cells have reported that in contrast to full-length vinculin, the head domain only of vinculin constitutively binds to talin at focal adhesions independently of actomyosin activity (Humphries et al. 2007 J Cell Biol; Carisey et al. 2013 Curr Biol; Hirata et al. 2014 Am J Physiol Cell Physiol). Please discuss why the vinculin head domain behaves differently in talin binding between in vitro and in cells and how relevant the results of vinculin head-talin binding obtained in this study to vinculin-talin binding at focal adhesions in cells.	Reviewers' comments: Reviewer #1 (Remarks to the Author): This technically very elegant study analyzes the force-dependent interaction of the two adhesion proteins RIAM and vinculin to talin, a crucial integrin activator and force transducer. By further developing a previously established in vitro reconstitution assay, the authors demonstrate the expected actomyosin-dependent binding of vinculin to talin’s R1–R3 domain but also provide direct evidence for a force-dependent interaction of vinculin with talin R11. The authors then show that actomyosin forces induce an unbinding of RIAM from talin R1–R3 and R11, while the interaction of RIAM with talin R7–R8 seems insensitive to actomyosin forces. Combining protein fragments from talin, vinculin and RIAM in one assay reveals a concomitant RIAM dissociation from and vinculin association with talin R1–R3 and talin R11 upon actomyosin contraction. Increasing concentrations of RIAM inhibit vinculin binding to R1–R3, but not to R11 suggesting a differential regulation of force-dependent vinculin binding to these domains. Overall, this is a well-written manuscript, the data are clear and statistically sound, and the results provide intriguing insights into the regulation of talin by mechanical force. Previous work already indicated that RIAM and vinculin undergo mutual exclusive interaction with R1–R3 (Goult et al. JBC, 2013), which is nicely confirmed in this paper. In addition, the manuscript shows how the additional RIAM/vinculin binding sites in R7–R8 and in R11 are regulated by actomyosin force. It would be nice to see whether increasing concentrations of vinculin (i.e. higher concentrations than used in Fig. 4E, G) can displace RIAM from talin, as this has been suggested before (Li e al, MBoC, 2013). However, the current data sets are convincing and do not require, from my perspective, any major adjustments. The only minor drawback of the study is the absence of any attempts to confirm the observations in an independent fashion. It would certainly increase the impact of the study if some of the predictions were being tested in cell culture experiments. I would not consider such cell culture experiments mandatory, because the data are conclusive and deserve to be published. Reviewer #2 (Remarks to the Author): In this manuscript, the authors use an elegant in vitro reconstitution system with time-lapse imaging to investigate mechanoresponses of talin in RIAM and vinculin bindings. They clearly show that RIAM dissociates from and vinculin binds to talin domains in response to assembly of the actomyosin cytoskeleton. They observe that different talin domains show distinct responses in actomyosin- dependent RIAM dissociation and vinculin association. With kinetics analyses, they show the sequential actions of talin in which actomyosin accumulation induces RIAM dissociation that is followed by vinculin binding. Furthermore, they reveal that RIAM inhibits actomyosin-dependent binding of vinculin to talin. This is an interesting and well-written paper that would contribute to our understanding of molecular basis for mechanoresponses of cell adhesion complexes. However, the reviewer has a few concerns that should be addressed before publication. 1) The authors assume that “force” is responsible for actomyosin-dependent RAIM dissociation from and vinculin association with talin. However, they applied actin and myosin II at a time and did not distinguish individual actions of these components, e.g., actin polymerization, F-actin bundling by myosin II, and force generation by actin-myosin II sliding. Therefore, it is currently unclear whether force is a true stimulus that induces the RAIM dissociation and the vinculin association in the system of this study. Several additional controls should be tested. For example, addition of 1) actin alone, and 2) actin together with myosin II that can crosslink F-actin but has no motor activity (e.g., N- ethylmaleimide-treated myosin II; Smith et al. 2007 Biophys J) would provide good controls. Furthermore, effects of inhibition of myosin II ATPase activity need to be tested. Even though conventional blebbistatin may not be compatible with time-lapse fluorescence imaging, a photostable and non-fluorescent derivative of blebbistatin is now commercially available. 2) Even in the presence of actomyosin, talin R7-R8 shows no apparent vinculin binding and only moderate RIAM dissociation. The authors discuss that the poor response of R7-R8 in vinculin binding and RIAM dissociation may arise from high mechanical stability of R7-R8. However, considering the fact that R7-R8 directly binds to F-actin (Atherton et al. 2015 Nat Commun), a simple possibility is that F-actin binding to R7-R8 may block RIAM dissociation from and vinculin binding to R7-R8. Alternatively, simultaneous pulling at R13 and at R7-R8 by actomyosin through F-actin binding to R13 and R7-R8 may result in no apparent tension development between R13 and R7-R8, leading to little RIAM dissociation/vinculin association. 3) In this in vitro study, the authors use the vinculin head domain as a vinculin model and show its actomyosin-dependent binding to talin domains. On the other hand, studies using cells have reported that in contrast to full-length vinculin, the head domain only of vinculin constitutively binds to talin at focal adhesions independently of actomyosin activity (Humphries et al. 2007 J Cell Biol; Carisey et al. 2013 Curr Biol; Hirata et al. 2014 Am J Physiol Cell Physiol). Please discuss why the vinculin head domain behaves differently in talin binding between in vitro and in cells and how relevant the results of vinculin head-talin binding obtained in this study to vinculin-talin binding at focal adhesions in cells.	TPR	NC_BS_BIOPH_004_RR1.txt	{ "criticism": 5, "example": 2, "importance_and_relevance": 3, "materials_and_methods": 20, "praise": 5, "presentation_and_reporting": 2, "results_and_discussion": 3, "suggestion_and_solution": 8, "total": 42 }	{ "criticism": 0.11904761904761904, "example": 0.047619047619047616, "importance_and_relevance": 0.07142857142857142, "materials_and_methods": 0.47619047619047616, "praise": 0.11904761904761904, "presentation_and_reporting": 0.047619047619047616, "results_and_discussion": 0.07142857142857142, "suggestion_and_solution": 0.19047619047619047 }	1.142857	[ { "content": "Please review this paper.", "role": "user" }, { "content": "Reviewers' comments:\nReviewer #1 (Remarks to the Author):\nThis technically very elegant study analyzes the force-dependent interaction of the two adhesion\nproteins RIAM and vinculin to talin, a crucial integrin activator and force transducer. By further\ndeveloping a previously established in vitro reconstitution assay, the authors demonstrate the\nexpected actomyosin-dependent binding of vinculin to talin’s R1–R3 domain but also provide direct\nevidence for a force-dependent interaction of vinculin with talin R11. The authors then show that\nactomyosin forces induce an unbinding of RIAM from talin R1–R3 and R11, while the interaction of\nRIAM with talin R7–R8 seems insensitive to actomyosin forces. Combining protein fragments from talin,\nvinculin and RIAM in one assay reveals a concomitant RIAM dissociation from and vinculin association\nwith talin R1–R3 and talin R11 upon actomyosin contraction. Increasing concentrations of RIAM inhibit\nvinculin binding to R1–R3, but not to R11 suggesting a differential regulation of force-dependent\nvinculin binding to these domains.\nOverall, this is a well-written manuscript, the data are clear and statistically sound, and the results\nprovide intriguing insights into the regulation of talin by mechanical force. Previous work already\nindicated that RIAM and vinculin undergo mutual exclusive interaction with R1–R3 (Goult et al. JBC,\n2013), which is nicely confirmed in this paper. In addition, the manuscript shows how the additional\nRIAM/vinculin binding sites in R7–R8 and in R11 are regulated by actomyosin force. It would be nice\nto see whether increasing concentrations of vinculin (i.e. higher concentrations than used in Fig. 4E, G)\ncan displace RIAM from talin, as this has been suggested before (Li e al, MBoC, 2013). However, the\ncurrent data sets are convincing and do not require, from my perspective, any major adjustments.\nThe only minor drawback of the study is the absence of any attempts to confirm the observations in\nan independent fashion. It would certainly increase the impact of the study if some of the predictions\nwere being tested in cell culture experiments. I would not consider such cell culture experiments\nmandatory, because the data are conclusive and deserve to be published.\nReviewer #2 (Remarks to the Author):\nIn this manuscript, the authors use an elegant in vitro reconstitution system with time-lapse imaging\nto investigate mechanoresponses of talin in RIAM and vinculin bindings. They clearly show that RIAM\ndissociates from and vinculin binds to talin domains in response to assembly of the actomyosin\ncytoskeleton. They observe that different talin domains show distinct responses in actomyosin-\ndependent RIAM dissociation and vinculin association. With kinetics analyses, they show the sequential\nactions of talin in which actomyosin accumulation induces RIAM dissociation that is followed by\nvinculin binding. Furthermore, they reveal that RIAM inhibits actomyosin-dependent binding of vinculin\nto talin.\nThis is an interesting and well-written paper that would contribute to our understanding of molecular\nbasis for mechanoresponses of cell adhesion complexes. However, the reviewer has a few concerns\nthat should be addressed before publication.\n1) The authors assume that “force” is responsible for actomyosin-dependent RAIM dissociation from\nand vinculin association with talin. However, they applied actin and myosin II at a time and did not\ndistinguish individual actions of these components, e.g., actin polymerization, F-actin bundling by\nmyosin II, and force generation by actin-myosin II sliding. Therefore, it is currently unclear whether\nforce is a true stimulus that induces the RAIM dissociation and the vinculin association in the system\nof this study. Several additional controls should be tested. For example, addition of 1) actin alone, and\n2) actin together with myosin II that can crosslink F-actin but has no motor activity (e.g., N-\n ethylmaleimide-treated myosin II; Smith et al. 2007 Biophys J) would provide good controls.\n Furthermore, effects of inhibition of myosin II ATPase activity need to be tested. Even though\n conventional blebbistatin may not be compatible with time-lapse fluorescence imaging, a photostable\n and non-fluorescent derivative of blebbistatin is now commercially available.\n 2) Even in the presence of actomyosin, talin R7-R8 shows no apparent vinculin binding and only\n moderate RIAM dissociation. The authors discuss that the poor response of R7-R8 in vinculin binding\n and RIAM dissociation may arise from high mechanical stability of R7-R8. However, considering the\n fact that R7-R8 directly binds to F-actin (Atherton et al. 2015 Nat Commun), a simple possibility is\n that F-actin binding to R7-R8 may block RIAM dissociation from and vinculin binding to R7-R8.\n Alternatively, simultaneous pulling at R13 and at R7-R8 by actomyosin through F-actin binding to R13\n and R7-R8 may result in no apparent tension development between R13 and R7-R8, leading to little\n RIAM dissociation/vinculin association.\n 3) In this in vitro study, the authors use the vinculin head domain as a vinculin model and show its\n actomyosin-dependent binding to talin domains. On the other hand, studies using cells have reported\n that in contrast to full-length vinculin, the head domain only of vinculin constitutively binds to talin at\n focal adhesions independently of actomyosin activity (Humphries et al. 2007 J Cell Biol; Carisey et al.\n 2013 Curr Biol; Hirata et al. 2014 Am J Physiol Cell Physiol). Please discuss why the vinculin head\n domain behaves differently in talin binding between in vitro and in cells and how relevant the results\n of vinculin head-talin binding obtained in this study to vinculin-talin binding at focal adhesions in cells.\n", "role": "assistant" } ]	2,020
A dietary fatty acid counteracts neuronal mechanical sensitization	Reviewers' Comments: Reviewer #1: Remarks to the Author: “A dietary fatty acid counteracts neuronal mechanical sensitization” SUMMARY: This manuscript presents a thoughtful and rigorous characterization of the effects of margaric acid (MA) enrichment of the plasma membrane on PIEZO2-dependent mechanotransduction in a variety of mammalian cell types: mouse and rat DRG neurons, N2A mouse neuroblastoma cell line, Merkel cell caricinoma cells, and human iPSC-derived sensory neurons. Prior to this study, MA was shown to inhibit mechanically evoked Piezo1 currents. In this current study, the authors observed that Piezo2 is less sensitive to MA than Piezo1. They also show that Piezo2 but not Piezo1 is perturbed by pharmacological disruption of the actin cytoskeleton, echoing earlier findings with Piezo1. Several of the authors’ key experiments, most notably the testing of Piezo1-Piezo2 chimeric channels, suggest that the beam domain of Piezo1 and Piezo2 ¬¬ confers differential dependence of the channels on cytoskeleton and their different sensitivities to MA. Using mouse, rat, and iPSC-derived human DRG neurons, the authors show that MA partially attenuates rapidly adapting currents (attributable to Piezo2) and that MA increases the apparent threshold for mechanically evoked currents in the cell indentation assay. They also demonstrate that MA decreases the ability of sensory neurons to recover from 1Hz indentation stimuli. They propose that MA affects mechanotransduction channels and not voltage- gated sodium or steady-state potassium channels, and show that resting membrane potential is not perturbed with MA treatment. Finally, the authors demonstrate that MA attenuates the effect of bradykinin, an inflammatory endogenous peptide that sensitizes Piezo2 and other cation channels to mechanical stimuli via G protein signaling, on mouse DRG neurons. Overall, this manuscript makes substantial strides to advance our understanding how Piezo1 and Piezo2 sense membrane tension and mechanical force, which is incompletely understood. The authors’ conclusions are generally supported by the data in its present form. We have several suggestions for additional analyses and/or experiments to better support the claims in the paper, in particular with regard to the proposed effects of MA on DRG neurons and the conclusions drawn from the Piezo2 chimera experiments. However, even in the absence of further revision experiments we believe this paper is suitable for publication with text revisions and/or analysis of existing data to address the items detailed below. Points of concern (note these are all relatively minor) 1. Electrophysiology of mouse DRG neurons (Figure 4 and Supplemental Figure 5): a. It is difficult to conclude from the data as-is that MA does not affect DRG neuronal excitability (and hence only affects mechanically-gated ion channels, i.e. Piezo2). For example, not all K+ channels modulate resting potential. It would be good to know if the input resistance of these cells is changed after MA, which could be calculated from your existing voltage-clamp experiments. From the methods, it looks like the bath solutions include many cations that could carry currents aside from stretch-activated ion channels. b. Furthermore, the AP waveform (4A and Supplemental Figure 5E) does look slightly different in the traces presented, contrary to the authors’ conclusions regarding MA and neuronal excitability/AP firing. Specifically, the after hyperpolarization and the shoulder are smaller after MA. This could be due to cell-to-cell variability (i.e. minor differences in the chosen traces). A dv/dt plot, or assessment of other AP waveform parameters in your analysis software could reveal differences due to MA effects on other channels, or convincingly demonstrate no difference. Any differences should be noted and addressed in the main text. c. You have some inactivating inward currents after the step from positive potential to resting/negative in Supplemental Figure 5A. Similar to concern 1a, is this observation reproducible over multiple cells and if so could this be due to effects of MA on voltage-dependent ion channels? d. The chosen 1Hz stimulus protocol is nice (Figure 4C and elsewhere). Have you observed differences in recovery from inactivation +/- MA using a voltage-clamp paradigm similar to Coste et al. PNAS (2013), varying ∆t and comparing the time constant of recovery? This is by no means an essential experiment. 2. Cytoskeleton disruption experiments with Piezo2, Piezo1, and beam chimera in N2A cells (Figure 2 and Supplemental Figure 3): a. The conferral of MA sensitivity in 2E and loss of LatA effect in 2F are striking. More specific discussion of how this directly relates to previous experimental findings, such as Cox et al. (2016) should be included. b. The Eijelkamp et al. (2013) paper on EPAC1 and Piezo2 by no means demonstrates that Piezo2 requires cytoskeletal elements for normal function (line 60), it suggests cytoskeletal involvement but does not prove direct interaction. Treatment with cytoskeletal-disrupting drugs such as colchicine and latrunculin A can have many other effects, for example trafficking and recycling of other membrane proteins. I would tone down this citation. Furthermore, the citation on line 122 cites the above article as well as Coste et al. (2010), Moroni et al. (2018), and Wang et al. (2019) as evidence that Piezo2 requires cytoskeleton. None of these other papers examined that directly. Reviewer #2: Remarks to the Author: The authors illustrate that Piezo2 currents are inhibited by Margaric Acid (MA) in N2A mouse neuroblastoma cells, rodent DRGs and human iPSC-derived DRGs. This study, a follow up to a previous study from the same group, finds that compared to Piezo1, greater amounts of MA are required for Piezo2 inhibition. However, disruption of the actin cytoskeleton by Latranculin A reduces the required amount. MA also inhibited action potentials from mouse and rat DRG neurons, and reversed sensitization of mouse DRG currents by bradykinin. The authors suggest that MA could be used to attenuate overactive touch responses as seen in tactile allodynia. This is a well-executed study, the data are of high quality, and the experiments are appropriately controlled and analyzed. Major Comments The authors suggest that the observations on the relationship between Piezo2 activity and MA supplementation suggest a novel method to reduce the effects of overactive sensory receptors in tactile allodynia. However, for translation to human DRGs and the proposed clinical applications, it is important to show the effect of MA on human iPSC-derived neurons electrophysiology in current- clamp mode - does MA treatment inhibit action potentials like in the rodent DRGs? Also, is the effect of MA on bradykinin-sensitized currents (as in Fig. 6) also observed in the human iPSC- derived neurons? These experiments are necessary to support the conclusions and proposed impact of the findings (e.g. “Our findings.....suggest new avenues to treat tactile allodynia”). If the authors are unable to perform these experiments, then the claims should be adjusted accordingly throughout the manuscript. Figure 2 - The authors conclude in this figure that the different beam regions between Piezo1 and Piezo2 is the determinant for their different responses to MA supplementation based on a chimeric Piezo2 channel that possesses a Piezo1 beam. Their argument would be more robust if they included another chimera of Piezo1 with a Piezo2 beam region. If this new chimera under the treatment of MA and Latrunculin A behaved more like Piezo2 in response to MA, the authors would be able strengthen their conclusion that the beam region in each protein is responsible for the individual protein’s response to MA. While this experiment would be highly informative in providing a mechanistic basis of the observed effect, in light of the current Covid-19 disruptions and lab closures, it may not be feasible to do these experiments. As such, I leave it to the author’s discretion. Minor Comments Line 116: “Thus, as with PIEZO1, PIEZO2 is less active in rigid membranes (>78 pN (ref 24)).” It is not clear what >78 pN refers to, nor where this number arises from. Supplemental Figure 3A- The colors and layout chosen for this figure make it confusing and hard to read. Please consider changing the layout and colors to make it easier to follow. Line 262 - 264. This speculative statement should be moved to the Discussion section.	Reviewers' Comments: Reviewer #1: Remarks to the Author: “A dietary fatty acid counteracts neuronal mechanical sensitization” SUMMARY: This manuscript presents a thoughtful and rigorous characterization of the effects of margaric acid (MA) enrichment of the plasma membrane on PIEZO2-dependent mechanotransduction in a variety of mammalian cell types: mouse and rat DRG neurons, N2A mouse neuroblastoma cell line, Merkel cell caricinoma cells, and human iPSC-derived sensory neurons. Prior to this study, MA was shown to inhibit mechanically evoked Piezo1 currents. In this current study, the authors observed that Piezo2 is less sensitive to MA than Piezo1. They also show that Piezo2 but not Piezo1 is perturbed by pharmacological disruption of the actin cytoskeleton, echoing earlier findings with Piezo1. Several of the authors’ key experiments, most notably the testing of Piezo1-Piezo2 chimeric channels, suggest that the beam domain of Piezo1 and Piezo2 ¬¬ confers differential dependence of the channels on cytoskeleton and their different sensitivities to MA. Using mouse, rat, and iPSC-derived human DRG neurons, the authors show that MA partially attenuates rapidly adapting currents (attributable to Piezo2) and that MA increases the apparent threshold for mechanically evoked currents in the cell indentation assay. They also demonstrate that MA decreases the ability of sensory neurons to recover from 1Hz indentation stimuli. They propose that MA affects mechanotransduction channels and not voltage- gated sodium or steady-state potassium channels, and show that resting membrane potential is not perturbed with MA treatment. Finally, the authors demonstrate that MA attenuates the effect of bradykinin, an inflammatory endogenous peptide that sensitizes Piezo2 and other cation channels to mechanical stimuli via G protein signaling, on mouse DRG neurons. Overall, this manuscript makes substantial strides to advance our understanding how Piezo1 and Piezo2 sense membrane tension and mechanical force, which is incompletely understood. The authors’ conclusions are generally supported by the data in its present form. We have several suggestions for additional analyses and/or experiments to better support the claims in the paper, in particular with regard to the proposed effects of MA on DRG neurons and the conclusions drawn from the Piezo2 chimera experiments. However, even in the absence of further revision experiments we believe this paper is suitable for publication with text revisions and/or analysis of existing data to address the items detailed below. Points of concern (note these are all relatively minor) 1. Electrophysiology of mouse DRG neurons (Figure 4 and Supplemental Figure 5): a. It is difficult to conclude from the data as-is that MA does not affect DRG neuronal excitability (and hence only affects mechanically-gated ion channels, i.e. Piezo2). For example, not all K+ channels modulate resting potential. It would be good to know if the input resistance of these cells is changed after MA, which could be calculated from your existing voltage-clamp experiments. From the methods, it looks like the bath solutions include many cations that could carry currents aside from stretch-activated ion channels. b. Furthermore, the AP waveform (4A and Supplemental Figure 5E) does look slightly different in the traces presented, contrary to the authors’ conclusions regarding MA and neuronal excitability/AP firing. Specifically, the after hyperpolarization and the shoulder are smaller after MA. This could be due to cell-to-cell variability (i.e. minor differences in the chosen traces). A dv/dt plot, or assessment of other AP waveform parameters in your analysis software could reveal differences due to MA effects on other channels, or convincingly demonstrate no difference. Any differences should be noted and addressed in the main text. c. You have some inactivating inward currents after the step from positive potential to resting/negative in Supplemental Figure 5A. Similar to concern 1a, is this observation reproducible over multiple cells and if so could this be due to effects of MA on voltage-dependent ion channels? d. The chosen 1Hz stimulus protocol is nice (Figure 4C and elsewhere). Have you observed differences in recovery from inactivation +/- MA using a voltage-clamp paradigm similar to Coste et al. PNAS (2013), varying ∆t and comparing the time constant of recovery? This is by no means an essential experiment. 2. Cytoskeleton disruption experiments with Piezo2, Piezo1, and beam chimera in N2A cells (Figure 2 and Supplemental Figure 3): a. The conferral of MA sensitivity in 2E and loss of LatA effect in 2F are striking. More specific discussion of how this directly relates to previous experimental findings, such as Cox et al. (2016) should be included. b. The Eijelkamp et al. (2013) paper on EPAC1 and Piezo2 by no means demonstrates that Piezo2 requires cytoskeletal elements for normal function (line 60), it suggests cytoskeletal involvement but does not prove direct interaction. Treatment with cytoskeletal-disrupting drugs such as colchicine and latrunculin A can have many other effects, for example trafficking and recycling of other membrane proteins. I would tone down this citation. Furthermore, the citation on line 122 cites the above article as well as Coste et al. (2010), Moroni et al. (2018), and Wang et al. (2019) as evidence that Piezo2 requires cytoskeleton. None of these other papers examined that directly. Reviewer #2: Remarks to the Author: The authors illustrate that Piezo2 currents are inhibited by Margaric Acid (MA) in N2A mouse neuroblastoma cells, rodent DRGs and human iPSC-derived DRGs. This study, a follow up to a previous study from the same group, finds that compared to Piezo1, greater amounts of MA are required for Piezo2 inhibition. However, disruption of the actin cytoskeleton by Latranculin A reduces the required amount. MA also inhibited action potentials from mouse and rat DRG neurons, and reversed sensitization of mouse DRG currents by bradykinin. The authors suggest that MA could be used to attenuate overactive touch responses as seen in tactile allodynia. This is a well-executed study, the data are of high quality, and the experiments are appropriately controlled and analyzed. Major Comments The authors suggest that the observations on the relationship between Piezo2 activity and MA supplementation suggest a novel method to reduce the effects of overactive sensory receptors in tactile allodynia. However, for translation to human DRGs and the proposed clinical applications, it is important to show the effect of MA on human iPSC-derived neurons electrophysiology in current- clamp mode - does MA treatment inhibit action potentials like in the rodent DRGs? Also, is the effect of MA on bradykinin-sensitized currents (as in Fig. 6) also observed in the human iPSC- derived neurons? These experiments are necessary to support the conclusions and proposed impact of the findings (e.g. “Our findings.....suggest new avenues to treat tactile allodynia”). If the authors are unable to perform these experiments, then the claims should be adjusted accordingly throughout the manuscript. Figure 2 - The authors conclude in this figure that the different beam regions between Piezo1 and Piezo2 is the determinant for their different responses to MA supplementation based on a chimeric Piezo2 channel that possesses a Piezo1 beam. Their argument would be more robust if they included another chimera of Piezo1 with a Piezo2 beam region. If this new chimera under the treatment of MA and Latrunculin A behaved more like Piezo2 in response to MA, the authors would be able strengthen their conclusion that the beam region in each protein is responsible for the individual protein’s response to MA. While this experiment would be highly informative in providing a mechanistic basis of the observed effect, in light of the current Covid-19 disruptions and lab closures, it may not be feasible to do these experiments. As such, I leave it to the author’s discretion. Minor Comments Line 116: “Thus, as with PIEZO1, PIEZO2 is less active in rigid membranes (>78 pN (ref 24)).” It is not clear what >78 pN refers to, nor where this number arises from. Supplemental Figure 3A- The colors and layout chosen for this figure make it confusing and hard to read. Please consider changing the layout and colors to make it easier to follow. Line 262 - 264. This speculative statement should be moved to the Discussion section.	TPR	NC_BS_BIOPH_005_RR1.txt	{ "criticism": 8, "example": 12, "importance_and_relevance": 4, "materials_and_methods": 32, "praise": 6, "presentation_and_reporting": 8, "results_and_discussion": 13, "suggestion_and_solution": 13, "total": 71 }	{ "criticism": 0.11267605633802817, "example": 0.16901408450704225, "importance_and_relevance": 0.056338028169014086, "materials_and_methods": 0.4507042253521127, "praise": 0.08450704225352113, "presentation_and_reporting": 0.11267605633802817, "results_and_discussion": 0.18309859154929578, "suggestion_and_solution": 0.18309859154929578 }	1.352113	[ { "content": "Please review this paper.", "role": "user" }, { "content": "Reviewers' Comments:\n Reviewer #1:\n Remarks to the Author:\n “A dietary fatty acid counteracts neuronal mechanical sensitization”\nSUMMARY: This manuscript presents a thoughtful and rigorous characterization of the effects of\nmargaric acid (MA) enrichment of the plasma membrane on PIEZO2-dependent\nmechanotransduction in a variety of mammalian cell types: mouse and rat DRG neurons, N2A\nmouse neuroblastoma cell line, Merkel cell caricinoma cells, and human iPSC-derived sensory\nneurons. Prior to this study, MA was shown to inhibit mechanically evoked Piezo1 currents. In this\ncurrent study, the authors observed that Piezo2 is less sensitive to MA than Piezo1. They also\nshow that Piezo2 but not Piezo1 is perturbed by pharmacological disruption of the actin\ncytoskeleton, echoing earlier findings with Piezo1. Several of the authors’ key experiments, most\nnotably the testing of Piezo1-Piezo2 chimeric channels, suggest that the beam domain of Piezo1\nand Piezo2 ¬¬ confers differential dependence of the channels on cytoskeleton and their different\nsensitivities to MA. Using mouse, rat, and iPSC-derived human DRG neurons, the authors show\nthat MA partially attenuates rapidly adapting currents (attributable to Piezo2) and that MA\nincreases the apparent threshold for mechanically evoked currents in the cell indentation assay.\nThey also demonstrate that MA decreases the ability of sensory neurons to recover from 1Hz\nindentation stimuli. They propose that MA affects mechanotransduction channels and not voltage-\ngated sodium or steady-state potassium channels, and show that resting membrane potential is\nnot perturbed with MA treatment. Finally, the authors demonstrate that MA attenuates the effect\nof bradykinin, an inflammatory endogenous peptide that sensitizes Piezo2 and other cation\nchannels to mechanical stimuli via G protein signaling, on mouse DRG neurons. Overall, this\nmanuscript makes substantial strides to advance our understanding how Piezo1 and Piezo2 sense\nmembrane tension and mechanical force, which is incompletely understood.\nThe authors’ conclusions are generally supported by the data in its present form. We have several\nsuggestions for additional analyses and/or experiments to better support the claims in the paper,\nin particular with regard to the proposed effects of MA on DRG neurons and the conclusions drawn\nfrom the Piezo2 chimera experiments. However, even in the absence of further revision\nexperiments we believe this paper is suitable for publication with text revisions and/or analysis of\nexisting data to address the items detailed below.\nPoints of concern (note these are all relatively minor)\n1. Electrophysiology of mouse DRG neurons (Figure 4 and Supplemental Figure 5):\na. It is difficult to conclude from the data as-is that MA does not affect DRG neuronal excitability\n(and hence only affects mechanically-gated ion channels, i.e. Piezo2). For example, not all K+\nchannels modulate resting potential. It would be good to know if the input resistance of these cells\nis changed after MA, which could be calculated from your existing voltage-clamp experiments.\nFrom the methods, it looks like the bath solutions include many cations that could carry currents\naside from stretch-activated ion channels.\nb. Furthermore, the AP waveform (4A and Supplemental Figure 5E) does look slightly different in\nthe traces presented, contrary to the authors’ conclusions regarding MA and neuronal\nexcitability/AP firing. Specifically, the after hyperpolarization and the shoulder are smaller after\nMA. This could be due to cell-to-cell variability (i.e. minor differences in the chosen traces). A\ndv/dt plot, or assessment of other AP waveform parameters in your analysis software could reveal\ndifferences due to MA effects on other channels, or convincingly demonstrate no difference. Any\ndifferences should be noted and addressed in the main text.\nc. You have some inactivating inward currents after the step from positive potential to\nresting/negative in Supplemental Figure 5A. Similar to concern 1a, is this observation reproducible\nover multiple cells and if so could this be due to effects of MA on voltage-dependent ion channels?\nd. The chosen 1Hz stimulus protocol is nice (Figure 4C and elsewhere). Have you observed\n differences in recovery from inactivation +/- MA using a voltage-clamp paradigm similar to Coste\n et al. PNAS (2013), varying ∆t and comparing the time constant of recovery? This is by no means\n an essential experiment.\n 2. Cytoskeleton disruption experiments with Piezo2, Piezo1, and beam chimera in N2A cells (Figure\n 2 and Supplemental Figure 3):\n a. The conferral of MA sensitivity in 2E and loss of LatA effect in 2F are striking. More specific\n discussion of how this directly relates to previous experimental findings, such as Cox et al. (2016)\n should be included.\n b. The Eijelkamp et al. (2013) paper on EPAC1 and Piezo2 by no means demonstrates that Piezo2\n requires cytoskeletal elements for normal function (line 60), it suggests cytoskeletal involvement\n but does not prove direct interaction. Treatment with cytoskeletal-disrupting drugs such as\n colchicine and latrunculin A can have many other effects, for example trafficking and recycling of\n other membrane proteins. I would tone down this citation. Furthermore, the citation on line 122\n cites the above article as well as Coste et al. (2010), Moroni et al. (2018), and Wang et al. (2019)\n as evidence that Piezo2 requires cytoskeleton. None of these other papers examined that directly.\n Reviewer #2:\n Remarks to the Author:\n The authors illustrate that Piezo2 currents are inhibited by Margaric Acid (MA) in N2A mouse\n neuroblastoma cells, rodent DRGs and human iPSC-derived DRGs. This study, a follow up to a\n previous study from the same group, finds that compared to Piezo1, greater amounts of MA are\n required for Piezo2 inhibition. However, disruption of the actin cytoskeleton by Latranculin A\n reduces the required amount. MA also inhibited action potentials from mouse and rat DRG\n neurons, and reversed sensitization of mouse DRG currents by bradykinin. The authors suggest\n that MA could be used to attenuate overactive touch responses as seen in tactile allodynia.\n This is a well-executed study, the data are of high quality, and the experiments are appropriately\n controlled and analyzed.\n Major Comments\n The authors suggest that the observations on the relationship between Piezo2 activity and MA\n supplementation suggest a novel method to reduce the effects of overactive sensory receptors in\n tactile allodynia. However, for translation to human DRGs and the proposed clinical applications, it\n is important to show the effect of MA on human iPSC-derived neurons electrophysiology in current-\n clamp mode - does MA treatment inhibit action potentials like in the rodent DRGs? Also, is the\n effect of MA on bradykinin-sensitized currents (as in Fig. 6) also observed in the human iPSC-\n derived neurons? These experiments are necessary to support the conclusions and proposed\n impact of the findings (e.g. “Our findings.....suggest new avenues to treat tactile allodynia”). If the\n authors are unable to perform these experiments, then the claims should be adjusted accordingly\n throughout the manuscript.\n Figure 2 - The authors conclude in this figure that the different beam regions between Piezo1 and\n Piezo2 is the determinant for their different responses to MA supplementation based on a chimeric\n Piezo2 channel that possesses a Piezo1 beam. Their argument would be more robust if they\n included another chimera of Piezo1 with a Piezo2 beam region. If this new chimera under the\n treatment of MA and Latrunculin A behaved more like Piezo2 in response to MA, the authors would\n be able strengthen their conclusion that the beam region in each protein is responsible for the\n individual protein’s response to MA. While this experiment would be highly informative in providing\n a mechanistic basis of the observed effect, in light of the current Covid-19 disruptions and lab\n closures, it may not be feasible to do these experiments. As such, I leave it to the author’s\n discretion.\n Minor Comments\n Line 116: “Thus, as with PIEZO1, PIEZO2 is less active in rigid membranes (>78 pN (ref 24)).”\n It is not clear what >78 pN refers to, nor where this number arises from.\n Supplemental Figure 3A- The colors and layout chosen for this figure make it confusing and hard\n to read. Please consider changing the layout and colors to make it easier to follow.\nLine 262 - 264. This speculative statement should be moved to the Discussion section.", "role": "assistant" } ]	2,020
A dietary fatty acid counteracts neuronal mechanical sensitization	Reviewers' Comments: Reviewer #1: Remarks to the Author: The authors have made substantial text changes and added new data analyses and results that greatly strengthen the manuscript. They have directly addressed all of our concerns highlighted in our initial review. It is our opinion that the data now fully support the claims, and no further revisions are required. Reviewer #2: Remarks to the Author: The authors have nicely addressed my comments, and further strengthened an already-strong manuscript. One remaining discrepancy arising from the edits is: Lines 331 - 333 (in Discussion): "Our data show that MA does not impair bradykinin-mediated sensitization, but rather reduces the mechano-currents similar to the ones measured during normal neuronal mechanical response or non-inflammatory-like levels." seems to be at odds with: Lines 84 - 86 (in Introduction): "Importantly, MA decreases PIEZO2 currents potentiated by the proalgesic agent bradykinin, indicating that it might be particularly useful for reducing heightened touch responses during inflammation."	Reviewers' Comments: Reviewer #1: Remarks to the Author: The authors have made substantial text changes and added new data analyses and results that greatly strengthen the manuscript. They have directly addressed all of our concerns highlighted in our initial review. It is our opinion that the data now fully support the claims, and no further revisions are required. Reviewer #2: Remarks to the Author: The authors have nicely addressed my comments, and further strengthened an already-strong manuscript. One remaining discrepancy arising from the edits is: Lines 331 - 333 (in Discussion): "Our data show that MA does not impair bradykinin-mediated sensitization, but rather reduces the mechano-currents similar to the ones measured during normal neuronal mechanical response or non-inflammatory-like levels." seems to be at odds with: Lines 84 - 86 (in Introduction): "Importantly, MA decreases PIEZO2 currents potentiated by the proalgesic agent bradykinin, indicating that it might be particularly useful for reducing heightened touch responses during inflammation."	TPR	NC_BS_BIOPH_005_RR2.txt	{ "criticism": 1, "example": 2, "importance_and_relevance": 0, "materials_and_methods": 3, "praise": 3, "presentation_and_reporting": 0, "results_and_discussion": 1, "suggestion_and_solution": 2, "total": 6 }	{ "criticism": 0.16666666666666666, "example": 0.3333333333333333, "importance_and_relevance": 0, "materials_and_methods": 0.5, "praise": 0.5, "presentation_and_reporting": 0, "results_and_discussion": 0.16666666666666666, "suggestion_and_solution": 0.3333333333333333 }	2	[ { "content": "Please review this paper.", "role": "user" }, { "content": "Reviewers' Comments:\nReviewer #1:\nRemarks to the Author:\nThe authors have made substantial text changes and added new data analyses and results that\ngreatly strengthen the manuscript. They have directly addressed all of our concerns highlighted in\nour initial review. It is our opinion that the data now fully support the claims, and no further\nrevisions are required.\n Reviewer #2:\n Remarks to the Author:\n The authors have nicely addressed my comments, and further strengthened an already-strong\n manuscript.\n One remaining discrepancy arising from the edits is:\n Lines 331 - 333 (in Discussion): \"Our data show that MA does not impair bradykinin-mediated\n sensitization, but rather reduces the mechano-currents similar to the ones measured during\n normal neuronal mechanical response or non-inflammatory-like levels.\"\nseems to be at odds with:\n Lines 84 - 86 (in Introduction): \"Importantly, MA decreases PIEZO2 currents potentiated by the\n proalgesic agent bradykinin, indicating that it might be particularly useful for reducing heightened\n touch responses during inflammation.\"", "role": "assistant" } ]	2,020
Redox-coupled proton pumping drives carbon concentration in the photosynthetic complex I	Reviewers' Comments: Reviewer #1: Remarks to the Author: The revised version of the work “Redox-Coupled Proton Pumping Drives Carbon Concentration in thePhotosynthetic Complex I “ presents a significant improved and corrected version of the manuscript. Authors have adressed and/or adressed most of my previous concerns as well as those presented by the other reviewers. Specifically, they have: i) added, as requested, a detailed methodology in the methods section now included as part of the main text, which properly describes theory levels and methodology ii) Clarified key elements of the three enezymes that are compared in the work (αCA, βCA and current enzyme). iii) Added proper description that energy calculations correspond to ΔH – TΔS + ΔZPE based on the electronic energy and the molecular Hessian. iv) Better explained the role of the titrable active site resiudes and pKa calculation v) Increased sampling and analysis of the waters present in the proton chanels vi) Provided estimate of the Free Energy Profile of CO2 migration along the proposed channel. Therefore, I believe the present work is suitable for publication. I only have a minor comment (described below) related to Figure 2b,. In the text desribing Figure 2 authors state (starting line 123) “ The CO 2 hydration is initiated by proton transfer (pT) from the Zn-bound water molecule followed by a nucleophilic attack.. of the hydroxide on CO 2 . In αCA (βCA), the rate limiting reaction barrier of 11 (12) kcal mol -1 is connected with pT to His64 (Tyr205), … In CupA, proton transfer from Zn-bound water to Tyr41 is slightly exergonic” Therefore, if I understood it correglty, in CupA Zn bound watre (which then acts as the nucelophile) transfer the proton to Tyr41 (in αCA /βCA the acceptor being His64/Tyr205. However, in Figure 2B, the Zn bound water transfer the proton to ANOTHER water molecule, whiwhc then transfers it to Tyr/His/Tyr. So, are there 1 or 2 water molecules?. In line with this point in Figure 2A waters are missing. I strongly beliebe they should be shown and the issue needs to be clarified. Reviewer #2: Remarks to the Author: I think that the authors have addressed the questions raised by all reviwers adequately, so that the revised version of the manuscript can be accepted. Reviewer #3: Remarks to the Author: The authors have answered all questions to my satisfaction and especially the figures have been improved considerably. I support publication of the paper.	Reviewers' Comments: Reviewer #1: Remarks to the Author: The revised version of the work “Redox-Coupled Proton Pumping Drives Carbon Concentration in thePhotosynthetic Complex I “ presents a significant improved and corrected version of the manuscript. Authors have adressed and/or adressed most of my previous concerns as well as those presented by the other reviewers. Specifically, they have: i) added, as requested, a detailed methodology in the methods section now included as part of the main text, which properly describes theory levels and methodology ii) Clarified key elements of the three enezymes that are compared in the work (αCA, βCA and current enzyme). iii) Added proper description that energy calculations correspond to ΔH – TΔS + ΔZPE based on the electronic energy and the molecular Hessian. iv) Better explained the role of the titrable active site resiudes and pKa calculation v) Increased sampling and analysis of the waters present in the proton chanels vi) Provided estimate of the Free Energy Profile of CO2 migration along the proposed channel. Therefore, I believe the present work is suitable for publication. I only have a minor comment (described below) related to Figure 2b,. In the text desribing Figure 2 authors state (starting line 123) “ The CO 2 hydration is initiated by proton transfer (pT) from the Zn-bound water molecule followed by a nucleophilic attack.. of the hydroxide on CO 2 . In αCA (βCA), the rate limiting reaction barrier of 11 (12) kcal mol -1 is connected with pT to His64 (Tyr205), … In CupA, proton transfer from Zn-bound water to Tyr41 is slightly exergonic” Therefore, if I understood it correglty, in CupA Zn bound watre (which then acts as the nucelophile) transfer the proton to Tyr41 (in αCA /βCA the acceptor being His64/Tyr205. However, in Figure 2B, the Zn bound water transfer the proton to ANOTHER water molecule, whiwhc then transfers it to Tyr/His/Tyr. So, are there 1 or 2 water molecules?. In line with this point in Figure 2A waters are missing. I strongly beliebe they should be shown and the issue needs to be clarified. Reviewer #2: Remarks to the Author: I think that the authors have addressed the questions raised by all reviwers adequately, so that the revised version of the manuscript can be accepted. Reviewer #3: Remarks to the Author: The authors have answered all questions to my satisfaction and especially the figures have been improved considerably. I support publication of the paper.	TPR	NC_BS_BIOPH_006_RR2.txt	{ "criticism": 1, "example": 3, "importance_and_relevance": 0, "materials_and_methods": 3, "praise": 4, "presentation_and_reporting": 6, "results_and_discussion": 0, "suggestion_and_solution": 4, "total": 16 }	{ "criticism": 0.0625, "example": 0.1875, "importance_and_relevance": 0, "materials_and_methods": 0.1875, "praise": 0.25, "presentation_and_reporting": 0.375, "results_and_discussion": 0, "suggestion_and_solution": 0.25 }	1.3125	[ { "content": "Please review this paper.", "role": "user" }, { "content": " \n \n\n\fReviewers' Comments: \n \nReviewer #1: \nRemarks to the Author: \nThe revised version of the work “Redox-Coupled Proton Pumping Drives Carbon Concentration in \nthePhotosynthetic Complex I “ presents a significant improved and corrected version of the \nmanuscript. Authors have adressed and/or adressed most of my previous concerns as well as \nthose presented by the other reviewers. \nSpecifically, they have: \ni) added, as requested, a detailed methodology in the methods section now included as part of the \nmain text, which properly describes theory levels and methodology \nii) Clarified key elements of the three enezymes that are compared in the work (αCA, βCA and \ncurrent enzyme). \niii) Added proper description that energy calculations correspond to ΔH – TΔS + ΔZPE based on the \nelectronic energy and the molecular Hessian. \niv) Better explained the role of the titrable active site resiudes and pKa calculation \nv) Increased sampling and analysis of the waters present in the proton chanels \nvi) Provided estimate of the Free Energy Profile of CO2 migration along the proposed channel. \n \nTherefore, I believe the present work is suitable for publication. I only have a minor comment \n(described below) related to Figure 2b,. \n \nIn the text desribing Figure 2 authors state (starting line 123) “ The CO 2 hydration is initiated by \nproton transfer (pT) from the Zn-bound water molecule followed by a nucleophilic attack.. of the \nhydroxide on CO 2 . In αCA (βCA), the rate limiting reaction barrier of 11 (12) kcal mol -1 is \nconnected with pT to His64 (Tyr205), … In CupA, proton transfer from Zn-bound water to Tyr41 is \nslightly exergonic” Therefore, if I understood it correglty, in CupA Zn bound watre (which then acts \nas the nucelophile) transfer the proton to Tyr41 (in αCA /βCA the acceptor being His64/Tyr205. \nHowever, in Figure 2B, the Zn bound water transfer the proton to ANOTHER water molecule, \nwhiwhc then transfers it to Tyr/His/Tyr. So, are there 1 or 2 water molecules?. \nIn line with this point in Figure 2A waters are missing. I strongly beliebe they should be shown and \nthe issue needs to be clarified. \n \n \n \nReviewer #2: \nRemarks to the Author: \nI think that the authors have addressed the questions raised by all reviwers adequately, so that \nthe revised version of the manuscript can be accepted. \n \n \n \nReviewer #3: \nRemarks to the Author: \nThe authors have answered all questions to my satisfaction and especially the figures have been \nimproved considerably. I support publication of the paper. \n \n", "role": "assistant" } ]	2,020
Redox-coupled proton pumping drives carbon concentration in the photosynthetic complex I	Reviewers' Comments: Reviewer #1: Remarks to the Author: The work entitled “Redox-Coupled Proton Pumping Drives Carbon Concentration in the Photosynthetic Complex I “ presents a structural study which combines CrioEM data, with Structural modeling, Molecular Dynamics (MD) and QM based calculations to analyze several important mechanistic aspects of structure function relationships in photosynthetic complex I. The results reveal a novel catalytic carbonic anhydrase module (different form alpha and beta carbonic anhydrases), with a unique CupA active site harboring a Zn 2+ coordinate to His and Arg residues and proposed to be connected thorugh proton exchange/transfer to the proton-pumping modules which are coupled by a 200 Å electrostatic wire to the electron transfer module of the enzyme. A proposed CO2 channel connecting the protein boundary with the Zn active site is also proposed and several mechanistic conclusions are drawn based on a interpretation of the structure and molecular simulation results. Although the work is novel, relevant and very interesting, it falls short on many technical aspects (particularly related to the molecular simulations) which prevent publication in its current form. Also, presentation should be improved since many important details have been left out in the main text which are important and necessary to understand and really gauge the impact of the work. Therefore I suggest publication only provided major revisions are performed and simulation aspects improved, Specific points follow: The pargraph beginning with “ To probe the catalytic properties of the CupA module, we performed quantum chemical calculations on the CO 2 hydration reaction and compared this to the reaction energetic in aCA and bCA (Fig. 2b, Extended Data Fig. 5, see Methods)... “ needs lots of work. First, in the main tecxt the authors give no detail on the performed calculations, just mentioned that they used QM/MM methods. Authors should properly described the level of theory, and the type of calculation, particular in relation as to how the energy barriers were obtained. Moreover, some brief explanation of the reaction mechanism for ll three enzymes bCA, aCa and CupA is needed in order to proper understand the differences. Key elements are missing, such as who is the key residue acting as base, that removes the proton from the Zn coordinated water that is supposed to initiate the reaction. Figure 2, needs more explanation. In particular concerning the energy diagram. Authors write DG, however they only computed “energies” with entropic corrections using single points?. Therefore, the profiles are not actual G but E+ZPE. To really compute free energy profiles authors need to perform QM/MM dynamics with some free energy (or PMF) determination method such as Umbrella sampling, Metadynamics or MSMD. Authors need to provide further detail on all three computed reactions. bCA, aCA and CupA. In supplementary data only the profile for CupA step 1 is shown, no data is provided for bCA and aCA and the remaining steps of CupA. Authors MUST provide free energy (or E+ZPE profiles if they want) of all steps in all three enzymes. Moreover, authors should clearly explain the reaction coordinate used in each reaction step and describe initial conditions. Also important, since the reaction involves proton transfer. The authors start with Y41 deprotonated, but who takes the proton from Y41?. Who acts as base in a/bCA??. Also related to Y41 and pka calculations (Figure 5F from extended data). Pka was computed using a continuum model which is not really accurate, authors should actually compute the pKa using better methods such as constant pH Molecular dynamics (Se Dr Roibergs works on the subject). Why do they report two pkas??? Concerning Extended data Figure 6. Authors performed MD simulation and looked at the presence of water molecules which they claim show potential proton transfer wires?. However no quantitative measure of the solvation is given. Authors could compute key position of water molecules inside the protein using IFST method, i.e determining the Water Sites and their properties (see for examples works by Gauto et. al. and/or WATCLUST method doi: 10.1093/bioinformatics/btv411). Finally, authors claim that protein hosts a tunnel that concentrates CO2. However, they only provide anecdotal evidence of CO2 moving inside the proposed tunnel. Authors should compute the free energy profile of CO2 migration along the tunnel to really measure how much the tunnel contributes (in relative amount) to concentrate CO2 inside the protein. See for example how this is performed by Myglobin to concentrate O2 inside the active site doi: 10.1074/jbc.M112.426056) Marcelo A. Marti Reviewer #2: Remarks to the Author: Energy-converting NADH:ubiquinone oxidoreductase, respiratory complex I, plays a major role in energy metabolism. In humans, defects in complex I are linked to severe neurodegenerative diseases. During the last years several sub-families of energy converting NADH:ubiquinone oxidoreductase have been identified that evolved from each other and that share a similar structure and a related mechanism. Cyanobacteria and plants contain an energy-converting ferredoxin:plastoquinone oxidoreductase consisting of several subunits with most of them being related to those of mitochondrial complex I. Very recently, the structure of this enzyme complex, also called photosynthetic complex I, from Thermosynechococcus elongatus was determined by means of cryo-electron microscopy (cryo-EM). The manuscript by Schuller et al. describes the cryo-EM structure of the NDH-1MS isoform of the Thermosynechococcus elongates photosynthetic complex I. This isoform is equipped with an additional module catalyzing a carbonic anhydrase (CA) reaction. The authors describe the ferredoxin and the plastoquinone binding sites and several lipids bound to the enzyme complex. Most importantly, they demonstrate that the CA domain is made up of CupA and S on the cytoplasmic side of the membrane. The authors unequivocally show that this domain contains a catalytically active Zn(2+) ion that is ligated in a different way compared to the well-known α and ß CAs leading to a different type of mechanism. The proposed mechanism is substantiated by theoretical energy calculations. It is nicely argued that proton translocation by the photosynthetic complex I facilitates the CA reaction by removing the proton, a reaction product. Furthermore, a mechanism is provided that could explain the propagation of the proton translocation along the membrane part of the complex. The manuscript is very well written, easy to follow and contains novel and highly interesting information. The experiments and the data are sound and well documented. The manuscript benefits very much from the impressive interplay of structural analysis and topical theoretical methods. I just have a few minor points to possibly be considered: Page 2, third para, first sentence: ‘The 0.5 MDa complex has an overall U-shape with 18 isolated subunits.’ However, in Extended Data Table 2, 19 subunits are listed. Page 3, first para, first sentence: ‘The CO2-concentrating CupA/S module (CO2 uptake/CO2 hydration protein, ChpY)….’ ‘up’ is underlined but does not contribute to the abbreviation. The abbreviation reads as CO2 hydration protein, maybe a better wording can be found in order to not confuse the non-experts. Page 3, second para, last sentence: Is there any evidence that the chlorophyll α/ß-carotene play a role in light-induced regulation? If not, I would delete the sentence in the main text and maybe place it as pure speculation in the legend of the corresponding figure. Extended Data Fig. 2: I would propose to also label the subunits above NdhF3 in 2a. It should be explained in the legend why NdhN and M appear twice in 2c/e. Reviewer #3: Remarks to the Author: Cyanobacteria contain a set of enzymes related to respiratory complex I, the NDH-1 family, with diverse functions that appear to have evolved to deal with the challenges of oxygenic photosynthesis. A number of papers appeared earlier this year that showed the structure of NDH- 1L, which is similar to respiratory complex I, but lacks the first three subunits of this complex that transfer electrons from NADH to the quinone binding site. Instead, a number of specific subunits allow electron transfer from ferredoxin. The paper by Schuller et al. concerns the cryo-EM structure and molecular dynamics studies of NDH-1MS from Thermosynechococcus elongatus, an enzyme involved in carbon concentration. The authors show that NDH-1MS contains a noncanonical carbonic anhydrase module, located at the end of the proton transfer chain, and develop a mechanism for its function in concentration CO2. The combination of cryo-EM, model building and molecular dynamics studies yields new insights in cyanobacterial photosynthesis that will be of high interest to experts in the field. The paper is overall well-written; it would however benefit from some modifications, in particular concerning the figures and figure legends, as detailed below. Specific points: 1. p. 3, assignment of the zinc ion in CupA: the cryo-EM density for this ion and its coordination should be shown in a supplementary figure. The zinc ion should also be shown in figure 3a. Further, the text states that the assignment is “supported by multi-element analysis data (Extended Data Fig. 2e)”. However, this figure does not show this. There is mention of Zn in Extended Data Fig. 2d, but this is “Inductively-Coupled Plasma Optical Emission Spectrometry”. It is totally unclear what this table (2d) shows, and neither this technique nor multi-element analysis are mentioned in the Methods. Further, it is stated that the Zn ligand Arg135 has a pKa <7, but Extended Data fig. 5f states pKa <0. 2. p. 3 third paragraph, the sentence starting “a non-polar tunnel…” is ungrammatical and can’t be understood. 3. p. 4 “global dynamics inferred from the cryo-EM map” would be clearer as “global dynamics inferred from the local resolution of the cryo-EM map”. The legends of Ext. Data Fig. 7, which show this, are inadequate and should state more explicitly what the color scheme represents. 4. p. 4 “charged residues in the broken helices TM7 and TM12 (Fig. 3)”. It is not mentioned which subunit these helices belong to. Figure 3d shows these helices in all three antiporter subunits NdhF3, D3 and B. It would be better to state this in the text and refer to figure 3d instead of figure 3 in full. 5. p. 5 “coupled protonation and/or conformational changes at the NdhF3/NdhD3 interface could close the gas channel and decouple the pump…”. This part is very speculative. It is not clear why the reverse reaction would close the channel that is presumably always open in the forward reaction. 6. Figure 1: This figure shows the cryo-EM map of the protein with a simulated, full atomic model of the membrane. Experimentally determined lipids (as seen in Ext. Data Fig. 3) are however not shown. The atomic model distracts from the protein map and partially obscures it (e.g. NdhL). It would be better to show the EM density map by itself, including observed non-protein density. Further, panel c does not show a back view, but a side view. It may be a good idea to reverse panel a and c, as a has the same orientation as the model in d, which would then be next to it. In panel c, it is not clear what subunit forms the horizontal surface helix (brownish) below NdhK. 7. Figure 3: presumably the CO2 channel is shown in purple in a, b and c, but this is never stated. It looks very different in b and c, how was the surface determined? Also the scale and orientation of these panels differs considerably and it is unclear how they relate to each other. 3d: what is an “experimentally-refined structure”? There are no densities shown in figure 3, so referring to “further example densities” is strange. 8. Figure 4: This figure could do with better legends to explain the colors and arrows. Why is one of the H+ in the lower panel grey? 9. Methods, image processing: “the motion correction algorithm” should be specified as MotionCorr2 and the reference added. FSC does not stand for “Fourier shell correction” but “Fourier shell correlation.” 10. Extended Data Fig. 1: twice refinement is misspelled as “refinment”. 11. Extended Data Fig. 3: PGT, SQD and DGD should be defined. It is better not to show hydrogens in the atomic models, they clutter the images. 12. Extended Data Fig. 4a: It is unclear what is shown here. Legend for the left panel is missing and the right panel does not show the interaction between the subunits, just the surface. 13. Extended Data Fig. 4b: It is unclear what is shown here. Is orange the pdb ID 2MXA, as stated in the legends, or the cryo-EM structure? The legend also states that the solution structure is cyan, which seems more likely. Further, helix α3, mentioned in the legend and in the text on page 3, is not shown. The text mentions movement of α2, α3, the legend α1, α2, α3. 14. Extended Data Fig. 5f: the legend mention “the latter values”. This makes no sense; presumably what is meant is the second figure in each column. 15. Extended Data Fig. 6e: it is not clear what is shown here. What color is mouse and what cyanobacterial? 16. Extended Data Fig. 7: “Dynamics from cryo-EM resolution map” should be “Dynamics from local resolution of the cryo-EM map”.	Reviewers' Comments: Reviewer #1: Remarks to the Author: The work entitled “Redox-Coupled Proton Pumping Drives Carbon Concentration in the Photosynthetic Complex I “ presents a structural study which combines CrioEM data, with Structural modeling, Molecular Dynamics (MD) and QM based calculations to analyze several important mechanistic aspects of structure function relationships in photosynthetic complex I. The results reveal a novel catalytic carbonic anhydrase module (different form alpha and beta carbonic anhydrases), with a unique CupA active site harboring a Zn 2+ coordinate to His and Arg residues and proposed to be connected thorugh proton exchange/transfer to the proton-pumping modules which are coupled by a 200 Å electrostatic wire to the electron transfer module of the enzyme. A proposed CO2 channel connecting the protein boundary with the Zn active site is also proposed and several mechanistic conclusions are drawn based on a interpretation of the structure and molecular simulation results. Although the work is novel, relevant and very interesting, it falls short on many technical aspects (particularly related to the molecular simulations) which prevent publication in its current form. Also, presentation should be improved since many important details have been left out in the main text which are important and necessary to understand and really gauge the impact of the work. Therefore I suggest publication only provided major revisions are performed and simulation aspects improved, Specific points follow: The pargraph beginning with “ To probe the catalytic properties of the CupA module, we performed quantum chemical calculations on the CO 2 hydration reaction and compared this to the reaction energetic in aCA and bCA (Fig. 2b, Extended Data Fig. 5, see Methods)... “ needs lots of work. First, in the main tecxt the authors give no detail on the performed calculations, just mentioned that they used QM/MM methods. Authors should properly described the level of theory, and the type of calculation, particular in relation as to how the energy barriers were obtained. Moreover, some brief explanation of the reaction mechanism for ll three enzymes bCA, aCa and CupA is needed in order to proper understand the differences. Key elements are missing, such as who is the key residue acting as base, that removes the proton from the Zn coordinated water that is supposed to initiate the reaction. Figure 2, needs more explanation. In particular concerning the energy diagram. Authors write DG, however they only computed “energies” with entropic corrections using single points?. Therefore, the profiles are not actual G but E+ZPE. To really compute free energy profiles authors need to perform QM/MM dynamics with some free energy (or PMF) determination method such as Umbrella sampling, Metadynamics or MSMD. Authors need to provide further detail on all three computed reactions. bCA, aCA and CupA. In supplementary data only the profile for CupA step 1 is shown, no data is provided for bCA and aCA and the remaining steps of CupA. Authors MUST provide free energy (or E+ZPE profiles if they want) of all steps in all three enzymes. Moreover, authors should clearly explain the reaction coordinate used in each reaction step and describe initial conditions. Also important, since the reaction involves proton transfer. The authors start with Y41 deprotonated, but who takes the proton from Y41?. Who acts as base in a/bCA??. Also related to Y41 and pka calculations (Figure 5F from extended data). Pka was computed using a continuum model which is not really accurate, authors should actually compute the pKa using better methods such as constant pH Molecular dynamics (Se Dr Roibergs works on the subject). Why do they report two pkas??? Concerning Extended data Figure 6. Authors performed MD simulation and looked at the presence of water molecules which they claim show potential proton transfer wires?. However no quantitative measure of the solvation is given. Authors could compute key position of water molecules inside the protein using IFST method, i.e determining the Water Sites and their properties (see for examples works by Gauto et. al. and/or WATCLUST method doi: 10.1093/bioinformatics/btv411). Finally, authors claim that protein hosts a tunnel that concentrates CO2. However, they only provide anecdotal evidence of CO2 moving inside the proposed tunnel. Authors should compute the free energy profile of CO2 migration along the tunnel to really measure how much the tunnel contributes (in relative amount) to concentrate CO2 inside the protein. See for example how this is performed by Myglobin to concentrate O2 inside the active site doi: 10.1074/jbc.M112.426056) Marcelo A. Marti Reviewer #2: Remarks to the Author: Energy-converting NADH:ubiquinone oxidoreductase, respiratory complex I, plays a major role in energy metabolism. In humans, defects in complex I are linked to severe neurodegenerative diseases. During the last years several sub-families of energy converting NADH:ubiquinone oxidoreductase have been identified that evolved from each other and that share a similar structure and a related mechanism. Cyanobacteria and plants contain an energy-converting ferredoxin:plastoquinone oxidoreductase consisting of several subunits with most of them being related to those of mitochondrial complex I. Very recently, the structure of this enzyme complex, also called photosynthetic complex I, from Thermosynechococcus elongatus was determined by means of cryo-electron microscopy (cryo-EM). The manuscript by Schuller et al. describes the cryo-EM structure of the NDH-1MS isoform of the Thermosynechococcus elongates photosynthetic complex I. This isoform is equipped with an additional module catalyzing a carbonic anhydrase (CA) reaction. The authors describe the ferredoxin and the plastoquinone binding sites and several lipids bound to the enzyme complex. Most importantly, they demonstrate that the CA domain is made up of CupA and S on the cytoplasmic side of the membrane. The authors unequivocally show that this domain contains a catalytically active Zn(2+) ion that is ligated in a different way compared to the well-known α and ß CAs leading to a different type of mechanism. The proposed mechanism is substantiated by theoretical energy calculations. It is nicely argued that proton translocation by the photosynthetic complex I facilitates the CA reaction by removing the proton, a reaction product. Furthermore, a mechanism is provided that could explain the propagation of the proton translocation along the membrane part of the complex. The manuscript is very well written, easy to follow and contains novel and highly interesting information. The experiments and the data are sound and well documented. The manuscript benefits very much from the impressive interplay of structural analysis and topical theoretical methods. I just have a few minor points to possibly be considered: Page 2, third para, first sentence: ‘The 0.5 MDa complex has an overall U-shape with 18 isolated subunits.’ However, in Extended Data Table 2, 19 subunits are listed. Page 3, first para, first sentence: ‘The CO2-concentrating CupA/S module (CO2 uptake/CO2 hydration protein, ChpY)….’ ‘up’ is underlined but does not contribute to the abbreviation. The abbreviation reads as CO2 hydration protein, maybe a better wording can be found in order to not confuse the non-experts. Page 3, second para, last sentence: Is there any evidence that the chlorophyll α/ß-carotene play a role in light-induced regulation? If not, I would delete the sentence in the main text and maybe place it as pure speculation in the legend of the corresponding figure. Extended Data Fig. 2: I would propose to also label the subunits above NdhF3 in 2a. It should be explained in the legend why NdhN and M appear twice in 2c/e. Reviewer #3: Remarks to the Author: Cyanobacteria contain a set of enzymes related to respiratory complex I, the NDH-1 family, with diverse functions that appear to have evolved to deal with the challenges of oxygenic photosynthesis. A number of papers appeared earlier this year that showed the structure of NDH- 1L, which is similar to respiratory complex I, but lacks the first three subunits of this complex that transfer electrons from NADH to the quinone binding site. Instead, a number of specific subunits allow electron transfer from ferredoxin. The paper by Schuller et al. concerns the cryo-EM structure and molecular dynamics studies of NDH-1MS from Thermosynechococcus elongatus, an enzyme involved in carbon concentration. The authors show that NDH-1MS contains a noncanonical carbonic anhydrase module, located at the end of the proton transfer chain, and develop a mechanism for its function in concentration CO2. The combination of cryo-EM, model building and molecular dynamics studies yields new insights in cyanobacterial photosynthesis that will be of high interest to experts in the field. The paper is overall well-written; it would however benefit from some modifications, in particular concerning the figures and figure legends, as detailed below. Specific points: 1. p. 3, assignment of the zinc ion in CupA: the cryo-EM density for this ion and its coordination should be shown in a supplementary figure. The zinc ion should also be shown in figure 3a. Further, the text states that the assignment is “supported by multi-element analysis data (Extended Data Fig. 2e)”. However, this figure does not show this. There is mention of Zn in Extended Data Fig. 2d, but this is “Inductively-Coupled Plasma Optical Emission Spectrometry”. It is totally unclear what this table (2d) shows, and neither this technique nor multi-element analysis are mentioned in the Methods. Further, it is stated that the Zn ligand Arg135 has a pKa <7, but Extended Data fig. 5f states pKa <0. 2. p. 3 third paragraph, the sentence starting “a non-polar tunnel…” is ungrammatical and can’t be understood. 3. p. 4 “global dynamics inferred from the cryo-EM map” would be clearer as “global dynamics inferred from the local resolution of the cryo-EM map”. The legends of Ext. Data Fig. 7, which show this, are inadequate and should state more explicitly what the color scheme represents. 4. p. 4 “charged residues in the broken helices TM7 and TM12 (Fig. 3)”. It is not mentioned which subunit these helices belong to. Figure 3d shows these helices in all three antiporter subunits NdhF3, D3 and B. It would be better to state this in the text and refer to figure 3d instead of figure 3 in full. 5. p. 5 “coupled protonation and/or conformational changes at the NdhF3/NdhD3 interface could close the gas channel and decouple the pump…”. This part is very speculative. It is not clear why the reverse reaction would close the channel that is presumably always open in the forward reaction. 6. Figure 1: This figure shows the cryo-EM map of the protein with a simulated, full atomic model of the membrane. Experimentally determined lipids (as seen in Ext. Data Fig. 3) are however not shown. The atomic model distracts from the protein map and partially obscures it (e.g. NdhL). It would be better to show the EM density map by itself, including observed non-protein density. Further, panel c does not show a back view, but a side view. It may be a good idea to reverse panel a and c, as a has the same orientation as the model in d, which would then be next to it. In panel c, it is not clear what subunit forms the horizontal surface helix (brownish) below NdhK. 7. Figure 3: presumably the CO2 channel is shown in purple in a, b and c, but this is never stated. It looks very different in b and c, how was the surface determined? Also the scale and orientation of these panels differs considerably and it is unclear how they relate to each other. 3d: what is an “experimentally-refined structure”? There are no densities shown in figure 3, so referring to “further example densities” is strange. 8. Figure 4: This figure could do with better legends to explain the colors and arrows. Why is one of the H+ in the lower panel grey? 9. Methods, image processing: “the motion correction algorithm” should be specified as MotionCorr2 and the reference added. FSC does not stand for “Fourier shell correction” but “Fourier shell correlation.” 10. Extended Data Fig. 1: twice refinement is misspelled as “refinment”. 11. Extended Data Fig. 3: PGT, SQD and DGD should be defined. It is better not to show hydrogens in the atomic models, they clutter the images. 12. Extended Data Fig. 4a: It is unclear what is shown here. Legend for the left panel is missing and the right panel does not show the interaction between the subunits, just the surface. 13. Extended Data Fig. 4b: It is unclear what is shown here. Is orange the pdb ID 2MXA, as stated in the legends, or the cryo-EM structure? The legend also states that the solution structure is cyan, which seems more likely. Further, helix α3, mentioned in the legend and in the text on page 3, is not shown. The text mentions movement of α2, α3, the legend α1, α2, α3. 14. Extended Data Fig. 5f: the legend mention “the latter values”. This makes no sense; presumably what is meant is the second figure in each column. 15. Extended Data Fig. 6e: it is not clear what is shown here. What color is mouse and what cyanobacterial? 16. Extended Data Fig. 7: “Dynamics from cryo-EM resolution map” should be “Dynamics from local resolution of the cryo-EM map”.	TPR	NC_BS_BIOPH_006_RR1.txt	{ "criticism": 26, "example": 32, "importance_and_relevance": 6, "materials_and_methods": 66, "praise": 8, "presentation_and_reporting": 50, "results_and_discussion": 5, "suggestion_and_solution": 31, "total": 144 }	{ "criticism": 0.18055555555555555, "example": 0.2222222222222222, "importance_and_relevance": 0.041666666666666664, "materials_and_methods": 0.4583333333333333, "praise": 0.05555555555555555, "presentation_and_reporting": 0.3472222222222222, "results_and_discussion": 0.034722222222222224, "suggestion_and_solution": 0.2152777777777778 }	1.555556	[ { "content": "Please review this paper.", "role": "user" }, { "content": "Reviewers' Comments: \n \nReviewer #1: \nRemarks to the Author: \nThe work entitled “Redox-Coupled Proton Pumping Drives Carbon Concentration in the \nPhotosynthetic Complex I “ presents a structural study which combines CrioEM data, with \nStructural modeling, Molecular Dynamics (MD) and QM based calculations to analyze several \nimportant mechanistic aspects of structure function relationships in photosynthetic complex I. The \nresults reveal a novel catalytic carbonic anhydrase module (different form alpha and beta carbonic \nanhydrases), with a unique CupA active site harboring a Zn 2+ coordinate to His and Arg residues \nand proposed to be connected thorugh proton exchange/transfer to the proton-pumping modules \nwhich are coupled by a 200 Å electrostatic wire to the electron transfer module of the enzyme. A \nproposed CO2 channel connecting the protein boundary with the Zn active site is also proposed \nand several mechanistic conclusions are drawn based on a interpretation of the structure and \nmolecular simulation results. \nAlthough the work is novel, relevant and very interesting, it falls short on many technical aspects \n(particularly related to the molecular simulations) which prevent publication in its current form. \nAlso, presentation should be improved since many important details have been left out in the main \ntext which are important and necessary to understand and really gauge the impact of the work. \nTherefore I suggest publication only provided major revisions are performed and simulation \naspects improved, Specific points follow: \n \n \nThe pargraph beginning with “ To probe the catalytic properties of the CupA module, we performed \nquantum chemical calculations on the CO 2 hydration reaction and compared this to the reaction \nenergetic in aCA and bCA (Fig. 2b, Extended Data Fig. 5, see Methods)... “ needs lots of work. \nFirst, in the main tecxt the authors give no detail on the performed calculations, just mentioned \nthat they used QM/MM methods. Authors should properly described the level of theory, and the \ntype of calculation, particular in relation as to how the energy barriers were obtained. \nMoreover, some brief explanation of the reaction mechanism for ll three enzymes bCA, aCa and \nCupA is needed in order to proper understand the differences. Key elements are missing, such as \nwho is the key residue acting as base, that removes the proton from the Zn coordinated water that \nis supposed to initiate the reaction. \n \nFigure 2, needs more explanation. In particular concerning the energy diagram. Authors write DG, \nhowever they only computed “energies” with entropic corrections using single points?. Therefore, \nthe profiles are not actual G but E+ZPE. To really compute free energy profiles authors need to \nperform QM/MM dynamics with some free energy (or PMF) determination method such as Umbrella \nsampling, Metadynamics or MSMD. \n \nAuthors need to provide further detail on all three computed reactions. bCA, aCA and CupA. In \nsupplementary data only the profile for CupA step 1 is shown, no data is provided for bCA and aCA \nand the remaining steps of CupA. Authors MUST provide free energy (or E+ZPE profiles if they \nwant) of all steps in all three enzymes. Moreover, authors should clearly explain the reaction \ncoordinate used in each reaction step and describe initial conditions. \n \nAlso important, since the reaction involves proton transfer. The authors start with Y41 \ndeprotonated, but who takes the proton from Y41?. Who acts as base in a/bCA??. Also related to \nY41 and pka calculations (Figure 5F from extended data). Pka was computed using a continuum \nmodel which is not really accurate, authors should actually compute the pKa using better methods \nsuch as constant pH Molecular dynamics (Se Dr Roibergs works on the subject). Why do they \nreport two pkas??? \n \nConcerning Extended data Figure 6. Authors performed MD simulation and looked at the presence \nof water molecules which they claim show potential proton transfer wires?. However no \n\n\fquantitative measure of the solvation is given. Authors could compute key position of water \nmolecules inside the protein using IFST method, i.e determining the Water Sites and their \nproperties (see for examples works by Gauto et. al. and/or WATCLUST method doi: \n10.1093/bioinformatics/btv411). \n \nFinally, authors claim that protein hosts a tunnel that concentrates CO2. However, they only \nprovide anecdotal evidence of CO2 moving inside the proposed tunnel. Authors should compute \nthe free energy profile of CO2 migration along the tunnel to really measure how much the tunnel \ncontributes (in relative amount) to concentrate CO2 inside the protein. See for example how this is \nperformed by Myglobin to concentrate O2 inside the active site doi: 10.1074/jbc.M112.426056) \n \nMarcelo A. Marti \n \n \n \nReviewer #2: \nRemarks to the Author: \nEnergy-converting NADH:ubiquinone oxidoreductase, respiratory complex I, plays a major role in \nenergy metabolism. In humans, defects in complex I are linked to severe neurodegenerative \ndiseases. During the last years several sub-families of energy converting NADH:ubiquinone \noxidoreductase have been identified that evolved from each other and that share a similar \nstructure and a related mechanism. Cyanobacteria and plants contain an energy-converting \nferredoxin:plastoquinone oxidoreductase consisting of several subunits with most of them being \nrelated to those of mitochondrial complex I. Very recently, the structure of this enzyme complex, \nalso called photosynthetic complex I, from Thermosynechococcus elongatus was determined by \nmeans of cryo-electron microscopy (cryo-EM). \n \nThe manuscript by Schuller et al. describes the cryo-EM structure of the NDH-1MS isoform of the \nThermosynechococcus elongates photosynthetic complex I. This isoform is equipped with an \nadditional module catalyzing a carbonic anhydrase (CA) reaction. The authors describe the \nferredoxin and the plastoquinone binding sites and several lipids bound to the enzyme complex. \nMost importantly, they demonstrate that the CA domain is made up of CupA and S on the \ncytoplasmic side of the membrane. The authors unequivocally show that this domain contains a \ncatalytically active Zn(2+) ion that is ligated in a different way compared to the well-known α and \nß CAs leading to a different type of mechanism. The proposed mechanism is substantiated by \ntheoretical energy calculations. It is nicely argued that proton translocation by the photosynthetic \ncomplex I facilitates the CA reaction by removing the proton, a reaction product. Furthermore, a \nmechanism is provided that could explain the propagation of the proton translocation along the \nmembrane part of the complex. \n \nThe manuscript is very well written, easy to follow and contains novel and highly interesting \ninformation. The experiments and the data are sound and well documented. The manuscript \nbenefits very much from the impressive interplay of structural analysis and topical theoretical \nmethods. I just have a few minor points to possibly be considered: \n \nPage 2, third para, first sentence: ‘The 0.5 MDa complex has an overall U-shape with 18 isolated \nsubunits.’ However, in Extended Data Table 2, 19 subunits are listed. \n \nPage 3, first para, first sentence: ‘The CO2-concentrating CupA/S module (CO2 uptake/CO2 \nhydration protein, ChpY)….’ ‘up’ is underlined but does not contribute to the abbreviation. The \nabbreviation reads as CO2 hydration protein, maybe a better wording can be found in order to not \nconfuse the non-experts. \n \nPage 3, second para, last sentence: Is there any evidence that the chlorophyll α/ß-carotene play a \nrole in light-induced regulation? If not, I would delete the sentence in the main text and maybe \n\n\fplace it as pure speculation in the legend of the corresponding figure. \n \nExtended Data Fig. 2: I would propose to also label the subunits above NdhF3 in 2a. It should be \nexplained in the legend why NdhN and M appear twice in 2c/e. \n \n \n \nReviewer #3: \nRemarks to the Author: \nCyanobacteria contain a set of enzymes related to respiratory complex I, the NDH-1 family, with \ndiverse functions that appear to have evolved to deal with the challenges of oxygenic \nphotosynthesis. A number of papers appeared earlier this year that showed the structure of NDH-\n1L, which is similar to respiratory complex I, but lacks the first three subunits of this complex that \ntransfer electrons from NADH to the quinone binding site. Instead, a number of specific subunits \nallow electron transfer from ferredoxin. The paper by Schuller et al. concerns the cryo-EM \nstructure and molecular dynamics studies of NDH-1MS from Thermosynechococcus elongatus, an \nenzyme involved in carbon concentration. The authors show that NDH-1MS contains a \nnoncanonical carbonic anhydrase module, located at the end of the proton transfer chain, and \ndevelop a mechanism for its function in concentration CO2. \n \nThe combination of cryo-EM, model building and molecular dynamics studies yields new insights in \ncyanobacterial photosynthesis that will be of high interest to experts in the field. The paper is \noverall well-written; it would however benefit from some modifications, in particular concerning \nthe figures and figure legends, as detailed below. \n \nSpecific points: \n1. p. 3, assignment of the zinc ion in CupA: the cryo-EM density for this ion and its coordination \nshould be shown in a supplementary figure. The zinc ion should also be shown in figure 3a. \nFurther, the text states that the assignment is “supported by multi-element analysis data \n(Extended Data Fig. 2e)”. However, this figure does not show this. There is mention of Zn in \nExtended Data Fig. 2d, but this is “Inductively-Coupled Plasma Optical Emission Spectrometry”. It \nis totally unclear what this table (2d) shows, and neither this technique nor multi-element analysis \nare mentioned in the Methods. Further, it is stated that the Zn ligand Arg135 has a pKa <7, but \nExtended Data fig. 5f states pKa <0. \n \n2. p. 3 third paragraph, the sentence starting “a non-polar tunnel…” is ungrammatical and can’t be \nunderstood. \n \n3. p. 4 “global dynamics inferred from the cryo-EM map” would be clearer as “global dynamics \ninferred from the local resolution of the cryo-EM map”. The legends of Ext. Data Fig. 7, which show \nthis, are inadequate and should state more explicitly what the color scheme represents. \n \n4. p. 4 “charged residues in the broken helices TM7 and TM12 (Fig. 3)”. It is not mentioned which \nsubunit these helices belong to. Figure 3d shows these helices in all three antiporter subunits \nNdhF3, D3 and B. It would be better to state this in the text and refer to figure 3d instead of figure \n3 in full. \n \n5. p. 5 “coupled protonation and/or conformational changes at the NdhF3/NdhD3 interface could \nclose the gas channel and decouple the pump…”. This part is very speculative. It is not clear why \nthe reverse reaction would close the channel that is presumably always open in the forward \nreaction. \n \n6. Figure 1: This figure shows the cryo-EM map of the protein with a simulated, full atomic model \nof the membrane. Experimentally determined lipids (as seen in Ext. Data Fig. 3) are however not \nshown. The atomic model distracts from the protein map and partially obscures it (e.g. NdhL). It \n\n\fwould be better to show the EM density map by itself, including observed non-protein density. \nFurther, panel c does not show a back view, but a side view. It may be a good idea to reverse \npanel a and c, as a has the same orientation as the model in d, which would then be next to it. In \npanel c, it is not clear what subunit forms the horizontal surface helix (brownish) below NdhK. \n \n7. Figure 3: presumably the CO2 channel is shown in purple in a, b and c, but this is never stated. \nIt looks very different in b and c, how was the surface determined? Also the scale and orientation \nof these panels differs considerably and it is unclear how they relate to each other. 3d: what is an \n“experimentally-refined structure”? There are no densities shown in figure 3, so referring to \n“further example densities” is strange. \n \n8. Figure 4: This figure could do with better legends to explain the colors and arrows. Why is one \nof the H+ in the lower panel grey? \n \n9. Methods, image processing: “the motion correction algorithm” should be specified as \nMotionCorr2 and the reference added. \nFSC does not stand for “Fourier shell correction” but “Fourier shell correlation.” \n \n10. Extended Data Fig. 1: twice refinement is misspelled as “refinment”. \n \n11. Extended Data Fig. 3: PGT, SQD and DGD should be defined. It is better not to show \nhydrogens in the atomic models, they clutter the images. \n \n12. Extended Data Fig. 4a: It is unclear what is shown here. Legend for the left panel is missing \nand the right panel does not show the interaction between the subunits, just the surface. \n \n13. Extended Data Fig. 4b: It is unclear what is shown here. Is orange the pdb ID 2MXA, as stated \nin the legends, or the cryo-EM structure? The legend also states that the solution structure is cyan, \nwhich seems more likely. Further, helix α3, mentioned in the legend and in the text on page 3, is \nnot shown. The text mentions movement of α2, α3, the legend α1, α2, α3. \n \n14. Extended Data Fig. 5f: the legend mention “the latter values”. This makes no sense; \npresumably what is meant is the second figure in each column. \n \n15. Extended Data Fig. 6e: it is not clear what is shown here. What color is mouse and what \ncyanobacterial? \n \n16. Extended Data Fig. 7: “Dynamics from cryo-EM resolution map” should be “Dynamics from \nlocal resolution of the cryo-EM map”. \n\n\f", "role": "assistant" } ]	2,020
Tug-of-war between actomyosin-driven antagonistic forces determines the positioning	Reports from round 1: Rev#1 The authors have done an excellent job addressing the reviewer concerns, including addition of significant new experiments, all of which support their model. All critiques have been addressed, this is a lovely paper. Rev#2 The authors have sufficiently addressed referee concerns. The manuscript is improved and appropriate for publication in Nature Communications Rev#3 The authors went a long way to answer my previous and other reviewer comments. Overall the manuscript has improved with the new experiments and the qualitative understanding of the process seems to be supported by the data. I only have two concerns left. One minor is that the authors provide new analysis to distinguish between bulk contraction vs ring contraction. For that they refer to the new supplementary figure 4 that shows that the intensity of the ring goes up as it contracts whereas the bulk intensity remain constant. This would indeed suggest that the contraction is happening in the ring. However, this seems to contradict their exact same measurements in supplementary figure 2 :”(f) Fluorescence intensity along the each broken line in e, which also corresponds to the intensity at the arrows with the same color in d. Both the F-actin densities on the rim and inside of the wave increased during the contraction. The thickness of the spacer was 100 μm. Scale bars, 1 mm. “ Could the authors clarify the discrepancy? Does it depend on day to day or example taken? My main concern is still related to the theory: For their theory the authors state: “We have proposed that the off-centering is mediated by the percolated actomyosin bridge”, and that the bridge emerges from “low number of remained F-actin and newly assembled F-actin stochastically form bridges between the cluster and the boundary.” The authors are assuming the bridge is pre-formed and just calculate the probability that cross linkers connect the filaments that form the bridge. But, why don’t we see the bridge than initially after the wave? From their images it looks that actin is actually polymerized as the bridge is form (See figure 4c. After laser ablation a new bridge is formed but no wave has formed before it). Since the authors only consider the top 5% of actin lengths, shouldn’t this have a dramatic effect on time scales too? Meaning that it may be very difficult to actually have a bridge made of actin filaments longer than the 5% population as opposed of having a bridge made of short actin filaments. Of course that time scale to percolate a bridge of shorter filaments is longer, but it may be compensated by the time scales (or probability required to create enough long (~6 um) filaments. Also, how can the authors really justify to use 6 um as length based on their measured length distributions? These distributions clearly show an average of ~1um, and almost no filaments larger than 5 um.I think it is necessary that the authors clarify this point, comment on the origin of actin in these bridges, what would be the time scales associated to having only the 5% population of actin, and recalculate their predictions based on the average actin length they measure. The answer the authors give to use ~6 um is flawed. They argue “In the proposed model, the probability that a bridge is formed by the length L of F-actin in a droplet with radius R is written as P=(1/2)R/L. If we compare the probability with different actin length, L=10 μm and L=5 μm, in a droplet with radius R=50 μm, we can estimate P=(1/2)5 and P=(1/2)10 respectively. Thus, only 2 times change in F-actin length decreases the bridge formation probability ~30 times.”. The authors are missing the probability of actually getting actin filaments of that particular length. With their measured length distribution, this can also change by a decade from ~1 to ~5 um so it could counteract their estimated probability, so the authors need to take this into account in their argument. The authors distinguish gelation vs percolation to explain the difference of timescales, which they argue leads to a faster gelation time. However, my understanding is that gelation is a connectivity transition, and that previous research has used percolation as a theory to explain the gelation process. In both cases, they are network-connectivity transitions. Gelation further implies the emergence of material properties such as a finite young modulus, whereas percolation theory makes no statement (or can’t explain) about it and it is just restricted to the emergence of a system size cluster. Thus, I don’t really understand what the authors imply by “Since percolation is possible even without gelation, it is natural to assume that the bridge formation does not have to rely on the gelation process.”, and why would we expect that the percolation would take longer? Could the authors explain what the explicit differences between percolation and gelation they are impling?	Reports from round 1: Rev#1 The authors have done an excellent job addressing the reviewer concerns, including addition of significant new experiments, all of which support their model. All critiques have been addressed, this is a lovely paper. Rev#2 The authors have sufficiently addressed referee concerns. The manuscript is improved and appropriate for publication in Nature Communications Rev#3 The authors went a long way to answer my previous and other reviewer comments. Overall the manuscript has improved with the new experiments and the qualitative understanding of the process seems to be supported by the data. I only have two concerns left. One minor is that the authors provide new analysis to distinguish between bulk contraction vs ring contraction. For that they refer to the new supplementary figure 4 that shows that the intensity of the ring goes up as it contracts whereas the bulk intensity remain constant. This would indeed suggest that the contraction is happening in the ring. However, this seems to contradict their exact same measurements in supplementary figure 2 :”(f) Fluorescence intensity along the each broken line in e, which also corresponds to the intensity at the arrows with the same color in d. Both the F-actin densities on the rim and inside of the wave increased during the contraction. The thickness of the spacer was 100 μm. Scale bars, 1 mm. “ Could the authors clarify the discrepancy? Does it depend on day to day or example taken? My main concern is still related to the theory: For their theory the authors state: “We have proposed that the off-centering is mediated by the percolated actomyosin bridge”, and that the bridge emerges from “low number of remained F-actin and newly assembled F-actin stochastically form bridges between the cluster and the boundary.” The authors are assuming the bridge is pre-formed and just calculate the probability that cross linkers connect the filaments that form the bridge. But, why don’t we see the bridge than initially after the wave? From their images it looks that actin is actually polymerized as the bridge is form (See figure 4c. After laser ablation a new bridge is formed but no wave has formed before it). Since the authors only consider the top 5% of actin lengths, shouldn’t this have a dramatic effect on time scales too? Meaning that it may be very difficult to actually have a bridge made of actin filaments longer than the 5% population as opposed of having a bridge made of short actin filaments. Of course that time scale to percolate a bridge of shorter filaments is longer, but it may be compensated by the time scales (or probability required to create enough long (~6 um) filaments. Also, how can the authors really justify to use 6 um as length based on their measured length distributions? These distributions clearly show an average of ~1um, and almost no filaments larger than 5 um.I think it is necessary that the authors clarify this point, comment on the origin of actin in these bridges, what would be the time scales associated to having only the 5% population of actin, and recalculate their predictions based on the average actin length they measure. The answer the authors give to use ~6 um is flawed. They argue “In the proposed model, the probability that a bridge is formed by the length L of F-actin in a droplet with radius R is written as P=(1/2)R/L. If we compare the probability with different actin length, L=10 μm and L=5 μm, in a droplet with radius R=50 μm, we can estimate P=(1/2)5 and P=(1/2)10 respectively. Thus, only 2 times change in F-actin length decreases the bridge formation probability ~30 times.”. The authors are missing the probability of actually getting actin filaments of that particular length. With their measured length distribution, this can also change by a decade from ~1 to ~5 um so it could counteract their estimated probability, so the authors need to take this into account in their argument. The authors distinguish gelation vs percolation to explain the difference of timescales, which they argue leads to a faster gelation time. However, my understanding is that gelation is a connectivity transition, and that previous research has used percolation as a theory to explain the gelation process. In both cases, they are network-connectivity transitions. Gelation further implies the emergence of material properties such as a finite young modulus, whereas percolation theory makes no statement (or can’t explain) about it and it is just restricted to the emergence of a system size cluster. Thus, I don’t really understand what the authors imply by “Since percolation is possible even without gelation, it is natural to assume that the bridge formation does not have to rely on the gelation process.”, and why would we expect that the percolation would take longer? Could the authors explain what the explicit differences between percolation and gelation they are impling?	TPR	NC_BS_BIOPH_007_RR1.txt	{ "criticism": 5, "example": 3, "importance_and_relevance": 2, "materials_and_methods": 15, "praise": 5, "presentation_and_reporting": 1, "results_and_discussion": 6, "suggestion_and_solution": 3, "total": 35 }	{ "criticism": 0.14285714285714285, "example": 0.08571428571428572, "importance_and_relevance": 0.05714285714285714, "materials_and_methods": 0.42857142857142855, "praise": 0.14285714285714285, "presentation_and_reporting": 0.02857142857142857, "results_and_discussion": 0.17142857142857143, "suggestion_and_solution": 0.08571428571428572 }	1.142857	[ { "content": "Please review this paper.", "role": "user" }, { "content": "Reports from round 1:\nRev#1\nThe authors have done an excellent job addressing the reviewer concerns, including addition of significant new experiments, all of which support their model. All critiques have been addressed, this is a lovely paper.\nRev#2\nThe authors have sufficiently addressed referee concerns. The manuscript is improved and appropriate for publication in Nature Communications\nRev#3\n The authors went a long way to answer my previous and other reviewer comments. Overall the\n manuscript has improved with the new experiments and the qualitative understanding of the\n process seems to be supported by the data. I only have two concerns left.\nOne minor is that the authors provide new analysis to distinguish between bulk contraction vs ring\ncontraction. For that they refer to the new supplementary figure 4 that shows that the intensity of\nthe ring goes up as it contracts whereas the bulk intensity remain constant. This would indeed\nsuggest that the contraction is happening in the ring. However, this seems to contradict their exact\nsame measurements in supplementary figure 2 :”(f) Fluorescence intensity along the each broken\nline in e, which also corresponds to the intensity at the arrows with the same color in d. Both the\nF-actin densities on the rim and inside of the wave increased during the contraction. The thickness\nof the spacer was 100 μm. Scale bars, 1 mm. “ Could the authors clarify the discrepancy? Does it\ndepend on day to day or example taken?\nMy main concern is still related to the theory:\nFor their theory the authors state: “We have proposed that the off-centering is mediated by the\npercolated actomyosin bridge”, and that the bridge emerges from “low number of remained F-actin\nand newly assembled F-actin stochastically form bridges between the cluster and the boundary.”\nThe authors are assuming the bridge is pre-formed and just calculate the probability that cross\nlinkers connect the filaments that form the bridge. But, why don’t we see the bridge than initially\nafter the wave? From their images it looks that actin is actually polymerized as the bridge is form\n(See figure 4c. After laser ablation a new bridge is formed but no wave has formed before it).\nSince the authors only consider the top 5% of actin lengths, shouldn’t this have a dramatic effect\non time scales too? Meaning that it may be very difficult to actually have a bridge made of actin\nfilaments longer than the 5% population as opposed of having a bridge made of short actin\nfilaments. Of course that time scale to percolate a bridge of shorter filaments is longer, but it may\nbe compensated by the time scales (or probability required to create enough long (~6 um)\nfilaments. Also, how can the authors really justify to use 6 um as length based on their measured\nlength distributions? These distributions clearly show an average of ~1um, and almost no\nfilaments larger than 5 um.I think it is necessary that the authors clarify this point, comment on\nthe origin of actin in these bridges, what would be the time scales associated to having only the\n5% population of actin, and recalculate their predictions based on the average actin length they\nmeasure.\nThe answer the authors give to use ~6 um is flawed. They argue “In the proposed model, the\nprobability that a bridge is formed by the length L of F-actin in a droplet with radius R is written as\nP=(1/2)R/L. If we compare the probability with different actin length, L=10 μm and L=5 μm, in a\ndroplet with radius R=50 μm, we can estimate P=(1/2)5 and P=(1/2)10 respectively. Thus, only 2\ntimes change in F-actin length decreases the bridge formation probability ~30 times.”. The authors\nare missing the probability of actually getting actin filaments of that particular length. With their\nmeasured length distribution, this can also change by a decade from ~1 to ~5 um so it could\ncounteract their estimated probability, so the authors need to take this into account in their\nargument.\nThe authors distinguish gelation vs percolation to explain the difference of timescales, which they\n argue leads to a faster gelation time. However, my understanding is that gelation is a connectivity\n transition, and that previous research has used percolation as a theory to explain the gelation\n process. In both cases, they are network-connectivity transitions. Gelation further implies the\n emergence of material properties such as a finite young modulus, whereas percolation theory\n makes no statement (or can’t explain) about it and it is just restricted to the emergence of a\n system size cluster. Thus, I don’t really understand what the authors imply by “Since percolation is\n possible even without gelation, it is natural to assume that the bridge formation does not have to\n rely on the gelation process.”, and why would we expect that the percolation would take longer?\n Could the authors explain what the explicit differences between percolation and gelation they are\n impling?\n", "role": "assistant" } ]	2,020
Aβ(1-42) tetramer and octamer structures reveal edge conductivity pores as a mechanism	Reviewers' comments first round: Reviewer #1 (Remarks to the Author): Carulla and co-workers employed various biophysical techniques - NMR, SEC/IM-MS (size exclusion chromatography coupled with ion-mobility mass spectrometry), MALDI-MS, electrical recordings, as well as MD (molecular dynamics) simulations - to determine a structure of micelle-bound Abeta(1-42) tetramers and octamers, which in case of the tetramer was submitted to the Protein Data Bank. This work is of high importance and novelty as in the amyloid field structural information on Abeta oligomers (in solution and membrane-bound) is currently still missing. Such information is of importance as Abeta oligomers are thought to play a key role in the development of Alzheimer's disease. While the importance of the current work is without doubt, some concerns have to be addressed before the manuscript may become publishable: 1) The authors need to mention that pH values between 8.5 and 9.5 are anything but physiological. It is known that the structures of monomeric and aggregated Abeta is highly susceptible to pH. Thus, the structures reported here might be of no relevance at all at physiological pH. Though having said this, it should also be mentioned that others publish Abeta fibril structures determined at a non-physiological pH of 2 in a solution incolving 30% acetonitril (doi: 10.1126/science.aao2825). Thus, when such a fibril structure can be published in Science, it should also be warranted to publish an amyloid oligomer structure at pH 9 in Nature Commun. But the pH issue must be openly discussed. 2) When discussing the pH problem, the most likely protonation states of the amino acids at pH 9 should be given. For sure, histidines will be neutral (unlike what the authors state on pp. 7 and 8). But also the lysines may become neutral at such high pH, also considering that a hydrophobic environment reduces the pKa auf lysine. In the Methods regarding the MD simulations the chosen protonation states of titratable amino acids need to be given. 3) It should be further stated that the current structures determined from the co-assembly of Abeta and DPC into micelles containing Abeta are not necessarily those which one might find in lipid bilayers (biomembranes). With the suggested structure (Fig. 1d), the N-terminus of two of the peptides, with all its charged residues, would need to traverse through the hydrophobic core of the bilayer. The authors need to state how likely this event would be, e.g., how much energy it would take for this process to happen, or whether this could be enabled by transporters. 4) The MD simulations do not fully support that the suggested structures would be likely in a lipid bilayer environment. Fig. S14 shows that DPC assembles with its headgroup region around beta1, i.e., this strand prefers a polar environment and therefore draws water into the bilayer as Fig. 5 shows. As the simulations were only conducted for 100 ns, which is too short for a system of that size and needs to be extended to 1 microscond or longer, it remains open how stable the suggested transmembrane tetramer structure indeed is. More analysis of the MD simulations needs to be done: RMSD/F, number of water molecules traversed, secondary structure per residue are some quantities that would be helpful. 5) While it is always expected that simulation scientists compare their results to existing experimental observations, it seems that the opposite is usually not expected. But almost 10 years ago models for transmembrane Abeta oligomers were published and their stability tested in MD simulations (doi: 10.1021/ja103725c; doi: 10.1016/j.bbamem.2012.09.001). Interestingly, later a very similar structure per Abeta peptide was determined by solid-state NMR as building block for Abeta hexamers: doi: 10.1002/anie.201406357. The authors should include a comparison of their structure model of membrane-bound Abeta oligomers with those previously published. 6) The CCS results in Fig. 4e and f are anything but convincing. Cutting off the flexible N-terminal ends to obtain agreemt between structure model and actual CCS results does not increase the trust in the suggested tetramer and octamer models. However, it is anyhow questionable whether the ionized oligomers in the gas phase are structurally similar to the Abeta structures in DPC micelles. The authors should do more to convince this reviewer of this correspondance. As some of the authors are simulation scientists, one approach could be to simulate the oligomers in the proper ionization state in the gas phase to see how they evolve. The current approach (ignoring some of the residues to obtain the experimental CCS results) is below the standard expected for a publication in Nat. Commun. Reviewer #2 (Remarks to the Author): The manuscript by Ciudad et al. describes the structural characterization of Abeta(1-42) tetrameric and octameric structures. The structure of the tetramer is derived from NMR measurements. Cross-linking and SEC/IM-MS corroborate the findings of an Abeta tetramer, octamer, and other species. The study brings a 3D structure of an Abeta tetramer to the fore. The introduction is composed of only two paragraphs and lacking background on Abeta or other oligomers that have been either modelled or determined experimentally. In short, the introduction would benefit from more background and context. Abeta oligomers were studied in dodecylphosphocholine (DPC), a detergent, and at non- physiological pH (8.5 and 9.5). The authors note in discussion that DPC is a membrane mimic and not a physiological environment. However, there is no mention in regards to pH. Would the proposed structures be stable at physiological pH? The 3D structure reveals an antiparallel arrangement with two of the three subunits exhibiting long flexible, solvent exposed N-termini. Given the structure, it would appear there would be large entropic barrier to insertion into the bilayer. Can the authors speculate on a plausible mechanism for insertion into the bilayer? Another issue not addressed, is the fact that charged and polar residues (such as His, Qln, and Lys) in their model (see Fig 1 and S17) are centrally located in the hydrophobic core of the bilayer. This simply does not make chemical sense and raises concerns to the validity of the proposed model. An explanation to rationalize exposure of such residues in the hydrophobic environment is warranted. In regards to the SEC/IM-MS measurements, increased activation conditions were used to disrupt the Ab complexes. The authors note tetramers broke into trimers and monomers, but this is not evident from the data presented (Fig S16 and S19). The assignment of oligomeric states by m/z and IM measurements is solid. The authors may consider isolating particular Ab species using the quadrupole prior to subjecting them to increased activation? This has been done before (see doi: 10.1038/nature20820 and doi: 10.1007/s13361-016-1555-1), which would provide a more robust evaluation of complex gas-phase stability/dissociation. IM measurements (CCS values) were compared to a limited set of Ab models. The CCS values indicate a compact CCS value that agrees with models where the flexible loops have been removed from the structure prior to calculation. This approach raises significant concerns; deleting residues to match measured CCS values is not acceptable nor rigorous. First, flexible regions can lead to an increase in CCS values (see DOI: 10.1002/anie.201203047), which is not observed here. Does this imply that the N-termini are more ordered than flexible? Second, have other models been compared to the measured CCS values? Such as, a six-stranded cylindrin-like structure? Or other models? At present, additional models should be explored and also considering gas-phase minimization prior to CCS calculation. No controls for the electrical recordings using planar lipid bilayers are presented. DPC is a detergent, which could disrupt the integrity of the bilayer giving rise to background signals. Were measurements of DPC in the absence of Ab performed? The section titled “preparation of ... enriched in Ab(1-42) octamers” seems out of place. Suggest moving up and before the SEC-MS analysis section. The manuscript could be improved by merging some figures. For example, Fig S17 can be incorporated into Fig S7. For denatured and native MS measurements, the measured and theoretical masses should be reported along with error. Supplementary Fig. 18 is mislabelled. I presume the data shown is for PFOsHIGH not PFOsLOW. Pg 8, “... using DPC at natural abundance.” This is unclear as DPC is non-natural. Pg 15, “.. we have precisely defined the structural ..” Remove precisely. Reviewer #3 (Remarks to the Author): Although I am not able to expertly comment on the quality of the NMR data, the study seems well designed and executed. This is a very interesting paper that presents a startling structure for the micelle-inserted amyloid- beta-42 polypeptide (AB42). AB42 is, of course, generally thought to be associated with the etiology of Alzheimer's disease. There is a body of evidence in the literature that AB42 can adopt some sort of transmembrane oligomer, with additional evidence that it can act (aberrantly) as an ion channel, potentially a mechanism that contributes to how AB42 promotes Alzheimer's disease. In this paper the authors used NMR spectroscopy to document that AB42 in dodecylphosphocholine micelles can form two closely related oligomeric structures. In one structure 4 subunits come togehther to form a single sheet with two subunits forming trans-micelle beta hairpins flank the two other subunits central in the tetrameric sheet as a single beta strands. What is amazing is that the 6-stranded tetatrameric sheet does not form a closed barrel, but instead is an "open" sheet with solvent exposed edges for strands 1 and 6. The other structure is a dimer of tetramers where the 6-stranded sheets form a sandwich in the micelle interior, again with exposed edges. The authors also present electrophysiology measurements in which ion channel activity is documented, apparently for both monomer and dimers, with the channel activity apparently being associated with the exposed edges of the monomeric or dimeric sheets. The NMR studies of this work were meticulously carried out and documented-- very convincing. Moreover, one can imagine how these structures might be stable in micelles: the horizontal diameter of the sheet is on the same order as the diameter of the micelle it sits in, so the need to form transmicelle beta barrels is relieved by the fact that the exposed edges of strands 1 and 6 are not exposed to a bilayer interior, but instead to water outside of the micelle water at the micelle- water interface. So, these structures seem reasonable as micelle structures. However, it seems less clear that they could be energetically accommodated as transmembrane structures in a sealed lipid bilayer, where the exposed edges of strands 1 and 6 would be exposed only to lipid, resulting in burial backbone amide groups with unsatisfied H-bonding potential. The text, figures, and supporting material are all very well written. I think this work is EXTREMELY interesting and should be published following revision, without any requirement for additional experiments. However, the authors should provide more detail on what evidence they have that the micellar structure they carefully document in this work is maintained in lipid vesicles. This closely relates to the question of whether the channel activity they document really stems from AB42 channels that resemble the micellar structures. Again, more textual justification seems needed. Finally, the sample composition, pH, and temperature need to be spelled out in detail in the caption to Figure 1. Reviewer #4 (Remarks to the Author): Review of NCOMMS-19-35967-T The manuscript by Ciudad et al deals with an important open question in our understanding of the Alzheimer’s Disease (AD) Pathology cascade: the mechanism by which amyloid-beta causes damages neurons. The authors cite the “amyloid hypothesis” in which Abeta is assumed to be the primary toxic agent in AD, and note that the mechanism of toxcity is unclear. They also cite works in which soluble forms of Abeta can induce pores of various sorts in membranes. This is the background for their basic hypothesis: that Abeta structures can form pores in planar lipid bilayers, and therefore can serve as a mechanism of neurotoxicity. The primary focus of the work is in establishing the feasibility of this idea, and the authors perform detailed work in establishing this to be the case. The evidence that Abeta structures can form pores in planar lipid bilayers, and that this is specific to certain forms of Abeta, seems to be well established by this study, and supported by beautiful and clear figures. I will address a different aspect of this study, which is the validity of their hypothesis: that Abeta can cause pores in the membranes of neurons adjacent to sufficient concentrations of Abeta, which will then disrupt neuronal activity. The principal evidence for that disruption is shown in Supplementary Figure 20, in which the membrane conductance is increased by the introduction of βPFOsAβ(1-42) into planar lipid bilayers. This makes a clear prediction of the effect of soluble Abeta on neuronal activity. The prediction is that the input conductance of these neurons will be increased. Intracellular recordings from several lines of APP+ transgenic mice (two of which are published) , in which both soluble Abeta and amyloid plaques are present in the surrounding neuropil, by measurement of average input resistance in the neurons, do not show the increase in RIN that would be expected to be associated with such a conductance increase (Stern et al, J. Neurosci 24, 4335-4340 (2004); Kellner et al, Neurobiology of Aging 35, 1982-1991 (2014)). I do not know of any studies which have shown different results. (In addition, measurement by several groups show hyperactivity in numerous lines of APP transgenic mice, which is the opposite effect that would be expected from a general conductance increase.) these are issues that should be addressd by the authors, if they wish to argue that increased membrane conductance is indeed a mechanism of neuronal disruption by Abeta. NOTE: This, however, is not necessarily evidence that the conductance increase does not occur. The direct measurements of input resistance are made using electrodes inserted into the soma. It may well be that the the toxic effects of Abeta on the membrane are local. Indeed, one study shows that there are specific effects on dendritic spines in proximity to plaques, where the concentrations of soluble Abeta are higher (Spires et al, J Neurosci. 25, 7278-7287 (2005)). This is an hypothesis which can be tested, using targeted patch recordings in slices of APP transgenic mouse cortex. The recordings would have to target dendrites in mice in which plaque aggregation is severe, so that dendrites would have a high probability of proximity to plaques. While difficult, targeted recordings are performed on a daily basis by numerous laboratories. If the input conductance is increased in these dendrites, compared to aged-match recordings in WT mice, it would indeed be strong evidence that the hypothesis proposed by the authors is correct. In many studies, including the one I cited, there is evidence for stripping of spines form these dendrites, which I would predict will show the opposite effect: raising the local dendritic input resistance. At present, however, there is no such evidence of which I am aware that the mechanism proposed by the authors disrupts neuronal activity in AD. And some evidence does exist that the overall conductance in the neuron is not raised in APP-transgenic mouse model neurons. The current study does show an interesting possible mechanism of action of one of the forms of Abeta: one interesting enough to be further investigated. Again, evidence for the specific mechanisms (note the plural) of the pathophysiological effects of Abeta is a “missing link” in our understanding of the cascade of effects of the Abeta in the disase process. Edward A. Stern Brain Research Center Bar Ilan University	Reviewers' comments first round: Reviewer #1 (Remarks to the Author): Carulla and co-workers employed various biophysical techniques - NMR, SEC/IM-MS (size exclusion chromatography coupled with ion-mobility mass spectrometry), MALDI-MS, electrical recordings, as well as MD (molecular dynamics) simulations - to determine a structure of micelle-bound Abeta(1-42) tetramers and octamers, which in case of the tetramer was submitted to the Protein Data Bank. This work is of high importance and novelty as in the amyloid field structural information on Abeta oligomers (in solution and membrane-bound) is currently still missing. Such information is of importance as Abeta oligomers are thought to play a key role in the development of Alzheimer's disease. While the importance of the current work is without doubt, some concerns have to be addressed before the manuscript may become publishable: 1) The authors need to mention that pH values between 8.5 and 9.5 are anything but physiological. It is known that the structures of monomeric and aggregated Abeta is highly susceptible to pH. Thus, the structures reported here might be of no relevance at all at physiological pH. Though having said this, it should also be mentioned that others publish Abeta fibril structures determined at a non-physiological pH of 2 in a solution incolving 30% acetonitril (doi: 10.1126/science.aao2825). Thus, when such a fibril structure can be published in Science, it should also be warranted to publish an amyloid oligomer structure at pH 9 in Nature Commun. But the pH issue must be openly discussed. 2) When discussing the pH problem, the most likely protonation states of the amino acids at pH 9 should be given. For sure, histidines will be neutral (unlike what the authors state on pp. 7 and 8). But also the lysines may become neutral at such high pH, also considering that a hydrophobic environment reduces the pKa auf lysine. In the Methods regarding the MD simulations the chosen protonation states of titratable amino acids need to be given. 3) It should be further stated that the current structures determined from the co-assembly of Abeta and DPC into micelles containing Abeta are not necessarily those which one might find in lipid bilayers (biomembranes). With the suggested structure (Fig. 1d), the N-terminus of two of the peptides, with all its charged residues, would need to traverse through the hydrophobic core of the bilayer. The authors need to state how likely this event would be, e.g., how much energy it would take for this process to happen, or whether this could be enabled by transporters. 4) The MD simulations do not fully support that the suggested structures would be likely in a lipid bilayer environment. Fig. S14 shows that DPC assembles with its headgroup region around beta1, i.e., this strand prefers a polar environment and therefore draws water into the bilayer as Fig. 5 shows. As the simulations were only conducted for 100 ns, which is too short for a system of that size and needs to be extended to 1 microscond or longer, it remains open how stable the suggested transmembrane tetramer structure indeed is. More analysis of the MD simulations needs to be done: RMSD/F, number of water molecules traversed, secondary structure per residue are some quantities that would be helpful. 5) While it is always expected that simulation scientists compare their results to existing experimental observations, it seems that the opposite is usually not expected. But almost 10 years ago models for transmembrane Abeta oligomers were published and their stability tested in MD simulations (doi: 10.1021/ja103725c; doi: 10.1016/j.bbamem.2012.09.001). Interestingly, later a very similar structure per Abeta peptide was determined by solid-state NMR as building block for Abeta hexamers: doi: 10.1002/anie.201406357. The authors should include a comparison of their structure model of membrane-bound Abeta oligomers with those previously published. 6) The CCS results in Fig. 4e and f are anything but convincing. Cutting off the flexible N-terminal ends to obtain agreemt between structure model and actual CCS results does not increase the trust in the suggested tetramer and octamer models. However, it is anyhow questionable whether the ionized oligomers in the gas phase are structurally similar to the Abeta structures in DPC micelles. The authors should do more to convince this reviewer of this correspondance. As some of the authors are simulation scientists, one approach could be to simulate the oligomers in the proper ionization state in the gas phase to see how they evolve. The current approach (ignoring some of the residues to obtain the experimental CCS results) is below the standard expected for a publication in Nat. Commun. Reviewer #2 (Remarks to the Author): The manuscript by Ciudad et al. describes the structural characterization of Abeta(1-42) tetrameric and octameric structures. The structure of the tetramer is derived from NMR measurements. Cross-linking and SEC/IM-MS corroborate the findings of an Abeta tetramer, octamer, and other species. The study brings a 3D structure of an Abeta tetramer to the fore. The introduction is composed of only two paragraphs and lacking background on Abeta or other oligomers that have been either modelled or determined experimentally. In short, the introduction would benefit from more background and context. Abeta oligomers were studied in dodecylphosphocholine (DPC), a detergent, and at non- physiological pH (8.5 and 9.5). The authors note in discussion that DPC is a membrane mimic and not a physiological environment. However, there is no mention in regards to pH. Would the proposed structures be stable at physiological pH? The 3D structure reveals an antiparallel arrangement with two of the three subunits exhibiting long flexible, solvent exposed N-termini. Given the structure, it would appear there would be large entropic barrier to insertion into the bilayer. Can the authors speculate on a plausible mechanism for insertion into the bilayer? Another issue not addressed, is the fact that charged and polar residues (such as His, Qln, and Lys) in their model (see Fig 1 and S17) are centrally located in the hydrophobic core of the bilayer. This simply does not make chemical sense and raises concerns to the validity of the proposed model. An explanation to rationalize exposure of such residues in the hydrophobic environment is warranted. In regards to the SEC/IM-MS measurements, increased activation conditions were used to disrupt the Ab complexes. The authors note tetramers broke into trimers and monomers, but this is not evident from the data presented (Fig S16 and S19). The assignment of oligomeric states by m/z and IM measurements is solid. The authors may consider isolating particular Ab species using the quadrupole prior to subjecting them to increased activation? This has been done before (see doi: 10.1038/nature20820 and doi: 10.1007/s13361-016-1555-1), which would provide a more robust evaluation of complex gas-phase stability/dissociation. IM measurements (CCS values) were compared to a limited set of Ab models. The CCS values indicate a compact CCS value that agrees with models where the flexible loops have been removed from the structure prior to calculation. This approach raises significant concerns; deleting residues to match measured CCS values is not acceptable nor rigorous. First, flexible regions can lead to an increase in CCS values (see DOI: 10.1002/anie.201203047), which is not observed here. Does this imply that the N-termini are more ordered than flexible? Second, have other models been compared to the measured CCS values? Such as, a six-stranded cylindrin-like structure? Or other models? At present, additional models should be explored and also considering gas-phase minimization prior to CCS calculation. No controls for the electrical recordings using planar lipid bilayers are presented. DPC is a detergent, which could disrupt the integrity of the bilayer giving rise to background signals. Were measurements of DPC in the absence of Ab performed? The section titled “preparation of ... enriched in Ab(1-42) octamers” seems out of place. Suggest moving up and before the SEC-MS analysis section. The manuscript could be improved by merging some figures. For example, Fig S17 can be incorporated into Fig S7. For denatured and native MS measurements, the measured and theoretical masses should be reported along with error. Supplementary Fig. 18 is mislabelled. I presume the data shown is for PFOsHIGH not PFOsLOW. Pg 8, “... using DPC at natural abundance.” This is unclear as DPC is non-natural. Pg 15, “.. we have precisely defined the structural ..” Remove precisely. Reviewer #3 (Remarks to the Author): Although I am not able to expertly comment on the quality of the NMR data, the study seems well designed and executed. This is a very interesting paper that presents a startling structure for the micelle-inserted amyloid- beta-42 polypeptide (AB42). AB42 is, of course, generally thought to be associated with the etiology of Alzheimer's disease. There is a body of evidence in the literature that AB42 can adopt some sort of transmembrane oligomer, with additional evidence that it can act (aberrantly) as an ion channel, potentially a mechanism that contributes to how AB42 promotes Alzheimer's disease. In this paper the authors used NMR spectroscopy to document that AB42 in dodecylphosphocholine micelles can form two closely related oligomeric structures. In one structure 4 subunits come togehther to form a single sheet with two subunits forming trans-micelle beta hairpins flank the two other subunits central in the tetrameric sheet as a single beta strands. What is amazing is that the 6-stranded tetatrameric sheet does not form a closed barrel, but instead is an "open" sheet with solvent exposed edges for strands 1 and 6. The other structure is a dimer of tetramers where the 6-stranded sheets form a sandwich in the micelle interior, again with exposed edges. The authors also present electrophysiology measurements in which ion channel activity is documented, apparently for both monomer and dimers, with the channel activity apparently being associated with the exposed edges of the monomeric or dimeric sheets. The NMR studies of this work were meticulously carried out and documented-- very convincing. Moreover, one can imagine how these structures might be stable in micelles: the horizontal diameter of the sheet is on the same order as the diameter of the micelle it sits in, so the need to form transmicelle beta barrels is relieved by the fact that the exposed edges of strands 1 and 6 are not exposed to a bilayer interior, but instead to water outside of the micelle water at the micelle- water interface. So, these structures seem reasonable as micelle structures. However, it seems less clear that they could be energetically accommodated as transmembrane structures in a sealed lipid bilayer, where the exposed edges of strands 1 and 6 would be exposed only to lipid, resulting in burial backbone amide groups with unsatisfied H-bonding potential. The text, figures, and supporting material are all very well written. I think this work is EXTREMELY interesting and should be published following revision, without any requirement for additional experiments. However, the authors should provide more detail on what evidence they have that the micellar structure they carefully document in this work is maintained in lipid vesicles. This closely relates to the question of whether the channel activity they document really stems from AB42 channels that resemble the micellar structures. Again, more textual justification seems needed. Finally, the sample composition, pH, and temperature need to be spelled out in detail in the caption to Figure 1. Reviewer #4 (Remarks to the Author): Review of NCOMMS-19-35967-T The manuscript by Ciudad et al deals with an important open question in our understanding of the Alzheimer’s Disease (AD) Pathology cascade: the mechanism by which amyloid-beta causes damages neurons. The authors cite the “amyloid hypothesis” in which Abeta is assumed to be the primary toxic agent in AD, and note that the mechanism of toxcity is unclear. They also cite works in which soluble forms of Abeta can induce pores of various sorts in membranes. This is the background for their basic hypothesis: that Abeta structures can form pores in planar lipid bilayers, and therefore can serve as a mechanism of neurotoxicity. The primary focus of the work is in establishing the feasibility of this idea, and the authors perform detailed work in establishing this to be the case. The evidence that Abeta structures can form pores in planar lipid bilayers, and that this is specific to certain forms of Abeta, seems to be well established by this study, and supported by beautiful and clear figures. I will address a different aspect of this study, which is the validity of their hypothesis: that Abeta can cause pores in the membranes of neurons adjacent to sufficient concentrations of Abeta, which will then disrupt neuronal activity. The principal evidence for that disruption is shown in Supplementary Figure 20, in which the membrane conductance is increased by the introduction of βPFOsAβ(1-42) into planar lipid bilayers. This makes a clear prediction of the effect of soluble Abeta on neuronal activity. The prediction is that the input conductance of these neurons will be increased. Intracellular recordings from several lines of APP+ transgenic mice (two of which are published) , in which both soluble Abeta and amyloid plaques are present in the surrounding neuropil, by measurement of average input resistance in the neurons, do not show the increase in RIN that would be expected to be associated with such a conductance increase (Stern et al, J. Neurosci 24, 4335-4340 (2004); Kellner et al, Neurobiology of Aging 35, 1982-1991 (2014)). I do not know of any studies which have shown different results. (In addition, measurement by several groups show hyperactivity in numerous lines of APP transgenic mice, which is the opposite effect that would be expected from a general conductance increase.) these are issues that should be addressd by the authors, if they wish to argue that increased membrane conductance is indeed a mechanism of neuronal disruption by Abeta. NOTE: This, however, is not necessarily evidence that the conductance increase does not occur. The direct measurements of input resistance are made using electrodes inserted into the soma. It may well be that the the toxic effects of Abeta on the membrane are local. Indeed, one study shows that there are specific effects on dendritic spines in proximity to plaques, where the concentrations of soluble Abeta are higher (Spires et al, J Neurosci. 25, 7278-7287 (2005)). This is an hypothesis which can be tested, using targeted patch recordings in slices of APP transgenic mouse cortex. The recordings would have to target dendrites in mice in which plaque aggregation is severe, so that dendrites would have a high probability of proximity to plaques. While difficult, targeted recordings are performed on a daily basis by numerous laboratories. If the input conductance is increased in these dendrites, compared to aged-match recordings in WT mice, it would indeed be strong evidence that the hypothesis proposed by the authors is correct. In many studies, including the one I cited, there is evidence for stripping of spines form these dendrites, which I would predict will show the opposite effect: raising the local dendritic input resistance. At present, however, there is no such evidence of which I am aware that the mechanism proposed by the authors disrupts neuronal activity in AD. And some evidence does exist that the overall conductance in the neuron is not raised in APP-transgenic mouse model neurons. The current study does show an interesting possible mechanism of action of one of the forms of Abeta: one interesting enough to be further investigated. Again, evidence for the specific mechanisms (note the plural) of the pathophysiological effects of Abeta is a “missing link” in our understanding of the cascade of effects of the Abeta in the disase process. Edward A. Stern Brain Research Center Bar Ilan University	TPR	NC_BS_BIOPH_008_RR1.txt	{ "criticism": 21, "example": 12, "importance_and_relevance": 11, "materials_and_methods": 64, "praise": 12, "presentation_and_reporting": 16, "results_and_discussion": 14, "suggestion_and_solution": 27, "total": 128 }	{ "criticism": 0.1640625, "example": 0.09375, "importance_and_relevance": 0.0859375, "materials_and_methods": 0.5, "praise": 0.09375, "presentation_and_reporting": 0.125, "results_and_discussion": 0.109375, "suggestion_and_solution": 0.2109375 }	1.382813	[ { "content": "Please review this paper.", "role": "user" }, { "content": "Reviewers' comments first round: Reviewer #1 (Remarks to the Author):\nCarulla and co-workers employed various biophysical techniques - NMR, SEC/IM-MS (size exclusion\nchromatography coupled with ion-mobility mass spectrometry), MALDI-MS, electrical recordings,\nas well as MD (molecular dynamics) simulations - to determine a structure of micelle-bound\nAbeta(1-42) tetramers and octamers, which in case of the tetramer was submitted to the Protein\nData Bank.\n This work is of high importance and novelty as in the amyloid field structural information on Abeta\n oligomers (in solution and membrane-bound) is currently still missing. Such information is of\n importance as Abeta oligomers are thought to play a key role in the development of Alzheimer's\n disease.\n While the importance of the current work is without doubt, some concerns have to be addressed\n before the manuscript may become publishable:\n1) The authors need to mention that pH values between 8.5 and 9.5 are anything but\nphysiological. It is known that the structures of monomeric and aggregated Abeta is highly\nsusceptible to pH. Thus, the structures reported here might be of no relevance at all at\nphysiological pH. Though having said this, it should also be mentioned that others publish Abeta\nfibril structures determined at a non-physiological pH of 2 in a solution incolving 30% acetonitril\n(doi: 10.1126/science.aao2825). Thus, when such a fibril structure can be published in Science, it\nshould also be warranted to publish an amyloid oligomer structure at pH 9 in Nature Commun. But\nthe pH issue must be openly discussed.\n2) When discussing the pH problem, the most likely protonation states of the amino acids at pH 9\nshould be given. For sure, histidines will be neutral (unlike what the authors state on pp. 7 and 8).\nBut also the lysines may become neutral at such high pH, also considering that a hydrophobic\nenvironment reduces the pKa auf lysine. In the Methods regarding the MD simulations the chosen\nprotonation states of titratable amino acids need to be given.\n3) It should be further stated that the current structures determined from the co-assembly of\nAbeta and DPC into micelles containing Abeta are not necessarily those which one might find in\nlipid bilayers (biomembranes). With the suggested structure (Fig. 1d), the N-terminus of two of\nthe peptides, with all its charged residues, would need to traverse through the hydrophobic core of\nthe bilayer. The authors need to state how likely this event would be, e.g., how much energy it\nwould take for this process to happen, or whether this could be enabled by transporters.\n4) The MD simulations do not fully support that the suggested structures would be likely in a lipid\nbilayer environment. Fig. S14 shows that DPC assembles with its headgroup region around beta1,\ni.e., this strand prefers a polar environment and therefore draws water into the bilayer as Fig. 5\nshows. As the simulations were only conducted for 100 ns, which is too short for a system of that\nsize and needs to be extended to 1 microscond or longer, it remains open how stable the\nsuggested transmembrane tetramer structure indeed is. More analysis of the MD simulations\nneeds to be done: RMSD/F, number of water molecules traversed, secondary structure per residue\nare some quantities that would be helpful.\n5) While it is always expected that simulation scientists compare their results to existing\nexperimental observations, it seems that the opposite is usually not expected. But almost 10 years\nago models for transmembrane Abeta oligomers were published and their stability tested in MD\nsimulations (doi: 10.1021/ja103725c; doi: 10.1016/j.bbamem.2012.09.001). Interestingly, later a\nvery similar structure per Abeta peptide was determined by solid-state NMR as building block for\nAbeta hexamers: doi: 10.1002/anie.201406357. The authors should include a comparison of their\nstructure model of membrane-bound Abeta oligomers with those previously published.\n 6) The CCS results in Fig. 4e and f are anything but convincing. Cutting off the flexible N-terminal\n ends to obtain agreemt between structure model and actual CCS results does not increase the\n trust in the suggested tetramer and octamer models. However, it is anyhow questionable whether\n the ionized oligomers in the gas phase are structurally similar to the Abeta structures in DPC\n micelles. The authors should do more to convince this reviewer of this correspondance. As some of\n the authors are simulation scientists, one approach could be to simulate the oligomers in the\n proper ionization state in the gas phase to see how they evolve. The current approach (ignoring\n some of the residues to obtain the experimental CCS results) is below the standard expected for a\n publication in Nat. Commun.\n Reviewer #2 (Remarks to the Author):\n The manuscript by Ciudad et al. describes the structural characterization of Abeta(1-42) tetrameric\n and octameric structures. The structure of the tetramer is derived from NMR measurements.\n Cross-linking and SEC/IM-MS corroborate the findings of an Abeta tetramer, octamer, and other\n species. The study brings a 3D structure of an Abeta tetramer to the fore.\n The introduction is composed of only two paragraphs and lacking background on Abeta or other\n oligomers that have been either modelled or determined experimentally. In short, the introduction\n would benefit from more background and context.\n Abeta oligomers were studied in dodecylphosphocholine (DPC), a detergent, and at non-\n physiological pH (8.5 and 9.5). The authors note in discussion that DPC is a membrane mimic and\n not a physiological environment. However, there is no mention in regards to pH. Would the\n proposed structures be stable at physiological pH?\n The 3D structure reveals an antiparallel arrangement with two of the three subunits exhibiting long\n flexible, solvent exposed N-termini. Given the structure, it would appear there would be large\n entropic barrier to insertion into the bilayer. Can the authors speculate on a plausible mechanism\n for insertion into the bilayer? Another issue not addressed, is the fact that charged and polar\n residues (such as His, Qln, and Lys) in their model (see Fig 1 and S17) are centrally located in the\n hydrophobic core of the bilayer. This simply does not make chemical sense and raises concerns to\n the validity of the proposed model. An explanation to rationalize exposure of such residues in the\n hydrophobic environment is warranted.\n In regards to the SEC/IM-MS measurements, increased activation conditions were used to disrupt\n the Ab complexes. The authors note tetramers broke into trimers and monomers, but this is not\n evident from the data presented (Fig S16 and S19). The assignment of oligomeric states by m/z\n and IM measurements is solid. The authors may consider isolating particular Ab species using the\n quadrupole prior to subjecting them to increased activation? This has been done before (see doi:\n 10.1038/nature20820 and doi: 10.1007/s13361-016-1555-1), which would provide a more robust\n evaluation of complex gas-phase stability/dissociation.\n IM measurements (CCS values) were compared to a limited set of Ab models. The CCS values\n indicate a compact CCS value that agrees with models where the flexible loops have been removed\n from the structure prior to calculation. This approach raises significant concerns; deleting residues\n to match measured CCS values is not acceptable nor rigorous. First, flexible regions can lead to an\n increase in CCS values (see DOI: 10.1002/anie.201203047), which is not observed here. Does this\n imply that the N-termini are more ordered than flexible? Second, have other models been\n compared to the measured CCS values? Such as, a six-stranded cylindrin-like structure? Or other\n models? At present, additional models should be explored and also considering gas-phase\n minimization prior to CCS calculation.\n No controls for the electrical recordings using planar lipid bilayers are presented. DPC is a\n detergent, which could disrupt the integrity of the bilayer giving rise to background signals. Were\n measurements of DPC in the absence of Ab performed?\n The section titled “preparation of ... enriched in Ab(1-42) octamers” seems out of place. Suggest\n moving up and before the SEC-MS analysis section.\n The manuscript could be improved by merging some figures. For example, Fig S17 can be\n incorporated into Fig S7.\n For denatured and native MS measurements, the measured and theoretical masses should be\n reported along with error.\n Supplementary Fig. 18 is mislabelled. I presume the data shown is for PFOsHIGH not PFOsLOW. Pg 8, “... using DPC at natural abundance.” This is unclear as DPC is non-natural.\nPg 15, “.. we have precisely defined the structural ..” Remove precisely.\nReviewer #3 (Remarks to the Author):\n Although I am not able to expertly comment on the quality of the NMR data, the study seems well\n designed and executed.\n This is a very interesting paper that presents a startling structure for the micelle-inserted amyloid-\n beta-42 polypeptide (AB42). AB42 is, of course, generally thought to be associated with the\n etiology of Alzheimer's disease. There is a body of evidence in the literature that AB42 can adopt\n some sort of transmembrane oligomer, with additional evidence that it can act (aberrantly) as an\n ion channel, potentially a mechanism that contributes to how AB42 promotes Alzheimer's disease.\n In this paper the authors used NMR spectroscopy to document that AB42 in dodecylphosphocholine\n micelles can form two closely related oligomeric structures. In one structure 4 subunits come\n togehther to form a single sheet with two subunits forming trans-micelle beta hairpins flank the\n two other subunits central in the tetrameric sheet as a single beta strands. What is amazing is that\n the 6-stranded tetatrameric sheet does not form a closed barrel, but instead is an \"open\" sheet\n with solvent exposed edges for strands 1 and 6. The other structure is a dimer of tetramers where\n the 6-stranded sheets form a sandwich in the micelle interior, again with exposed edges. The\n authors also present electrophysiology measurements in which ion channel activity is documented,\n apparently for both monomer and dimers, with the channel activity apparently being associated\n with the exposed edges of the monomeric or dimeric sheets.\n The NMR studies of this work were meticulously carried out and documented-- very convincing.\n Moreover, one can imagine how these structures might be stable in micelles: the horizontal\n diameter of the sheet is on the same order as the diameter of the micelle it sits in, so the need to\n form transmicelle beta barrels is relieved by the fact that the exposed edges of strands 1 and 6 are\n not exposed to a bilayer interior, but instead to water outside of the micelle water at the micelle-\n water interface. So, these structures seem reasonable as micelle structures. However, it seems\n less clear that they could be energetically accommodated as transmembrane structures in a sealed\n lipid bilayer, where the exposed edges of strands 1 and 6 would be exposed only to lipid, resulting\n in burial backbone amide groups with unsatisfied H-bonding potential.\n The text, figures, and supporting material are all very well written. I think this work is EXTREMELY\n interesting and should be published following revision, without any requirement for additional\n experiments. However, the authors should provide more detail on what evidence they have that\n the micellar structure they carefully document in this work is maintained in lipid vesicles. This\n closely relates to the question of whether the channel activity they document really stems from\n AB42 channels that resemble the micellar structures. Again, more textual justification seems\n needed.\n Finally, the sample composition, pH, and temperature need to be spelled out in detail in the\n caption to Figure 1.\n Reviewer #4 (Remarks to the Author): Review of NCOMMS-19-35967-T\n The manuscript by Ciudad et al deals with an important open question in our understanding of the\n Alzheimer’s Disease (AD) Pathology cascade: the mechanism by which amyloid-beta causes\n damages neurons. The authors cite the “amyloid hypothesis” in which Abeta is assumed to be the\n primary toxic agent in AD, and note that the mechanism of toxcity is unclear. They also cite works\n in which soluble forms of Abeta can induce pores of various sorts in membranes. This is the\n background for their basic hypothesis: that Abeta structures can form pores in planar lipid\n bilayers, and therefore can serve as a mechanism of neurotoxicity. The primary focus of the work\n is in establishing the feasibility of this idea, and the authors perform detailed work in establishing\n this to be the case. The evidence that Abeta structures can form pores in planar lipid bilayers, and\n that this is specific to certain forms of Abeta, seems to be well established by this study, and\n supported by beautiful and clear figures.\n I will address a different aspect of this study, which is the validity of their hypothesis: that Abeta\n can cause pores in the membranes of neurons adjacent to sufficient concentrations of Abeta, which\n will then disrupt neuronal activity. The principal evidence for that disruption is shown in\n Supplementary Figure 20, in which the membrane conductance is increased by the introduction of\n βPFOsAβ(1-42) into planar lipid bilayers.\n This makes a clear prediction of the effect of soluble Abeta on neuronal activity. The prediction is\n that the input conductance of these neurons will be increased. Intracellular recordings from several\n lines of APP+ transgenic mice (two of which are published) , in which both soluble Abeta and\n amyloid plaques are present in the surrounding neuropil, by measurement of average input\n resistance in the neurons, do not show the increase in RIN that would be expected to be\n associated with such a conductance increase (Stern et al, J. Neurosci 24, 4335-4340 (2004);\n Kellner et al, Neurobiology of Aging 35, 1982-1991 (2014)). I do not know of any studies which\n have shown different results. (In addition, measurement by several groups show hyperactivity in\n numerous lines of APP transgenic mice, which is the opposite effect that would be expected from a\n general conductance increase.) these are issues that should be addressd by the authors, if they\n wish to argue that increased membrane conductance is indeed a mechanism of neuronal disruption\n by Abeta.\n NOTE: This, however, is not necessarily evidence that the conductance increase does not occur.\n The direct measurements of input resistance are made using electrodes inserted into the soma. It\n may well be that the the toxic effects of Abeta on the membrane are local. Indeed, one study\n shows that there are specific effects on dendritic spines in proximity to plaques, where the\n concentrations of soluble Abeta are higher (Spires et al, J Neurosci. 25, 7278-7287 (2005)).\n This is an hypothesis which can be tested, using targeted patch recordings in slices of APP\n transgenic mouse cortex. The recordings would have to target dendrites in mice in which plaque\n aggregation is severe, so that dendrites would have a high probability of proximity to plaques.\n While difficult, targeted recordings are performed on a daily basis by numerous laboratories. If the\n input conductance is increased in these dendrites, compared to aged-match recordings in WT\n mice, it would indeed be strong evidence that the hypothesis proposed by the authors is correct.\n In many studies, including the one I cited, there is evidence for stripping of spines form these\n dendrites, which I would predict will show the opposite effect: raising the local dendritic input\n resistance.\n At present, however, there is no such evidence of which I am aware that the mechanism proposed\n by the authors disrupts neuronal activity in AD. And some evidence does exist that the overall\n conductance in the neuron is not raised in APP-transgenic mouse model neurons. The current\n study does show an interesting possible mechanism of action of one of the forms of Abeta: one\n interesting enough to be further investigated. Again, evidence for the specific mechanisms (note\n the plural) of the pathophysiological effects of Abeta is a “missing link” in our understanding of the\n cascade of effects of the Abeta in the disase process.\n Edward A. Stern\n Brain Research Center\n Bar Ilan University\n", "role": "assistant" } ]	2,020
Aβ(1-42) tetramer and octamer structures reveal edge conductivity pores as a mechanism	REVIEWERS' COMMENTS second round: Reviewer #1 (Remarks to the Author): Reviewer #2 (Remarks to the Author): Reviewer #3 (Remarks to the Author): The authors addressed the comments of all reviewers with great care. With regard to my comments I can confirm that they were also addressed satisfactorily. I have no further suggestions and recommend publication of the manuscript as is. The revised manuscript by Ciudad et al. has greatly improved. They have addressed my concerns and recommend publication. The authors have done a good job of addressing the reviewer concerns. As indicated in my original review I think this is an interesting, important, and novel work. I am in favor of publishing the revised manuscript as is.	REVIEWERS' COMMENTS second round: Reviewer #1 (Remarks to the Author): Reviewer #2 (Remarks to the Author): Reviewer #3 (Remarks to the Author): The authors addressed the comments of all reviewers with great care. With regard to my comments I can confirm that they were also addressed satisfactorily. I have no further suggestions and recommend publication of the manuscript as is. The revised manuscript by Ciudad et al. has greatly improved. They have addressed my concerns and recommend publication. The authors have done a good job of addressing the reviewer concerns. As indicated in my original review I think this is an interesting, important, and novel work. I am in favor of publishing the revised manuscript as is.	TPR	NC_BS_BIOPH_008_RR2.txt	{ "criticism": 1, "example": 0, "importance_and_relevance": 1, "materials_and_methods": 0, "praise": 6, "presentation_and_reporting": 0, "results_and_discussion": 0, "suggestion_and_solution": 2, "total": 9 }	{ "criticism": 0.1111111111111111, "example": 0, "importance_and_relevance": 0.1111111111111111, "materials_and_methods": 0, "praise": 0.6666666666666666, "presentation_and_reporting": 0, "results_and_discussion": 0, "suggestion_and_solution": 0.2222222222222222 }	1.111111	[ { "content": "Please review this paper.", "role": "user" }, { "content": "REVIEWERS' COMMENTS second round: Reviewer #1 (Remarks to the Author):\nReviewer #2 (Remarks to the Author):\nReviewer #3 (Remarks to the Author):\n The authors addressed the comments of all reviewers with great care. With regard to my\n comments I can confirm that they were also addressed satisfactorily. I have no further suggestions\n and recommend publication of the manuscript as is.\n The revised manuscript by Ciudad et al. has greatly improved. They have addressed my concerns\n and recommend publication.\n The authors have done a good job of addressing the reviewer concerns. As indicated in my original\n review I think this is an interesting, important, and novel work. I am in favor of publishing the\n revised manuscript as is.", "role": "assistant" } ]	2,020
Single molecule secondary structure determination of proteins through infrared absorption	Reviewers' comments: Reviewer #1 (Remarks to the Author): The manuscript by Ruggeri and co-authors reports on structural determination of single proteins using AFM-IR. Specifically, the aurthors showed that using this highly sensitive technique they could probe secondary structure of thyroglobulin and apoferritin determining amounts of a-helix, b-turns and coil in their structures. The manuscript is well-written and should eventually be accepted to the publication. However, prior to it, the authors must address the most important question of their work that is not entirely clear. How do the researchers know that they dealt with one protein molecule rather than with a protein aggregate. The reviewer suggests to use alternative analytical technique, such as DLS, to convince the audience that single molecules of the mentioned above proteins were studied. Once demonstrated, the reviewer suggests to accept the manuscript. Reviewer #2 (Remarks to the Author): The article by Ruggeri et al addresses the investigation of isolated large biomolecules by the photothermal AFM-IR technique. For increased sensitivity a slightly off-resonant measurement modus was used in a study of isolated apoferritin and thyroglobulin molecules on a gold substrate. Conclusions on the secondary structure of the investigated proteins are drawn with the interpretation of measured AFM-IR and supplementary CD spectra. Although the thyroglobulin molecules are with dimensions of about 8 nm x 10 nm x 5 nm quite large, to my knowledge this is the first single molecule study by an mid-infrared spectroscopic technique. As such new experimental IR spectroscopic possibilities give access to the analysis of structural and chemical properties of single large biomolecules and macromolecules, they could be in particular relevant for biomedical research. I recommend the publication of the article after reasonable revision. Major points: I) How do you know that really only one molecule was measured? Please comment on the tip convolution effects in the text. II) It is claimed that physical properties of thermomechanical detection and field enhancement are unraveled, where? III) Potential future possibilities of the technique should be discussed better. In this context: e.g., could the internal structure of the large biomolecules be studied? Why is it not studied yet? It is written that AFM-IR has spatial resolutions from about 1 to 10 nm. This should be maybe sufficient to investigate e.g. the two parts of the homodimer thyroglobulin? IV) What was the spatial resolution in the presented experiments? Please add. If the spatial resolution was below 10 nm, why can the molecules not be studied without probing any parts of the bare gold substrate? Please comment. V) With respect to the used measurement modus, why didn’t the authors use the shown higher order resonance at around 400 kHz in Fig. 1b? For this one protein and gold resonance should be better separated than for the used resonance frequency at 194 kHz. VI) Make a statement if the spectrum of the substrate is influenced by the QCl emission profile related background as shown in Fig S1. How does this background and low sensitivity in particular influence the amide 2 band shape in the spectra of the proteins, can this be related to the significant differences between FTIR and AFM-IR spectra as seen in Figure 4. If not, what is the reason? Minor: - Introduction: Thus, the IR absorption ... are not affected by plasmonic effects. I guess plasmonic resonances are meant here?. Please clarify. - Result sections: Add the technical details of the used mesh filter. - Caption Figure 1: ... high values can cause instabilities .. Comment: .. Maybe the signal is saturated?.. - Figure 2 b: as it can be barely identified or mistaken please mark the measurement line by an other colour, e.g. green. - Materials section: What means a 20 μM of apoferritin.. ; solution? - Please control referencing of Ref. [32] after (Figure 4b) in the Results section. - A reference for the assignments used in the deconvolution of the amide I band (supplementary part) is missing. Supplementary: - 1st sentence: typo: ... we aimed at we aimed .... - Line 33: 10 mW is given as the lower limit for typical power of the laser. This sounds quite high for the study of an organic samples. - Line 56: contact resonance at higher angle? - A reference for the assignments used in the deconvolution of the amide I band (supplementary part) is missing. Reviewer #3 (Remarks to the Author): Here are my comments regarding the manuscript: Manuscript untitled: « Single Molecule Secondary Structure Determination of Protein through Off- Resonance and Short Pulse Infrared Absorption Nanospectroscopy » I don’t recommend its publication in nature communication: it’s not innovative enough and the format is not appropriated to this technical article. Here are some reasons and the list is not comprehensive. General comments: The authors introduce off-resonance, low power and short pulse AFM-IR to acquire infrared absorption spectra of thyroglobulin at the single molecule level, with a high signal-to-noise ratio. Without any doubt, it’s essential to the community to have access to new protocols to improve the sensitivity and reliability of this technique but it’s not innovative enough to be published in a journal like nature communication. I don't want to minimize the interest of this publication. Since the analysis of single protein using AFM-IR is a real challenge, I suggest to the authors to publish it in a more specialized journal where they can bring a real discussion, develop the different ambiguous points and illustrate it with more figures. This way, I think it should be appreciated to its full value by giving more details (by merging the main text and the supplementary documents) and complementary information and by highlighting the results with more proteins. -You said “To date, however, the sensitivity of AFM-IR has been limited to the measurements of large (>0.3 m) and flat (~2-10 nm) self-assembled monolayers or biomolecular aggregates composed of hundreds of molecules and has not demonstrated the ability to characterise single molecules”: check other publications and update your bibliography. Some points are a little bit confusing: -Figure 1 d) is really difficult to understand. Are you sure of your legend (blue dots 100ns ? red dots 7%) I don’t understand the link. -Figure e): if your system is well optimized (in term of power and pulse width) the amplitude of the resonance of the substrate should be lower than the one of the protein. It’s not the case. Why? -in the caption of the figure 1 “(e) IR amplitude as a function of frequency pulse for signal on gold and on a thyroglobulin molecule at a laser power of 7% and at 1655 cm-1” the power used for the experiments is 7 % but in the text it’s written p5 that “already at the low laser power of 0.5 mW (1-5% of commercial IR sources), the enhancement of the field causes non-linearity and instabilities” may be I have missed something but the power used for the experiments should be lower than 7% ? I suppose it’s 7% of the power already reduce after the addition of the filter. I think it’s better to use only mW and not %. -Then figure 2 and 3 no power nor pulse width is indicated. -What about the discrepancies observed between FTIR results and AFM-IR? For example why is your amide II lower? May be there is a physical reason behind it: discuss it. But my main concern is about the interpretation of the local IR spectra: “Apoferritin has an - helical content of 75% (Figure 4a),34 while thyroglobulin has major contributions from -helix and coils (55%), -turn (17%) and intramolecular -sheet structures (28%) (Figure 4b).” How you can propose such repartition without showing a curve fitting? the reference extract from FTIR curve fitting (reference) should be in the main document.	Reviewers' comments: Reviewer #1 (Remarks to the Author): The manuscript by Ruggeri and co-authors reports on structural determination of single proteins using AFM-IR. Specifically, the aurthors showed that using this highly sensitive technique they could probe secondary structure of thyroglobulin and apoferritin determining amounts of a-helix, b-turns and coil in their structures. The manuscript is well-written and should eventually be accepted to the publication. However, prior to it, the authors must address the most important question of their work that is not entirely clear. How do the researchers know that they dealt with one protein molecule rather than with a protein aggregate. The reviewer suggests to use alternative analytical technique, such as DLS, to convince the audience that single molecules of the mentioned above proteins were studied. Once demonstrated, the reviewer suggests to accept the manuscript. Reviewer #2 (Remarks to the Author): The article by Ruggeri et al addresses the investigation of isolated large biomolecules by the photothermal AFM-IR technique. For increased sensitivity a slightly off-resonant measurement modus was used in a study of isolated apoferritin and thyroglobulin molecules on a gold substrate. Conclusions on the secondary structure of the investigated proteins are drawn with the interpretation of measured AFM-IR and supplementary CD spectra. Although the thyroglobulin molecules are with dimensions of about 8 nm x 10 nm x 5 nm quite large, to my knowledge this is the first single molecule study by an mid-infrared spectroscopic technique. As such new experimental IR spectroscopic possibilities give access to the analysis of structural and chemical properties of single large biomolecules and macromolecules, they could be in particular relevant for biomedical research. I recommend the publication of the article after reasonable revision. Major points: I) How do you know that really only one molecule was measured? Please comment on the tip convolution effects in the text. II) It is claimed that physical properties of thermomechanical detection and field enhancement are unraveled, where? III) Potential future possibilities of the technique should be discussed better. In this context: e.g., could the internal structure of the large biomolecules be studied? Why is it not studied yet? It is written that AFM-IR has spatial resolutions from about 1 to 10 nm. This should be maybe sufficient to investigate e.g. the two parts of the homodimer thyroglobulin? IV) What was the spatial resolution in the presented experiments? Please add. If the spatial resolution was below 10 nm, why can the molecules not be studied without probing any parts of the bare gold substrate? Please comment. V) With respect to the used measurement modus, why didn’t the authors use the shown higher order resonance at around 400 kHz in Fig. 1b? For this one protein and gold resonance should be better separated than for the used resonance frequency at 194 kHz. VI) Make a statement if the spectrum of the substrate is influenced by the QCl emission profile related background as shown in Fig S1. How does this background and low sensitivity in particular influence the amide 2 band shape in the spectra of the proteins, can this be related to the significant differences between FTIR and AFM-IR spectra as seen in Figure 4. If not, what is the reason? Minor: - Introduction: Thus, the IR absorption ... are not affected by plasmonic effects. I guess plasmonic resonances are meant here?. Please clarify. - Result sections: Add the technical details of the used mesh filter. - Caption Figure 1: ... high values can cause instabilities .. Comment: .. Maybe the signal is saturated?.. - Figure 2 b: as it can be barely identified or mistaken please mark the measurement line by an other colour, e.g. green. - Materials section: What means a 20 μM of apoferritin.. ; solution? - Please control referencing of Ref. [32] after (Figure 4b) in the Results section. - A reference for the assignments used in the deconvolution of the amide I band (supplementary part) is missing. Supplementary: - 1st sentence: typo: ... we aimed at we aimed .... - Line 33: 10 mW is given as the lower limit for typical power of the laser. This sounds quite high for the study of an organic samples. - Line 56: contact resonance at higher angle? - A reference for the assignments used in the deconvolution of the amide I band (supplementary part) is missing. Reviewer #3 (Remarks to the Author): Here are my comments regarding the manuscript: Manuscript untitled: « Single Molecule Secondary Structure Determination of Protein through Off- Resonance and Short Pulse Infrared Absorption Nanospectroscopy » I don’t recommend its publication in nature communication: it’s not innovative enough and the format is not appropriated to this technical article. Here are some reasons and the list is not comprehensive. General comments: The authors introduce off-resonance, low power and short pulse AFM-IR to acquire infrared absorption spectra of thyroglobulin at the single molecule level, with a high signal-to-noise ratio. Without any doubt, it’s essential to the community to have access to new protocols to improve the sensitivity and reliability of this technique but it’s not innovative enough to be published in a journal like nature communication. I don't want to minimize the interest of this publication. Since the analysis of single protein using AFM-IR is a real challenge, I suggest to the authors to publish it in a more specialized journal where they can bring a real discussion, develop the different ambiguous points and illustrate it with more figures. This way, I think it should be appreciated to its full value by giving more details (by merging the main text and the supplementary documents) and complementary information and by highlighting the results with more proteins. -You said “To date, however, the sensitivity of AFM-IR has been limited to the measurements of large (>0.3 m) and flat (~2-10 nm) self-assembled monolayers or biomolecular aggregates composed of hundreds of molecules and has not demonstrated the ability to characterise single molecules”: check other publications and update your bibliography. Some points are a little bit confusing: -Figure 1 d) is really difficult to understand. Are you sure of your legend (blue dots 100ns ? red dots 7%) I don’t understand the link. -Figure e): if your system is well optimized (in term of power and pulse width) the amplitude of the resonance of the substrate should be lower than the one of the protein. It’s not the case. Why? -in the caption of the figure 1 “(e) IR amplitude as a function of frequency pulse for signal on gold and on a thyroglobulin molecule at a laser power of 7% and at 1655 cm-1” the power used for the experiments is 7 % but in the text it’s written p5 that “already at the low laser power of 0.5 mW (1-5% of commercial IR sources), the enhancement of the field causes non-linearity and instabilities” may be I have missed something but the power used for the experiments should be lower than 7% ? I suppose it’s 7% of the power already reduce after the addition of the filter. I think it’s better to use only mW and not %. -Then figure 2 and 3 no power nor pulse width is indicated. -What about the discrepancies observed between FTIR results and AFM-IR? For example why is your amide II lower? May be there is a physical reason behind it: discuss it. But my main concern is about the interpretation of the local IR spectra: “Apoferritin has an - helical content of 75% (Figure 4a),34 while thyroglobulin has major contributions from -helix and coils (55%), -turn (17%) and intramolecular -sheet structures (28%) (Figure 4b).” How you can propose such repartition without showing a curve fitting? the reference extract from FTIR curve fitting (reference) should be in the main document.	TPR	NC_BS_BIOPH_009_RR1.txt	{ "criticism": 9, "example": 13, "importance_and_relevance": 7, "materials_and_methods": 29, "praise": 2, "presentation_and_reporting": 25, "results_and_discussion": 7, "suggestion_and_solution": 23, "total": 71 }	{ "criticism": 0.1267605633802817, "example": 0.18309859154929578, "importance_and_relevance": 0.09859154929577464, "materials_and_methods": 0.4084507042253521, "praise": 0.028169014084507043, "presentation_and_reporting": 0.352112676056338, "results_and_discussion": 0.09859154929577464, "suggestion_and_solution": 0.323943661971831 }	1.619718	[ { "content": "Please review this paper.", "role": "user" }, { "content": "Reviewers' comments:\nReviewer #1 (Remarks to the Author):\nThe manuscript by Ruggeri and co-authors reports on structural determination of single proteins\nusing AFM-IR. Specifically, the aurthors showed that using this highly sensitive technique they\ncould probe secondary structure of thyroglobulin and apoferritin determining amounts of a-helix,\nb-turns and coil in their structures. The manuscript is well-written and should eventually be\naccepted to the publication. However, prior to it, the authors must address the most important\nquestion of their work that is not entirely clear. How do the researchers know that they dealt with\none protein molecule rather than with a protein aggregate. The reviewer suggests to use\nalternative analytical technique, such as DLS, to convince the audience that single molecules of the\nmentioned above proteins were studied. Once demonstrated, the reviewer suggests to accept the\nmanuscript.\nReviewer #2 (Remarks to the Author):\nThe article by Ruggeri et al addresses the investigation of isolated large biomolecules by the\nphotothermal AFM-IR technique. For increased sensitivity a slightly off-resonant measurement\nmodus was used in a study of isolated apoferritin and thyroglobulin molecules on a gold substrate.\nConclusions on the secondary structure of the investigated proteins are drawn with the\ninterpretation of measured AFM-IR and supplementary CD spectra.\nAlthough the thyroglobulin molecules are with dimensions of about 8 nm x 10 nm x 5 nm quite\nlarge, to my knowledge this is the first single molecule study by an mid-infrared spectroscopic\ntechnique. As such new experimental IR spectroscopic possibilities give access to the analysis of\nstructural and chemical properties of single large biomolecules and macromolecules, they could be\nin particular relevant for biomedical research. I recommend the publication of the article after\nreasonable revision.\nMajor points:\nI) How do you know that really only one molecule was measured? Please comment on the tip\nconvolution effects in the text.\nII) It is claimed that physical properties of thermomechanical detection and field enhancement are\nunraveled, where?\nIII) Potential future possibilities of the technique should be discussed better. In this context: e.g.,\ncould the internal structure of the large biomolecules be studied? Why is it not studied yet? It is\nwritten that AFM-IR has spatial resolutions from about 1 to 10 nm. This should be maybe sufficient\nto investigate e.g. the two parts of the homodimer thyroglobulin?\nIV) What was the spatial resolution in the presented experiments? Please add. If the spatial\nresolution was below 10 nm, why can the molecules not be studied without probing any parts of\nthe bare gold substrate? Please comment.\nV) With respect to the used measurement modus, why didn’t the authors use the shown higher\norder resonance at around 400 kHz in Fig. 1b? For this one protein and gold resonance should be\nbetter separated than for the used resonance frequency at 194 kHz.\nVI) Make a statement if the spectrum of the substrate is influenced by the QCl emission profile\nrelated background as shown in Fig S1. How does this background and low sensitivity in particular\ninfluence the amide 2 band shape in the spectra of the proteins, can this be related to the\nsignificant differences between FTIR and AFM-IR spectra as seen in Figure 4. If not, what is the\nreason?\n Minor:\n - Introduction: Thus, the IR absorption ... are not affected by plasmonic effects.\n I guess plasmonic resonances are meant here?. Please clarify.\n - Result sections:\n Add the technical details of the used mesh filter.\n - Caption Figure 1: ... high values can cause instabilities ..\n Comment: .. Maybe the signal is saturated?..\n - Figure 2 b: as it can be barely identified or mistaken please mark the measurement line by an\n other colour, e.g. green.\n - Materials section:\n What means a 20 μM of apoferritin.. ; solution?\n - Please control referencing of Ref. [32] after (Figure 4b) in the Results section.\n - A reference for the assignments used in the deconvolution of the amide I band (supplementary\n part) is missing.\n Supplementary:\n - 1st sentence: typo: ... we aimed at we aimed ....\n - Line 33: 10 mW is given as the lower limit for typical power of the laser. This sounds quite high\n for the study of an organic samples.\n - Line 56: contact resonance at higher angle?\n - A reference for the assignments used in the deconvolution of the amide I band (supplementary\n part) is missing.\n Reviewer #3 (Remarks to the Author):\nHere are my comments regarding the manuscript:\n Manuscript untitled: « Single Molecule Secondary Structure Determination of Protein through Off-\n Resonance and Short Pulse Infrared Absorption Nanospectroscopy »\n I don’t recommend its publication in nature communication: it’s not innovative enough and the\n format is not appropriated to this technical article.\n Here are some reasons and the list is not comprehensive.\n General comments:\n The authors introduce off-resonance, low power and short pulse AFM-IR to acquire infrared\n absorption spectra of thyroglobulin at the single molecule level, with a high signal-to-noise ratio.\n Without any doubt, it’s essential to the community to have access to new protocols to improve the\n sensitivity and reliability of this technique but it’s not innovative enough to be published in a\n journal like nature communication.\n I don't want to minimize the interest of this publication. Since the analysis of single protein using\n AFM-IR is a real challenge, I suggest to the authors to publish it in a more specialized journal\n where they can bring a real discussion, develop the different ambiguous points and illustrate it\n with more figures. This way, I think it should be appreciated to its full value by giving more details\n (by merging the main text and the supplementary documents) and complementary information\n and by highlighting the results with more proteins.\n -You said “To date, however, the sensitivity of AFM-IR has been limited to the measurements of\n large (>0.3 m) and flat (~2-10 nm) self-assembled monolayers or biomolecular aggregates\n composed of hundreds of molecules and has not demonstrated the ability to characterise single\n molecules”: check other publications and update your bibliography.\n Some points are a little bit confusing:\n -Figure 1 d) is really difficult to understand. Are you sure of your legend (blue dots 100ns ? red\n dots 7%) I don’t understand the link.\n -Figure e): if your system is well optimized (in term of power and pulse width) the amplitude of\n the resonance of the substrate should be lower than the one of the protein. It’s not the case. Why?\n -in the caption of the figure 1 “(e) IR amplitude as a function of frequency pulse for signal on gold\n and on a thyroglobulin molecule at a laser power of 7% and at 1655 cm-1” the power used for the\n experiments is 7 % but in the text it’s written p5 that “already at the low laser power of 0.5 mW\n (1-5% of commercial IR sources), the enhancement of the field causes non-linearity and\n instabilities” may be I have missed something but the power used for the experiments should be\n lower than 7% ? I suppose it’s 7% of the power already reduce after the addition of the filter. I\n think it’s better to use only mW and not %.\n -Then figure 2 and 3 no power nor pulse width is indicated.\n -What about the discrepancies observed between FTIR results and AFM-IR? For example why is\n your amide II lower? May be there is a physical reason behind it: discuss it.\n But my main concern is about the interpretation of the local IR spectra: “Apoferritin has an -\n helical content of 75% (Figure 4a),34 while thyroglobulin has major contributions from -helix and\n coils (55%), -turn (17%) and intramolecular -sheet structures (28%) (Figure 4b).” How you can\n propose such repartition without showing a curve fitting? the reference extract from FTIR curve\n fitting (reference) should be in the main document.\n", "role": "assistant" } ]	2,020
Single molecule secondary structure determination of proteins through infrared absorption	REVIEWERS' COMMENTS: Reviewer #2 (Remarks to the Author): The manuscript was revised very well. I recommend to accept it for publication. technical comments: - The authors should consider to check if the maximum of signal as given in the text in connection with Figure 1 f) is at ~-2kHz off-resonance. - Fig S4; "FTIR" is found in the inset and caption, but the axis label is "IR amplitude"?	REVIEWERS' COMMENTS: Reviewer #2 (Remarks to the Author): The manuscript was revised very well. I recommend to accept it for publication. technical comments: - The authors should consider to check if the maximum of signal as given in the text in connection with Figure 1 f) is at ~-2kHz off-resonance. - Fig S4; "FTIR" is found in the inset and caption, but the axis label is "IR amplitude"?	TPR	NC_BS_BIOPH_009_RR2.txt	{ "criticism": 0, "example": 1, "importance_and_relevance": 0, "materials_and_methods": 1, "praise": 2, "presentation_and_reporting": 2, "results_and_discussion": 0, "suggestion_and_solution": 2, "total": 4 }	{ "criticism": 0, "example": 0.25, "importance_and_relevance": 0, "materials_and_methods": 0.25, "praise": 0.5, "presentation_and_reporting": 0.5, "results_and_discussion": 0, "suggestion_and_solution": 0.5 }	2	[ { "content": "Please review this paper.", "role": "user" }, { "content": "REVIEWERS' COMMENTS:\nReviewer #2 (Remarks to the Author):\n The manuscript was revised very well.\n I recommend to accept it for publication.\n technical comments:\n - The authors should consider to check if the maximum of signal as given in the text in connection\n with Figure 1 f) is at ~-2kHz off-resonance.\n - Fig S4; \"FTIR\" is found in the inset and caption, but the axis label is \"IR amplitude\"?", "role": "assistant" } ]	2,020
Polyamine regulation of ion channel assembly and implications for nicotinic acetylcholine	Reviewers' Comments: Reviewer #1: Remarks to the Author: Novel genome-wide screening was used to discover an unlikely candidate for selectively and efficiently promoting expression of alpha 7 and alpha 4 beta 2 AChRs through a novel mechanism. The small cytoplasmic protein SAT1 encodes the rate limiting enzyme for polyamine catabolism, and degrading polyamines turns out to promote assembly of these AChRs through interactions with their large cytoplasmic domains. The polyamine spermine can block the AChR cation channel, but does not alter the up regulation of AChRs by SAT1. Upregulation by SAT 1 depends on the presence of the alpha 4 or alpha 7 AChR large cytoplasmic domain. A cell-permeant inhibitor of SAT1 can block its effects. Blocking polyamine synthesis with DMFO (a drug that inhibits polyamide synthesis) was shown to up regulate alpha 4 beta 2 and alpha 7 AChR amounts and thereby nicotine-mediated protection of postsynaptic neurons from excitotoxicity, thereby suggesting a new approach to developing neuroprotective drugs. Mutations that reduce SAT1 expression are associated with increased suicide. Alpha 7 and alpha 4 beta 2 AChR agonists are thought to improve depressive behavior in patients and animal models. These investigations have significant implications for drug development. These investigations also reveal previously unknown mechanisms for regulating expression of these AChR subtypes. The data presented are extensive, clearly presented, and compelling. Reviewer #2: Remarks to the Author: Dhara et al show for the first time a new function of polyamines, distinct from that previously known of controlling channel gating. They demonstrate that the intracellular reduction of polyamine levels, by means of different approaches, increases the assembly and plasma membrane trafficking of the alpha7 and alpha 4beta2 subtypes. They also clearly demonstrate that the polyamine effects on assembly involve the M3-M4 subunit cytosolic loop. The effect of polyamine on nAChRs is present in neurons and the polyamine-induced nAChR upregulation is protective towards glutamate induce excitotoxicity; however, this protective effect is not present in neurons of NACHO KO mice. The work is well done, clearly described and discussed. The results are very interesting and deserve publication, although the cellular mechanisms acting on the different subtypes are not completely elucidated Minor but important points 1) the title does not reflect properly the content of the paper. In the title there is no mention of nicotinic receptor subtypes, but in this paper the effect of polyamines is only on some of these subtypes. Indeed the reference to neuropharmacology is too vague. 2) In the introduction it is written that “polyamine levels selectively up-regulates cerebrocortical α4β2 and α7 levels”. Whereas in neurons it has been shown, by using non permeable Bgtx647, an increase in alpha7-containing receptors, in the case of alpha4beta2 receptors this has not been demonstrated. The increase in 3H-epibatidine binding seen in figure 6d may be due to binding to both receptor subtypes. It will be important to perform the binding of 3H-Epibatidine in presence and absence of 100 nM Cytisine (that inhibits the binding to the alpha4beta2 subtype), in order to quantify in neurons the increase in alpha7 and alpha4beta2 subtypes, respectively. Moreover by performing in neurons the Ca+ response in presence of an alpha4beta2 specific allosteric modulator it is also possible to know whether the functional alpha4beta2 receptors are increased at the cell surface. 3) At page 9 in discussion it is written that “Nicotine and other orthosteric ligands promote assembly through α7’s extracellular ACh binding domain43”. Please delete “extracellular” because Lester et al (43) show that the assembly is favoured by the intracellular action of nicotine and orthosteric ligands.	Reviewers' Comments: Reviewer #1: Remarks to the Author: Novel genome-wide screening was used to discover an unlikely candidate for selectively and efficiently promoting expression of alpha 7 and alpha 4 beta 2 AChRs through a novel mechanism. The small cytoplasmic protein SAT1 encodes the rate limiting enzyme for polyamine catabolism, and degrading polyamines turns out to promote assembly of these AChRs through interactions with their large cytoplasmic domains. The polyamine spermine can block the AChR cation channel, but does not alter the up regulation of AChRs by SAT1. Upregulation by SAT 1 depends on the presence of the alpha 4 or alpha 7 AChR large cytoplasmic domain. A cell-permeant inhibitor of SAT1 can block its effects. Blocking polyamine synthesis with DMFO (a drug that inhibits polyamide synthesis) was shown to up regulate alpha 4 beta 2 and alpha 7 AChR amounts and thereby nicotine-mediated protection of postsynaptic neurons from excitotoxicity, thereby suggesting a new approach to developing neuroprotective drugs. Mutations that reduce SAT1 expression are associated with increased suicide. Alpha 7 and alpha 4 beta 2 AChR agonists are thought to improve depressive behavior in patients and animal models. These investigations have significant implications for drug development. These investigations also reveal previously unknown mechanisms for regulating expression of these AChR subtypes. The data presented are extensive, clearly presented, and compelling. Reviewer #2: Remarks to the Author: Dhara et al show for the first time a new function of polyamines, distinct from that previously known of controlling channel gating. They demonstrate that the intracellular reduction of polyamine levels, by means of different approaches, increases the assembly and plasma membrane trafficking of the alpha7 and alpha 4beta2 subtypes. They also clearly demonstrate that the polyamine effects on assembly involve the M3-M4 subunit cytosolic loop. The effect of polyamine on nAChRs is present in neurons and the polyamine-induced nAChR upregulation is protective towards glutamate induce excitotoxicity; however, this protective effect is not present in neurons of NACHO KO mice. The work is well done, clearly described and discussed. The results are very interesting and deserve publication, although the cellular mechanisms acting on the different subtypes are not completely elucidated Minor but important points 1) the title does not reflect properly the content of the paper. In the title there is no mention of nicotinic receptor subtypes, but in this paper the effect of polyamines is only on some of these subtypes. Indeed the reference to neuropharmacology is too vague. 2) In the introduction it is written that “polyamine levels selectively up-regulates cerebrocortical α4β2 and α7 levels”. Whereas in neurons it has been shown, by using non permeable Bgtx647, an increase in alpha7-containing receptors, in the case of alpha4beta2 receptors this has not been demonstrated. The increase in 3H-epibatidine binding seen in figure 6d may be due to binding to both receptor subtypes. It will be important to perform the binding of 3H-Epibatidine in presence and absence of 100 nM Cytisine (that inhibits the binding to the alpha4beta2 subtype), in order to quantify in neurons the increase in alpha7 and alpha4beta2 subtypes, respectively. Moreover by performing in neurons the Ca+ response in presence of an alpha4beta2 specific allosteric modulator it is also possible to know whether the functional alpha4beta2 receptors are increased at the cell surface. 3) At page 9 in discussion it is written that “Nicotine and other orthosteric ligands promote assembly through α7’s extracellular ACh binding domain43”. Please delete “extracellular” because Lester et al (43) show that the assembly is favoured by the intracellular action of nicotine and orthosteric ligands.	TPR	NC_BS_BIOPH_010_RR1.txt	{ "criticism": 4, "example": 2, "importance_and_relevance": 3, "materials_and_methods": 12, "praise": 4, "presentation_and_reporting": 5, "results_and_discussion": 3, "suggestion_and_solution": 1, "total": 26 }	{ "criticism": 0.15384615384615385, "example": 0.07692307692307693, "importance_and_relevance": 0.11538461538461539, "materials_and_methods": 0.46153846153846156, "praise": 0.15384615384615385, "presentation_and_reporting": 0.19230769230769232, "results_and_discussion": 0.11538461538461539, "suggestion_and_solution": 0.038461538461538464 }	1.307692	[ { "content": "Please review this paper.", "role": "user" }, { "content": "Reviewers' Comments:\nReviewer #1:\nRemarks to the Author:\nNovel genome-wide screening was used to discover an unlikely candidate for selectively and efficiently\npromoting expression of alpha 7 and alpha 4 beta 2 AChRs through a novel mechanism. The small\ncytoplasmic protein SAT1 encodes the rate limiting enzyme for polyamine catabolism, and degrading\npolyamines turns out to promote assembly of these AChRs through interactions with their large\ncytoplasmic domains. The polyamine spermine can block the AChR cation channel, but does not alter\nthe up regulation of AChRs by SAT1. Upregulation by SAT 1 depends on the presence of the alpha 4 or\nalpha 7 AChR large cytoplasmic domain. A cell-permeant inhibitor of SAT1 can block its effects.\nBlocking polyamine synthesis with DMFO (a drug that inhibits polyamide synthesis) was shown to up\nregulate alpha 4 beta 2 and alpha 7 AChR amounts and thereby nicotine-mediated protection of\npostsynaptic neurons from excitotoxicity, thereby suggesting a new approach to developing\nneuroprotective drugs. Mutations that reduce SAT1 expression are associated with increased suicide.\nAlpha 7 and alpha 4 beta 2 AChR agonists are thought to improve depressive behavior in patients and\nanimal models.\n These investigations have significant implications for drug development. These investigations also\n reveal previously unknown mechanisms for regulating expression of these AChR subtypes. The data\n presented are extensive, clearly presented, and compelling.\nReviewer #2:\nRemarks to the Author:\nDhara et al show for the first time a new function of polyamines, distinct from that previously known\nof controlling channel gating.\nThey demonstrate that the intracellular reduction of polyamine levels, by means of different\napproaches, increases the assembly and plasma membrane trafficking of the alpha7 and alpha 4beta2\nsubtypes. They also clearly demonstrate that the polyamine effects on assembly involve the M3-M4\nsubunit cytosolic loop.\nThe effect of polyamine on nAChRs is present in neurons and the polyamine-induced nAChR\nupregulation is protective towards glutamate induce excitotoxicity; however, this protective effect is\nnot present in neurons of NACHO KO mice.\nThe work is well done, clearly described and discussed.\n The results are very interesting and deserve publication, although the cellular mechanisms acting on\n the different subtypes are not completely elucidated\nMinor but important points\n1) the title does not reflect properly the content of the paper. In the title there is no mention of\nnicotinic receptor subtypes, but in this paper the effect of polyamines is only on some of these\nsubtypes. Indeed the reference to neuropharmacology is too vague.\n2) In the introduction it is written that “polyamine levels selectively up-regulates cerebrocortical α4β2\nand α7 levels”. Whereas in neurons it has been shown, by using non permeable Bgtx647, an increase\nin alpha7-containing receptors, in the case of alpha4beta2 receptors this has not been demonstrated.\nThe increase in 3H-epibatidine binding seen in figure 6d may be due to binding to both receptor\nsubtypes. It will be important to perform the binding of 3H-Epibatidine in presence and absence of 100\nnM Cytisine (that inhibits the binding to the alpha4beta2 subtype), in order to quantify in neurons the\nincrease in alpha7 and alpha4beta2 subtypes, respectively. Moreover by performing in neurons the\nCa+ response in presence of an alpha4beta2 specific allosteric modulator it is also possible to know\nwhether the functional alpha4beta2 receptors are increased at the cell surface.\n3) At page 9 in discussion it is written that “Nicotine and other orthosteric ligands promote assembly\nthrough α7’s extracellular ACh binding domain43”. Please delete “extracellular” because Lester et al\n(43) show that the assembly is favoured by the intracellular action of nicotine and orthosteric ligands.", "role": "assistant" } ]	2,020
Polyamine regulation of ion channel assembly and implications for nicotinic acetylcholine	Reviewers' Comments: Reviewer #1: Remarks to the Author: This paper reports novel discoveries about the role of polyamines in regulating the expression of alpha 7 and alpha 4 beta 2 AChRs but not alpha 6 beta 4 AChRs. Extensive data are provided. The authors were responsive to reviewer comments. This is an interesting and important paper with substantial implications for new therapeutic approaches. Reviewer #2: Remarks to the Author: The authors have addressed all the issues that I have raised and inserted them in the revised version of the paper.	Reviewers' Comments: Reviewer #1: Remarks to the Author: This paper reports novel discoveries about the role of polyamines in regulating the expression of alpha 7 and alpha 4 beta 2 AChRs but not alpha 6 beta 4 AChRs. Extensive data are provided. The authors were responsive to reviewer comments. This is an interesting and important paper with substantial implications for new therapeutic approaches. Reviewer #2: Remarks to the Author: The authors have addressed all the issues that I have raised and inserted them in the revised version of the paper.	TPR	NC_BS_BIOPH_010_RR2.txt	{ "criticism": 0, "example": 0, "importance_and_relevance": 2, "materials_and_methods": 2, "praise": 3, "presentation_and_reporting": 1, "results_and_discussion": 0, "suggestion_and_solution": 0, "total": 5 }	{ "criticism": 0, "example": 0, "importance_and_relevance": 0.4, "materials_and_methods": 0.4, "praise": 0.6, "presentation_and_reporting": 0.2, "results_and_discussion": 0, "suggestion_and_solution": 0 }	1.6	[ { "content": "Please review this paper.", "role": "user" }, { "content": "Reviewers' Comments:\nReviewer #1:\nRemarks to the Author:\nThis paper reports novel discoveries about the role of polyamines in regulating the expression of alpha\n7 and alpha 4 beta 2 AChRs but not alpha 6 beta 4 AChRs. Extensive data are provided. The authors\nwere responsive to reviewer comments. This is an interesting and important paper with substantial\nimplications for new therapeutic approaches.\n Reviewer #2:\n Remarks to the Author:\n The authors have addressed all the issues that I have raised and inserted them in the revised version\n of the paper.", "role": "assistant" } ]	2,020
Kinesin-14 motors drive a right-handed helical motion of antiparallel microtubules around	REVIEWERS' COMMENTS: Reviewer #1 (Remarks to the Author): Reviewer #2 (Remarks to the Author): The revised manuscript by Mitra et al. has now been greatly improved by editing and new data. The authors now demonstrated that Ncd can actually generate torque that is large enough to cause microtubule twisting and coiling. Although the data is qualitative, this demonstration is highly relevant to the chirality of the mitotic spindle in vivo. All comments of me have been addressed in great detail to the full satisfaction. I have read the answers of the authors as well as new data and find it to be satisfactory. I would also like to thank the authors for invested time and effort to improve on their work. I have no further objection for the article to be published in Nature Communications.	REVIEWERS' COMMENTS: Reviewer #1 (Remarks to the Author): Reviewer #2 (Remarks to the Author): The revised manuscript by Mitra et al. has now been greatly improved by editing and new data. The authors now demonstrated that Ncd can actually generate torque that is large enough to cause microtubule twisting and coiling. Although the data is qualitative, this demonstration is highly relevant to the chirality of the mitotic spindle in vivo. All comments of me have been addressed in great detail to the full satisfaction. I have read the answers of the authors as well as new data and find it to be satisfactory. I would also like to thank the authors for invested time and effort to improve on their work. I have no further objection for the article to be published in Nature Communications.	TPR	NC_BS_BIOPH_011_RR2.txt	{ "criticism": 1, "example": 0, "importance_and_relevance": 2, "materials_and_methods": 3, "praise": 5, "presentation_and_reporting": 1, "results_and_discussion": 0, "suggestion_and_solution": 1, "total": 8 }	{ "criticism": 0.125, "example": 0, "importance_and_relevance": 0.25, "materials_and_methods": 0.375, "praise": 0.625, "presentation_and_reporting": 0.125, "results_and_discussion": 0, "suggestion_and_solution": 0.125 }	1.625	[ { "content": "Please review this paper.", "role": "user" }, { "content": "REVIEWERS' COMMENTS:\nReviewer #1 (Remarks to the Author):\nReviewer #2 (Remarks to the Author):\nThe revised manuscript by Mitra et al. has now been greatly improved by editing and new data.\nThe authors now demonstrated that Ncd can actually generate torque that is large enough to\ncause microtubule twisting and coiling. Although the data is qualitative, this demonstration is\nhighly relevant to the chirality of the mitotic spindle in vivo. All comments of me have been\naddressed in great detail to the full satisfaction.\n I have read the answers of the authors as well as new data and find it to be satisfactory. I would\n also like to thank the authors for invested time and effort to improve on their work. I have no\n further objection for the article to be published in Nature Communications.", "role": "assistant" } ]	2,020
Kinesin-14 motors drive a right-handed helical motion of antiparallel microtubules around	Reviewers' comments: Reviewer #1 (Remarks to the Author): The authors systematically compare the rotational motion of microtubules that Ncd motors crosslink in vitro. It is well known that several mitotic kinesins such as kinesin-5 and kinesin-14 crosslink two microtubules. An emerging view is that they exhibit off-axis components and generate the rotation of a microtubule around the long axis of the microtubule as they move on the other microtubule, suggesting an important role in generating twisting of the mitotic spindle. Despite its importance of Ncd in the spindle organization, the rotational motion of microtubules driven by Ncd has not been sufficiently addressed so far. The authors presented a discovery that Ncd can generate the coiling of microtubule bundles, leading to the generation of chirality in the mitotic spindle possibly in collaboration with kinesin-5. A weak point of this paper is that the authors were not able to assess the dependence of the motor density on the helical pitches in 3D sliding motility assays. However, this point was carefully addressed by 2D microtubule motility assays performed on a planar glass surface using a FLIC-based technique. This paper will be an important contribution to the biophysics and cell biology fields, and can also be a useful guide for those who are interested in applying the 3D measurement of nanometer-scale objects to their own research fields. The manuscript is well organized and described. I recommend that it be accepted for publication in Nature Communications. Additional comments: - The difference between the current study and the previous work (Nitzsche et al., PNAS 2016) seems a bit unclear since both papers describe microtubule gliding on Ncd-coated surfaces that rotated around their longitudinal axes. What is novel in the current paper is the possibility that Ncd can generate the coiling of microtubule bundles, which is relevant to the recent finding about the chirality of the mitotic spindle. Although, in principle, this result could be anticipated from the previous results that Ncd exhibits off-axis components in the microtubule gliding assays, the actual experimental data about crosslinked microtubule bundles are of critical importance when building a molecular model of the mitotic spindle assembly. - If possible, it might be better to mention briefly the quantitative relationship between the size/direction of the powerstroke of Ncd that descrived in the previous study, and the resultant rotational motion of microtubules. Although this point might be discussed in the previous study (Nitzsche et al., PNAS 2016), the discussion must be too complicated for readers to follow. Reviewer #2 (Remarks to the Author): In this paper Mitra et al. have upgraded an in vitro system developed previously by the same authors (Mitra et al. PNAS 2018) to show that Ncd, a non processive minus end directed motor protein that has been shown to slide microtubules (Fink et al. Nat Cel Biol, 2009), in a three- dimensional setting can carry a short stabilised antiparallel microtubule around a longer one following a helical, right-handed trajectory. Even though this is the first time such a specific setting has been shown, the amount of previously published data that already shows helical motion of different kinesin motors (as cited by authors themselves) and observations that spindle is a helical structure in both human and yeast cells (Novak et al. Nat. Comm 2018; Winey et al. J. Cell Biol 1995; Ding et al. J Cell Biol. 1993) in my opinion do not make this paper in its current state acceptable for publication in Nature Communication. Even though there is a difference between a motor protein carrying a quantum dot and several motor proteins carrying a short microtubule, I was not convinced by the authors’ arguments to see how this work represents a significant jump in our current understanding of microtubule sliding, especially in the case of the mammalian mitotic spindle where microtubules of similar lengths seem to coil around each other in order to produce torque of microtubule bundles. Rotation of a short microtubule around bigger one is not the same as twisting of microtubules around each other. This being said I believe the work can be accepted for publication with some additional experimental data. As I mentioned the biological motivation of the authors seems to be the mitotic spindle. In that case, as proposed in the cartoon of the Figure 5F, the authors should do an experiment involving two long and fixed microtubules to see whether the motor proteins can exert enough force to cause the coiling of two microtubules. This data would in my opinion be much more informative for the mitotic spindle context, and would actually mirror what is believed to be happening in vivo, where microtubules are fixed at the spindle poles. Further comments: The authors on more than one occasion misquote the Novak et al. article as saying the spindle has right-handed chirality. This is not true, as the article clearly states that the spindle has left-handed chirality. in the last section of Discussion, the authors discuss how the interplay of Eg5 and Ncd might produce such chirality of the spindle. Taking into account that the spindle is left-handed and not right-handed, the authors should adjust their arguments and rewrite the discussion accordingly. Statement that Eg5 and Ncd with their movements of opposite handiness and direction result in torques that add up would benefit from a graphical depiction to make it more understandable. This statement in my opinion is rather important even though it is part of discussion, since it tries to explain the possible interplay of motors with different directionalities and torques present in vivo. Probably my biggest concern with this work is that In Figure 2A the pitches vary considerably with the same concentration at 4nM Ncd. Is seems to me they vary much more within this one concentration, than in the Figure 4D where the concentration of the Ncd was changed? Also in Figure 4D there seems to be no difference between 2.4 nM and 6 nM Ncd and no pitch is close to 3 um, observable at 4nM in Figure 2A. How do authors explain these inconsistencies apart from the fact that in the latter experiment they are using a 2D motility assay and not the suspended microtubule assay? In a situation with parallel transport microtubules related to Figure 2 and Supplementary Figure 2, the comparison between suspended template (Figure 2C) and surface-immobilized template (Supplementary Figure 2C) is confusing. The distribution on the surface-immobilized template in my opinion does not show a” minor bias to the right” as stated by authors but shows a quite big bias to the right. In any case, for me this difference to the bimodal distribution seen in suspended template seems unexpected, why the authors think this is happening? Also, a clearer way of quantifying this bias would help to avoid imprecise wording. The authors attribute to autofluorescence the inability to measure the concentration of Ncd-GFP. Would it be possible to measure it with another fluorescent tag? Would a time-gated imaging approach be helpful to eliminate autofluorescence in this setup? In supplementary figure 4C caption text, I am unconvinced by the hard claim that “At 0.6nM Ncd the microtubule did not show any clear rotational movement”. Looking at the example trajectory shown below the kymographs it is clearly visible that at 0.6 nM there is rotational movement. To me it looks quite similar to what is observed after the switch of concentration considering its much shorter duration. Morover, the authors quantify this rotational movement at 0.6 nM Ncd as having the rotational frequency similar to 6 nM Ncd (supplementary figure 4B right panel). Also, in Figure 4 at 0.6 nM Ncd, the rotational pitch is close to 2 whereas at 6 nM Ncd it is close to 0.8. In the main text however the authors are more prudent with their wording and talk about “few transport microtubules” that rotate at 0.6 nM. I noticed in these experiments the very low N for 0.6 nM Ncd experiments (N=7) compared to N=29 for 6 nM Ncd. Why is this the case? On the same note both Figure 4 and Supplementary Figure 4 lack p-values. The mention of chirality observed in yeast spindles in my opinion also merits a mention in the discussion, specifically Winey et al. J. Cell Biol 1995; Ding et al. J Cell Biol. 1993. Supplementary Movie 3 is not referenced in the main text. Supplementary Figure 4E is referenced in the text, yet there is no Supplementary Figure 4E.	Reviewers' comments: Reviewer #1 (Remarks to the Author): The authors systematically compare the rotational motion of microtubules that Ncd motors crosslink in vitro. It is well known that several mitotic kinesins such as kinesin-5 and kinesin-14 crosslink two microtubules. An emerging view is that they exhibit off-axis components and generate the rotation of a microtubule around the long axis of the microtubule as they move on the other microtubule, suggesting an important role in generating twisting of the mitotic spindle. Despite its importance of Ncd in the spindle organization, the rotational motion of microtubules driven by Ncd has not been sufficiently addressed so far. The authors presented a discovery that Ncd can generate the coiling of microtubule bundles, leading to the generation of chirality in the mitotic spindle possibly in collaboration with kinesin-5. A weak point of this paper is that the authors were not able to assess the dependence of the motor density on the helical pitches in 3D sliding motility assays. However, this point was carefully addressed by 2D microtubule motility assays performed on a planar glass surface using a FLIC-based technique. This paper will be an important contribution to the biophysics and cell biology fields, and can also be a useful guide for those who are interested in applying the 3D measurement of nanometer-scale objects to their own research fields. The manuscript is well organized and described. I recommend that it be accepted for publication in Nature Communications. Additional comments: - The difference between the current study and the previous work (Nitzsche et al., PNAS 2016) seems a bit unclear since both papers describe microtubule gliding on Ncd-coated surfaces that rotated around their longitudinal axes. What is novel in the current paper is the possibility that Ncd can generate the coiling of microtubule bundles, which is relevant to the recent finding about the chirality of the mitotic spindle. Although, in principle, this result could be anticipated from the previous results that Ncd exhibits off-axis components in the microtubule gliding assays, the actual experimental data about crosslinked microtubule bundles are of critical importance when building a molecular model of the mitotic spindle assembly. - If possible, it might be better to mention briefly the quantitative relationship between the size/direction of the powerstroke of Ncd that descrived in the previous study, and the resultant rotational motion of microtubules. Although this point might be discussed in the previous study (Nitzsche et al., PNAS 2016), the discussion must be too complicated for readers to follow. Reviewer #2 (Remarks to the Author): In this paper Mitra et al. have upgraded an in vitro system developed previously by the same authors (Mitra et al. PNAS 2018) to show that Ncd, a non processive minus end directed motor protein that has been shown to slide microtubules (Fink et al. Nat Cel Biol, 2009), in a three- dimensional setting can carry a short stabilised antiparallel microtubule around a longer one following a helical, right-handed trajectory. Even though this is the first time such a specific setting has been shown, the amount of previously published data that already shows helical motion of different kinesin motors (as cited by authors themselves) and observations that spindle is a helical structure in both human and yeast cells (Novak et al. Nat. Comm 2018; Winey et al. J. Cell Biol 1995; Ding et al. J Cell Biol. 1993) in my opinion do not make this paper in its current state acceptable for publication in Nature Communication. Even though there is a difference between a motor protein carrying a quantum dot and several motor proteins carrying a short microtubule, I was not convinced by the authors’ arguments to see how this work represents a significant jump in our current understanding of microtubule sliding, especially in the case of the mammalian mitotic spindle where microtubules of similar lengths seem to coil around each other in order to produce torque of microtubule bundles. Rotation of a short microtubule around bigger one is not the same as twisting of microtubules around each other. This being said I believe the work can be accepted for publication with some additional experimental data. As I mentioned the biological motivation of the authors seems to be the mitotic spindle. In that case, as proposed in the cartoon of the Figure 5F, the authors should do an experiment involving two long and fixed microtubules to see whether the motor proteins can exert enough force to cause the coiling of two microtubules. This data would in my opinion be much more informative for the mitotic spindle context, and would actually mirror what is believed to be happening in vivo, where microtubules are fixed at the spindle poles. Further comments: The authors on more than one occasion misquote the Novak et al. article as saying the spindle has right-handed chirality. This is not true, as the article clearly states that the spindle has left-handed chirality. in the last section of Discussion, the authors discuss how the interplay of Eg5 and Ncd might produce such chirality of the spindle. Taking into account that the spindle is left-handed and not right-handed, the authors should adjust their arguments and rewrite the discussion accordingly. Statement that Eg5 and Ncd with their movements of opposite handiness and direction result in torques that add up would benefit from a graphical depiction to make it more understandable. This statement in my opinion is rather important even though it is part of discussion, since it tries to explain the possible interplay of motors with different directionalities and torques present in vivo. Probably my biggest concern with this work is that In Figure 2A the pitches vary considerably with the same concentration at 4nM Ncd. Is seems to me they vary much more within this one concentration, than in the Figure 4D where the concentration of the Ncd was changed? Also in Figure 4D there seems to be no difference between 2.4 nM and 6 nM Ncd and no pitch is close to 3 um, observable at 4nM in Figure 2A. How do authors explain these inconsistencies apart from the fact that in the latter experiment they are using a 2D motility assay and not the suspended microtubule assay? In a situation with parallel transport microtubules related to Figure 2 and Supplementary Figure 2, the comparison between suspended template (Figure 2C) and surface-immobilized template (Supplementary Figure 2C) is confusing. The distribution on the surface-immobilized template in my opinion does not show a” minor bias to the right” as stated by authors but shows a quite big bias to the right. In any case, for me this difference to the bimodal distribution seen in suspended template seems unexpected, why the authors think this is happening? Also, a clearer way of quantifying this bias would help to avoid imprecise wording. The authors attribute to autofluorescence the inability to measure the concentration of Ncd-GFP. Would it be possible to measure it with another fluorescent tag? Would a time-gated imaging approach be helpful to eliminate autofluorescence in this setup? In supplementary figure 4C caption text, I am unconvinced by the hard claim that “At 0.6nM Ncd the microtubule did not show any clear rotational movement”. Looking at the example trajectory shown below the kymographs it is clearly visible that at 0.6 nM there is rotational movement. To me it looks quite similar to what is observed after the switch of concentration considering its much shorter duration. Morover, the authors quantify this rotational movement at 0.6 nM Ncd as having the rotational frequency similar to 6 nM Ncd (supplementary figure 4B right panel). Also, in Figure 4 at 0.6 nM Ncd, the rotational pitch is close to 2 whereas at 6 nM Ncd it is close to 0.8. In the main text however the authors are more prudent with their wording and talk about “few transport microtubules” that rotate at 0.6 nM. I noticed in these experiments the very low N for 0.6 nM Ncd experiments (N=7) compared to N=29 for 6 nM Ncd. Why is this the case? On the same note both Figure 4 and Supplementary Figure 4 lack p-values. The mention of chirality observed in yeast spindles in my opinion also merits a mention in the discussion, specifically Winey et al. J. Cell Biol 1995; Ding et al. J Cell Biol. 1993. Supplementary Movie 3 is not referenced in the main text. Supplementary Figure 4E is referenced in the text, yet there is no Supplementary Figure 4E.	TPR	NC_BS_BIOPH_011_RR1.txt	{ "criticism": 14, "example": 10, "importance_and_relevance": 4, "materials_and_methods": 22, "praise": 4, "presentation_and_reporting": 12, "results_and_discussion": 7, "suggestion_and_solution": 13, "total": 66 }	{ "criticism": 0.21212121212121213, "example": 0.15151515151515152, "importance_and_relevance": 0.06060606060606061, "materials_and_methods": 0.3333333333333333, "praise": 0.06060606060606061, "presentation_and_reporting": 0.18181818181818182, "results_and_discussion": 0.10606060606060606, "suggestion_and_solution": 0.19696969696969696 }	1.30303	[ { "content": "Please review this paper.", "role": "user" }, { "content": "Reviewers' comments:\nReviewer #1 (Remarks to the Author):\nThe authors systematically compare the rotational motion of microtubules that Ncd motors\ncrosslink in vitro. It is well known that several mitotic kinesins such as kinesin-5 and kinesin-14\ncrosslink two microtubules. An emerging view is that they exhibit off-axis components and\ngenerate the rotation of a microtubule around the long axis of the microtubule as they move on\nthe other microtubule, suggesting an important role in generating twisting of the mitotic spindle.\nDespite its importance of Ncd in the spindle organization, the rotational motion of microtubules\ndriven by Ncd has not been sufficiently addressed so far. The authors presented a discovery that\nNcd can generate the coiling of microtubule bundles, leading to the generation of chirality in the\nmitotic spindle possibly in collaboration with kinesin-5. A weak point of this paper is that the\nauthors were not able to assess the dependence of the motor density on the helical pitches in 3D\nsliding motility assays. However, this point was carefully addressed by 2D microtubule motility\nassays performed on a planar glass surface using a FLIC-based technique. This paper will be an\nimportant contribution to the biophysics and cell biology fields, and can also be a useful guide for\nthose who are interested in applying the 3D measurement of nanometer-scale objects to their own\nresearch fields. The manuscript is well organized and described. I recommend that it be accepted\nfor publication in Nature Communications.\nAdditional comments:\n- The difference between the current study and the previous work (Nitzsche et al., PNAS 2016)\nseems a bit unclear since both papers describe microtubule gliding on Ncd-coated surfaces that\nrotated around their longitudinal axes. What is novel in the current paper is the possibility that Ncd\ncan generate the coiling of microtubule bundles, which is relevant to the recent finding about the\nchirality of the mitotic spindle. Although, in principle, this result could be anticipated from the\nprevious results that Ncd exhibits off-axis components in the microtubule gliding assays, the actual\nexperimental data about crosslinked microtubule bundles are of critical importance when building a\nmolecular model of the mitotic spindle assembly.\n - If possible, it might be better to mention briefly the quantitative relationship between the\n size/direction of the powerstroke of Ncd that descrived in the previous study, and the resultant\n rotational motion of microtubules. Although this point might be discussed in the previous study\n (Nitzsche et al., PNAS 2016), the discussion must be too complicated for readers to follow.\nReviewer #2 (Remarks to the Author):\nIn this paper Mitra et al. have upgraded an in vitro system developed previously by the same\nauthors (Mitra et al. PNAS 2018) to show that Ncd, a non processive minus end directed motor\nprotein that has been shown to slide microtubules (Fink et al. Nat Cel Biol, 2009), in a three-\ndimensional setting can carry a short stabilised antiparallel microtubule around a longer one\nfollowing a helical, right-handed trajectory. Even though this is the first time such a specific setting\nhas been shown, the amount of previously published data that already shows helical motion of\ndifferent kinesin motors (as cited by authors themselves) and observations that spindle is a helical\nstructure in both human and yeast cells (Novak et al. Nat. Comm 2018; Winey et al. J. Cell Biol\n1995; Ding et al. J Cell Biol. 1993) in my opinion do not make this paper in its current state\nacceptable for publication in Nature Communication. Even though there is a difference between a\nmotor protein carrying a quantum dot and several motor proteins carrying a short microtubule, I\nwas not convinced by the authors’ arguments to see how this work represents a significant jump in\nour current understanding of microtubule sliding, especially in the case of the mammalian mitotic\nspindle where microtubules of similar lengths seem to coil around each other in order to produce\ntorque of microtubule bundles. Rotation of a short microtubule around bigger one is not the same\n as twisting of microtubules around each other.\n This being said I believe the work can be accepted for publication with some additional\n experimental data. As I mentioned the biological motivation of the authors seems to be the mitotic\n spindle. In that case, as proposed in the cartoon of the Figure 5F, the authors should do an\n experiment involving two long and fixed microtubules to see whether the motor proteins can exert\n enough force to cause the coiling of two microtubules. This data would in my opinion be much\n more informative for the mitotic spindle context, and would actually mirror what is believed to be\n happening in vivo, where microtubules are fixed at the spindle poles.\n Further comments:\n The authors on more than one occasion misquote the Novak et al. article as saying the spindle has\n right-handed chirality. This is not true, as the article clearly states that the spindle has left-handed\n chirality. in the last section of Discussion, the authors discuss how the interplay of Eg5 and Ncd\n might produce such chirality of the spindle. Taking into account that the spindle is left-handed and\n not right-handed, the authors should adjust their arguments and rewrite the discussion\n accordingly. Statement that Eg5 and Ncd with their movements of opposite handiness and\n direction result in torques that add up would benefit from a graphical depiction to make it more\n understandable. This statement in my opinion is rather important even though it is part of\n discussion, since it tries to explain the possible interplay of motors with different directionalities\n and torques present in vivo.\n Probably my biggest concern with this work is that In Figure 2A the pitches vary considerably with\n the same concentration at 4nM Ncd. Is seems to me they vary much more within this one\n concentration, than in the Figure 4D where the concentration of the Ncd was changed? Also in\n Figure 4D there seems to be no difference between 2.4 nM and 6 nM Ncd and no pitch is close to 3\n um, observable at 4nM in Figure 2A. How do authors explain these inconsistencies apart from the\n fact that in the latter experiment they are using a 2D motility assay and not the suspended\n microtubule assay?\n In a situation with parallel transport microtubules related to Figure 2 and Supplementary Figure 2,\n the comparison between suspended template (Figure 2C) and surface-immobilized template\n (Supplementary Figure 2C) is confusing. The distribution on the surface-immobilized template in\n my opinion does not show a” minor bias to the right” as stated by authors but shows a quite big\n bias to the right. In any case, for me this difference to the bimodal distribution seen in suspended\n template seems unexpected, why the authors think this is happening?\n Also, a clearer way of quantifying this bias would help to avoid imprecise wording.\n The authors attribute to autofluorescence the inability to measure the concentration of Ncd-GFP.\n Would it be possible to measure it with another fluorescent tag? Would a time-gated imaging\n approach be helpful to eliminate autofluorescence in this setup?\n In supplementary figure 4C caption text, I am unconvinced by the hard claim that “At 0.6nM Ncd\n the microtubule did not show any clear rotational movement”. Looking at the example trajectory\n shown below the kymographs it is clearly visible that at 0.6 nM there is rotational movement. To\n me it looks quite similar to what is observed after the switch of concentration considering its much\n shorter duration. Morover, the authors quantify this rotational movement at 0.6 nM Ncd as having\n the rotational frequency similar to 6 nM Ncd (supplementary figure 4B right panel). Also, in Figure\n 4 at 0.6 nM Ncd, the rotational pitch is close to 2 whereas at 6 nM Ncd it is close to 0.8. In the\n main text however the authors are more prudent with their wording and talk about “few transport\n microtubules” that rotate at 0.6 nM.\n I noticed in these experiments the very low N for 0.6 nM Ncd experiments (N=7) compared to\n N=29 for 6 nM Ncd. Why is this the case? On the same note both Figure 4 and Supplementary\n Figure 4 lack p-values.\n The mention of chirality observed in yeast spindles in my opinion also merits a mention in the\n discussion, specifically Winey et al. J. Cell Biol 1995; Ding et al. J Cell Biol. 1993.\nSupplementary Movie 3 is not referenced in the main text.\nSupplementary Figure 4E is referenced in the text, yet there is no Supplementary Figure 4E.", "role": "assistant" } ]	2,020
Cooperative transport mechanism of human monocarboxylate transporter 2	REVIEWERS' COMMENTS: Reviewer #1 (Remarks to the Author): I think authors have addressed the issues regarding functional characterization of the transporters to a reasonable extent, including new experiements. A pity that pCMBS assays were not possible. Structural aspects are beyond my expertise. Reviewer #3 (Remarks to the Author): The authors have adequately addressed most of the comments in particular those concerning protein expression levels and localization. At the time, the authors have explained in their rebuttal letter that the HEK293 expression system is not fully suitable for all conclusions drawn. In the light of this, my recommendation is to appropriately use "might"/"may"/"could" in the respective parts of the text to properly indicate uncertainty. Otherwise, the reviewer is satisfied with the revision.	REVIEWERS' COMMENTS: Reviewer #1 (Remarks to the Author): I think authors have addressed the issues regarding functional characterization of the transporters to a reasonable extent, including new experiements. A pity that pCMBS assays were not possible. Structural aspects are beyond my expertise. Reviewer #3 (Remarks to the Author): The authors have adequately addressed most of the comments in particular those concerning protein expression levels and localization. At the time, the authors have explained in their rebuttal letter that the HEK293 expression system is not fully suitable for all conclusions drawn. In the light of this, my recommendation is to appropriately use "might"/"may"/"could" in the respective parts of the text to properly indicate uncertainty. Otherwise, the reviewer is satisfied with the revision.	TPR	NC_BS_BIOPH_012_RR2.txt	{ "criticism": 3, "example": 1, "importance_and_relevance": 0, "materials_and_methods": 1, "praise": 3, "presentation_and_reporting": 1, "results_and_discussion": 2, "suggestion_and_solution": 1, "total": 7 }	{ "criticism": 0.42857142857142855, "example": 0.14285714285714285, "importance_and_relevance": 0, "materials_and_methods": 0.14285714285714285, "praise": 0.42857142857142855, "presentation_and_reporting": 0.14285714285714285, "results_and_discussion": 0.2857142857142857, "suggestion_and_solution": 0.14285714285714285 }	1.714286	[ { "content": "Please review this paper.", "role": "user" }, { "content": "\n \n \n \n\n\fREVIEWERS' COMMENTS: \n\n \n\nReviewer #1 (Remarks to the Author): \n\n \n\nI think authors have addressed the issues regarding functional characterization of the transporters to \na reasonable extent, including new experiements. A pity that pCMBS assays were not possible. \nStructural aspects are beyond my expertise. \n\n \n\n \n\nReviewer #3 (Remarks to the Author): \n\n \n\nThe authors have adequately addressed most of the comments in particular those concerning \nprotein expression levels and localization. At the time, the authors have explained in their rebuttal \nletter that the HEK293 expression system is not fully suitable for all conclusions drawn. In the light of \nthis, my recommendation is to appropriately use \"might\"/\"may\"/\"could\" in the respective parts of \nthe text to properly indicate uncertainty. Otherwise, the reviewer is satisfied with the revision. \n\n\f\n", "role": "assistant" } ]	2,020
Backmapping triangulated surfaces to coarse-grained membrane models	REVIEWERS' COMMENTS: Reviewer #1 (Remarks to the Author): All of my points have been more than adequately addressed. Reviewer #2 (Remarks to the Author): The authors have done a great job replying to reviews and the paper is suitable for publication.	REVIEWERS' COMMENTS: Reviewer #1 (Remarks to the Author): All of my points have been more than adequately addressed. Reviewer #2 (Remarks to the Author): The authors have done a great job replying to reviews and the paper is suitable for publication.	TPR	NC_BS_BIOPH_013_RR2.txt	{ "criticism": 0, "example": 0, "importance_and_relevance": 0, "materials_and_methods": 0, "praise": 1, "presentation_and_reporting": 0, "results_and_discussion": 0, "suggestion_and_solution": 0, "total": 2 }	{ "criticism": 0, "example": 0, "importance_and_relevance": 0, "materials_and_methods": 0, "praise": 0.5, "presentation_and_reporting": 0, "results_and_discussion": 0, "suggestion_and_solution": 0 }	0.5	[ { "content": "Please review this paper.", "role": "user" }, { "content": "\n \n\n\fREVIEWERS' COMMENTS: \n\nReviewer #1 (Remarks to the Author): \n\nAll of my points have been more than adequately addressed. \n\nReviewer #2 (Remarks to the Author): \n\nThe authors have done a great job replying to reviews and the paper is suitable for publication.\n\n\f\n", "role": "assistant" } ]	2,020
Backmapping triangulated surfaces to coarse-grained membrane models	Reviewers' comments: Reviewer #1 (Remarks to the Author): Overall this is an excellent manuscript on what I think is a game-changing tool; it will be highly useful in taking molecular level simulations to the next level in terms of system size. I have no problem with the methods - they are demonstrated in the manuscript to work. My comments are minor, but ones I feel. should be addressed. They concern the context of the present work. The authors provide comparisons to other approaches and indicate how theirs is better, but I feel the references to 50 and 51 are confusing- are you saying that the CG/DTS implementation is better than the methods employed in these papers? If so please explain how. Secondly it would be good to give examples of how current CG studies in the literature would benefit from using this approach, for example Jefferies and Khalid J Mol Biol 2020 already showed membrane undulations around vesicles, without CG-DTS, what would DTS add here? This is not a criticism of the present paper but an opportunity to show what extra the method could do. Also could the method be applied to the Fluctuating Finite Element approach of S.A. Harris for example see PloS Comp Biol, 2018. Finally, it would be good to have a comment on how portions of the CG system could then be converted to all-atom detail for fine-grained studies of small patches. Overall, I think this is an excellent & timely manuscript that will be undoubtedly be very useful and highly cited. Reviewer #2 (Remarks to the Author): Overall, this manuscript describes a timely and necessary tool to build CG models from triangulated meshes. The developers have created an algorithm that improves upon existing lipid packing software such as LipidWrapper. By using a pointillism approach and constructing the membrane with monolayers, rather than bilayers, the algorithm accounts for the increased number of lipids in the outer leaflet compared to the inner leaflet, and thus reduces the amount of time it would otherwise take to equilibrate a curved structure. The authors show (in at least one case) that their method is able to produce stable conformations ready for production runs. The work provides a much needed tool for computational biologists / biophysicists / chemists. The figures are clear, the data support the conclusions. The software was provided for review during the review process. Considerations for additional elaboration in the manuscript: 1) Do the authors plan to maintain and continue to develop the code? 2) Could the authors please elaborate on the file types output by DTS and thus required as inputs for their algorithm? 3) Could the authors elaborate on ease of user application, whether the user must modify the scripts with the appropriate parameters (such as APL) or whether the parameters are executed as arguments in the command line? 4) The authors describe using an EM tomography map to create their triangulated mesh. Could the authors elaborate on the method of triangulation they used to generate their mesh from the tomography, and how easily the program could be modified to accept other triangulated mesh files (such as: dae, stl, or obj)? 5) Could the algorithm be easily modified to generate monolayer (rather than bilayer) structures? 6) The authors claim that they were able to achieve “stable” structures ready for simulation with the MARTINI model. For the mitochondrial membrane model, the authors do not comment on whether this resulting configuration was stable. Could the authors provide evidence that the structure was able to achieve an energy-minimized conformation? Additionally, is the system also stable if the headgroup restraints are removed? It seems like a model this complex with such high variation in curvature would not be stable with short equilibrations and a very short production (2ns). If the system does not achieve a stable conformation, could the authors explain how the system evolves and any additional precautions users should take when working with systems like this? Could they also provide a general discussion of the challenges and additional considerations of working with bilayer systems of “unprecedented” size and shape? 7) There is a brief discussion on membrane heterogeneity and curvature. Did the authors then use an average APL to build the complex systems with heterogeneous membrane compositions? Did these membranes take longer to converge? Was there any evidence that the variation in APL caused local deformations from the starting mesh or the formation of pores? Figure 1 does not have any lettering (A-E) as suggested by the text. In the SI-F: first paragraph, final sentence: “backmapped” is misspelled as “backmaped” This review was provided by Rommie E. Amaro and Abigail Dommer. Reviewer #3 (Remarks to the Author): In their manuscript "Backmapping triangulated surfaces to coarse-grained membrane models" Pezeshkian et al. describe, with three examples, a computational tool they call TS2CG. The tool takes as its input a triangulated surface in 3D and outputs a particle-based description of a biomembrane with a corresponding 3D geometry. The latter can then be used for molecular dynamics simulations of the biomembrane. I strongly believe that software developers should be given the credit they deserve. However, I also think that the TS2CG tool does not represent an advance in scientific understanding that is likely to influence thinking in the field. To thus end, I do not find the manuscript suitable to be published in Nature Communications. More suitable venues would be, e.g., the journals SoftwareX or Molecular Simulation. With regards to the TS2CG tool itself, I would like to draw the authors' attention to the following two points: 1) Currently the tool matches the areas of the two leaflets. This, however, does not guarantee that the preferred local curvature of the CG model agrees with that of the triangulated surface, because the area difference of the leaflets does not in fact predict the preferred curvature of the membrane, see Biophys. J. 115 1638 (2018), Nano Lett. 19 5011 (2019), and Biophys. J. 118 624 (2020). 2) Instead of a one-way script to map a triangulated surface to CG, it would be greatly more impactful to create a two-way mapping of the two Hamiltonians, such that the CG simulation would have a way of influencing the triangulated surface simulation and vice versa. Connecting the two simulations in such a rigorous way into a true multiscale simulation would be highly desirable.	Reviewers' comments: Reviewer #1 (Remarks to the Author): Overall this is an excellent manuscript on what I think is a game-changing tool; it will be highly useful in taking molecular level simulations to the next level in terms of system size. I have no problem with the methods - they are demonstrated in the manuscript to work. My comments are minor, but ones I feel. should be addressed. They concern the context of the present work. The authors provide comparisons to other approaches and indicate how theirs is better, but I feel the references to 50 and 51 are confusing- are you saying that the CG/DTS implementation is better than the methods employed in these papers? If so please explain how. Secondly it would be good to give examples of how current CG studies in the literature would benefit from using this approach, for example Jefferies and Khalid J Mol Biol 2020 already showed membrane undulations around vesicles, without CG-DTS, what would DTS add here? This is not a criticism of the present paper but an opportunity to show what extra the method could do. Also could the method be applied to the Fluctuating Finite Element approach of S.A. Harris for example see PloS Comp Biol, 2018. Finally, it would be good to have a comment on how portions of the CG system could then be converted to all-atom detail for fine-grained studies of small patches. Overall, I think this is an excellent & timely manuscript that will be undoubtedly be very useful and highly cited. Reviewer #2 (Remarks to the Author): Overall, this manuscript describes a timely and necessary tool to build CG models from triangulated meshes. The developers have created an algorithm that improves upon existing lipid packing software such as LipidWrapper. By using a pointillism approach and constructing the membrane with monolayers, rather than bilayers, the algorithm accounts for the increased number of lipids in the outer leaflet compared to the inner leaflet, and thus reduces the amount of time it would otherwise take to equilibrate a curved structure. The authors show (in at least one case) that their method is able to produce stable conformations ready for production runs. The work provides a much needed tool for computational biologists / biophysicists / chemists. The figures are clear, the data support the conclusions. The software was provided for review during the review process. Considerations for additional elaboration in the manuscript: 1) Do the authors plan to maintain and continue to develop the code? 2) Could the authors please elaborate on the file types output by DTS and thus required as inputs for their algorithm? 3) Could the authors elaborate on ease of user application, whether the user must modify the scripts with the appropriate parameters (such as APL) or whether the parameters are executed as arguments in the command line? 4) The authors describe using an EM tomography map to create their triangulated mesh. Could the authors elaborate on the method of triangulation they used to generate their mesh from the tomography, and how easily the program could be modified to accept other triangulated mesh files (such as: dae, stl, or obj)? 5) Could the algorithm be easily modified to generate monolayer (rather than bilayer) structures? 6) The authors claim that they were able to achieve “stable” structures ready for simulation with the MARTINI model. For the mitochondrial membrane model, the authors do not comment on whether this resulting configuration was stable. Could the authors provide evidence that the structure was able to achieve an energy-minimized conformation? Additionally, is the system also stable if the headgroup restraints are removed? It seems like a model this complex with such high variation in curvature would not be stable with short equilibrations and a very short production (2ns). If the system does not achieve a stable conformation, could the authors explain how the system evolves and any additional precautions users should take when working with systems like this? Could they also provide a general discussion of the challenges and additional considerations of working with bilayer systems of “unprecedented” size and shape? 7) There is a brief discussion on membrane heterogeneity and curvature. Did the authors then use an average APL to build the complex systems with heterogeneous membrane compositions? Did these membranes take longer to converge? Was there any evidence that the variation in APL caused local deformations from the starting mesh or the formation of pores? Figure 1 does not have any lettering (A-E) as suggested by the text. In the SI-F: first paragraph, final sentence: “backmapped” is misspelled as “backmaped” This review was provided by Rommie E. Amaro and Abigail Dommer. Reviewer #3 (Remarks to the Author): In their manuscript "Backmapping triangulated surfaces to coarse-grained membrane models" Pezeshkian et al. describe, with three examples, a computational tool they call TS2CG. The tool takes as its input a triangulated surface in 3D and outputs a particle-based description of a biomembrane with a corresponding 3D geometry. The latter can then be used for molecular dynamics simulations of the biomembrane. I strongly believe that software developers should be given the credit they deserve. However, I also think that the TS2CG tool does not represent an advance in scientific understanding that is likely to influence thinking in the field. To thus end, I do not find the manuscript suitable to be published in Nature Communications. More suitable venues would be, e.g., the journals SoftwareX or Molecular Simulation. With regards to the TS2CG tool itself, I would like to draw the authors' attention to the following two points: 1) Currently the tool matches the areas of the two leaflets. This, however, does not guarantee that the preferred local curvature of the CG model agrees with that of the triangulated surface, because the area difference of the leaflets does not in fact predict the preferred curvature of the membrane, see Biophys. J. 115 1638 (2018), Nano Lett. 19 5011 (2019), and Biophys. J. 118 624 (2020). 2) Instead of a one-way script to map a triangulated surface to CG, it would be greatly more impactful to create a two-way mapping of the two Hamiltonians, such that the CG simulation would have a way of influencing the triangulated surface simulation and vice versa. Connecting the two simulations in such a rigorous way into a true multiscale simulation would be highly desirable.	TPR	NC_BS_BIOPH_013_RR1.txt	{ "criticism": 9, "example": 2, "importance_and_relevance": 6, "materials_and_methods": 38, "praise": 5, "presentation_and_reporting": 6, "results_and_discussion": 4, "suggestion_and_solution": 13, "total": 55 }	{ "criticism": 0.16363636363636364, "example": 0.03636363636363636, "importance_and_relevance": 0.10909090909090909, "materials_and_methods": 0.6909090909090909, "praise": 0.09090909090909091, "presentation_and_reporting": 0.10909090909090909, "results_and_discussion": 0.07272727272727272, "suggestion_and_solution": 0.23636363636363636 }	1.509091	[ { "content": "Please review this paper.", "role": "user" }, { "content": "Reviewers' comments: \n\nReviewer #1 (Remarks to the Author): \n\nOverall this is an excellent manuscript on what I think is a game-changing tool; it will be highly \nuseful in taking molecular level simulations to the next level in terms of system size. I have no \nproblem with the methods - they are demonstrated in the manuscript to work. \n\nMy comments are minor, but ones I feel. should be addressed. They concern the context of the \npresent work. The authors provide comparisons to other approaches and indicate how theirs is \nbetter, but I feel the references to 50 and 51 are confusing- are you saying that the CG/DTS \nimplementation is better than the methods employed in these papers? If so please explain how. \nSecondly it would be good to give examples of how current CG studies in the literature would \nbenefit from using this approach, for example Jefferies and Khalid J Mol Biol 2020 already showed \nmembrane undulations around vesicles, without CG-DTS, what would DTS add here? This is not a \ncriticism of the present paper but an opportunity to show what extra the method could do. Also \ncould the method be applied to the Fluctuating Finite Element approach of S.A. Harris for example \nsee PloS Comp Biol, 2018. \n\nFinally, it would be good to have a comment on how portions of the CG system could then be \nconverted to all-atom detail for fine-grained studies of small patches. \n\nOverall, I think this is an excellent & timely manuscript that will be undoubtedly be very useful and \nhighly cited. \n\nReviewer #2 (Remarks to the Author): \n\nOverall, this manuscript describes a timely and necessary tool to build CG models from \ntriangulated meshes. The developers have created an algorithm that improves upon existing lipid \npacking software such as LipidWrapper. By using a pointillism approach and constructing the \nmembrane with monolayers, rather than bilayers, the algorithm accounts for the increased number \nof lipids in the outer leaflet compared to the inner leaflet, and thus reduces the amount of time it \nwould otherwise take to equilibrate a curved structure. The authors show (in at least one case) \nthat their method is able to produce stable conformations ready for production runs. The work \nprovides a much needed tool for computational biologists / biophysicists / chemists. The figures \nare clear, the data support the conclusions. The software was provided for review during the \nreview process. \n\nConsiderations for additional elaboration in the manuscript: \n\n1) Do the authors plan to maintain and continue to develop the code? \n2) Could the authors please elaborate on the file types output by DTS and thus required as inputs \nfor their algorithm? \n3) Could the authors elaborate on ease of user application, whether the user must modify the \nscripts with the appropriate parameters (such as APL) or whether the parameters are executed as \narguments in the command line? \n4) The authors describe using an EM tomography map to create their triangulated mesh. Could the \nauthors elaborate on the method of triangulation they used to generate their mesh from the \ntomography, and how easily the program could be modified to accept other triangulated mesh files \n(such as: dae, stl, or obj)? \n5) Could the algorithm be easily modified to generate monolayer (rather than bilayer) structures? \n6) The authors claim that they were able to achieve “stable” structures ready for simulation with \nthe MARTINI model. For the mitochondrial membrane model, the authors do not comment on \nwhether this resulting configuration was stable. Could the authors provide evidence that the \n\n\fstructure was able to achieve an energy-minimized conformation? Additionally, is the system also \nstable if the headgroup restraints are removed? It seems like a model this complex with such high \nvariation in curvature would not be stable with short equilibrations and a very short production \n(2ns). If the system does not achieve a stable conformation, could the authors explain how the \nsystem evolves and any additional precautions users should take when working with systems like \nthis? Could they also provide a general discussion of the challenges and additional considerations \nof working with bilayer systems of “unprecedented” size and shape? \n7) There is a brief discussion on membrane heterogeneity and curvature. Did the authors then use \nan average APL to build the complex systems with heterogeneous membrane compositions? Did \nthese membranes take longer to converge? Was there any evidence that the variation in APL \ncaused local deformations from the starting mesh or the formation of pores? \n\nFigure 1 does not have any lettering (A-E) as suggested by the text. \nIn the SI-F: first paragraph, final sentence: “backmapped” is misspelled as “backmaped” \n\nThis review was provided by Rommie E. Amaro and Abigail Dommer. \n\nReviewer #3 (Remarks to the Author): \n\nIn their manuscript \"Backmapping triangulated surfaces to coarse-grained membrane models\" \nPezeshkian et al. describe, with three examples, a computational tool they call TS2CG. The tool \ntakes as its input a triangulated surface in 3D and outputs a particle-based description of a \nbiomembrane with a corresponding 3D geometry. The latter can then be used for molecular \ndynamics simulations of the biomembrane. \n\nI strongly believe that software developers should be given the credit they deserve. However, I \nalso think that the TS2CG tool does not represent an advance in scientific understanding that is \nlikely to influence thinking in the field. To thus end, I do not find the manuscript suitable to be \npublished in Nature Communications. More suitable venues would be, e.g., the journals SoftwareX \nor Molecular Simulation. \n\nWith regards to the TS2CG tool itself, I would like to draw the authors' attention to the following \ntwo points: \n\n1) Currently the tool matches the areas of the two leaflets. This, however, does not guarantee that \nthe preferred local curvature of the CG model agrees with that of the triangulated surface, because \nthe area difference of the leaflets does not in fact predict the preferred curvature of the \nmembrane, see Biophys. J. 115 1638 (2018), Nano Lett. 19 5011 (2019), and Biophys. J. 118 624 \n(2020). \n\n2) Instead of a one-way script to map a triangulated surface to CG, it would be greatly more \nimpactful to create a two-way mapping of the two Hamiltonians, such that the CG simulation \nwould have a way of influencing the triangulated surface simulation and vice versa. Connecting the \ntwo simulations in such a rigorous way into a true multiscale simulation would be highly desirable.\n\n\f", "role": "assistant" } ]	2,020
ucleoplasmic signals promote directed transmembrane protein import simultaneously via	Reviewers' comments: Reviewer #1 (Remarks to the Author): Mudumbi et al combine a toolbox of advanced microscopy methods aiming to resolve the long-lasting debate about the nature of different INM protein groups in the peripheral channels in the NPC and their integration within the nuclear pore membrane. Overall, I find the manuscript exciting, well written and presented (professionally presented Figures), technical sound, and important to be published. I therefore recommend to accept the manuscript for publication in Nature Communications after addressing the following minor comments. (1) Could the authors please discuss the impact of the mechanical rigidity of INM proteins in the revised version of the manuscript? I appreciate the significance of their length/height. However, the reaction diffusion dynamics and their hydrodynamic radius will also depend on the mechanical status of the molecules. (2) I think that the manuscript would be strengthened if the authors could show dynamic data, which allows the readership to judge the quality of the measurements. The post-processed data and presentation is very well executed but the raw data acquisition is difficult to judge. (3) Could the authors please provide statistical significance of the different mobilities as measure by smFRAP in Table 1? Table 1 provides standard deviations but no p-values. (4) I believe that Nat Com does not allow abbreviations in the abstract. Reviewer #2 (Remarks to the Author): (5) Could the authors please rephrase the sentence starting with "Here" (line 87)? The reader could be confused that the summary paragraph of the end of the introduction starts at this location, which only follows line 130. In this paper Mudumbi et al, show exciting data providing structural evidence how transmembrane proteins are imported via the Nuclear Pore Complex to the inner nuclear membrane. With their advanced and unique single-molecule techniques they make an important contribution to the understanding of the import pathway of several membrane proteins. The import route that membrane proteins take has been long debated. While the relevance of sorting signals, RanGTP, and NTRS have all been a subject of debate, the fiercest debates concerned the transport route through the NPC. First, multiple labs provided support for the idea that most yeast and human INM proteins use the peripheral channels observed in several EM studies. However, this transport route through the lateral channels was never directly shown. In addition to this canonical route, two yeast INM proteins were proposed to take a different route: they use an active import mechanism which involves the passage of their NTR-bound end of the molecule through the central channel, while their transmembrane domains and the C-terminus travel the peripheral channels. The proposed transport route implied that, at least transiently, openings must exist between the space immediately aligning the pore membrane and the central channel. Only indirect evidence was available so far to support this transport route. According to many in the field, this proposed transport route was incompatible with the current structural data, and hence regarded highly unlikely. Another concern was that it could be an oddity of these two proteins, and/or an oddity of yeast. In the current manuscript this longstanding debate about where these proteins translocate is addressed using suitable and sophisticated single molecule methods developed by Prof Yang and coworkers and using a range of native proteins. Five human inner nuclear membrane proteins are studied (NET51, NET59, NET99, LBR, Lap2beta). The existence of the peripheral channel as well as passage of Net51/59/99 via the peripheral channels is confirmed. The overall dimensions of the NPC scaffold and the peripheral channels are is good agreement with EM studies, which provides confidence in the accuracy of the SPEED method (which I cannot evaluate in detail). Most exciting is the data on LBR and Lap2beta which show that their NLS and disordered regions pass the central channel, at the same place as soluble proteins with transportins, while the transmembrane and membrane proximal regions pass through the lateral channels. Most convincing are those measurements where the LBR proteins is simultaneously tagged on both termini, so that the recordings of transport through the lateral channels and the central channel reflects one and the same molecule. These measurements provide the first direct evidence that the NPC can accommodate for the hotly debated transport route through the central channel. In fact, the studies by Mudumbi et al align well with the emerging view that the structure of the NPC is much more dynamic than previously thought. In a set of elegant studies the authors additionally show how perturbations of the NPC channels or the sorting signals can alter the transport path of a membrane proteins. This is a surprising finding that may turn out to have important biological implications. Also, the authors provide kinetic information about both transport routes, which is roughly in line with previous estimates from yeast and human systems. In addition to the experimental data the authors attempt to address the question how prevalent both transport mechanisms are. They estimate that only some 10% have ID linkers and NLSs. This paper is in my opinion an important and novel paper that would be very suitable for publication in nature communications after a few controls and textual changes are considered. Main concerns: 1) the dual-channel co-tracking SPEED microscopy of the double tagged LBR is important since it shows that single molecules of LBR pass the NPC as a full-length protein and membrane embedded. Similar experiments are not included for Lap2β. This is a potential problem in the interpretation of the data for Lap2β, as, obviously, the presence of truncated versions encoding only the nucleoplasmic, soluble domain of LBR could easily explain the signal the central channel. I realize that doing dual channel co-tracking of Lap2β would involve a very significant extra effort and this is not required in my opinion. I would however like to see western blot and microscopy images of the cells expressing the lagged Lap2β, so that readers can be convinced that the data very likely reflects full length proteins and not a degradation product of Lap2β. 2) The estimate that ~9% of the inner nuclear membrane proteins may encode an ID domain and an NLS is really only an estimate as should be referred to as such (also in the abstract). The assumption that all these “extend intrinsically disordered (ID) domains containing nuclear localization signals (NLSs) into the central channel for directed nuclear transport” is also really only an assumption and should be referred to as such. To clarify the analysis done, I have three more suggestions: Please provide the list of 46 INM and 11 ONM proteins that were included in the analysis. Line 190 (methods) Please indicate the rationale for taking SeqNLS scores higher than 0.86 as ‘sufficiently strong’ NLSs. Please elaborate what “made sense in a biological manner for facilitated transport through the center of the NPC “ means. Minor comments 3) I have few suggestions towards placing the work in the context of the literature. a. The two papers that were published in 2015, by the Ellenberg and by the Kutay lab (Boni et al, JCB 2015, Ungricht et al JCB 2015) should be discussed. Both groups used life imaging and computational techniques to show that a diffusion-retention mechanism can explain the import of INM proteins. In both papers Lap2β and LBR are studied, like in this manuscript. The estimates of transport kinetics in Boni et al match nicely the ones reported here (5.6 molecules per minute per NPC versus 4.6 molecules/NPC/min), but while the numbers match, the conclusion of Boni et al is that their data supports a diffusion retention model for INM protein transport. b. In the introduction Line 107-109 is an incorrect statement: In Meinema et al (2011) the endogenous ID region of the yeast proteins was used and in Kralt et al. MolBiol Cell 2015 the full length and endogenously expressed protein (native promotor, native chromosomal location) was used to study the NLS. If the authors meant to say that that thus far there were no papers that studied the relevance of an ID linker in the context of the full length and natively expressed protein, that would be correct. c. In the Discussion: Line 341-343. The experiments do not address this concern directly (different proteins, different cells). Laba et al (Cells, 2015) confirms the transmembrane nature of the Heh2- derived truncated proteins. 3) Line 255: there is no data to support that the R74T mutation indeed results in the disruption of NTR binding. Is there biochemical evidence to support this statement? Or is there data to show mislocalisation of GFP-NLS(R74T) fusions in cells? Ma et al 2007 J Cell Sci can be cited to support that imp beta is bound. 4) Other: Line 220-223: Lap2β and Lapβ are used both throughout the manuscript. Line 261: Lapβ ΔNLS: actually the NLS is not deleted but mutated: R319T and R320T. Please mention. Line 160 (methods) range not rage. Line 181 (methods) In the case of LBR it reads that the no of copies per cell was based on a previous estimate; what was used for the other proteins? Line 218-223 (methods). I do not understand which figures or experiments this related to. Line 265-268: this paragraph is slightly confusing. Maybe add: ...and roughly 80% of the length of the nucleoplasmic region still present, LBR required these ID Linker regions to transit....etc. Smovie 2 does not work Reviewer #3 (Remarks to the Author): This paper has studied how transmembrane proteins in their inner nuclear membrane are transported to the inner nuclear membrane using live cell high-speed super-resolution single-molecule microscopy. Authors found that either just the peripheral channels or both central and peripheral channels together are used for the transport depending on their nucleoplasmic signals. They succeeded in visualizing pathways and single-molecule trajectories going through the central and peripheral channels distinguishing the positions of the central and peripheral channels and pore membrane lumen. However, the paper has substantive issues in description, which makes the paper hard to comprehend. Major: 1. There is a problem with the logical description. In addition, some paragraphs are too long in the Results and Introduction sections. The Discussion section has insufficient contents. 2. Abstract is inappropriate as summary. It does not reflect the entire contents, and combining Abstract and Significance Statement makes a better summary. Furthermore, for example, the value of “~9%” is described only once in the Results section, and Fig. 4E is referred. However, Figure 4E and its legend has no description of the clear basis for calculation, and the accuracy of the numerical value of 9 is not described. 3. The title is not suitable. The word “Nucleoplasmic signals” of the title subject is used only once in the Significance Statement section through the whole manuscript, and no explanation is described even though the word is not in common use. The word “simultaneously” is also used, but it is not clear with which nucleoplasmic signals promote simultaneously. 4. It is written that the route precision is ~2 nm at line 164 in the Result section, and 0.90 ~ 1.42 in Table S2 based on the simulations (Figs. S8-10). However, the validity of using the values is not clear. It is considered appropriate to use the standard deviation of the distribution or that of data from independent measurements (not standard error of the mean SEM; more precisely, Student's t value multiplied by SEM) as the precision and error of the route and diameter. Minor: 1. The sentence “~9% additionally extend ...” in Abstract is unclear, since the subject of the sentence is not clear.	Reviewers' comments: Reviewer #1 (Remarks to the Author): Mudumbi et al combine a toolbox of advanced microscopy methods aiming to resolve the long-lasting debate about the nature of different INM protein groups in the peripheral channels in the NPC and their integration within the nuclear pore membrane. Overall, I find the manuscript exciting, well written and presented (professionally presented Figures), technical sound, and important to be published. I therefore recommend to accept the manuscript for publication in Nature Communications after addressing the following minor comments. (1) Could the authors please discuss the impact of the mechanical rigidity of INM proteins in the revised version of the manuscript? I appreciate the significance of their length/height. However, the reaction diffusion dynamics and their hydrodynamic radius will also depend on the mechanical status of the molecules. (2) I think that the manuscript would be strengthened if the authors could show dynamic data, which allows the readership to judge the quality of the measurements. The post-processed data and presentation is very well executed but the raw data acquisition is difficult to judge. (3) Could the authors please provide statistical significance of the different mobilities as measure by smFRAP in Table 1? Table 1 provides standard deviations but no p-values. (4) I believe that Nat Com does not allow abbreviations in the abstract. Reviewer #2 (Remarks to the Author): (5) Could the authors please rephrase the sentence starting with "Here" (line 87)? The reader could be confused that the summary paragraph of the end of the introduction starts at this location, which only follows line 130. In this paper Mudumbi et al, show exciting data providing structural evidence how transmembrane proteins are imported via the Nuclear Pore Complex to the inner nuclear membrane. With their advanced and unique single-molecule techniques they make an important contribution to the understanding of the import pathway of several membrane proteins. The import route that membrane proteins take has been long debated. While the relevance of sorting signals, RanGTP, and NTRS have all been a subject of debate, the fiercest debates concerned the transport route through the NPC. First, multiple labs provided support for the idea that most yeast and human INM proteins use the peripheral channels observed in several EM studies. However, this transport route through the lateral channels was never directly shown. In addition to this canonical route, two yeast INM proteins were proposed to take a different route: they use an active import mechanism which involves the passage of their NTR-bound end of the molecule through the central channel, while their transmembrane domains and the C-terminus travel the peripheral channels. The proposed transport route implied that, at least transiently, openings must exist between the space immediately aligning the pore membrane and the central channel. Only indirect evidence was available so far to support this transport route. According to many in the field, this proposed transport route was incompatible with the current structural data, and hence regarded highly unlikely. Another concern was that it could be an oddity of these two proteins, and/or an oddity of yeast. In the current manuscript this longstanding debate about where these proteins translocate is addressed using suitable and sophisticated single molecule methods developed by Prof Yang and coworkers and using a range of native proteins. Five human inner nuclear membrane proteins are studied (NET51, NET59, NET99, LBR, Lap2beta). The existence of the peripheral channel as well as passage of Net51/59/99 via the peripheral channels is confirmed. The overall dimensions of the NPC scaffold and the peripheral channels are is good agreement with EM studies, which provides confidence in the accuracy of the SPEED method (which I cannot evaluate in detail). Most exciting is the data on LBR and Lap2beta which show that their NLS and disordered regions pass the central channel, at the same place as soluble proteins with transportins, while the transmembrane and membrane proximal regions pass through the lateral channels. Most convincing are those measurements where the LBR proteins is simultaneously tagged on both termini, so that the recordings of transport through the lateral channels and the central channel reflects one and the same molecule. These measurements provide the first direct evidence that the NPC can accommodate for the hotly debated transport route through the central channel. In fact, the studies by Mudumbi et al align well with the emerging view that the structure of the NPC is much more dynamic than previously thought. In a set of elegant studies the authors additionally show how perturbations of the NPC channels or the sorting signals can alter the transport path of a membrane proteins. This is a surprising finding that may turn out to have important biological implications. Also, the authors provide kinetic information about both transport routes, which is roughly in line with previous estimates from yeast and human systems. In addition to the experimental data the authors attempt to address the question how prevalent both transport mechanisms are. They estimate that only some 10% have ID linkers and NLSs. This paper is in my opinion an important and novel paper that would be very suitable for publication in nature communications after a few controls and textual changes are considered. Main concerns: 1) the dual-channel co-tracking SPEED microscopy of the double tagged LBR is important since it shows that single molecules of LBR pass the NPC as a full-length protein and membrane embedded. Similar experiments are not included for Lap2β. This is a potential problem in the interpretation of the data for Lap2β, as, obviously, the presence of truncated versions encoding only the nucleoplasmic, soluble domain of LBR could easily explain the signal the central channel. I realize that doing dual channel co-tracking of Lap2β would involve a very significant extra effort and this is not required in my opinion. I would however like to see western blot and microscopy images of the cells expressing the lagged Lap2β, so that readers can be convinced that the data very likely reflects full length proteins and not a degradation product of Lap2β. 2) The estimate that ~9% of the inner nuclear membrane proteins may encode an ID domain and an NLS is really only an estimate as should be referred to as such (also in the abstract). The assumption that all these “extend intrinsically disordered (ID) domains containing nuclear localization signals (NLSs) into the central channel for directed nuclear transport” is also really only an assumption and should be referred to as such. To clarify the analysis done, I have three more suggestions: Please provide the list of 46 INM and 11 ONM proteins that were included in the analysis. Line 190 (methods) Please indicate the rationale for taking SeqNLS scores higher than 0.86 as ‘sufficiently strong’ NLSs. Please elaborate what “made sense in a biological manner for facilitated transport through the center of the NPC “ means. Minor comments 3) I have few suggestions towards placing the work in the context of the literature. a. The two papers that were published in 2015, by the Ellenberg and by the Kutay lab (Boni et al, JCB 2015, Ungricht et al JCB 2015) should be discussed. Both groups used life imaging and computational techniques to show that a diffusion-retention mechanism can explain the import of INM proteins. In both papers Lap2β and LBR are studied, like in this manuscript. The estimates of transport kinetics in Boni et al match nicely the ones reported here (5.6 molecules per minute per NPC versus 4.6 molecules/NPC/min), but while the numbers match, the conclusion of Boni et al is that their data supports a diffusion retention model for INM protein transport. b. In the introduction Line 107-109 is an incorrect statement: In Meinema et al (2011) the endogenous ID region of the yeast proteins was used and in Kralt et al. MolBiol Cell 2015 the full length and endogenously expressed protein (native promotor, native chromosomal location) was used to study the NLS. If the authors meant to say that that thus far there were no papers that studied the relevance of an ID linker in the context of the full length and natively expressed protein, that would be correct. c. In the Discussion: Line 341-343. The experiments do not address this concern directly (different proteins, different cells). Laba et al (Cells, 2015) confirms the transmembrane nature of the Heh2- derived truncated proteins. 3) Line 255: there is no data to support that the R74T mutation indeed results in the disruption of NTR binding. Is there biochemical evidence to support this statement? Or is there data to show mislocalisation of GFP-NLS(R74T) fusions in cells? Ma et al 2007 J Cell Sci can be cited to support that imp beta is bound. 4) Other: Line 220-223: Lap2β and Lapβ are used both throughout the manuscript. Line 261: Lapβ ΔNLS: actually the NLS is not deleted but mutated: R319T and R320T. Please mention. Line 160 (methods) range not rage. Line 181 (methods) In the case of LBR it reads that the no of copies per cell was based on a previous estimate; what was used for the other proteins? Line 218-223 (methods). I do not understand which figures or experiments this related to. Line 265-268: this paragraph is slightly confusing. Maybe add: ...and roughly 80% of the length of the nucleoplasmic region still present, LBR required these ID Linker regions to transit....etc. Smovie 2 does not work Reviewer #3 (Remarks to the Author): This paper has studied how transmembrane proteins in their inner nuclear membrane are transported to the inner nuclear membrane using live cell high-speed super-resolution single-molecule microscopy. Authors found that either just the peripheral channels or both central and peripheral channels together are used for the transport depending on their nucleoplasmic signals. They succeeded in visualizing pathways and single-molecule trajectories going through the central and peripheral channels distinguishing the positions of the central and peripheral channels and pore membrane lumen. However, the paper has substantive issues in description, which makes the paper hard to comprehend. Major: 1. There is a problem with the logical description. In addition, some paragraphs are too long in the Results and Introduction sections. The Discussion section has insufficient contents. 2. Abstract is inappropriate as summary. It does not reflect the entire contents, and combining Abstract and Significance Statement makes a better summary. Furthermore, for example, the value of “~9%” is described only once in the Results section, and Fig. 4E is referred. However, Figure 4E and its legend has no description of the clear basis for calculation, and the accuracy of the numerical value of 9 is not described. 3. The title is not suitable. The word “Nucleoplasmic signals” of the title subject is used only once in the Significance Statement section through the whole manuscript, and no explanation is described even though the word is not in common use. The word “simultaneously” is also used, but it is not clear with which nucleoplasmic signals promote simultaneously. 4. It is written that the route precision is ~2 nm at line 164 in the Result section, and 0.90 ~ 1.42 in Table S2 based on the simulations (Figs. S8-10). However, the validity of using the values is not clear. It is considered appropriate to use the standard deviation of the distribution or that of data from independent measurements (not standard error of the mean SEM; more precisely, Student's t value multiplied by SEM) as the precision and error of the route and diameter. Minor: 1. The sentence “~9% additionally extend ...” in Abstract is unclear, since the subject of the sentence is not clear.	TPR	NC_BS_BIOPH_014_RR1.txt	{ "criticism": 22, "example": 16, "importance_and_relevance": 9, "materials_and_methods": 48, "praise": 8, "presentation_and_reporting": 29, "results_and_discussion": 15, "suggestion_and_solution": 19, "total": 99 }	{ "criticism": 0.2222222222222222, "example": 0.16161616161616163, "importance_and_relevance": 0.09090909090909091, "materials_and_methods": 0.48484848484848486, "praise": 0.08080808080808081, "presentation_and_reporting": 0.29292929292929293, "results_and_discussion": 0.15151515151515152, "suggestion_and_solution": 0.1919191919191919 }	1.676768	[ { "content": "Please review this paper.", "role": "user" }, { "content": "Reviewers' comments:\nReviewer #1 (Remarks to the Author):\nMudumbi et al combine a toolbox of advanced microscopy methods aiming to resolve the long-lasting\n debate about the nature of different INM protein groups in the peripheral channels in the NPC and\n their integration within the nuclear pore membrane.\n Overall, I find the manuscript exciting, well written and presented (professionally presented Figures),\n technical sound, and important to be published. I therefore recommend to accept the manuscript for\n publication in Nature Communications after addressing the following minor comments.\n (1) Could the authors please discuss the impact of the mechanical rigidity of INM proteins in the\n revised version of the manuscript? I appreciate the significance of their length/height. However, the\n reaction diffusion dynamics and their hydrodynamic radius will also depend on the mechanical status\n of the molecules.\n (2) I think that the manuscript would be strengthened if the authors could show dynamic data, which\n allows the readership to judge the quality of the measurements. The post-processed data and\n presentation is very well executed but the raw data acquisition is difficult to judge.\n (3) Could the authors please provide statistical significance of the different mobilities as measure by\n smFRAP in Table 1? Table 1 provides standard deviations but no p-values.\n (4) I believe that Nat Com does not allow abbreviations in the abstract.\nReviewer #2 (Remarks to the Author):\n(5) Could the authors please rephrase the sentence starting with \"Here\" (line 87)? The reader could be\n confused that the summary paragraph of the end of the introduction starts at this location, which only\n follows line 130.\n \nIn this paper Mudumbi et al, show exciting data providing structural evidence how transmembrane\nproteins are imported via the Nuclear Pore Complex to the inner nuclear membrane. With their\nadvanced and unique single-molecule techniques they make an important contribution to the\nunderstanding of the import pathway of several membrane proteins.\nThe import route that membrane proteins take has been long debated. While the relevance of sorting\nsignals, RanGTP, and NTRS have all been a subject of debate, the fiercest debates concerned the\ntransport route through the NPC. First, multiple labs provided support for the idea that most yeast and\nhuman INM proteins use the peripheral channels observed in several EM studies. However, this\ntransport route through the lateral channels was never directly shown. In addition to this canonical\nroute, two yeast INM proteins were proposed to take a different route: they use an active import\nmechanism which involves the passage of their NTR-bound end of the molecule through the central\nchannel, while their transmembrane domains and the C-terminus travel the peripheral channels. The\nproposed transport route implied that, at least transiently, openings must exist between the space\nimmediately aligning the pore membrane and the central channel. Only indirect evidence was available\nso far to support this transport route. According to many in the field, this proposed transport route\nwas incompatible with the current structural data, and hence regarded highly unlikely. Another\nconcern was that it could be an oddity of these two proteins, and/or an oddity of yeast.\nIn the current manuscript this longstanding debate about where these proteins translocate is\n addressed using suitable and sophisticated single molecule methods developed by Prof Yang and\n coworkers and using a range of native proteins. Five human inner nuclear membrane proteins are\n studied (NET51, NET59, NET99, LBR, Lap2beta). The existence of the peripheral channel as well as\n passage of Net51/59/99 via the peripheral channels is confirmed. The overall dimensions of the NPC\n scaffold and the peripheral channels are is good agreement with EM studies, which provides\n confidence in the accuracy of the SPEED method (which I cannot evaluate in detail). Most exciting is\n the data on LBR and Lap2beta which show that their NLS and disordered regions pass the central\n channel, at the same place as soluble proteins with transportins, while the transmembrane and\n membrane proximal regions pass through the lateral channels. Most convincing are those\n measurements where the LBR proteins is simultaneously tagged on both termini, so that the\n recordings of transport through the lateral channels and the central channel reflects one and the same\n molecule. These measurements provide the first direct evidence that the NPC can accommodate for\n the hotly debated transport route through the central channel. In fact, the studies by Mudumbi et al\n align well with the emerging view that the structure of the NPC is much more dynamic than previously\n thought. In a set of elegant studies the authors additionally show how perturbations of the NPC\n channels or the sorting signals can alter the transport path of a membrane proteins. This is a\n surprising finding that may turn out to have important biological implications. Also, the authors\n provide kinetic information about both transport routes, which is roughly in line with previous\n estimates from yeast and human systems. In addition to the experimental data the authors attempt to\n address the question how prevalent both transport mechanisms are. They estimate that only some 10%\n have ID linkers and NLSs.\n This paper is in my opinion an important and novel paper that would be very suitable for publication in\n nature communications after a few controls and textual changes are considered.\n Main concerns:\n 1) the dual-channel co-tracking SPEED microscopy of the double tagged LBR is important since it\n shows that single molecules of LBR pass the NPC as a full-length protein and membrane embedded.\n Similar experiments are not included for Lap2β. This is a potential problem in the interpretation of the\n data for Lap2β, as, obviously, the presence of truncated versions encoding only the nucleoplasmic,\n soluble domain of LBR could easily explain the signal the central channel. I realize that doing dual\n channel co-tracking of Lap2β would involve a very significant extra effort and this is not required in\n my opinion. I would however like to see western blot and microscopy images of the cells expressing\n the lagged Lap2β, so that readers can be convinced that the data very likely reflects full length\n proteins and not a degradation product of Lap2β.\n 2) The estimate that ~9% of the inner nuclear membrane proteins may encode an ID domain and an\n NLS is really only an estimate as should be referred to as such (also in the abstract). The assumption\n that all these “extend intrinsically disordered (ID) domains containing nuclear localization signals\n (NLSs) into the central channel for directed nuclear transport” is also really only an assumption and\n should be referred to as such. To clarify the analysis done, I have three more suggestions: Please\n provide the list of 46 INM and 11 ONM proteins that were included in the analysis. Line 190 (methods)\n Please indicate the rationale for taking SeqNLS scores higher than 0.86 as ‘sufficiently strong’ NLSs.\n Please elaborate what “made sense in a biological manner for facilitated transport through the center\n of the NPC “ means.\n Minor comments\n 3) I have few suggestions towards placing the work in the context of the literature.\n a. The two papers that were published in 2015, by the Ellenberg and by the Kutay lab (Boni et al, JCB\n 2015, Ungricht et al JCB 2015) should be discussed. Both groups used life imaging and computational\n \ntechniques to show that a diffusion-retention mechanism can explain the import of INM proteins. In\n both papers Lap2β and LBR are studied, like in this manuscript. The estimates of transport kinetics in\n Boni et al match nicely the ones reported here (5.6 molecules per minute per NPC versus 4.6\n molecules/NPC/min), but while the numbers match, the conclusion of Boni et al is that their data\n supports a diffusion retention model for INM protein transport.\n b. In the introduction Line 107-109 is an incorrect statement: In Meinema et al (2011) the\n endogenous ID region of the yeast proteins was used and in Kralt et al. MolBiol Cell 2015 the full\n length and endogenously expressed protein (native promotor, native chromosomal location) was used\n to study the NLS. If the authors meant to say that that thus far there were no papers that studied the\n relevance of an ID linker in the context of the full length and natively expressed protein, that would be\n correct.\n c. In the Discussion: Line 341-343. The experiments do not address this concern directly (different\n proteins, different cells). Laba et al (Cells, 2015) confirms the transmembrane nature of the Heh2-\n derived truncated proteins.\n 3) Line 255: there is no data to support that the R74T mutation indeed results in the disruption of NTR\n binding. Is there biochemical evidence to support this statement? Or is there data to show\n mislocalisation of GFP-NLS(R74T) fusions in cells? Ma et al 2007 J Cell Sci can be cited to support that\n imp beta is bound.\n 4) Other:\n Line 220-223: Lap2β and Lapβ are used both throughout the manuscript.\n Line 261: Lapβ ΔNLS: actually the NLS is not deleted but mutated: R319T and R320T. Please mention.\n Line 160 (methods) range not rage.\n Line 181 (methods) In the case of LBR it reads that the no of copies per cell was based on a previous\n estimate; what was used for the other proteins?\n Line 218-223 (methods). I do not understand which figures or experiments this related to.\n Line 265-268: this paragraph is slightly confusing. Maybe add: ...and roughly 80% of the length of the\n nucleoplasmic region still present, LBR required these ID Linker regions to transit....etc.\n Smovie 2 does not work\n Reviewer #3 (Remarks to the Author):\n This paper has studied how transmembrane proteins in their inner nuclear membrane are transported\n to the inner nuclear membrane using live cell high-speed super-resolution single-molecule microscopy.\n Authors found that either just the peripheral channels or both central and peripheral channels together\n are used for the transport depending on their nucleoplasmic signals. They succeeded in visualizing\n pathways and single-molecule trajectories going through the central and peripheral channels\n distinguishing the positions of the central and peripheral channels and pore membrane lumen.\n However, the paper has substantive issues in description, which makes the paper hard to comprehend.\n Major:\n 1. There is a problem with the logical description. In addition, some paragraphs are too long in the\n Results and Introduction sections. The Discussion section has insufficient contents.\n 2. Abstract is inappropriate as summary. It does not reflect the entire contents, and combining\n Abstract and Significance Statement makes a better summary. Furthermore, for example, the value of\n “~9%” is described only once in the Results section, and Fig. 4E is referred. However, Figure 4E and\n its legend has no description of the clear basis for calculation, and the accuracy of the numerical value\n of 9 is not described.\n 3. The title is not suitable. The word “Nucleoplasmic signals” of the title subject is used only once in\n \nthe Significance Statement section through the whole manuscript, and no explanation is described\n even though the word is not in common use. The word “simultaneously” is also used, but it is not clear\n with which nucleoplasmic signals promote simultaneously.\n 4. It is written that the route precision is ~2 nm at line 164 in the Result section, and 0.90 ~ 1.42 in\nTable S2 based on the simulations (Figs. S8-10). However, the validity of using the values is not clear.\nIt is considered appropriate to use the standard deviation of the distribution or that of data from\nindependent measurements (not standard error of the mean SEM; more precisely, Student's t value\nmultiplied by SEM) as the precision and error of the route and diameter.\nMinor:\n 1. The sentence “~9% additionally extend ...” in Abstract is unclear, since the subject of the sentence\n is not clear.", "role": "assistant" } ]	2,020
ucleoplasmic signals promote directed transmembrane protein import simultaneously via	REVIEWERS' COMMENTS: Reviewer #1 (Remarks to the Author): The authors addressed satisfactorily my previously raised concerns, and therefore recommend to accept the manuscript for publication in Nature Communications. Reviewer #2 (Remarks to the Author): My concerns have been dealt with satisfactory and I support publication. I have one remaining suggestions: To show the complete western in S13 and not just a section reflecting where the full length protein runs, as this control is meant to take away the concern of (soluble) breakdown products complicating the interpretation of the data.	REVIEWERS' COMMENTS: Reviewer #1 (Remarks to the Author): The authors addressed satisfactorily my previously raised concerns, and therefore recommend to accept the manuscript for publication in Nature Communications. Reviewer #2 (Remarks to the Author): My concerns have been dealt with satisfactory and I support publication. I have one remaining suggestions: To show the complete western in S13 and not just a section reflecting where the full length protein runs, as this control is meant to take away the concern of (soluble) breakdown products complicating the interpretation of the data.	TPR	NC_BS_BIOPH_014_RR2.txt	{ "criticism": 0, "example": 0, "importance_and_relevance": 0, "materials_and_methods": 1, "praise": 1, "presentation_and_reporting": 1, "results_and_discussion": 0, "suggestion_and_solution": 1, "total": 3 }	{ "criticism": 0, "example": 0, "importance_and_relevance": 0, "materials_and_methods": 0.3333333333333333, "praise": 0.3333333333333333, "presentation_and_reporting": 0.3333333333333333, "results_and_discussion": 0, "suggestion_and_solution": 0.3333333333333333 }	1.333333	[ { "content": "Please review this paper.", "role": "user" }, { "content": "\n\n\fREVIEWERS' COMMENTS: \n \nReviewer #1 (Remarks to the Author): \n \nThe authors addressed satisfactorily my previously raised concerns, and therefore recommend to \naccept the manuscript for publication in Nature Communications. \n \n \nReviewer #2 (Remarks to the Author): \n \nMy concerns have been dealt with satisfactory and I support publication. I have one remaining \nsuggestions: \n \nTo show the complete western in S13 and not just a section reflecting where the full length protein \nruns, as this control is meant to take away the concern of (soluble) breakdown products complicating \nthe interpretation of the data. \n\n\f\n\f\n", "role": "assistant" } ]	2,020
Lipid-protein interactions modulate the conformational equilibrium of a potassium channel	REVIEWERS' COMMENTS: Reviewer #1 (Remarks to the Author): All questions are answered satisfactory. I have no further comment and therefore recommend this paper to be published in Nature Communications. Reviewer #2 (Remarks to the Author): In the revised manuscript the authors addressed to all reviewer comments, and I can recommend the paper for publication Reviewer #3 (Remarks to the Author): I am satisfied with all changes made to the manuscript. Congratulations on an excellent study.	REVIEWERS' COMMENTS: Reviewer #1 (Remarks to the Author): All questions are answered satisfactory. I have no further comment and therefore recommend this paper to be published in Nature Communications. Reviewer #2 (Remarks to the Author): In the revised manuscript the authors addressed to all reviewer comments, and I can recommend the paper for publication Reviewer #3 (Remarks to the Author): I am satisfied with all changes made to the manuscript. Congratulations on an excellent study.	TPR	NC_BS_BIOPH_015_RR2.txt	{ "criticism": 1, "example": 0, "importance_and_relevance": 0, "materials_and_methods": 0, "praise": 3, "presentation_and_reporting": 0, "results_and_discussion": 0, "suggestion_and_solution": 1, "total": 4 }	{ "criticism": 0.25, "example": 0, "importance_and_relevance": 0, "materials_and_methods": 0, "praise": 0.75, "presentation_and_reporting": 0, "results_and_discussion": 0, "suggestion_and_solution": 0.25 }	1.25	[ { "content": "Please review this paper.", "role": "user" }, { "content": "\fREVIEWERS' COMMENTS: \n \nReviewer #1 (Remarks to the Author): \n \nAll questions are answered satisfactory. I have no further comment and therefore recommend this \npaper to be published in Nature Communications. \n \n \nReviewer #2 (Remarks to the Author): \n \nIn the revised manuscript the authors addressed to all reviewer comments, and I can recommend \nthe paper for publication \n \n \nReviewer #3 (Remarks to the Author): \n \nI am satisfied with all changes made to the manuscript. Congratulations on an excellent study. \n\n\f", "role": "assistant" } ]	2,020
Lipid-protein interactions modulate the conformational equilibrium of a potassium channel	Reviewers' comments: Reviewer #1 (Remarks to the Author): In this study, the authors investigated the effect of different membrane environment on the channel permeation rates using molecular dynamics simulations. The authors found that the thickness of the bilayer, lipid unsaturation as well as cholesterol content all influence the ion permeation rate of MthK. Furthermore, from their restrained and unrestrained simulations, they identified two gates at the selectivity filter and the central cavity that are linked to each other. In the field of molecular dynamics, the influence of different membranes and lipids on the channel permeation has been questioned for quite some time. Therefore I highly appreciate this systematic investigation by the authors and I believe this study deserves publication in Nature Communications. I have some minor suggestions and questions: 1. The authors showed a strong correlation between the Thr59 O(gamma)-O(gamma) distance and the Ile84-Thr59 side chain distance. Does the interaction between the Ile84-Thr59 also influence the ion occupancy at the S4 in the selectivity filter? 2. The authors stated that conductive states are found for asymmetrical structures when some subunits are in the kinked states whereas the remaining ones are in the bent state. Is the conductance the same when only one of the subunit is in the kinked state or two and three subunits are in the kinked state? 3. In a recent publication in Nature Communications (Hydrophobic gating in BK channels) a hydrophobic gate was identified in the pore. Since MthK has been considered as a bacterial homologue of the BK channel, do the authors think that both channels share in a way same mechanism in gating? 4. Is MthK a mechanosensitive channel? If the channel conductance and open probability are sensitive to the lateral pressure in the membrane as proposed by the authors, I would naively think that the channel would be also sensitive to the mechanical force. Is this the case? Reviewer #2 (Remarks to the Author): The paper demonstrates, by molecular dynamics computer simulations, that lipid membrane composition modulates conductivity of MthK potasium ion channel, and reveals molecular details of this phenomena. Particularly it was convinsingly demonstrated that two states of the transmembrane helix near residues 81-86, "kink" and "bent", are related to lower respectively higher conductivity of the channel, while the equlibrium between these two states are affected by the composition of surrounding lipids. The study is very well designed and the text is clearly written. The main conclusion are supported by the use of different force fields showing simular results, as well as by repeating simulations significant amount of times. The study is also a nice example on how computer modeling can contribute to getting novel insight into membrane protein functioning. I favor publication of this paper, subject to some revision accountings for the points below. 1. The temperature of the simulations, as follows from the SI, was set to 323 K. I guess the reason for his was to provide conditions for some lipids (particularly DPPC) to be in a liquid crystalline phase, but question arise how the temperature difference (compared to the physiological temperature) may affect results of the paper. The authors need to discuss possible effects of the temperature shift. 2. More details should be given of how the initial state of the system was prepared and equilibrated. E.g., whether protein was fixed while lipids and water were equilibrated, when the pressure coupling was switched on, etc. Also, whether the same initial state was used in all replicas of the same system? 3. What was the electric field applied to the system (in units like V/m)? It is not enough to say "to mimic a transmembrane voltage of 300 mv" since this can be interpreted in different ways. 4. Location of double bond in unsaturated lipids should be given (like POPC 16:0/18:1(n-9) ) Reviewer #3 (Remarks to the Author): In this manuscript, the author’s describe the regulation of MthK channel ion conductance by a variety of membrane lipids. The experiments were well performed, and I particularly appreciated the validation of results by using 2 unrelated force-fields. The authors conclude that in MthK, membrane thickness alters ion conductance in part through affecting the equilibrium of bent to kinked conformations of the inner helix. The authors’ data suggests that kinking of this inner helix affects both the opening degree of the selectivity filter and the dehydration degree of the central. The authors propose specific atomic interactions that enable the inner helix to impact the structure of the selectivity filter. Therefore, the authors provide specific molecular details about how the lipid conditions examined can alter ion conductance in MthK channels. Overall, this paper provides novel insights into the molecular details of lipid regulation of MthK channels. Experiments are appropriate, detailed, and well executed. The manuscript is well written. I offer only these minor points to address. Minor Points: Lines 111 – 116: Results on Cholesterol. Since cholesterol can have multiple effects on protein function through both changes in bilayer properties and through direct interactions with the protein, it would be useful to briefly mention in this paragraph whether or not cholesterol was interacting with any of the channel subunits in any of your simulations, and whether or not such interactions may contribute to the observed changes in MthK’s conductance in that group of simulations. The authors do finally addressed in the section of lines 387 – 388, however, there it is somewhat of a late display of important results. Line 12: “Membranes do not only provide a matrix for a variety…” should say “Membranes not only provide a matrix for a variety…” Lines 52 and 53: Did you mean “responds” instead of “response” in this sentence “However, it remains unclear how the channel structure response to the membrane environment at the atomistic level.”?? Line 53: The sentence “This is where MD simulations can provide the missing information.” seems a bit awkward. Perhaps something like “This is where MD simulations can provide valuable insights.” Lines 75-77: This sentence is poorly structured. Please reword. “More importantly, the conformational changes of the inner helix revealed by our simulations define two gates at the selectivity filter and the central cavity, respectively, and mediate a crosstalk between these two gates, suggesting a complex way of channel gating.” Line 77: “Our work provided” should say “Our work provides” Line 392: “The thickness mismatch is mainly resolved by the distortion of the membranes in simulations.” I think it’s worth noting that this is similar to what was observed for hydrophobic mistmatch in KcsA by Callahan et al., 2019. Supp. Doc. Line 33. “We conducted a series of simulations with restraints…” should say “We conducted a series of simulations of the MthK potassium channel with restraints…”	Reviewers' comments: Reviewer #1 (Remarks to the Author): In this study, the authors investigated the effect of different membrane environment on the channel permeation rates using molecular dynamics simulations. The authors found that the thickness of the bilayer, lipid unsaturation as well as cholesterol content all influence the ion permeation rate of MthK. Furthermore, from their restrained and unrestrained simulations, they identified two gates at the selectivity filter and the central cavity that are linked to each other. In the field of molecular dynamics, the influence of different membranes and lipids on the channel permeation has been questioned for quite some time. Therefore I highly appreciate this systematic investigation by the authors and I believe this study deserves publication in Nature Communications. I have some minor suggestions and questions: 1. The authors showed a strong correlation between the Thr59 O(gamma)-O(gamma) distance and the Ile84-Thr59 side chain distance. Does the interaction between the Ile84-Thr59 also influence the ion occupancy at the S4 in the selectivity filter? 2. The authors stated that conductive states are found for asymmetrical structures when some subunits are in the kinked states whereas the remaining ones are in the bent state. Is the conductance the same when only one of the subunit is in the kinked state or two and three subunits are in the kinked state? 3. In a recent publication in Nature Communications (Hydrophobic gating in BK channels) a hydrophobic gate was identified in the pore. Since MthK has been considered as a bacterial homologue of the BK channel, do the authors think that both channels share in a way same mechanism in gating? 4. Is MthK a mechanosensitive channel? If the channel conductance and open probability are sensitive to the lateral pressure in the membrane as proposed by the authors, I would naively think that the channel would be also sensitive to the mechanical force. Is this the case? Reviewer #2 (Remarks to the Author): The paper demonstrates, by molecular dynamics computer simulations, that lipid membrane composition modulates conductivity of MthK potasium ion channel, and reveals molecular details of this phenomena. Particularly it was convinsingly demonstrated that two states of the transmembrane helix near residues 81-86, "kink" and "bent", are related to lower respectively higher conductivity of the channel, while the equlibrium between these two states are affected by the composition of surrounding lipids. The study is very well designed and the text is clearly written. The main conclusion are supported by the use of different force fields showing simular results, as well as by repeating simulations significant amount of times. The study is also a nice example on how computer modeling can contribute to getting novel insight into membrane protein functioning. I favor publication of this paper, subject to some revision accountings for the points below. 1. The temperature of the simulations, as follows from the SI, was set to 323 K. I guess the reason for his was to provide conditions for some lipids (particularly DPPC) to be in a liquid crystalline phase, but question arise how the temperature difference (compared to the physiological temperature) may affect results of the paper. The authors need to discuss possible effects of the temperature shift. 2. More details should be given of how the initial state of the system was prepared and equilibrated. E.g., whether protein was fixed while lipids and water were equilibrated, when the pressure coupling was switched on, etc. Also, whether the same initial state was used in all replicas of the same system? 3. What was the electric field applied to the system (in units like V/m)? It is not enough to say "to mimic a transmembrane voltage of 300 mv" since this can be interpreted in different ways. 4. Location of double bond in unsaturated lipids should be given (like POPC 16:0/18:1(n-9) ) Reviewer #3 (Remarks to the Author): In this manuscript, the author’s describe the regulation of MthK channel ion conductance by a variety of membrane lipids. The experiments were well performed, and I particularly appreciated the validation of results by using 2 unrelated force-fields. The authors conclude that in MthK, membrane thickness alters ion conductance in part through affecting the equilibrium of bent to kinked conformations of the inner helix. The authors’ data suggests that kinking of this inner helix affects both the opening degree of the selectivity filter and the dehydration degree of the central. The authors propose specific atomic interactions that enable the inner helix to impact the structure of the selectivity filter. Therefore, the authors provide specific molecular details about how the lipid conditions examined can alter ion conductance in MthK channels. Overall, this paper provides novel insights into the molecular details of lipid regulation of MthK channels. Experiments are appropriate, detailed, and well executed. The manuscript is well written. I offer only these minor points to address. Minor Points: Lines 111 – 116: Results on Cholesterol. Since cholesterol can have multiple effects on protein function through both changes in bilayer properties and through direct interactions with the protein, it would be useful to briefly mention in this paragraph whether or not cholesterol was interacting with any of the channel subunits in any of your simulations, and whether or not such interactions may contribute to the observed changes in MthK’s conductance in that group of simulations. The authors do finally addressed in the section of lines 387 – 388, however, there it is somewhat of a late display of important results. Line 12: “Membranes do not only provide a matrix for a variety…” should say “Membranes not only provide a matrix for a variety…” Lines 52 and 53: Did you mean “responds” instead of “response” in this sentence “However, it remains unclear how the channel structure response to the membrane environment at the atomistic level.”?? Line 53: The sentence “This is where MD simulations can provide the missing information.” seems a bit awkward. Perhaps something like “This is where MD simulations can provide valuable insights.” Lines 75-77: This sentence is poorly structured. Please reword. “More importantly, the conformational changes of the inner helix revealed by our simulations define two gates at the selectivity filter and the central cavity, respectively, and mediate a crosstalk between these two gates, suggesting a complex way of channel gating.” Line 77: “Our work provided” should say “Our work provides” Line 392: “The thickness mismatch is mainly resolved by the distortion of the membranes in simulations.” I think it’s worth noting that this is similar to what was observed for hydrophobic mistmatch in KcsA by Callahan et al., 2019. Supp. Doc. Line 33. “We conducted a series of simulations with restraints…” should say “We conducted a series of simulations of the MthK potassium channel with restraints…”	TPR	NC_BS_BIOPH_015_RR1.txt	{ "criticism": 5, "example": 9, "importance_and_relevance": 4, "materials_and_methods": 21, "praise": 8, "presentation_and_reporting": 11, "results_and_discussion": 11, "suggestion_and_solution": 11, "total": 57 }	{ "criticism": 0.08771929824561403, "example": 0.15789473684210525, "importance_and_relevance": 0.07017543859649122, "materials_and_methods": 0.3684210526315789, "praise": 0.14035087719298245, "presentation_and_reporting": 0.19298245614035087, "results_and_discussion": 0.19298245614035087, "suggestion_and_solution": 0.19298245614035087 }	1.403509	[ { "content": "Please review this paper.", "role": "user" }, { "content": "Reviewers' comments: \n \nReviewer #1 (Remarks to the Author): \n \nIn this study, the authors investigated the effect of different membrane environment on the \nchannel permeation rates using molecular dynamics simulations. The authors found that the \nthickness of the bilayer, lipid unsaturation as well as cholesterol content all influence the ion \npermeation rate of MthK. Furthermore, from their restrained and unrestrained simulations, they \nidentified two gates at the selectivity filter and the central cavity that are linked to each other. In \nthe field of molecular dynamics, the influence of different membranes and lipids on the channel \npermeation has been questioned for quite some time. Therefore I highly appreciate this systematic \ninvestigation by the authors and I believe this study deserves publication in Nature \nCommunications. \n \nI have some minor suggestions and questions: \n \n1. The authors showed a strong correlation between the Thr59 O(gamma)-O(gamma) distance and \nthe Ile84-Thr59 side chain distance. Does the interaction between the Ile84-Thr59 also influence \nthe ion occupancy at the S4 in the selectivity filter? \n \n2. The authors stated that conductive states are found for asymmetrical structures when some \nsubunits are in the kinked states whereas the remaining ones are in the bent state. Is the \nconductance the same when only one of the subunit is in the kinked state or two and three \nsubunits are in the kinked state? \n \n3. In a recent publication in Nature Communications (Hydrophobic gating in BK channels) a \nhydrophobic gate was identified in the pore. Since MthK has been considered as a bacterial \nhomologue of the BK channel, do the authors think that both channels share in a way same \nmechanism in gating? \n \n4. Is MthK a mechanosensitive channel? If the channel conductance and open probability are \nsensitive to the lateral pressure in the membrane as proposed by the authors, I would naively \nthink that the channel would be also sensitive to the mechanical force. Is this the case? \n \n \n \nReviewer #2 (Remarks to the Author): \n \nThe paper demonstrates, by molecular dynamics computer simulations, that \nlipid membrane composition modulates conductivity of MthK potasium ion \nchannel, and reveals molecular details of this phenomena. \nParticularly it was convinsingly demonstrated that \ntwo states of the transmembrane helix near residues 81-86, \"kink\" and \"bent\", \nare related to lower respectively higher conductivity of the channel, while \nthe equlibrium between these two states are affected by the composition of \nsurrounding lipids. The study is very well designed and the text is clearly \nwritten. The main conclusion are supported by the use of different force fields \nshowing simular results, as well as by repeating simulations significant \namount of times. The study is also a nice example on how computer modeling can \ncontribute to getting novel insight into membrane protein functioning. \nI favor publication of this paper, subject to some revision \naccountings for the points below. \n \n1. The temperature of the simulations, as follows from the SI, was set to 323 K. I guess the \nreason for his was to provide conditions for \nsome lipids (particularly DPPC) to be in a liquid crystalline phase, but \nquestion arise how the temperature difference (compared to the physiological \ntemperature) may affect results of the paper. The authors need to discuss \npossible effects of the temperature shift. \n\n\f \n2. More details should be given of how the initial state of the system \nwas prepared and equilibrated. E.g., whether protein was fixed while lipids \nand water were equilibrated, when the pressure coupling was switched on, etc. \nAlso, whether the same initial state was used in all replicas of the \nsame system? \n \n3. What was the electric field applied to the system (in units like V/m)? \nIt is not enough to say \"to mimic a transmembrane voltage of 300 mv\" \nsince this can be interpreted in different ways. \n \n4. Location of double bond in unsaturated lipids should be given (like POPC \n16:0/18:1(n-9) ) \n \n \nReviewer #3 (Remarks to the Author): \n \nIn this manuscript, the author’s describe the regulation of MthK channel ion conductance by a \nvariety of membrane lipids. The experiments were well performed, and I particularly appreciated \nthe validation of results by using 2 unrelated force-fields. The authors conclude that in MthK, \nmembrane thickness alters ion conductance in part through affecting the equilibrium of bent to \nkinked conformations of the inner helix. The authors’ data suggests that kinking of this inner helix \naffects both the opening degree of the selectivity filter and the dehydration degree of the central. \nThe authors propose specific atomic interactions that enable the inner helix to impact the structure \nof the selectivity filter. Therefore, the authors provide specific molecular details about how the lipid \nconditions examined can alter ion conductance in MthK channels. \n \nOverall, this paper provides novel insights into the molecular details of lipid regulation of MthK \nchannels. Experiments are appropriate, detailed, and well executed. The manuscript is well \nwritten. \n \nI offer only these minor points to address. \n \nMinor Points: \n \nLines 111 – 116: Results on Cholesterol. Since cholesterol can have multiple effects on protein \nfunction through both changes in bilayer properties and through direct interactions with the \nprotein, it would be useful to briefly mention in this paragraph whether or not cholesterol was \ninteracting with any of the channel subunits in any of your simulations, and whether or not such \ninteractions may contribute to the observed changes in MthK’s conductance in that group of \nsimulations. The authors do finally addressed in the section of lines 387 – 388, however, there it is \nsomewhat of a late display of important results. \n \nLine 12: “Membranes do not only provide a matrix for a variety…” should say “Membranes not only \nprovide a matrix for a variety…” \n \nLines 52 and 53: Did you mean “responds” instead of “response” in this sentence “However, it \nremains unclear how the channel structure response to the membrane environment at the \natomistic level.”?? \n \nLine 53: The sentence “This is where MD simulations can provide the missing information.” seems \na bit awkward. Perhaps something like “This is where MD simulations can provide valuable \ninsights.” \n \nLines 75-77: This sentence is poorly structured. Please reword. “More importantly, the \nconformational changes of the inner helix revealed by our simulations define two gates at the \nselectivity filter and the central cavity, respectively, and mediate a crosstalk between these two \ngates, suggesting a complex way of channel gating.” \n \n\n\fLine 77: “Our work provided” should say “Our work provides” \n \nLine 392: “The thickness mismatch is mainly resolved by the distortion of the membranes in \nsimulations.” I think it’s worth noting that this is similar to what was observed for hydrophobic \nmistmatch in KcsA by Callahan et al., 2019. \n \nSupp. Doc. \n \nLine 33. “We conducted a series of simulations with restraints…” should say “We conducted a \nseries of simulations of the MthK potassium channel with restraints…” \n\n\f", "role": "assistant" } ]	2,020
Selective flexible packaging pathways of the segmented genome of influenza A virus	Reviewers' Comments: Reviewer #1: Remarks to the Author: In the revised manuscript entitled, “Selective, flexible packaging pathways of the segmented genome of Influenza A Virus”, the authors utilized sequential FISH staining to visualize all 8 segments within single cells at a single timepoint (10hpi) and quantify the colocalization of segments. Visualizing all 8 vRNA segments is a powerful tool for studying IAV replication and assembly. The overall conclusions remain largely unchanged, 1) overall proportions of each segment are equal in a given cell, 2) the most abundant vRNP complexes are those with either 1 segment (rank 1) or 8 segments (rank 8), and 3) higher ranked complexes are found more often in intact virions. Overall, the reviewers have revised their initial manuscript for clarity and have added the necessary caveats to the manuscript when describing the data. However, a few points should still be addressed. 1. The authors provided clarification on the infection of A/Mallard in A549 cells and revealed “impaired nuclear export of NP” (lines 264-266 and in the rebuttal letter), which would assume imparired nuclear export of vRNP as well. In addition, the authors report a significant defect in viral replication of A/mallard in A459 cells compare to A/Panama (ie a 3 log reduction in viral titer at 72 hpi). If vRNA segments are found in the cytoplasm, but NP nuclear export is defective that would also suggest that the NP – vRNA binding is altered in these cells and could then impact the RNA-RNA interactions. Proper NP oligermization, binding, and post-translational modifications are known to be important for vRNP packaging, so it unclear why that authors state that host proteins may be the restrictive feature (lines 271-273). While host proteins may be important for IAV vRNP assembly through a variety of mechanisms that could impact vRNP bundling, there is no data presented in this manuscript to support the assertion. Given the complexity of comparing data from A/mallard and A/Panama – the authors should remove the data as it is misleading to compare the two when the amount of cytoplasmic vRNP are not equal at this time point. 2. The authors provided the code for colocalization to the reviewers, but should make this freely available via github site to ensure that others can compare the colocalization algorithms. It is unclear, without extensive testing, whether the code uses ellipse centers for each segment spot to calculate the colocalization to nearby spots and whether the colocalization must fit all criteria - ie be within 300nm xy and 1000nm z or if a spot within 1000nm in z would still be counted in the MSC even if it was > than 300nm in xy. 3. In addition, the authors should acknowledge somewhere that not all 20-30 FISH probes will bind each segment. Chou et al (PMID: 22547828) very clearly demonstrate using photo-bleaching studies that only 7 FISH probes bound a single segment within a purified virion. This could be different within an infected cell as the accessibility of the FISH probe can change during assembly. This is the justification for why it is difficult to conclude that less than 2 copies of the same segment are located within a diffraction limited spot. Reviewer #2: Remarks to the Author: Haralampiev et al. submitted a revised version of the manuscript titled “Selective flexible packaging pathways of the segmented genome of Influenza A virus” that includes improvements to data analysis and more comprehensive discussion. However, in its current state the narrative remains unclear. • The application of statistics improved the analysis of the data (seen in figure legends), but the text describing the presentation and interpretation of results requires revision. The results section currently reads like a descriptive discussion. In some cases, the authors have included statistical analyses for parts of the figures and include explanation in the figure legends, but they do not use the statistics to make meaningful conclusions about the data. See examples in lines 132-141 and 185-188. • Figures and figure panels should be labeled separately in the order of occurrence in the text of the manuscript. The authors need to restructure the text to address all panels of the figures in the order that they appear. It is disjointed and hard to follow in its current format. • Lines 53-57 seem unnecessary to include since the paper does not address reassortment. • Lines 62-65 appear to be contradictory and should be edited for clarity. • Interpretation of data should not be included in the figure legends. Lines 785-787, 799-800, 825- 827 should be moved to the appropriate results section. Statistical significance should be able to be determined from the figure itself (* or ns) with appropriate strata listed in the legends. • Fig. 5c scale is incorrect (25 instead of 20). The colored bars are distracting, unless the color correlates to something else to improve understanding. Reviewer #3: Remarks to the Author: Influenza A viruses co-package eight different genome segments into virions. Previous studies suggested that the eight genomic segments bundle inside infected cells prior to their co-packaging into new virus particles, however this process has remained to be visualized in detail. The current revised manuscript offers a novel FISH technique to analyze genome bundling inside infected cells. While the manuscript does not test whether previously suspected determinants such as vRNA-vRNA interactions among packaging signals are required for efficient genome bundling, it suggest three novel mechanisms: 1. balanced segment expression 2. balanced accumulation of vRNPs in the cytoplasm (nuclear export) 3. certain host factors On the one hand, the data on A/Panama suggest that low expression levels of particular genome segments (segment 1 or segment 8) may abrogate the formation of high-rank MSCs at different stages of their assembly. The authors frankly admit that this hypothesis is only based on two analyzed cells. Furthermore, the data on A/Mallard suggest that even in the presence of similar cytosolic levels of the eight segments, the formation of high-rank MSCs is inefficient. They further suggest that other mechanism(s) may influence genome bundling. This assumption is based on two older papers, where the authors identified that A/Mallard inefficiently exports the vRNPs from the nucleus to the cytoplasm and induces the expression of different host factors. Taken together, all three mechanisms could affect genome bundling, yet the current data are too limited and vague to draw any conclusions. The manuscript requires additional data to support either of these three mechanisms. The data provided by the authors in the refined manuscript suggest that the formation of high-rank MSCs is a dynamic process that changes with time. The authors provide a new experiment with A/Panama where they show that the formation of high-rank MSCs (Suppl. Fig 9) is very poor at 6 hpi, yet efficient at 10 hpi. In an older publication (Bogdanow et al., Nat. Com, 2019), the authors showed that at 6 hpi, A/Panama poorly exports its vRNPs from the nucleus, while they efficiently localize to the cytoplasm at 10 hpi. This suggests, that the accumulation of (certain levels of) the eight vRNPs in the cytoplasm triggers genome bundling. Importantly, this publication also revealed that the nuclear vRNP export of A/Mallard is substantially delayed compared to that of A/Panama. At 10 hpi, A/Mallard largely fails to export its genome from the nucleus, while at 16 hpi, the vRNPs partially accumulate in the cytoplasm. This may indicate that A/Mallard fails to bundle the genome as a consequence of the poor accumulation of vRNPs in the cytoplasm but is in principle able to bundle the genome when vRNPs are exported to the cytoplasm. To test this hypothesis, the authors should investigate MSC formation of A/Mallard at a later time point of infection. Ideally, the vRNPs should accumulate in the cytoplasm to similar levels as observed with A/Panama at 10 hpi. If efficient vRNP accumulation triggers genome bundling, the eight MSC ranks would shift towards a U-shaped distribution. The authors are encouraged to perform this experiment as all experimental tools should be available. Furthermore, reviewer 1 has raised a similar concern (concern 6, reviewer 1). Minor points: 1) line 268-269: “the comparatively increased amount”: there is no quantification and only a picture is shown. 2) line 396-398: what is the rational to compare RNA-RNA networks with MSC formation? Please specify. 3) line 413: what are transient segment shortages?	Reviewers' Comments: Reviewer #1: Remarks to the Author: In the revised manuscript entitled, “Selective, flexible packaging pathways of the segmented genome of Influenza A Virus”, the authors utilized sequential FISH staining to visualize all 8 segments within single cells at a single timepoint (10hpi) and quantify the colocalization of segments. Visualizing all 8 vRNA segments is a powerful tool for studying IAV replication and assembly. The overall conclusions remain largely unchanged, 1) overall proportions of each segment are equal in a given cell, 2) the most abundant vRNP complexes are those with either 1 segment (rank 1) or 8 segments (rank 8), and 3) higher ranked complexes are found more often in intact virions. Overall, the reviewers have revised their initial manuscript for clarity and have added the necessary caveats to the manuscript when describing the data. However, a few points should still be addressed. 1. The authors provided clarification on the infection of A/Mallard in A549 cells and revealed “impaired nuclear export of NP” (lines 264-266 and in the rebuttal letter), which would assume imparired nuclear export of vRNP as well. In addition, the authors report a significant defect in viral replication of A/mallard in A459 cells compare to A/Panama (ie a 3 log reduction in viral titer at 72 hpi). If vRNA segments are found in the cytoplasm, but NP nuclear export is defective that would also suggest that the NP – vRNA binding is altered in these cells and could then impact the RNA-RNA interactions. Proper NP oligermization, binding, and post-translational modifications are known to be important for vRNP packaging, so it unclear why that authors state that host proteins may be the restrictive feature (lines 271-273). While host proteins may be important for IAV vRNP assembly through a variety of mechanisms that could impact vRNP bundling, there is no data presented in this manuscript to support the assertion. Given the complexity of comparing data from A/mallard and A/Panama – the authors should remove the data as it is misleading to compare the two when the amount of cytoplasmic vRNP are not equal at this time point. 2. The authors provided the code for colocalization to the reviewers, but should make this freely available via github site to ensure that others can compare the colocalization algorithms. It is unclear, without extensive testing, whether the code uses ellipse centers for each segment spot to calculate the colocalization to nearby spots and whether the colocalization must fit all criteria - ie be within 300nm xy and 1000nm z or if a spot within 1000nm in z would still be counted in the MSC even if it was > than 300nm in xy. 3. In addition, the authors should acknowledge somewhere that not all 20-30 FISH probes will bind each segment. Chou et al (PMID: 22547828) very clearly demonstrate using photo-bleaching studies that only 7 FISH probes bound a single segment within a purified virion. This could be different within an infected cell as the accessibility of the FISH probe can change during assembly. This is the justification for why it is difficult to conclude that less than 2 copies of the same segment are located within a diffraction limited spot. Reviewer #2: Remarks to the Author: Haralampiev et al. submitted a revised version of the manuscript titled “Selective flexible packaging pathways of the segmented genome of Influenza A virus” that includes improvements to data analysis and more comprehensive discussion. However, in its current state the narrative remains unclear. • The application of statistics improved the analysis of the data (seen in figure legends), but the text describing the presentation and interpretation of results requires revision. The results section currently reads like a descriptive discussion. In some cases, the authors have included statistical analyses for parts of the figures and include explanation in the figure legends, but they do not use the statistics to make meaningful conclusions about the data. See examples in lines 132-141 and 185-188. • Figures and figure panels should be labeled separately in the order of occurrence in the text of the manuscript. The authors need to restructure the text to address all panels of the figures in the order that they appear. It is disjointed and hard to follow in its current format. • Lines 53-57 seem unnecessary to include since the paper does not address reassortment. • Lines 62-65 appear to be contradictory and should be edited for clarity. • Interpretation of data should not be included in the figure legends. Lines 785-787, 799-800, 825- 827 should be moved to the appropriate results section. Statistical significance should be able to be determined from the figure itself (* or ns) with appropriate strata listed in the legends. • Fig. 5c scale is incorrect (25 instead of 20). The colored bars are distracting, unless the color correlates to something else to improve understanding. Reviewer #3: Remarks to the Author: Influenza A viruses co-package eight different genome segments into virions. Previous studies suggested that the eight genomic segments bundle inside infected cells prior to their co-packaging into new virus particles, however this process has remained to be visualized in detail. The current revised manuscript offers a novel FISH technique to analyze genome bundling inside infected cells. While the manuscript does not test whether previously suspected determinants such as vRNA-vRNA interactions among packaging signals are required for efficient genome bundling, it suggest three novel mechanisms: 1. balanced segment expression 2. balanced accumulation of vRNPs in the cytoplasm (nuclear export) 3. certain host factors On the one hand, the data on A/Panama suggest that low expression levels of particular genome segments (segment 1 or segment 8) may abrogate the formation of high-rank MSCs at different stages of their assembly. The authors frankly admit that this hypothesis is only based on two analyzed cells. Furthermore, the data on A/Mallard suggest that even in the presence of similar cytosolic levels of the eight segments, the formation of high-rank MSCs is inefficient. They further suggest that other mechanism(s) may influence genome bundling. This assumption is based on two older papers, where the authors identified that A/Mallard inefficiently exports the vRNPs from the nucleus to the cytoplasm and induces the expression of different host factors. Taken together, all three mechanisms could affect genome bundling, yet the current data are too limited and vague to draw any conclusions. The manuscript requires additional data to support either of these three mechanisms. The data provided by the authors in the refined manuscript suggest that the formation of high-rank MSCs is a dynamic process that changes with time. The authors provide a new experiment with A/Panama where they show that the formation of high-rank MSCs (Suppl. Fig 9) is very poor at 6 hpi, yet efficient at 10 hpi. In an older publication (Bogdanow et al., Nat. Com, 2019), the authors showed that at 6 hpi, A/Panama poorly exports its vRNPs from the nucleus, while they efficiently localize to the cytoplasm at 10 hpi. This suggests, that the accumulation of (certain levels of) the eight vRNPs in the cytoplasm triggers genome bundling. Importantly, this publication also revealed that the nuclear vRNP export of A/Mallard is substantially delayed compared to that of A/Panama. At 10 hpi, A/Mallard largely fails to export its genome from the nucleus, while at 16 hpi, the vRNPs partially accumulate in the cytoplasm. This may indicate that A/Mallard fails to bundle the genome as a consequence of the poor accumulation of vRNPs in the cytoplasm but is in principle able to bundle the genome when vRNPs are exported to the cytoplasm. To test this hypothesis, the authors should investigate MSC formation of A/Mallard at a later time point of infection. Ideally, the vRNPs should accumulate in the cytoplasm to similar levels as observed with A/Panama at 10 hpi. If efficient vRNP accumulation triggers genome bundling, the eight MSC ranks would shift towards a U-shaped distribution. The authors are encouraged to perform this experiment as all experimental tools should be available. Furthermore, reviewer 1 has raised a similar concern (concern 6, reviewer 1). Minor points: 1) line 268-269: “the comparatively increased amount”: there is no quantification and only a picture is shown. 2) line 396-398: what is the rational to compare RNA-RNA networks with MSC formation? Please specify. 3) line 413: what are transient segment shortages?	TPR	NC_BS_BIOTEC_001_RR2.txt	{ "criticism": 13, "example": 11, "importance_and_relevance": 1, "materials_and_methods": 36, "praise": 3, "presentation_and_reporting": 17, "results_and_discussion": 8, "suggestion_and_solution": 18, "total": 66 }	{ "criticism": 0.19696969696969696, "example": 0.16666666666666666, "importance_and_relevance": 0.015151515151515152, "materials_and_methods": 0.5454545454545454, "praise": 0.045454545454545456, "presentation_and_reporting": 0.25757575757575757, "results_and_discussion": 0.12121212121212122, "suggestion_and_solution": 0.2727272727272727 }	1.621212	[ { "content": "Please review this paper.", "role": "user" }, { "content": "\n\n \n \n\n\fReviewers' Comments: \n \nReviewer #1: \nRemarks to the Author: \nIn the revised manuscript entitled, “Selective, flexible packaging pathways of the segmented genome \nof Influenza A Virus”, the authors utilized sequential FISH staining to visualize all 8 segments within \nsingle cells at a single timepoint (10hpi) and quantify the colocalization of segments. Visualizing all 8 \nvRNA segments is a powerful tool for studying IAV replication and assembly. The overall conclusions \nremain largely unchanged, 1) overall proportions of each segment are equal in a given cell, 2) the \nmost abundant vRNP complexes are those with either 1 segment (rank 1) or 8 segments (rank 8), and \n3) higher ranked complexes are found more often in intact virions. \nOverall, the reviewers have revised their initial manuscript for clarity and have added the necessary \ncaveats to the manuscript when describing the data. However, a few points should still be addressed. \n1. The authors provided clarification on the infection of A/Mallard in A549 cells and revealed “impaired \nnuclear export of NP” (lines 264-266 and in the rebuttal letter), which would assume imparired nuclear \nexport of vRNP as well. In addition, the authors report a significant defect in viral replication of \nA/mallard in A459 cells compare to A/Panama (ie a 3 log reduction in viral titer at 72 hpi). If vRNA \nsegments are found in the cytoplasm, but NP nuclear export is defective that would also suggest that \nthe NP – vRNA binding is altered in these cells and could then impact the RNA-RNA interactions. \nProper NP oligermization, binding, and post-translational modifications are known to be important for \nvRNP packaging, so it unclear why that authors state that host proteins may be the restrictive feature \n(lines 271-273). While host proteins may be important for IAV vRNP assembly through a variety of \nmechanisms that could impact vRNP bundling, there is no data presented in this manuscript to support \nthe assertion. Given the complexity of comparing data from A/mallard and A/Panama – the authors \nshould remove the data as it is misleading to compare the two when the amount of cytoplasmic vRNP \nare not equal at this time point. \n2. The authors provided the code for colocalization to the reviewers, but should make this freely \navailable via github site to ensure that others can compare the colocalization algorithms. It is unclear, \nwithout extensive testing, whether the code uses ellipse centers for each segment spot to calculate \nthe colocalization to nearby spots and whether the colocalization must fit all criteria - ie be within \n300nm xy and 1000nm z or if a spot within 1000nm in z would still be counted in the MSC even if it \nwas > than 300nm in xy. \n3. In addition, the authors should acknowledge somewhere that not all 20-30 FISH probes will bind \neach segment. Chou et al (PMID: 22547828) very clearly demonstrate using photo-bleaching studies \nthat only 7 FISH probes bound a single segment within a purified virion. This could be different within \nan infected cell as the accessibility of the FISH probe can change during assembly. This is the \njustification for why it is difficult to conclude that less than 2 copies of the same segment are located \nwithin a diffraction limited spot. \n \n \n \nReviewer #2: \nRemarks to the Author: \nHaralampiev et al. submitted a revised version of the manuscript titled “Selective flexible packaging \npathways of the segmented genome of Influenza A virus” that includes improvements to data analysis \nand more comprehensive discussion. However, in its current state the narrative remains unclear. \n• The application of statistics improved the analysis of the data (seen in figure legends), but the text \ndescribing the presentation and interpretation of results requires revision. The results section currently \nreads like a descriptive discussion. In some cases, the authors have included statistical analyses for \nparts of the figures and include explanation in the figure legends, but they do not use the statistics to \nmake meaningful conclusions about the data. See examples in lines 132-141 and 185-188. \n\n\f• Figures and figure panels should be labeled separately in the order of occurrence in the text of the \nmanuscript. The authors need to restructure the text to address all panels of the figures in the order \nthat they appear. It is disjointed and hard to follow in its current format. \n• Lines 53-57 seem unnecessary to include since the paper does not address reassortment. \n• Lines 62-65 appear to be contradictory and should be edited for clarity. \n• Interpretation of data should not be included in the figure legends. Lines 785-787, 799-800, 825-\n827 should be moved to the appropriate results section. Statistical significance should be able to be \ndetermined from the figure itself (* or ns) with appropriate strata listed in the legends. \n• Fig. 5c scale is incorrect (25 instead of 20). The colored bars are distracting, unless the color \ncorrelates to something else to improve understanding. \n \n \n \nReviewer #3: \nRemarks to the Author: \nInfluenza A viruses co-package eight different genome segments into virions. Previous studies \nsuggested that the eight genomic segments bundle inside infected cells prior to their co-packaging \ninto new virus particles, however this process has remained to be visualized in detail. The current \nrevised manuscript offers a novel FISH technique to analyze genome bundling inside infected cells. \nWhile the manuscript does not test whether previously suspected determinants such as vRNA-vRNA \ninteractions among packaging signals are required for efficient genome bundling, it suggest three \nnovel mechanisms: \n1. balanced segment expression \n2. balanced accumulation of vRNPs in the cytoplasm (nuclear export) \n3. certain host factors \nOn the one hand, the data on A/Panama suggest that low expression levels of particular genome \nsegments (segment 1 or segment 8) may abrogate the formation of high-rank MSCs at different \nstages of their assembly. The authors frankly admit that this hypothesis is only based on two analyzed \ncells. \nFurthermore, the data on A/Mallard suggest that even in the presence of similar cytosolic levels of the \neight segments, the formation of high-rank MSCs is inefficient. They further suggest that other \nmechanism(s) may influence genome bundling. This assumption is based on two older papers, where \nthe authors identified that A/Mallard inefficiently exports the vRNPs from the nucleus to the cytoplasm \nand induces the expression of different host factors. \nTaken together, all three mechanisms could affect genome bundling, yet the current data are too \nlimited and vague to draw any conclusions. The manuscript requires additional data to support either \nof these three mechanisms. \n \nThe data provided by the authors in the refined manuscript suggest that the formation of high-rank \nMSCs is a dynamic process that changes with time. The authors provide a new experiment with \nA/Panama where they show that the formation of high-rank MSCs (Suppl. Fig 9) is very poor at 6 hpi, \nyet efficient at 10 hpi. In an older publication (Bogdanow et al., Nat. Com, 2019), the authors showed \nthat at 6 hpi, A/Panama poorly exports its vRNPs from the nucleus, while they efficiently localize to \nthe cytoplasm at 10 hpi. This suggests, that the accumulation of (certain levels of) the eight vRNPs in \nthe cytoplasm triggers genome bundling. Importantly, this publication also revealed that the nuclear \nvRNP export of A/Mallard is substantially delayed compared to that of A/Panama. At 10 hpi, A/Mallard \nlargely fails to export its genome from the nucleus, while at 16 hpi, the vRNPs partially accumulate in \nthe cytoplasm. This may indicate that A/Mallard fails to bundle the genome as a consequence of the \npoor accumulation of vRNPs in the cytoplasm but is in principle able to bundle the genome when \nvRNPs are exported to the cytoplasm. To test this hypothesis, the authors should investigate MSC \nformation of A/Mallard at a later time point of infection. Ideally, the vRNPs should accumulate in the \n\n\fcytoplasm to similar levels as observed with A/Panama at 10 hpi. If efficient vRNP accumulation \ntriggers genome bundling, the eight MSC ranks would shift towards a U-shaped distribution. The \nauthors are encouraged to perform this experiment as all experimental tools should be available. \nFurthermore, reviewer 1 has raised a similar concern (concern 6, reviewer 1). \n \nMinor points: \n1) line 268-269: “the comparatively increased amount”: there is no quantification and only a picture is \nshown. \n2) line 396-398: what is the rational to compare RNA-RNA networks with MSC formation? Please \nspecify. \n3) line 413: what are transient segment shortages? \n\n\f", "role": "assistant" } ]	2,020
Selective flexible packaging pathways of the segmented genome of influenza A virus	Reviewers' Comments: Reviewer #1: Remarks to the Author: In the manuscript entitled, “Selective, flexible packaging pathways of the segmented genome of Influenza A Virus”, the authors utilized sequential FISH staining to visualize all eight influenza viral segments within a single infected cell at a single timepoint during infection. In this approach, the authors used two dyes per vRNA segment in random order to validate their staining. Importantly, the authors imaged each segment twice with FISH probe sets conjugated to two different fluorophores to isolate the reproducible intracellular localization of each segment. The overall conclusions of this study are that, 1) the most abundant vRNP complexes are those with either 1 segment (rank 1) or 8 segments (rank 8) while Intermediate ranks are found at lower abundances, 2) higher ranked complexes are found more often in intact virions, 3) complexes rarely contain multiple copies of the same segment, and 4) disruption of multi-segment complexes (MSC) in human cells infected with an avian virus. Uncovering the mechanism by which IAV genome segments assemble is crucial to understanding reassortment and virion production. Visualizing these vRNA segments by FISH, especially all 8, is a powerful tool to study influenza selective assembly. The data presented supports the current model of IAV genome assembly where vRNP segments form bundles, the study also supports a newer model that the vRNP interaction clusters are flexible in nature as demonstrated with RNA-RNA interaction mapping in purified virions (PMID: 31332385). However, the study presented here has the advantage of knowing the precise three-dimensional location and composition of each vRNP bundle with in a cell yet they do not utilize this information to further refine the vRNP networks at a cellular level. Given the recent data demonstrating that heterogeneity in viral infection at a single cell level (PMID 29451492 and PMID 30181264), the authors should consider whether the multiple bundling routes are a consequence of including cells that do not contain all 8 segments into their analysis. Overall the visualization of all 8 segments within infected cells and in bound virions is unique and this technique serves as an additional strategy to confirm a model of influenza assembly where the vRNP form subcomplexes that are flexible as demonstrated by biochemical and recent machine learning approaches (PMID: 30689627). Overall, the clarity of the study could be improved with attention to the following issues. 1. The authors should provide experimental validation of the specificity for each probe set against other segments as well as any cRNA/vmRNA. Precise probe specificity is critical for accurate conclusions to be drawn from these data sets. It is also important to note that the authors state that the colocalization algorithm was a custom script but have not publicly shared the code or imaging data sets. 2. How do the networks defined by the authors in Fig 6 compare to the recently published RNA-RNA interactions for the Udorn or Wyoming/03/2003 H3N2 virus in PMID: 31332385. In particular, the lack of segments 7 and 8 in any of the initial complexes is intriguing and it may be due to the use of “all cells” in this analysis rather than cells that had an equal proportion of all 8 segments. Based on the data in Fig 3, segment 8 was under represented more often compared to the other segments and that may bias the networks constructed in Fig 6. 3. On page 8, the authors stated that, “we never obtained evidence for more than two copies of the same segment in an MSC; less than 5% of the MSCs contained two copies of the same segment”. However, volume binning for vRNA spots in this study was 300 x 1000 nm, when they reference that a vRNP is 50-150 nm. If this statement was derived from the FISH experiments, it would be impossible that single vRNA of a specific segment could be differentiated given the limits of their methods and imaging resolution of standard confocal microscopy. This conclusion should be experimentally validated using approaches similar to PMID: 22547828 or else do not make such a claim. 4. The current model of IAV MSC assembly suggests that segments assemble as MSCs translocate from the nuclei to the plasma membrane. If so, lower ranked complexes would be observed near the nucleus with higher ranked complexes near the plasma membrane. The authors should analyze ranks of complexes within various distances from the nuclei and/or plasma membrane. 5. In Fig. 5C, it appears that the frequency of MSCs containing S1 is greater than S2 and so on (frequency of S1>S2>S3>…S8). In virions with MSCs ranked less than 8, is the probability of which vRNP segments make it into these particles linked to segment size? In this study, are larger segments found more often in lower ranked MSCs compared to smaller segments during infection? This could be an artifact of the number of probes capable of binding to a given segment and should be acknowledged in the text. 6. In this study, the authors used a low-pathogenic avian IAV, A/Mallard, to study MSC assembly under non-permissive IAV infections. Given that infection with this virus is abortive in human A549 cells, the rationale for this experimental design is unclear. The replication kinetics of A/Mallard versus A/Panama are likely different and thus a 10 hpi cross-section between these infections may not be comparable. The complexity of this comparison makes these data difficult to interpret why the pattern of vRNP colocalization are different between these two virus infections. Reviewer #2: Remarks to the Author: Haralampiev, Prisner, et al. used a multiplexed FISH assay to investigate all 8 influenza vRNA segments in parallel within A549 cells infected with human or avian influenza A virus. These data support the prevailing model of influenza virus genome packaging in which each one of eight-influenza virus vRNPs are selectively bundled into multi-segment complexes and then subsequently incorporated into newly forming virions. The authors developed significant improvements to an existing approach in order to answer an important and interesting question pertinent to the influenza field with relevance to other viruses with segmented genomes. The data have the potential to advance current understaning of influenza A virus genome assembly, an area of high interest in the field. However, a more rigorous analysis and systematic presentation of the data is greatly needed. The current version of the manuscript is unclear and poorly organized, to the extent that it is difficult to fully evaluate the results. The authors should consider rectifying the following major and minor concerns: 1. The introduction would benefit from a more comprehensive summary regarding several topics: i) if you are going to present reassortment as an evolutionary benefit of genome segmentation, you should also mention fitness costs of reassortment ii) packaging model—if you are going to mention both random vs selective packaging models, you should present evidence for the prevailing model (selective) but note the limitations of the previous studies, iii) expand on the limitations of current understanding of influenza virus genome packaging mechanism and the signals that direct packaging. 2. Quantification of observed differences were included throughout the manuscript. For example, avoid vague phrasing such as “mostly a rather balanced ratio”, “approximately equal”, “slightly lower than average” and “highly similar” and quantify these differences using statistics. 3. In sub-section 2 of the results section, expand on the rationale for determining vRNA spot dimensions and exclusion of nuclear spots. Also reference data to support the statement that <5% of MSCs showed 2 copies of the same segment. 4. An important aspect of this study is the inclusion of human vs avian influenza A strains infecting a human cell line. The results would be more meaningful if presented in a way that cohesively and linearly presents these data (i.e., the figures should be presented chronologically in the text, e.g., figure 2C/D should be interpreted before analyzing Figure 3) with emphasis on comparing/contrasting human vs avian influenza A-infected cells. 5. Figure 3. Presentation of the data that shows the variance among the different cells (with statistical analysis) is needed. It is unclear whether row 1 represents an average over all cells (as mentioned in the text) or total number of MSCs or segments. Showing averages, with bars including error bars, would allow the authors to make meaningful comparisons between ranks rather than referring to the u-shaped distribution 6. Fig. 4. Rather than showing segment counts for each MSC for each cell, it would be more impactful to consolidate these data to show major trends (or lack thereof) to more clearly articulate the conclusion that there is no dominant MSC composition within each rank. 7. ‘Reliability of colocalization analysis’ sub-section of the results should be included earlier on in the text. These are important controls that should be presented before the data, but do not necessarily need a separate section and can be integrated throughout the text. 8. ‘Implications for IAV genome assembly’ sub-section should be moved to the Discussion/Conclusions section of the manuscript. 9. Please include panel letters on all figures for clarity throughout the figure legends and text. 10. Figure legends should not contain results or interpretation of the data shown, except for a descriptive title if warranted. 11. Include more descriptive y axis titles for Figure 2A/C (e.g., Cells infected with A/Panama or Cells infected with A/Mallard). 12. Include axis labels for Fig. 5C. If represented on the same graph, scale of y-axis needs to be consistent. Purple color for NP vRNA is not distinguishable in image in Panel A. Reviewer #3: Remarks to the Author: The paper by Haralampiev et al. investigates the bundling of the 8 influenza A virus (IAV) genome segments by a multiplexed FISH assay. The authors could provide first insight into the cytoplasmic segment composition and possible genome complex intermediates. The paper describes for the first time a robust FISH-based assay to study genome bundling of all IAV segments and thus provides an excellent basis for future studies. However, besides developing a robust FISH assays, the authors did not really apply their method to address some open questions in the field of genome packaging. Major points: 1) To convincingly demonstrate that the avian IAV used in this study is able to bundle correctly the authors should study the vRNP composition also in avian cells. 2) Packaging sequences in the viral genome are required for correct packaging of the 8 different genome segments into a virus particle. It is believed that these sequences orchestrate genome bundling. Thus, it is not clear why the authors did not investigate a packaging mutant virus. This would provide valuable information. 3) There are published data available suggesting that vRNP bundling increases en route to the plasma membrane. This information is lacking and if technical possible should be included. 4) There are Panama virus-infected cells were 8 bundles are not efficiently formed, because some segments are poorly expressed. This suggests that balanced genome expression levels are important. To address this point experimentally the authors might consoder to generate a virus that express less efficiently only one segment. Minor points: 1) Control experiments (page 15: Reliability….) is very difficult to understand. 2) In the supplement all analyzed cells (Panama and Mallard) can be shown. 3) Show pictures of genome distribution in Mallard-infected cells 4) Infection of human cells with the Mallard virus: are semi-infectious virus particles released from the cells? 5) Is the introduction of the human PB2 signature in the Mallard strain sufficient to restore genome bundling? 6) ...mediating interactions between vRNPs (10,13-23)Ref https://doi.org/10.1073/pnas.0437772100 is missing. 7) ...Fluorescence in situ hybridization (FISH) studies (24-26). Ref 24 is not accurate because that was FISH in particles. 8) Definition of packaging in the footnote: For the reviewer packaging is the incorporation of the genomes into viral particles. 9) ...In fact, considering all possible combinations of vRNPs, the random packaging model predicts that 98% of the MSCs would contain two or more copies of a distinct segment species: How do the authors come to 98%? 10) ... To assess the vRNP composition of mature A/Panama virions by MuSeq-FISH, virions present in virus stocks were allowed to bind to the surface of human A549 cells on ice followed by immediate chemical fixation. Interestingly, the frequency distribution of MSCs in A/Panama virions was clearly different to that found in A/Panama-infected cells being shifted to MSCs of high rank (Fig. 5): How valid is the comparison? Which virus stock was used? One after infection with a high MOI or one after infection with a low MOI? 11) The authors analyzed only one time point post infection and one infection dose. Is the vRNP composition different at earlier time points and at lower MOIs?	Reviewers' Comments: Reviewer #1: Remarks to the Author: In the manuscript entitled, “Selective, flexible packaging pathways of the segmented genome of Influenza A Virus”, the authors utilized sequential FISH staining to visualize all eight influenza viral segments within a single infected cell at a single timepoint during infection. In this approach, the authors used two dyes per vRNA segment in random order to validate their staining. Importantly, the authors imaged each segment twice with FISH probe sets conjugated to two different fluorophores to isolate the reproducible intracellular localization of each segment. The overall conclusions of this study are that, 1) the most abundant vRNP complexes are those with either 1 segment (rank 1) or 8 segments (rank 8) while Intermediate ranks are found at lower abundances, 2) higher ranked complexes are found more often in intact virions, 3) complexes rarely contain multiple copies of the same segment, and 4) disruption of multi-segment complexes (MSC) in human cells infected with an avian virus. Uncovering the mechanism by which IAV genome segments assemble is crucial to understanding reassortment and virion production. Visualizing these vRNA segments by FISH, especially all 8, is a powerful tool to study influenza selective assembly. The data presented supports the current model of IAV genome assembly where vRNP segments form bundles, the study also supports a newer model that the vRNP interaction clusters are flexible in nature as demonstrated with RNA-RNA interaction mapping in purified virions (PMID: 31332385). However, the study presented here has the advantage of knowing the precise three-dimensional location and composition of each vRNP bundle with in a cell yet they do not utilize this information to further refine the vRNP networks at a cellular level. Given the recent data demonstrating that heterogeneity in viral infection at a single cell level (PMID 29451492 and PMID 30181264), the authors should consider whether the multiple bundling routes are a consequence of including cells that do not contain all 8 segments into their analysis. Overall the visualization of all 8 segments within infected cells and in bound virions is unique and this technique serves as an additional strategy to confirm a model of influenza assembly where the vRNP form subcomplexes that are flexible as demonstrated by biochemical and recent machine learning approaches (PMID: 30689627). Overall, the clarity of the study could be improved with attention to the following issues. 1. The authors should provide experimental validation of the specificity for each probe set against other segments as well as any cRNA/vmRNA. Precise probe specificity is critical for accurate conclusions to be drawn from these data sets. It is also important to note that the authors state that the colocalization algorithm was a custom script but have not publicly shared the code or imaging data sets. 2. How do the networks defined by the authors in Fig 6 compare to the recently published RNA-RNA interactions for the Udorn or Wyoming/03/2003 H3N2 virus in PMID: 31332385. In particular, the lack of segments 7 and 8 in any of the initial complexes is intriguing and it may be due to the use of “all cells” in this analysis rather than cells that had an equal proportion of all 8 segments. Based on the data in Fig 3, segment 8 was under represented more often compared to the other segments and that may bias the networks constructed in Fig 6. 3. On page 8, the authors stated that, “we never obtained evidence for more than two copies of the same segment in an MSC; less than 5% of the MSCs contained two copies of the same segment”. However, volume binning for vRNA spots in this study was 300 x 1000 nm, when they reference that a vRNP is 50-150 nm. If this statement was derived from the FISH experiments, it would be impossible that single vRNA of a specific segment could be differentiated given the limits of their methods and imaging resolution of standard confocal microscopy. This conclusion should be experimentally validated using approaches similar to PMID: 22547828 or else do not make such a claim. 4. The current model of IAV MSC assembly suggests that segments assemble as MSCs translocate from the nuclei to the plasma membrane. If so, lower ranked complexes would be observed near the nucleus with higher ranked complexes near the plasma membrane. The authors should analyze ranks of complexes within various distances from the nuclei and/or plasma membrane. 5. In Fig. 5C, it appears that the frequency of MSCs containing S1 is greater than S2 and so on (frequency of S1>S2>S3>…S8). In virions with MSCs ranked less than 8, is the probability of which vRNP segments make it into these particles linked to segment size? In this study, are larger segments found more often in lower ranked MSCs compared to smaller segments during infection? This could be an artifact of the number of probes capable of binding to a given segment and should be acknowledged in the text. 6. In this study, the authors used a low-pathogenic avian IAV, A/Mallard, to study MSC assembly under non-permissive IAV infections. Given that infection with this virus is abortive in human A549 cells, the rationale for this experimental design is unclear. The replication kinetics of A/Mallard versus A/Panama are likely different and thus a 10 hpi cross-section between these infections may not be comparable. The complexity of this comparison makes these data difficult to interpret why the pattern of vRNP colocalization are different between these two virus infections. Reviewer #2: Remarks to the Author: Haralampiev, Prisner, et al. used a multiplexed FISH assay to investigate all 8 influenza vRNA segments in parallel within A549 cells infected with human or avian influenza A virus. These data support the prevailing model of influenza virus genome packaging in which each one of eight-influenza virus vRNPs are selectively bundled into multi-segment complexes and then subsequently incorporated into newly forming virions. The authors developed significant improvements to an existing approach in order to answer an important and interesting question pertinent to the influenza field with relevance to other viruses with segmented genomes. The data have the potential to advance current understaning of influenza A virus genome assembly, an area of high interest in the field. However, a more rigorous analysis and systematic presentation of the data is greatly needed. The current version of the manuscript is unclear and poorly organized, to the extent that it is difficult to fully evaluate the results. The authors should consider rectifying the following major and minor concerns: 1. The introduction would benefit from a more comprehensive summary regarding several topics: i) if you are going to present reassortment as an evolutionary benefit of genome segmentation, you should also mention fitness costs of reassortment ii) packaging model—if you are going to mention both random vs selective packaging models, you should present evidence for the prevailing model (selective) but note the limitations of the previous studies, iii) expand on the limitations of current understanding of influenza virus genome packaging mechanism and the signals that direct packaging. 2. Quantification of observed differences were included throughout the manuscript. For example, avoid vague phrasing such as “mostly a rather balanced ratio”, “approximately equal”, “slightly lower than average” and “highly similar” and quantify these differences using statistics. 3. In sub-section 2 of the results section, expand on the rationale for determining vRNA spot dimensions and exclusion of nuclear spots. Also reference data to support the statement that <5% of MSCs showed 2 copies of the same segment. 4. An important aspect of this study is the inclusion of human vs avian influenza A strains infecting a human cell line. The results would be more meaningful if presented in a way that cohesively and linearly presents these data (i.e., the figures should be presented chronologically in the text, e.g., figure 2C/D should be interpreted before analyzing Figure 3) with emphasis on comparing/contrasting human vs avian influenza A-infected cells. 5. Figure 3. Presentation of the data that shows the variance among the different cells (with statistical analysis) is needed. It is unclear whether row 1 represents an average over all cells (as mentioned in the text) or total number of MSCs or segments. Showing averages, with bars including error bars, would allow the authors to make meaningful comparisons between ranks rather than referring to the u-shaped distribution 6. Fig. 4. Rather than showing segment counts for each MSC for each cell, it would be more impactful to consolidate these data to show major trends (or lack thereof) to more clearly articulate the conclusion that there is no dominant MSC composition within each rank. 7. ‘Reliability of colocalization analysis’ sub-section of the results should be included earlier on in the text. These are important controls that should be presented before the data, but do not necessarily need a separate section and can be integrated throughout the text. 8. ‘Implications for IAV genome assembly’ sub-section should be moved to the Discussion/Conclusions section of the manuscript. 9. Please include panel letters on all figures for clarity throughout the figure legends and text. 10. Figure legends should not contain results or interpretation of the data shown, except for a descriptive title if warranted. 11. Include more descriptive y axis titles for Figure 2A/C (e.g., Cells infected with A/Panama or Cells infected with A/Mallard). 12. Include axis labels for Fig. 5C. If represented on the same graph, scale of y-axis needs to be consistent. Purple color for NP vRNA is not distinguishable in image in Panel A. Reviewer #3: Remarks to the Author: The paper by Haralampiev et al. investigates the bundling of the 8 influenza A virus (IAV) genome segments by a multiplexed FISH assay. The authors could provide first insight into the cytoplasmic segment composition and possible genome complex intermediates. The paper describes for the first time a robust FISH-based assay to study genome bundling of all IAV segments and thus provides an excellent basis for future studies. However, besides developing a robust FISH assays, the authors did not really apply their method to address some open questions in the field of genome packaging. Major points: 1) To convincingly demonstrate that the avian IAV used in this study is able to bundle correctly the authors should study the vRNP composition also in avian cells. 2) Packaging sequences in the viral genome are required for correct packaging of the 8 different genome segments into a virus particle. It is believed that these sequences orchestrate genome bundling. Thus, it is not clear why the authors did not investigate a packaging mutant virus. This would provide valuable information. 3) There are published data available suggesting that vRNP bundling increases en route to the plasma membrane. This information is lacking and if technical possible should be included. 4) There are Panama virus-infected cells were 8 bundles are not efficiently formed, because some segments are poorly expressed. This suggests that balanced genome expression levels are important. To address this point experimentally the authors might consoder to generate a virus that express less efficiently only one segment. Minor points: 1) Control experiments (page 15: Reliability….) is very difficult to understand. 2) In the supplement all analyzed cells (Panama and Mallard) can be shown. 3) Show pictures of genome distribution in Mallard-infected cells 4) Infection of human cells with the Mallard virus: are semi-infectious virus particles released from the cells? 5) Is the introduction of the human PB2 signature in the Mallard strain sufficient to restore genome bundling? 6) ...mediating interactions between vRNPs (10,13-23)Ref https://doi.org/10.1073/pnas.0437772100 is missing. 7) ...Fluorescence in situ hybridization (FISH) studies (24-26). Ref 24 is not accurate because that was FISH in particles. 8) Definition of packaging in the footnote: For the reviewer packaging is the incorporation of the genomes into viral particles. 9) ...In fact, considering all possible combinations of vRNPs, the random packaging model predicts that 98% of the MSCs would contain two or more copies of a distinct segment species: How do the authors come to 98%? 10) ... To assess the vRNP composition of mature A/Panama virions by MuSeq-FISH, virions present in virus stocks were allowed to bind to the surface of human A549 cells on ice followed by immediate chemical fixation. Interestingly, the frequency distribution of MSCs in A/Panama virions was clearly different to that found in A/Panama-infected cells being shifted to MSCs of high rank (Fig. 5): How valid is the comparison? Which virus stock was used? One after infection with a high MOI or one after infection with a low MOI? 11) The authors analyzed only one time point post infection and one infection dose. Is the vRNP composition different at earlier time points and at lower MOIs?	TPR	NC_BS_BIOTEC_001_RR1.txt	{ "criticism": 12, "example": 13, "importance_and_relevance": 6, "materials_and_methods": 66, "praise": 6, "presentation_and_reporting": 24, "results_and_discussion": 16, "suggestion_and_solution": 30, "total": 112 }	{ "criticism": 0.10714285714285714, "example": 0.11607142857142858, "importance_and_relevance": 0.05357142857142857, "materials_and_methods": 0.5892857142857143, "praise": 0.05357142857142857, "presentation_and_reporting": 0.21428571428571427, "results_and_discussion": 0.14285714285714285, "suggestion_and_solution": 0.26785714285714285 }	1.544643	[ { "content": "Please review this paper.", "role": "user" }, { "content": "Reviewers' Comments: \n \nReviewer #1: \nRemarks to the Author: \nIn the manuscript entitled, “Selective, flexible packaging pathways of the segmented genome of \nInfluenza A Virus”, the authors utilized sequential FISH staining to visualize all eight influenza viral \nsegments within a single infected cell at a single timepoint during infection. In this approach, the \nauthors used two dyes per vRNA segment in random order to validate their staining. Importantly, the \nauthors imaged each segment twice with FISH probe sets conjugated to two different fluorophores to \nisolate the reproducible intracellular localization of each segment. The overall conclusions of this study \nare that, 1) the most abundant vRNP complexes are those with either 1 segment (rank 1) or 8 \nsegments (rank 8) while Intermediate ranks are found at lower abundances, 2) higher ranked \ncomplexes are found more often in intact virions, 3) complexes rarely contain multiple copies of the \nsame segment, and 4) disruption of multi-segment complexes (MSC) in human cells infected with an \navian virus. \nUncovering the mechanism by which IAV genome segments assemble is crucial to understanding \nreassortment and virion production. Visualizing these vRNA segments by FISH, especially all 8, is a \npowerful tool to study influenza selective assembly. The data presented supports the current model of \nIAV genome assembly where vRNP segments form bundles, the study also supports a newer model \nthat the vRNP interaction clusters are flexible in nature as demonstrated with RNA-RNA interaction \nmapping in purified virions (PMID: 31332385). However, the study presented here has the advantage \nof knowing the precise three-dimensional location and composition of each vRNP bundle with in a cell \nyet they do not utilize this information to further refine the vRNP networks at a cellular level. Given \nthe recent data demonstrating that heterogeneity in viral infection at a single cell level (PMID \n29451492 and PMID 30181264), the authors should consider whether the multiple bundling routes are \na consequence of including cells that do not contain all 8 segments into their analysis. Overall the \nvisualization of all 8 segments within infected cells and in bound virions is unique and this technique \nserves as an additional strategy to confirm a model of influenza assembly where the vRNP form \nsubcomplexes that are flexible as demonstrated by biochemical and recent machine learning \napproaches (PMID: 30689627). \nOverall, the clarity of the study could be improved with attention to the following issues. \n1. The authors should provide experimental validation of the specificity for each probe set against \nother segments as well as any cRNA/vmRNA. Precise probe specificity is critical for accurate \nconclusions to be drawn from these data sets. It is also important to note that the authors state that \nthe colocalization algorithm was a custom script but have not publicly shared the code or imaging data \nsets. \n2. How do the networks defined by the authors in Fig 6 compare to the recently published RNA-RNA \ninteractions for the Udorn or Wyoming/03/2003 H3N2 virus in PMID: 31332385. In particular, the lack \nof segments 7 and 8 in any of the initial complexes is intriguing and it may be due to the use of “all \ncells” in this analysis rather than cells that had an equal proportion of all 8 segments. Based on the \ndata in Fig 3, segment 8 was under represented more often compared to the other segments and that \nmay bias the networks constructed in Fig 6. \n3. On page 8, the authors stated that, “we never obtained evidence for more than two copies of the \nsame segment in an MSC; less than 5% of the MSCs contained two copies of the same segment”. \nHowever, volume binning for vRNA spots in this study was 300 x 1000 nm, when they reference that a \nvRNP is 50-150 nm. If this statement was derived from the FISH experiments, it would be impossible \nthat single vRNA of a specific segment could be differentiated given the limits of their methods and \nimaging resolution of standard confocal microscopy. This conclusion should be experimentally \nvalidated using approaches similar to PMID: 22547828 or else do not make such a claim. \n4. The current model of IAV MSC assembly suggests that segments assemble as MSCs translocate \nfrom the nuclei to the plasma membrane. If so, lower ranked complexes would be observed near the \n\n\fnucleus with higher ranked complexes near the plasma membrane. The authors should analyze ranks \nof complexes within various distances from the nuclei and/or plasma membrane. \n5. In Fig. 5C, it appears that the frequency of MSCs containing S1 is greater than S2 and so on \n(frequency of S1>S2>S3>…S8). In virions with MSCs ranked less than 8, is the probability of which \nvRNP segments make it into these particles linked to segment size? In this study, are larger segments \nfound more often in lower ranked MSCs compared to smaller segments during infection? This could be \nan artifact of the number of probes capable of binding to a given segment and should be \nacknowledged in the text. \n6. In this study, the authors used a low-pathogenic avian IAV, A/Mallard, to study MSC assembly \nunder non-permissive IAV infections. Given that infection with this virus is abortive in human A549 \ncells, the rationale for this experimental design is unclear. The replication kinetics of A/Mallard versus \nA/Panama are likely different and thus a 10 hpi cross-section between these infections may not be \ncomparable. The complexity of this comparison makes these data difficult to interpret why the pattern \nof vRNP colocalization are different between these two virus infections. \n \n \n \nReviewer #2: \nRemarks to the Author: \nHaralampiev, Prisner, et al. used a multiplexed FISH assay to investigate all 8 influenza vRNA \nsegments in parallel within A549 cells infected with human or avian influenza A virus. These data \nsupport the prevailing model of influenza virus genome packaging in which each one of eight-influenza \nvirus vRNPs are selectively bundled into multi-segment complexes and then subsequently incorporated \ninto newly forming virions. \n \nThe authors developed significant improvements to an existing approach in order to answer an \nimportant and interesting question pertinent to the influenza field with relevance to other viruses with \nsegmented genomes. The data have the potential to advance current understaning of influenza A virus \ngenome assembly, an area of high interest in the field. However, a more rigorous analysis and \nsystematic presentation of the data is greatly needed. The current version of the manuscript is unclear \nand poorly organized, to the extent that it is difficult to fully evaluate the results. The authors should \nconsider rectifying the following major and minor concerns: \n \n1. The introduction would benefit from a more comprehensive summary regarding several topics: i) if \nyou are going to present reassortment as an evolutionary benefit of genome segmentation, you should \nalso mention fitness costs of reassortment ii) packaging model—if you are going to mention both \nrandom vs selective packaging models, you should present evidence for the prevailing model \n(selective) but note the limitations of the previous studies, iii) expand on the limitations of current \nunderstanding of influenza virus genome packaging mechanism and the signals that direct packaging. \n2. Quantification of observed differences were included throughout the manuscript. For example, avoid \nvague phrasing such as “mostly a rather balanced ratio”, “approximately equal”, “slightly lower than \naverage” and “highly similar” and quantify these differences using statistics. \n3. In sub-section 2 of the results section, expand on the rationale for determining vRNA spot \ndimensions and exclusion of nuclear spots. Also reference data to support the statement that <5% of \nMSCs showed 2 copies of the same segment. \n4. An important aspect of this study is the inclusion of human vs avian influenza A strains infecting a \nhuman cell line. The results would be more meaningful if presented in a way that cohesively and \nlinearly presents these data (i.e., the figures should be presented chronologically in the text, e.g., \nfigure 2C/D should be interpreted before analyzing Figure 3) with emphasis on comparing/contrasting \nhuman vs avian influenza A-infected cells. \n5. Figure 3. Presentation of the data that shows the variance among the different cells (with statistical \n\n\fanalysis) is needed. It is unclear whether row 1 represents an average over all cells (as mentioned in \nthe text) or total number of MSCs or segments. Showing averages, with bars including error bars, \nwould allow the authors to make meaningful comparisons between ranks rather than referring to the \nu-shaped distribution \n6. Fig. 4. Rather than showing segment counts for each MSC for each cell, it would be more impactful \nto consolidate these data to show major trends (or lack thereof) to more clearly articulate the \nconclusion that there is no dominant MSC composition within each rank. \n7. ‘Reliability of colocalization analysis’ sub-section of the results should be included earlier on in the \ntext. These are important controls that should be presented before the data, but do not necessarily \nneed a separate section and can be integrated throughout the text. \n8. ‘Implications for IAV genome assembly’ sub-section should be moved to the Discussion/Conclusions \nsection of the manuscript. \n9. Please include panel letters on all figures for clarity throughout the figure legends and text. \n10. Figure legends should not contain results or interpretation of the data shown, except for a \ndescriptive title if warranted. \n11. Include more descriptive y axis titles for Figure 2A/C (e.g., Cells infected with A/Panama or Cells \ninfected with A/Mallard). \n12. Include axis labels for Fig. 5C. If represented on the same graph, scale of y-axis needs to be \nconsistent. Purple color for NP vRNA is not distinguishable in image in Panel A. \n \n \n \nReviewer #3: \nRemarks to the Author: \nThe paper by Haralampiev et al. investigates the bundling of the 8 influenza A virus (IAV) genome \nsegments by a multiplexed FISH assay. The authors could provide first insight into the cytoplasmic \nsegment composition and possible genome complex intermediates. \n \nThe paper describes for the first time a robust FISH-based assay to study genome bundling of all IAV \nsegments and thus provides an excellent basis for future studies. However, besides developing a \nrobust FISH assays, the authors did not really apply their method to address some open questions in \nthe field of genome packaging. \n \nMajor points: \n \n1) To convincingly demonstrate that the avian IAV used in this study is able to bundle correctly the \nauthors should study the vRNP composition also in avian cells. \n \n2) Packaging sequences in the viral genome are required for correct packaging of the 8 different \ngenome segments into a virus particle. It is believed that these sequences orchestrate genome \nbundling. Thus, it is not clear why the authors did not investigate a packaging mutant virus. This \nwould provide valuable information. \n \n3) There are published data available suggesting that vRNP bundling increases en route to the plasma \nmembrane. This information is lacking and if technical possible should be included. \n \n4) There are Panama virus-infected cells were 8 bundles are not efficiently formed, because some \nsegments are poorly expressed. This suggests that balanced genome expression levels are important. \nTo address this point experimentally the authors might consoder to generate a virus that express less \nefficiently only one segment. \n \n\n\fMinor points: \n \n1) Control experiments (page 15: Reliability….) is very difficult to understand. \n2) In the supplement all analyzed cells (Panama and Mallard) can be shown. \n3) Show pictures of genome distribution in Mallard-infected cells \n4) Infection of human cells with the Mallard virus: are semi-infectious virus particles released from the \ncells? \n5) Is the introduction of the human PB2 signature in the Mallard strain sufficient to restore genome \nbundling? \n6) ...mediating interactions between vRNPs (10,13-23)Ref https://doi.org/10.1073/pnas.0437772100 \nis missing. \n7) ...Fluorescence in situ hybridization (FISH) studies (24-26). Ref 24 is not accurate because that \nwas FISH in particles. \n8) Definition of packaging in the footnote: For the reviewer packaging is the incorporation of the \ngenomes into viral particles. \n9) ...In fact, considering all possible combinations of vRNPs, the random packaging model predicts \nthat 98% of the MSCs would contain two or more copies of a distinct segment species: How do the \nauthors come to 98%? \n10) ... To assess the vRNP composition of mature A/Panama virions by MuSeq-FISH, virions present in \nvirus stocks were allowed to bind to the surface of human A549 cells on ice followed by immediate \nchemical fixation. Interestingly, the frequency distribution of MSCs in A/Panama virions was clearly \ndifferent to that found in A/Panama-infected cells being \nshifted to MSCs of high rank (Fig. 5): How valid is the comparison? Which virus stock was used? One \nafter infection with a high MOI or one after infection with a low MOI? \n11) The authors analyzed only one time point post infection and one infection dose. Is the vRNP \ncomposition different at earlier time points and at lower MOIs? \n\n\f\f\n", "role": "assistant" } ]	2,020
BRCA1-associated structural variations are a consequence of polymerase theta-mediated	Reviewer #1 (Remarks to the Author): In this investigation, Kamp et al. report that C. elegans lacking the mammalian BRCA1 and BARD1 orthologs accumulate mutations similar to those seen in human brca1-deficient tumors. These mutations include relative small (<30 bp) deletions and large (1-1000kb) tandem duplications, both of which have microhomologous sequences at the breakpoints and/or small insertions. In addition, an increase in base substitutions is observed. They also show that polymerase theta mediated end joining, but not classical non-homologous end joining, is responsible for the majority of the deletions and tandem duplications. While the role of POLQ in promoting deletions with accompanying microhomology is well-established, its involvement in the formation of tandem duplications has not been reported. This is the novel aspect of this paper. The authors propose that the tandem duplications could result from aborted homologous recombination, when one strand invades into a homologous template and DNA synthesis occurs, but the nascent strand is unable to anneal with the other broken end. They argue that this would be more likely in the absence of BRC-1 and BRD-1, due to a defect in resection. My major concern is that while this model is attractive, the authors provide no direct support. They do cite other analyses showing that tandem duplications are common in human tumors with brca1 mutations, but not with brca2 mutations, but this is only correlative data. Because the main novelty of this study lies in the proposed mechanism of tandem duplication formation, some experimental support of their model is needed. The most obvious prediction, that a brc-2 mutation would prevent their formation, cannot be tested due to the necessity of brc-2 for meiotic recombination. Perhaps the authors can look in an mre11 brc-1 double mutant (although these have reduced fertility)—the prediction would be that tandem duplications should also be more frequent due to a defect in resection. Alternatively, a mutation or condition that increases the amount of resection at double- strand breaks should decrease the frequency of tandem duplications. The inclusion of additional data to elucidate the conditions which allow TMEJ-mediated tandem duplication formation would significantly strengthen the manuscript. Other points: 1. The Scully lab proposed an alternative model for tandem duplication formation involving replication restart-bypass in brca1 mutant cells (Willis et al., 2017, PMID 29168504). While the current manuscript does reference this work, a more through discussion of this model is warranted. 2. The model for TMEJ repair following aborted HR is not novel. This was suggested at least two previous publications: Wyatt et al., 2016 (PMID 27453047) and Chan et al., 2010 (PMID 20617203). These papers should be cited when the model is proposed. 3. A synergism between HR and TMEJ in the repair of IR-induced damage was previously reported in Chan et al., 2010. In this instance, mutation of both RAD51 and POLQ resulted in hypersensitivity to IR, similar to that reported in the current study. This should be cited. 4. Lines 98-99: More detail about the type of somatic cells and how HDR was monitored should be provided here, so the readers don’t have to rely only on Figure S1. 5. Line 103: Are the accompanying insertions templated from flanking sequences, as has been reported for other TMEJ junctions? 6. Line 110: I seem to recall that earlier papers from this group showed that deletions that occur at G4 sequences almost always involved 1 nt of microhomology. Is the proportion different in brc-1 mutants and if so, why might this be? 7. Lines 174-175: this statement is oversimplified, as it is true only for relatively small deletions (Fig. 3). 8. Lines 189-198: This paragraph is confusing as written. Are the authors arguing that TMEJ is more common in C. elegans than in mammals? The phrase “provides an unlikely explanation” makes this unclear. 9. Line 263: The authors claim that the reliance on short microhomologies sets TMEJ apart from MMEJ, but there are no consistent guidelines for MMEJ in the literature-- particularly as it is defined differently in yeast vs. other organisms, as the next sentence points out. MMEJ is often used as a general term for the use of any amount of microhomology during break repair. Therefore, it would be better to use different terminology here. Reviewer #2 (Remarks to the Author): The work of Kamp et al. proposes a key role for the DNA polymerase theta in generating mutation patterns that are known to be characteristic of homologous recombination (HR)-deficient human cells, using a worm germline mutagenesis model. These are important findings because they shed light on the mechanisms underlying mutagenesis under HR deficiency. An implication of these results is that they further suggest relevance of therapeutically targeting DNA polymerase theta in those HR- deficient tumors that exhibit certain mutation patterns. Overall the methodology appears quite solid and the presentation balanced. I do have a few queries that need to be addressed. 1. The conclusions have some potential to be affected by technical issues stemming from mutation calling from short reads, which is error-prone in case of calling short indels and moreso in case of larger structural variants. - only a single tool to call structural variants is used (Pindel); more modern tools (e.g. Manta, Gridss etc) may give more accurate results, e.g. in terms of placing the breakpoint, or in terms of false negative rates; note that the latter cannot be prevented by manual inspection of calls which they have done. Running an additional caller would help ensure the results are not skewed towards biases of a single algorithm. In particular different tools may have different biases with respect to lengths/types of variants they can detect, which can be relevant for the statements such as line 117 "The rate of TDs in brc-1 and brd-1 mutants is approximately tenfold lower than the rate of deletions in these mutants." - would the conclusions be affected if limiting to variant calls with higher confidence scores and/or in less difficult to align (less repetitive) genome regions? - please state average and/or median depth of sequencing in the methods. 2. Data presentation issues. - there is much overplotting of the dots in 1A/1B/2A/2B/3A/3B which makes it hard to judge the density of points of different colors (microhomology lengths), affecting interpretation of results. I suggest to sort the points (x axis) by # nucleotides of microhomology, within each genotype. - line 142 "we observed a small increase in deletion mutagenesis in cku-80 and lig-4" and line 146 "base substitution rate is mildly increased in cku-80 and lig-4 mutants as compared to wildtype". I would not say this is a small nor mild increase (>=2 fold in Figs 2C/2E). In 2E, the SNV rate increase from NHEJ KOs is bigger than for brc-1 KO. Consider rephrasing. - please draw two-sided error bars on bar charts - mutation rates flanking the breaks do not appear to be elevated, arguing against common use of an error-prone polymerase to extend past pol theta activity. Pol theta itself does not appear to contribute many SNVs according to Fig 3e. Thus currently the cause of higher SNV mutation rates in brc-1/brd-1 backgrounds is unclear. Would looking at the trinucleotide mutation spectrum, and examining correlations to known mutation signatures from human tumors, be helpful in elucidating this? 3. Multiple lines of animals k.o. for various genes were propagated, and mutation data pooled together for analysis. Are the results overall consistent between the independant lines? Potentially some of the observations might be driven by an acquired mutator in one of the lines (e.g. the increased SNV rate in NHEJ deficient lines). 4. Regarding negative selection. Lines 179-181 "Deletions of that size will affect multiple genes in a gene-dense organism suchas C. elegans and their presence are likely counterselected for in animal propagation experiments". It is well possible that many of the smaller deletions are also under selection. Please discuss and/or address with analyses if this has the potential to bias results (e.g. there might be confounding between chromosomal locations of fitness genes and the chromosomal location bias of activity of the various HR failure-related mutational processes). 5. A 1 nt microhomology has a rather high probability of occuring at random (depending on the local nucleotide composition). On the one hand, this affects interpretation of the data shown in plots, where 1 nt microhomologies, often arising at random, are lumped together with 2 and 3 nt microhomologies, which are more likely to stem from real signal. On the other hand, this can affect interpretation of statements such as line 110: "These deletions are also characterized by an overrepresentation of micro-homology: 58% of deletions had at least one nucleotide that could be mapped to either junction". A baseline estimate of microhomology should be provided, if relying on 1 nt matches to call MHs. Minor. - typos: "undistinguisable", "we have here describe (~10 kb)". Line 169 might need to refer to Fig 3A. - line 174: "Our data demonstrate that all deletions with ≤6 nt of homology are the result of polymerase theta on DSBs". This is correct, however if I understand correctly this 6nt threshold might be better placed at a higher number of nucleotides, depending on the lengths of the microhomology segments in polq-1 mutants in Fig 3A, which are not evident from the figure. - line 107: "We found no significant difference in deletion rate between brc-1 and brd-1 (p=0.236), arguing against a BRC-1-independent role for BRD-1 in deletion prevention, and vice versa". I would not say this is necessarily very strong evidence, in the absence of experiments on the brc-1 brd-1 double mutant, which they did not perform.	Reviewer #1 (Remarks to the Author): In this investigation, Kamp et al. report that C. elegans lacking the mammalian BRCA1 and BARD1 orthologs accumulate mutations similar to those seen in human brca1-deficient tumors. These mutations include relative small (<30 bp) deletions and large (1-1000kb) tandem duplications, both of which have microhomologous sequences at the breakpoints and/or small insertions. In addition, an increase in base substitutions is observed. They also show that polymerase theta mediated end joining, but not classical non-homologous end joining, is responsible for the majority of the deletions and tandem duplications. While the role of POLQ in promoting deletions with accompanying microhomology is well-established, its involvement in the formation of tandem duplications has not been reported. This is the novel aspect of this paper. The authors propose that the tandem duplications could result from aborted homologous recombination, when one strand invades into a homologous template and DNA synthesis occurs, but the nascent strand is unable to anneal with the other broken end. They argue that this would be more likely in the absence of BRC-1 and BRD-1, due to a defect in resection. My major concern is that while this model is attractive, the authors provide no direct support. They do cite other analyses showing that tandem duplications are common in human tumors with brca1 mutations, but not with brca2 mutations, but this is only correlative data. Because the main novelty of this study lies in the proposed mechanism of tandem duplication formation, some experimental support of their model is needed. The most obvious prediction, that a brc-2 mutation would prevent their formation, cannot be tested due to the necessity of brc-2 for meiotic recombination. Perhaps the authors can look in an mre11 brc-1 double mutant (although these have reduced fertility)—the prediction would be that tandem duplications should also be more frequent due to a defect in resection. Alternatively, a mutation or condition that increases the amount of resection at double- strand breaks should decrease the frequency of tandem duplications. The inclusion of additional data to elucidate the conditions which allow TMEJ-mediated tandem duplication formation would significantly strengthen the manuscript. Other points: 1. The Scully lab proposed an alternative model for tandem duplication formation involving replication restart-bypass in brca1 mutant cells (Willis et al., 2017, PMID 29168504). While the current manuscript does reference this work, a more through discussion of this model is warranted. 2. The model for TMEJ repair following aborted HR is not novel. This was suggested at least two previous publications: Wyatt et al., 2016 (PMID 27453047) and Chan et al., 2010 (PMID 20617203). These papers should be cited when the model is proposed. 3. A synergism between HR and TMEJ in the repair of IR-induced damage was previously reported in Chan et al., 2010. In this instance, mutation of both RAD51 and POLQ resulted in hypersensitivity to IR, similar to that reported in the current study. This should be cited. 4. Lines 98-99: More detail about the type of somatic cells and how HDR was monitored should be provided here, so the readers don’t have to rely only on Figure S1. 5. Line 103: Are the accompanying insertions templated from flanking sequences, as has been reported for other TMEJ junctions? 6. Line 110: I seem to recall that earlier papers from this group showed that deletions that occur at G4 sequences almost always involved 1 nt of microhomology. Is the proportion different in brc-1 mutants and if so, why might this be? 7. Lines 174-175: this statement is oversimplified, as it is true only for relatively small deletions (Fig. 3). 8. Lines 189-198: This paragraph is confusing as written. Are the authors arguing that TMEJ is more common in C. elegans than in mammals? The phrase “provides an unlikely explanation” makes this unclear. 9. Line 263: The authors claim that the reliance on short microhomologies sets TMEJ apart from MMEJ, but there are no consistent guidelines for MMEJ in the literature-- particularly as it is defined differently in yeast vs. other organisms, as the next sentence points out. MMEJ is often used as a general term for the use of any amount of microhomology during break repair. Therefore, it would be better to use different terminology here. Reviewer #2 (Remarks to the Author): The work of Kamp et al. proposes a key role for the DNA polymerase theta in generating mutation patterns that are known to be characteristic of homologous recombination (HR)-deficient human cells, using a worm germline mutagenesis model. These are important findings because they shed light on the mechanisms underlying mutagenesis under HR deficiency. An implication of these results is that they further suggest relevance of therapeutically targeting DNA polymerase theta in those HR- deficient tumors that exhibit certain mutation patterns. Overall the methodology appears quite solid and the presentation balanced. I do have a few queries that need to be addressed. 1. The conclusions have some potential to be affected by technical issues stemming from mutation calling from short reads, which is error-prone in case of calling short indels and moreso in case of larger structural variants. - only a single tool to call structural variants is used (Pindel); more modern tools (e.g. Manta, Gridss etc) may give more accurate results, e.g. in terms of placing the breakpoint, or in terms of false negative rates; note that the latter cannot be prevented by manual inspection of calls which they have done. Running an additional caller would help ensure the results are not skewed towards biases of a single algorithm. In particular different tools may have different biases with respect to lengths/types of variants they can detect, which can be relevant for the statements such as line 117 "The rate of TDs in brc-1 and brd-1 mutants is approximately tenfold lower than the rate of deletions in these mutants." - would the conclusions be affected if limiting to variant calls with higher confidence scores and/or in less difficult to align (less repetitive) genome regions? - please state average and/or median depth of sequencing in the methods. 2. Data presentation issues. - there is much overplotting of the dots in 1A/1B/2A/2B/3A/3B which makes it hard to judge the density of points of different colors (microhomology lengths), affecting interpretation of results. I suggest to sort the points (x axis) by # nucleotides of microhomology, within each genotype. - line 142 "we observed a small increase in deletion mutagenesis in cku-80 and lig-4" and line 146 "base substitution rate is mildly increased in cku-80 and lig-4 mutants as compared to wildtype". I would not say this is a small nor mild increase (>=2 fold in Figs 2C/2E). In 2E, the SNV rate increase from NHEJ KOs is bigger than for brc-1 KO. Consider rephrasing. - please draw two-sided error bars on bar charts - mutation rates flanking the breaks do not appear to be elevated, arguing against common use of an error-prone polymerase to extend past pol theta activity. Pol theta itself does not appear to contribute many SNVs according to Fig 3e. Thus currently the cause of higher SNV mutation rates in brc-1/brd-1 backgrounds is unclear. Would looking at the trinucleotide mutation spectrum, and examining correlations to known mutation signatures from human tumors, be helpful in elucidating this? 3. Multiple lines of animals k.o. for various genes were propagated, and mutation data pooled together for analysis. Are the results overall consistent between the independant lines? Potentially some of the observations might be driven by an acquired mutator in one of the lines (e.g. the increased SNV rate in NHEJ deficient lines). 4. Regarding negative selection. Lines 179-181 "Deletions of that size will affect multiple genes in a gene-dense organism suchas C. elegans and their presence are likely counterselected for in animal propagation experiments". It is well possible that many of the smaller deletions are also under selection. Please discuss and/or address with analyses if this has the potential to bias results (e.g. there might be confounding between chromosomal locations of fitness genes and the chromosomal location bias of activity of the various HR failure-related mutational processes). 5. A 1 nt microhomology has a rather high probability of occuring at random (depending on the local nucleotide composition). On the one hand, this affects interpretation of the data shown in plots, where 1 nt microhomologies, often arising at random, are lumped together with 2 and 3 nt microhomologies, which are more likely to stem from real signal. On the other hand, this can affect interpretation of statements such as line 110: "These deletions are also characterized by an overrepresentation of micro-homology: 58% of deletions had at least one nucleotide that could be mapped to either junction". A baseline estimate of microhomology should be provided, if relying on 1 nt matches to call MHs. Minor. - typos: "undistinguisable", "we have here describe (~10 kb)". Line 169 might need to refer to Fig 3A. - line 174: "Our data demonstrate that all deletions with ≤6 nt of homology are the result of polymerase theta on DSBs". This is correct, however if I understand correctly this 6nt threshold might be better placed at a higher number of nucleotides, depending on the lengths of the microhomology segments in polq-1 mutants in Fig 3A, which are not evident from the figure. - line 107: "We found no significant difference in deletion rate between brc-1 and brd-1 (p=0.236), arguing against a BRC-1-independent role for BRD-1 in deletion prevention, and vice versa". I would not say this is necessarily very strong evidence, in the absence of experiments on the brc-1 brd-1 double mutant, which they did not perform.	TPR	NC_BS_BIOTEC_002_RR1.txt	{ "criticism": 12, "example": 17, "importance_and_relevance": 6, "materials_and_methods": 40, "praise": 4, "presentation_and_reporting": 16, "results_and_discussion": 14, "suggestion_and_solution": 23, "total": 96 }	{ "criticism": 0.125, "example": 0.17708333333333334, "importance_and_relevance": 0.0625, "materials_and_methods": 0.4166666666666667, "praise": 0.041666666666666664, "presentation_and_reporting": 0.16666666666666666, "results_and_discussion": 0.14583333333333334, "suggestion_and_solution": 0.23958333333333334 }	1.375	[ { "content": "Please review this paper.", "role": "user" }, { "content": "\n \nReviewer #1 (Remarks to the Author): \n \nIn this investigation, Kamp et al. report that C. elegans lacking the mammalian BRCA1 and BARD1 \northologs accumulate mutations similar to those seen in human brca1-deficient tumors. These \nmutations include relative small (<30 bp) deletions and large (1-1000kb) tandem duplications, both of \nwhich have microhomologous sequences at the breakpoints and/or small insertions. In addition, an \nincrease in base substitutions is observed. They also show that polymerase theta mediated end \njoining, but not classical non-homologous end joining, is responsible for the majority of the deletions \nand tandem duplications. \n \nWhile the role of POLQ in promoting deletions with accompanying microhomology is well-established, \nits involvement in the formation of tandem duplications has not been reported. This is the novel \naspect of this paper. The authors propose that the tandem duplications could result from aborted \nhomologous recombination, when one strand invades into a homologous template and DNA synthesis \noccurs, but the nascent strand is unable to anneal with the other broken end. They argue that this \nwould be more likely in the absence of BRC-1 and BRD-1, due to a defect in resection. \n \nMy major concern is that while this model is attractive, the authors provide no direct support. They do \ncite other analyses showing that tandem duplications are common in human tumors with brca1 \nmutations, but not with brca2 mutations, but this is only correlative data. Because the main novelty of \nthis study lies in the proposed mechanism of tandem duplication formation, some experimental \nsupport of their model is needed. The most obvious prediction, that a brc-2 mutation would prevent \ntheir formation, cannot be tested due to the necessity of brc-2 for meiotic recombination. Perhaps the \nauthors can look in an mre11 brc-1 double mutant (although these have reduced fertility)—the \nprediction would be that tandem duplications should also be more frequent due to a defect in \nresection. Alternatively, a mutation or condition that increases the amount of resection at double-\nstrand breaks should decrease the frequency of tandem duplications. The inclusion of additional data \nto elucidate the conditions which allow TMEJ-mediated tandem duplication formation would \nsignificantly strengthen the manuscript. \n \nOther points: \n \n1. The Scully lab proposed an alternative model for tandem duplication formation involving replication \nrestart-bypass in brca1 mutant cells (Willis et al., 2017, PMID 29168504). While the current \nmanuscript does reference this work, a more through discussion of this model is warranted. \n \n2. The model for TMEJ repair following aborted HR is not novel. This was suggested at least two \nprevious publications: Wyatt et al., 2016 (PMID 27453047) and Chan et al., 2010 (PMID 20617203). \nThese papers should be cited when the model is proposed. \n \n3. A synergism between HR and TMEJ in the repair of IR-induced damage was previously reported in \nChan et al., 2010. In this instance, mutation of both RAD51 and POLQ resulted in hypersensitivity to \nIR, similar to that reported in the current study. This should be cited. \n \n4. Lines 98-99: More detail about the type of somatic cells and how HDR was monitored should be \nprovided here, so the readers don’t have to rely only on Figure S1. \n \n5. Line 103: Are the accompanying insertions templated from flanking sequences, as has been \nreported for other TMEJ junctions? \n \n6. Line 110: I seem to recall that earlier papers from this group showed that deletions that occur at \nG4 sequences almost always involved 1 nt of microhomology. Is the proportion different in brc-1 \nmutants and if so, why might this be? \n\n\f \n7. Lines 174-175: this statement is oversimplified, as it is true only for relatively small deletions (Fig. \n3). \n \n8. Lines 189-198: This paragraph is confusing as written. Are the authors arguing that TMEJ is more \ncommon in C. elegans than in mammals? The phrase “provides an unlikely explanation” makes this \nunclear. \n \n9. Line 263: The authors claim that the reliance on short microhomologies sets TMEJ apart from MMEJ, \nbut there are no consistent guidelines for MMEJ in the literature-- particularly as it is defined \ndifferently in yeast vs. other organisms, as the next sentence points out. MMEJ is often used as a \ngeneral term for the use of any amount of microhomology during break repair. Therefore, it would be \nbetter to use different terminology here. \n \n \nReviewer #2 (Remarks to the Author): \n \nThe work of Kamp et al. proposes a key role for the DNA polymerase theta in generating mutation \npatterns that are known to be characteristic of homologous recombination (HR)-deficient human cells, \nusing a worm germline mutagenesis model. These are important findings because they shed light on \nthe mechanisms underlying mutagenesis under HR deficiency. An implication of these results is that \nthey further suggest relevance of therapeutically targeting DNA polymerase theta in those HR-\ndeficient tumors that exhibit certain mutation patterns. Overall the methodology appears quite solid \nand the presentation balanced. I do have a few queries that need to be addressed. \n \n1. The conclusions have some potential to be affected by technical issues stemming from mutation \ncalling from short reads, which is error-prone in case of calling short indels and moreso in case of \nlarger structural variants. \n- only a single tool to call structural variants is used (Pindel); more modern tools (e.g. Manta, Gridss \netc) may give more accurate results, e.g. in terms of placing the breakpoint, or in terms of false \nnegative rates; note that the latter cannot be prevented by manual inspection of calls which they have \ndone. Running an additional caller would help ensure the results are not skewed towards biases of a \nsingle algorithm. In particular different tools may have different biases with respect to lengths/types \nof variants they can detect, which can be relevant for the statements such as line 117 \"The rate of \nTDs in brc-1 and brd-1 mutants is approximately tenfold lower than the rate of deletions in these \nmutants.\" \n- would the conclusions be affected if limiting to variant calls with higher confidence scores and/or in \nless difficult to align (less repetitive) genome regions? \n- please state average and/or median depth of sequencing in the methods. \n \n2. Data presentation issues. \n- there is much overplotting of the dots in 1A/1B/2A/2B/3A/3B which makes it hard to judge the \ndensity of points of different colors (microhomology lengths), affecting interpretation of results. I \nsuggest to sort the points (x axis) by # nucleotides of microhomology, within each genotype. \n- line 142 \"we observed a small increase in deletion mutagenesis in cku-80 and lig-4\" and line 146 \n\"base substitution rate is mildly increased in cku-80 and lig-4 mutants as compared to wildtype\". I \nwould not say this is a small nor mild increase (>=2 fold in Figs 2C/2E). In 2E, the SNV rate increase \nfrom NHEJ KOs is bigger than for brc-1 KO. Consider rephrasing. \n- please draw two-sided error bars on bar charts \n- mutation rates flanking the breaks do not appear to be elevated, arguing against common use of an \nerror-prone polymerase to extend past pol theta activity. Pol theta itself does not appear to contribute \nmany SNVs according to Fig 3e. Thus currently the cause of higher SNV mutation rates in brc-1/brd-1 \nbackgrounds is unclear. Would looking at the trinucleotide mutation spectrum, and examining \ncorrelations to known mutation signatures from human tumors, be helpful in elucidating this? \n \n\n\f3. Multiple lines of animals k.o. for various genes were propagated, and mutation data pooled together \nfor analysis. Are the results overall consistent between the independant lines? Potentially some of the \nobservations might be driven by an acquired mutator in one of the lines (e.g. the increased SNV rate \nin NHEJ deficient lines). \n \n4. Regarding negative selection. Lines 179-181 \"Deletions of that size will affect multiple genes in a \ngene-dense organism suchas C. elegans and their presence are likely counterselected for in animal \npropagation experiments\". It is well possible that many of the smaller deletions are also under \nselection. Please discuss and/or address with analyses if this has the potential to bias results (e.g. \nthere might be confounding between chromosomal locations of fitness genes and the chromosomal \nlocation bias of activity of the various HR failure-related mutational processes). \n \n5. A 1 nt microhomology has a rather high probability of occuring at random (depending on the local \nnucleotide composition). On the one hand, this affects interpretation of the data shown in plots, where \n1 nt microhomologies, often arising at random, are lumped together with 2 and 3 nt microhomologies, \nwhich are more likely to stem from real signal. On the other hand, this can affect interpretation of \nstatements such as line 110: \"These deletions are also characterized by an overrepresentation of \nmicro-homology: 58% of deletions had at least one nucleotide that could be mapped to either \njunction\". A baseline estimate of microhomology should be provided, if relying on 1 nt matches to call \nMHs. \n \nMinor. \n- typos: \"undistinguisable\", \"we have here describe (~10 kb)\". Line 169 might need to refer to Fig 3A. \n- line 174: \"Our data demonstrate that all deletions with ≤6 nt of homology are the result of \npolymerase theta on DSBs\". This is correct, however if I understand correctly this 6nt threshold might \nbe better placed at a higher number of nucleotides, depending on the lengths of the microhomology \nsegments in polq-1 mutants in Fig 3A, which are not evident from the figure. \n- line 107: \"We found no significant difference in deletion rate between brc-1 and brd-1 (p=0.236), \narguing against a BRC-1-independent role for BRD-1 in deletion prevention, and vice versa\". I would \nnot say this is necessarily very strong evidence, in the absence of experiments on the brc-1 brd-1 \ndouble mutant, which they did not perform. \n\n\f\f\n", "role": "assistant" } ]	2,020
Fast and multiplexed superresolution imaging with DNA-PAINT-ERS	Reviewers' comments: Reviewer #1 (Remarks to the Author): Civitci et al. report an effort to improve the imaging speed of DNA-PAINT by the addition of ethylene carbonate (EC) into the imaging buffer. EC helped increase the unbinding speed of the imaging strand from the docking strand, and with the addition of repeating sequences and PEG spacers in the docking strand, they increased the binding kinetics. Thus they achieved DNA-PAINT with improved speeds at several minutes per image, and demonstrated two-color imaging. The work is generally interesting, and I recommend the publication of this work in Nature Communications, with the below comments. 1. The authors should describe what their method is in the abstract. Now the name “DNA-PAINT- ERS” just comes up from nowhere. 2. What are the actual frame rates used to record the different data? Some of this information is hidden in the captions of their videos, but this should be more clearly stated in the text, captions, and the method section. For example, as they show good DNA-PAINT images in 150 s in Fig. 2a, how many frames were accumulated? If the frame rate was 67 frames/s as stated in the video caption, then 10,000 frame seems a bit low for typical SMLM? Also, in the videos shown, please add in timestamps and scale bars in each frame. 3. In figure 1B, bottom panel, for the differential images, can the authors show both positive and negative changes in signal, e.g., using different colors? Then it could show which single molecules have unbound. 4. The comparison of with and without EC in Fig. 3AB may not be very fair, since in this case it was with a docking strain optimized for performance in the presence of EC. I suggest either moving them to supplement or be more cautious in the discussion. 5. It would be useful to add in data to show whether the single-molecule brightness (photon count per unit time) is varied due to the addition of EC into the buffer. Fig. S1 alludes to this, but it would be useful to get histograms of photon counts. 6. In the introduction part, “other SMLM (t on ~0.1 s)6”: 10 ms time scale may be more common for STORM/PALM. Reviewer #2 (Remarks to the Author): in this work, the authors propose and test a number of technical improvements for DNA-PAINT, an approach to single molecule localisation microscopy that is based on the controlled, transient hybridisation of dye-coupled DNA oligomers with their reverse-complementary counterparts that are bound to a traget structure. This technique is promising as it allows for multiplexed labelling and higher resolution than other approaches such as (d) STORM and PALM. The authors present: a) an optimisation of the binding period for shorter, but equally specific binding events using addition of ethylene-carbonate, a reagent known from other DNA-based techniques such as FISH. The authors present very convincing, high quality data to support their claim that this speeds up DNA-PAINT imaging. b) a multiplication of the target sequence on the DNA to increase local concentration and faster on-rate. This significantly increases signal, although the quantification is not entirely clear and more controlled quantitative data, maybe acquired in vitro should be presented. c) the introduction of a polyethylene glycol linker between the antibody and the oligomer to make the oligomer more accessible. This furthermore increases the signal. Overall the authors present three tweaks to DNA-PAINT that will make this technique more efficient and accessible to many and this is an important and worthwhile contribution. The experimental setup is designed nicely and the results convincing. The only problem is that the quantification for b) and c) is not clear and should be demonstrated using a clear in vitro setup rather than intrinsically uneven biological images. I thus suggest minor revisions.	Reviewers' comments: Reviewer #1 (Remarks to the Author): Civitci et al. report an effort to improve the imaging speed of DNA-PAINT by the addition of ethylene carbonate (EC) into the imaging buffer. EC helped increase the unbinding speed of the imaging strand from the docking strand, and with the addition of repeating sequences and PEG spacers in the docking strand, they increased the binding kinetics. Thus they achieved DNA-PAINT with improved speeds at several minutes per image, and demonstrated two-color imaging. The work is generally interesting, and I recommend the publication of this work in Nature Communications, with the below comments. 1. The authors should describe what their method is in the abstract. Now the name “DNA-PAINT- ERS” just comes up from nowhere. 2. What are the actual frame rates used to record the different data? Some of this information is hidden in the captions of their videos, but this should be more clearly stated in the text, captions, and the method section. For example, as they show good DNA-PAINT images in 150 s in Fig. 2a, how many frames were accumulated? If the frame rate was 67 frames/s as stated in the video caption, then 10,000 frame seems a bit low for typical SMLM? Also, in the videos shown, please add in timestamps and scale bars in each frame. 3. In figure 1B, bottom panel, for the differential images, can the authors show both positive and negative changes in signal, e.g., using different colors? Then it could show which single molecules have unbound. 4. The comparison of with and without EC in Fig. 3AB may not be very fair, since in this case it was with a docking strain optimized for performance in the presence of EC. I suggest either moving them to supplement or be more cautious in the discussion. 5. It would be useful to add in data to show whether the single-molecule brightness (photon count per unit time) is varied due to the addition of EC into the buffer. Fig. S1 alludes to this, but it would be useful to get histograms of photon counts. 6. In the introduction part, “other SMLM (t on ~0.1 s)6”: 10 ms time scale may be more common for STORM/PALM. Reviewer #2 (Remarks to the Author): in this work, the authors propose and test a number of technical improvements for DNA-PAINT, an approach to single molecule localisation microscopy that is based on the controlled, transient hybridisation of dye-coupled DNA oligomers with their reverse-complementary counterparts that are bound to a traget structure. This technique is promising as it allows for multiplexed labelling and higher resolution than other approaches such as (d) STORM and PALM. The authors present: a) an optimisation of the binding period for shorter, but equally specific binding events using addition of ethylene-carbonate, a reagent known from other DNA-based techniques such as FISH. The authors present very convincing, high quality data to support their claim that this speeds up DNA-PAINT imaging. b) a multiplication of the target sequence on the DNA to increase local concentration and faster on-rate. This significantly increases signal, although the quantification is not entirely clear and more controlled quantitative data, maybe acquired in vitro should be presented. c) the introduction of a polyethylene glycol linker between the antibody and the oligomer to make the oligomer more accessible. This furthermore increases the signal. Overall the authors present three tweaks to DNA-PAINT that will make this technique more efficient and accessible to many and this is an important and worthwhile contribution. The experimental setup is designed nicely and the results convincing. The only problem is that the quantification for b) and c) is not clear and should be demonstrated using a clear in vitro setup rather than intrinsically uneven biological images. I thus suggest minor revisions.	TPR	NC_BS_BIOTEC_003_RR1.txt	{ "criticism": 2, "example": 3, "importance_and_relevance": 2, "materials_and_methods": 22, "praise": 6, "presentation_and_reporting": 10, "results_and_discussion": 2, "suggestion_and_solution": 12, "total": 40 }	{ "criticism": 0.05, "example": 0.075, "importance_and_relevance": 0.05, "materials_and_methods": 0.55, "praise": 0.15, "presentation_and_reporting": 0.25, "results_and_discussion": 0.05, "suggestion_and_solution": 0.3 }	1.475	[ { "content": "Please review this paper.", "role": "user" }, { "content": "Reviewers' comments: \n \nReviewer #1 (Remarks to the Author): \n \nCivitci et al. report an effort to improve the imaging speed of DNA-PAINT by the addition of \nethylene carbonate (EC) into the imaging buffer. EC helped increase the unbinding speed of the \nimaging strand from the docking strand, and with the addition of repeating sequences and PEG \nspacers in the docking strand, they increased the binding kinetics. Thus they achieved DNA-PAINT \nwith improved speeds at several minutes per image, and demonstrated two-color imaging. \n \nThe work is generally interesting, and I recommend the publication of this work in Nature \nCommunications, with the below comments. \n \n1. The authors should describe what their method is in the abstract. Now the name “DNA-PAINT-\nERS” just comes up from nowhere. \n \n2. What are the actual frame rates used to record the different data? Some of this information is \nhidden in the captions of their videos, but this should be more clearly stated in the text, captions, \nand the method section. For example, as they show good DNA-PAINT images in 150 s in Fig. 2a, \nhow many frames were accumulated? If the frame rate was 67 frames/s as stated in the video \ncaption, then 10,000 frame seems a bit low for typical SMLM? Also, in the videos shown, please \nadd in timestamps and scale bars in each frame. \n \n3. In figure 1B, bottom panel, for the differential images, can the authors show both positive and \nnegative changes in signal, e.g., using different colors? Then it could show which single molecules \nhave unbound. \n \n4. The comparison of with and without EC in Fig. 3AB may not be very fair, since in this case it was \nwith a docking strain optimized for performance in the presence of EC. I suggest either moving \nthem to supplement or be more cautious in the discussion. \n \n5. It would be useful to add in data to show whether the single-molecule brightness (photon count \nper unit time) is varied due to the addition of EC into the buffer. Fig. S1 alludes to this, but it \nwould be useful to get histograms of photon counts. \n \n6. In the introduction part, “other SMLM (t on ~0.1 s)6”: 10 ms time scale may be more common \nfor STORM/PALM. \n \n \n \n \nReviewer #2 (Remarks to the Author): \n \nin this work, the authors propose and test a number of technical improvements for DNA-PAINT, an \napproach to single molecule localisation microscopy that is based on the controlled, transient \nhybridisation of dye-coupled DNA oligomers with their reverse-complementary counterparts that \nare bound to a traget structure. This technique is promising as it allows for multiplexed labelling \nand higher resolution than other approaches such as (d) STORM and PALM. \nThe authors present: \n \na) an optimisation of the binding period for shorter, but equally specific binding events using \naddition of ethylene-carbonate, a reagent known from other DNA-based techniques such as FISH. \nThe authors present very convincing, high quality data to support their claim that this speeds up \nDNA-PAINT imaging. \n \n\n\fb) a multiplication of the target sequence on the DNA to increase local concentration and faster \non-rate. This significantly increases signal, although the quantification is not entirely clear and \nmore controlled quantitative data, maybe acquired in vitro should be presented. \n \nc) the introduction of a polyethylene glycol linker between the antibody and the oligomer to make \nthe oligomer more accessible. This furthermore increases the signal. \n \nOverall the authors present three tweaks to DNA-PAINT that will make this technique more \nefficient and accessible to many and this is an important and worthwhile contribution. The \nexperimental setup is designed nicely and the results convincing. The only problem is that the \nquantification for b) and c) is not clear and should be demonstrated using a clear in vitro setup \nrather than intrinsically uneven biological images. \n \nI thus suggest minor revisions. \n\n\f \n\n\f\n", "role": "assistant" } ]	2,020
Multiplexed single-cell transcriptional response profiling to define cancer vulnerabilities and	Reviewers' Comments: Reviewer #1: Remarks to the Author: The is a very mature and clearly written manuscript. The authors present an approach for SNP based demultiplexing of single cell RNA-seq data to measure transcriptional effects of perturbations. This is a very active research field and the contribution is very timely. The methods proposed and used are all very solid and state-of-the art. I really like the didactic way of presenting the linear model in Fig2, which really explains what is being modelled here. I only have two minor comments: - for consistency, would it hurt to decide on either tSNE or UMAP for visualisation, rather than switching from one to the other between figures? - in the methods section, in the multiplexing model, the authors use a logistic regression model - this is a supervised model, so on top of the X_ij you also need a label for each SNP i (=alternative allele or not). I did not see a clear description on how these labels (= the ground truth) were assigned and ideally would like to see this clarified. Florian Markowetz Reviewer #2: Remarks to the Author: McFarland et al developed a method (MIX-Seq) to measure post-perturbational transcriptomic response at single-cell resolution multiplexed across a wide range of cell lines by resolving the identity of each cell based on single nucleotide polymorphism (SNP) profiles from mRNA. Utilizing this method, they profiled transcriptomic response of 13 drugs at multiple time points across a range of cancer cell lines characterized previously in DepMap. Based on these profiles, the authors first identified drug-specific and shared transcriptional response signatures that: 1. Are informative about the drug’s mechanism of action (MOA), via benchmarking on the subset with known MOAs, 2. Are associated or are independent of sensitivity, 3. could predict long-term viability response, and finally, 4. Could predict heterogeneity of the response. Finally, combining this approach with cell-hashing they performed dual-multiplexing across time points or dosages. Overall, this is a well-written manuscript describing the experimental and statistical methods clearly and briefly. Comments: 1. An additional validation for resolving the identity of each cell based on SNP profiles from mRNA could be performed by integrating pre/post- treatment transcriptomic profiles from CMAP (Subramanian et al 2017, for ~20K small molecules for 1000 genes across 9 cancer cell lines from DepMap across two time points and multiple dosages) for the subset drugs and cell lines used in this study that are common in CMAP. In some more detail, we propose: (for every drug separately) Each cell transcriptomic response from MIX-seq can be compared based on the similarity (by a correlation test) to CMAP-based nine cell lines transcriptomic response for the matched drug. Identification of cell lines (out of nine) can be done based on this correlation rho which can act as labels to compare against SNP-based method predictions. Importantly this framework could be used to integrate other CMAP like datasets like DRUG-seq (Chaoyang Ye et al 2018) and PLATE-seq (Bush, E. C. et 2017). 2. It would also help to compare the long-term viability prediction power of treatment response transcriptome profile from single-cell seq based MIX-seq vs bulk-seq based from CMAP. 3. The last section highlights a later-onset regulation of viability-related genes vs viability- independent genes. We would like to point out that viability-related genes could have a later onset due to the identification methodology rather of underlying biology. Viability-related genes are differentially expressed in sensitive vs non-sensitive cell lines. This is because when the sensitive cells subset is targeted it leads to a population decrease with treatment time at a higher extent than non-sensitive cell lines. Thus, the extent of differential expression of viability-related would increase with time. Hence, need to be corrected for during this analysis.	Reviewers' Comments: Reviewer #1: Remarks to the Author: The is a very mature and clearly written manuscript. The authors present an approach for SNP based demultiplexing of single cell RNA-seq data to measure transcriptional effects of perturbations. This is a very active research field and the contribution is very timely. The methods proposed and used are all very solid and state-of-the art. I really like the didactic way of presenting the linear model in Fig2, which really explains what is being modelled here. I only have two minor comments: - for consistency, would it hurt to decide on either tSNE or UMAP for visualisation, rather than switching from one to the other between figures? - in the methods section, in the multiplexing model, the authors use a logistic regression model - this is a supervised model, so on top of the X_ij you also need a label for each SNP i (=alternative allele or not). I did not see a clear description on how these labels (= the ground truth) were assigned and ideally would like to see this clarified. Florian Markowetz Reviewer #2: Remarks to the Author: McFarland et al developed a method (MIX-Seq) to measure post-perturbational transcriptomic response at single-cell resolution multiplexed across a wide range of cell lines by resolving the identity of each cell based on single nucleotide polymorphism (SNP) profiles from mRNA. Utilizing this method, they profiled transcriptomic response of 13 drugs at multiple time points across a range of cancer cell lines characterized previously in DepMap. Based on these profiles, the authors first identified drug-specific and shared transcriptional response signatures that: 1. Are informative about the drug’s mechanism of action (MOA), via benchmarking on the subset with known MOAs, 2. Are associated or are independent of sensitivity, 3. could predict long-term viability response, and finally, 4. Could predict heterogeneity of the response. Finally, combining this approach with cell-hashing they performed dual-multiplexing across time points or dosages. Overall, this is a well-written manuscript describing the experimental and statistical methods clearly and briefly. Comments: 1. An additional validation for resolving the identity of each cell based on SNP profiles from mRNA could be performed by integrating pre/post- treatment transcriptomic profiles from CMAP (Subramanian et al 2017, for ~20K small molecules for 1000 genes across 9 cancer cell lines from DepMap across two time points and multiple dosages) for the subset drugs and cell lines used in this study that are common in CMAP. In some more detail, we propose: (for every drug separately) Each cell transcriptomic response from MIX-seq can be compared based on the similarity (by a correlation test) to CMAP-based nine cell lines transcriptomic response for the matched drug. Identification of cell lines (out of nine) can be done based on this correlation rho which can act as labels to compare against SNP-based method predictions. Importantly this framework could be used to integrate other CMAP like datasets like DRUG-seq (Chaoyang Ye et al 2018) and PLATE-seq (Bush, E. C. et 2017). 2. It would also help to compare the long-term viability prediction power of treatment response transcriptome profile from single-cell seq based MIX-seq vs bulk-seq based from CMAP. 3. The last section highlights a later-onset regulation of viability-related genes vs viability- independent genes. We would like to point out that viability-related genes could have a later onset due to the identification methodology rather of underlying biology. Viability-related genes are differentially expressed in sensitive vs non-sensitive cell lines. This is because when the sensitive cells subset is targeted it leads to a population decrease with treatment time at a higher extent than non-sensitive cell lines. Thus, the extent of differential expression of viability-related would increase with time. Hence, need to be corrected for during this analysis.	TPR	NC_BS_BIOTEC_004_RR1.txt	{ "criticism": 1, "example": 1, "importance_and_relevance": 1, "materials_and_methods": 22, "praise": 4, "presentation_and_reporting": 4, "results_and_discussion": 6, "suggestion_and_solution": 9, "total": 30 }	{ "criticism": 0.03333333333333333, "example": 0.03333333333333333, "importance_and_relevance": 0.03333333333333333, "materials_and_methods": 0.7333333333333333, "praise": 0.13333333333333333, "presentation_and_reporting": 0.13333333333333333, "results_and_discussion": 0.2, "suggestion_and_solution": 0.3 }	1.6	[ { "content": "Please review this paper.", "role": "user" }, { "content": "Reviewers' Comments: \n \nReviewer #1: \nRemarks to the Author: \nThe is a very mature and clearly written manuscript. The authors present an approach for SNP \nbased demultiplexing of single cell RNA-seq data to measure transcriptional effects of \nperturbations. This is a very active research field and the contribution is very timely. The methods \nproposed and used are all very solid and state-of-the art. \n \nI really like the didactic way of presenting the linear model in Fig2, which really explains what is \nbeing modelled here. \n \nI only have two minor comments: \n- for consistency, would it hurt to decide on either tSNE or UMAP for visualisation, rather than \nswitching from one to the other between figures? \n- in the methods section, in the multiplexing model, the authors use a logistic regression model - \nthis is a supervised model, so on top of the X_ij you also need a label for each SNP i (=alternative \nallele or not). I did not see a clear description on how these labels (= the ground truth) were \nassigned and ideally would like to see this clarified. \n \nFlorian Markowetz \n \n \n \nReviewer #2: \nRemarks to the Author: \nMcFarland et al developed a method (MIX-Seq) to measure post-perturbational transcriptomic \nresponse at single-cell resolution multiplexed across a wide range of cell lines by resolving the \nidentity of each cell based on single nucleotide polymorphism (SNP) profiles from mRNA. Utilizing \nthis method, they profiled transcriptomic response of 13 drugs at multiple time points across a \nrange of cancer cell lines characterized previously in DepMap. Based on these profiles, the authors \nfirst identified drug-specific and shared transcriptional response signatures that: 1. Are informative \nabout the drug’s mechanism of action (MOA), via benchmarking on the subset with known MOAs, \n2. Are associated or are independent of sensitivity, 3. could predict long-term viability response, \nand finally, 4. Could predict heterogeneity of the response. Finally, combining this approach with \ncell-hashing they performed dual-multiplexing across time points or dosages. \n \nOverall, this is a well-written manuscript describing the experimental and statistical methods \nclearly and briefly. \n \nComments: \n \n1. An additional validation for resolving the identity of each cell based on SNP profiles from mRNA \ncould be performed by integrating pre/post- treatment transcriptomic profiles from CMAP \n(Subramanian et al 2017, for ~20K small molecules for 1000 genes across 9 cancer cell lines from \nDepMap across two time points and multiple dosages) for the subset drugs and cell lines used in \nthis study that are common in CMAP. In some more detail, we propose: (for every drug \nseparately) Each cell transcriptomic response from MIX-seq can be compared based on the \nsimilarity (by a correlation test) to CMAP-based nine cell lines transcriptomic response for the \nmatched drug. Identification of cell lines (out of nine) can be done based on this correlation rho \nwhich can act as labels to compare against SNP-based method predictions. Importantly this \nframework could be used to integrate other CMAP like datasets like DRUG-seq (Chaoyang Ye et al \n2018) and PLATE-seq (Bush, E. C. et 2017). \n \n2. It would also help to compare the long-term viability prediction power of treatment response \n\n\ftranscriptome profile from single-cell seq based MIX-seq vs bulk-seq based from CMAP. \n \n3. The last section highlights a later-onset regulation of viability-related genes vs viability-\nindependent genes. We would like to point out that viability-related genes could have a later onset \ndue to the identification methodology rather of underlying biology. Viability-related genes are \ndifferentially expressed in sensitive vs non-sensitive cell lines. This is because when the sensitive \ncells subset is targeted it leads to a population decrease with treatment time at a higher extent \nthan non-sensitive cell lines. Thus, the extent of differential expression of viability-related would \nincrease with time. Hence, need to be corrected for during this analysis. \n\n\f \n\f\n", "role": "assistant" } ]	2,019
Metabolic cross-feeding in imbalanced diets allows gut microbes to improve reproduction	Reviewers' comments: Reviewer #1 (Remarks to the Author): Silivia et al. utilized isotope tracing based metabolomics to discover a syntrophic relationship between Acetobacter pomorum (Ap) and Lactobacillus plantarum (Lp). In general, it is found that Lp provided lactate for Ap as a precursor to synthesize and secrete AAs. These AAs could allow Lp to overcome the imbalanced diary. The story is potentially interesting. I was asked to carefully look at the metabolomics technical part. And I found there are severe technical issues with isotope tracing based metabolomics. 1. The authors did not mention in Methods whether they have corrected nature isotope abundances in the final isotopologue (not isotopomer) results. Some of the unlabeled results are not easy to explain. For example, in Figure 3 of 12C-glucose labeled samples, isoleucine has an observable M+4 peak but no M+3/M+5 peaks. If nature isotope abundances were corrected, M+1 to M+6 peaks of isoleucine should be close to 0. I would suspect the M+4 peak is a contamination of other metabolites. The authors should provide the raw MS peaks and details on this. The similar issues were also observed in Figure 5 and other figures. 2. For each metabolite, M+0 to M+n (n=6 for isoleucine) isotopologues should be demonstrated. But the authors only demonstrated M+3 to M+5 without explanations. I would like to see all isotopologue results with the correction of nature isotope abundances. It is common practice in isotope-tracing experiments. 3. The metabolic pathways from glucose or lactate to isoleucine are not explained or proposed. Since lactate has only 3 carbons, while isoleucine has 6 carbons. In general, in U-13C-lactate labeling experiment, it is reasonable that isoleucine has 3 carbons labelled from lactate. But in Figure 5, M+4 is the major isotopologue peak. How is lactate used for synthesize isoleucine (M+4)? It might be possible, but a metabolic pathway and explanation is required. 4. Similar question for U-13C-glucose labeling experiment, glucose has 6 carbons, but the authors only demonstrated isoleucine (M+5)? How is glucose used for synthesize isoleucine? Converted to lactate first? Then the same question repeated as comment 3. 5. In Figure 3, for the third and the fourth bars, the summed peak areas should be equal to each other (since they have the same experimental condition), but with different distributions on isotopologues. How to explain the differences on summed peak areas??? 6. In addition, the MID pattern of isoleucine in figure 3 (second bar) and figure 5 (third bar) are quite different. How to explain the difference? 7. Page 11, line 269, “the LpWCFS1∆ldh mutant strain failed to suppress yeast appetite (Fig. 4a)”. But actually, as the result showed, compared with the LpWCFS result, the LpWCFS1∆ldh did not have significant changes for increased yeast appetite. 8. Figure 5, the authors should provide a scheme to show how M+3, M+4 and M+5 isoleucine were produced. 9. In the isotope-tracer experiments, could author use mole percentage enrichment (MPE) to quantify the new synthesized metabolites (isoleucine and other AAs)? It would help to clearly demonstrate the percentage of new synthesized metabolites. 10. It is very challenging to read supplementary Figure S4, please change to a color scheme. I will re- evaluate this figure in revision. 11. The details for identification of metabolite should be given. Reviewer #2 (Remarks to the Author): This well-written manuscript by Henriques et all examines uses a multiprong approach to understand how two bacteria, Acetobacter Pomorum (Ap) and Lactobacillus plantarum (Lp) metabolically interact within the context of Drosophila melanogaster host diet. By using an liquid chemically defined diet that mimics the Drosophila holidic diet, the authors examined how these two bacterial species grow in the absence of two essential amino acids – isoleucine and histidine. These experiments revealed that although Lp is an isoleucine auxotroph, coculture of Lp with Ap allows Lp to grow in the absence of isoleucine. Using stable isotope tracers, the authors then demonstrate that in the presence of Lp, Ap synthesized and secreted labeled isoleucine. Moreover, the authors demonstrate the Lp synthesizes isoleucine using lactate provided by Ap, thus revealing an elegant symbiosis between the two bacteria. Finally, the authors provide evidence that this symbiotic relationship is likely not responsible for the ability of these two bacteria to jointly regulate Drosophila feeding behavior. Overall, I found this a clear, logical, and important study. While the mechanism by which these two bacteria jointly regulate feeding behavior remains elusive, it doesn’t diminish my interest in this story. In fact, I found it refreshing that the authors pointedly demonstrated that isoleucine production is essential for Lp growth but not fly behavior – a breath of fresh air in a field where these sort of metabolic interactions are usually the basis for wild speculation. While I can think of additional experiments that could be conducted by the authors, none would improve the manuscript. Reviewer #3 (Remarks to the Author): In this manuscript, Henriques et al. showed an intriguing metabolic connection between two dominant commensal bacterial species in Drosophila gut, Acetobacter pomorum (Ap) and Lactobacillus plantarum (Lp). Given their previous finding that these two gut microbes are required for controlling the appetite of the host, the authors attempted to identify the underlying mechanism. The careful gnotobiotic and metabolic experiments demonstrated the clear metabolic relay to produce essential amino acids from glucose via lactate. The in vitro data of bacterial culture using liquid holidic medium (prepared based on the fly diet) was convincing. Isotopic tracer experiments have beautifully demonstrated lactate is a product of glucose in Lp, and amino acids are produced from lactate in Ap. Lp is therefore required for the provision of lactate. The authors successfully showed that adding lactate is sufficient for Ap to influence the host behaviour. Amino acid production in Ap might also help Lp to grow under detrimental medium that lacks isoleucine. Overall, the manuscript is written well, addressing the essential question of "community-wise" bacterial metabolism and its effect on host physiology. However, it has still too many questions unanswered. Major concerns: 1, Lack of identification of the factor regulating host appetite, as it is mentioned in the manuscript, is critical especially when the authors' take-home message seemed to be "host behaviour" altered by the microbial metabolic interaction shown in the title as well as in the abstract. What is only revealed in the study is that lactate mediates Lp's effect on the behaviour. This reviewer assumes that the rescue of increased appetite with Ile(-) diet by Ap/Lp would be because it promotes the growth of Ap. If it is the case, high abundance of Ap (~10^9 CFU) but not Lp, during gnotobiotic preparation may rescue the behaviour. 2, Strikingly, the study did not reveal the mechanism of synergistic growth of the two bacterial species in vivo (fly gut/vial). Liquid culture nicely showed the unidirectional effect of Ap on Lp growth on Ile(-) medium. However, the authors did not demonstrate that this was because of the provision of Ile by Ap, especially given that the level of Ile production by Ap is very low. If possible, they need to use the loss of function mutant of Ap, which cannot utilise lactate or synthesise Ile. Besides, if lactate from Lp can also "feed" Ap, the growth of Ap in the liquid culture should be faster, as it was evident in the fly gut. Is it possible that lactate can increase Ap's growth in the fly gut, as well as in the liquid? The growth measurement in Fig.2 needs more time course, especially for the first 24 hours; one day of bacterial culture is too long to observe the upregulation of the growth speed. If this is not the case, then the bacterial growth in liquid is irrelevant to that in solid diet or in fly gut. 3, The major finding of the present study would be the production of Ile by Ap from Lp-derived lactate. However, this is expecting, given that Acetobactereceae is proved to produce amino acids from lactate by using stable isotope labelling technique (Adler et al., Appl Environ, Microbiol 80, 4702, 2014). Considering the study did not identify the Ap-derived factor controlling the host behaviour, the present may not significantly advance our understanding of microbial metabolism in the context of the host-microbe interaction. It should also be explained why Ap does not use glucose directly to produce Ile, while it produces (secretes) Ile by exogenous lactate. Other concerns: Fig.1a, It is possible that the quantity of bacteria, rather than quality, is important for the appetite control. It might be better to test the various amount of bacterial feeding on behaviour. Fig.1c, Lp in Ile(-) is greater in amount (~10^5) compared to that in complete medium (Lp alone conditions). What is the reason for this? There should be no growth at all under Ile (-) medium for this bacterial species, but the data suggests the opposite. Fig.2b, Can the authors deny the possibility that the rescue of Lp growth by Ap is due to the provision of Ile by dead Ap? Fig.3, How much did the total Ile content increased by co-culture? Fig.3, Is the production of Ile abolished by co-culture of Ap with the ldh mutant Lp? 1	Reviewers' comments: Reviewer #1 (Remarks to the Author): Silivia et al. utilized isotope tracing based metabolomics to discover a syntrophic relationship between Acetobacter pomorum (Ap) and Lactobacillus plantarum (Lp). In general, it is found that Lp provided lactate for Ap as a precursor to synthesize and secrete AAs. These AAs could allow Lp to overcome the imbalanced diary. The story is potentially interesting. I was asked to carefully look at the metabolomics technical part. And I found there are severe technical issues with isotope tracing based metabolomics. 1. The authors did not mention in Methods whether they have corrected nature isotope abundances in the final isotopologue (not isotopomer) results. Some of the unlabeled results are not easy to explain. For example, in Figure 3 of 12C-glucose labeled samples, isoleucine has an observable M+4 peak but no M+3/M+5 peaks. If nature isotope abundances were corrected, M+1 to M+6 peaks of isoleucine should be close to 0. I would suspect the M+4 peak is a contamination of other metabolites. The authors should provide the raw MS peaks and details on this. The similar issues were also observed in Figure 5 and other figures. 2. For each metabolite, M+0 to M+n (n=6 for isoleucine) isotopologues should be demonstrated. But the authors only demonstrated M+3 to M+5 without explanations. I would like to see all isotopologue results with the correction of nature isotope abundances. It is common practice in isotope-tracing experiments. 3. The metabolic pathways from glucose or lactate to isoleucine are not explained or proposed. Since lactate has only 3 carbons, while isoleucine has 6 carbons. In general, in U-13C-lactate labeling experiment, it is reasonable that isoleucine has 3 carbons labelled from lactate. But in Figure 5, M+4 is the major isotopologue peak. How is lactate used for synthesize isoleucine (M+4)? It might be possible, but a metabolic pathway and explanation is required. 4. Similar question for U-13C-glucose labeling experiment, glucose has 6 carbons, but the authors only demonstrated isoleucine (M+5)? How is glucose used for synthesize isoleucine? Converted to lactate first? Then the same question repeated as comment 3. 5. In Figure 3, for the third and the fourth bars, the summed peak areas should be equal to each other (since they have the same experimental condition), but with different distributions on isotopologues. How to explain the differences on summed peak areas??? 6. In addition, the MID pattern of isoleucine in figure 3 (second bar) and figure 5 (third bar) are quite different. How to explain the difference? 7. Page 11, line 269, “the LpWCFS1∆ldh mutant strain failed to suppress yeast appetite (Fig. 4a)”. But actually, as the result showed, compared with the LpWCFS result, the LpWCFS1∆ldh did not have significant changes for increased yeast appetite. 8. Figure 5, the authors should provide a scheme to show how M+3, M+4 and M+5 isoleucine were produced. 9. In the isotope-tracer experiments, could author use mole percentage enrichment (MPE) to quantify the new synthesized metabolites (isoleucine and other AAs)? It would help to clearly demonstrate the percentage of new synthesized metabolites. 10. It is very challenging to read supplementary Figure S4, please change to a color scheme. I will re- evaluate this figure in revision. 11. The details for identification of metabolite should be given. Reviewer #2 (Remarks to the Author): This well-written manuscript by Henriques et all examines uses a multiprong approach to understand how two bacteria, Acetobacter Pomorum (Ap) and Lactobacillus plantarum (Lp) metabolically interact within the context of Drosophila melanogaster host diet. By using an liquid chemically defined diet that mimics the Drosophila holidic diet, the authors examined how these two bacterial species grow in the absence of two essential amino acids – isoleucine and histidine. These experiments revealed that although Lp is an isoleucine auxotroph, coculture of Lp with Ap allows Lp to grow in the absence of isoleucine. Using stable isotope tracers, the authors then demonstrate that in the presence of Lp, Ap synthesized and secreted labeled isoleucine. Moreover, the authors demonstrate the Lp synthesizes isoleucine using lactate provided by Ap, thus revealing an elegant symbiosis between the two bacteria. Finally, the authors provide evidence that this symbiotic relationship is likely not responsible for the ability of these two bacteria to jointly regulate Drosophila feeding behavior. Overall, I found this a clear, logical, and important study. While the mechanism by which these two bacteria jointly regulate feeding behavior remains elusive, it doesn’t diminish my interest in this story. In fact, I found it refreshing that the authors pointedly demonstrated that isoleucine production is essential for Lp growth but not fly behavior – a breath of fresh air in a field where these sort of metabolic interactions are usually the basis for wild speculation. While I can think of additional experiments that could be conducted by the authors, none would improve the manuscript. Reviewer #3 (Remarks to the Author): In this manuscript, Henriques et al. showed an intriguing metabolic connection between two dominant commensal bacterial species in Drosophila gut, Acetobacter pomorum (Ap) and Lactobacillus plantarum (Lp). Given their previous finding that these two gut microbes are required for controlling the appetite of the host, the authors attempted to identify the underlying mechanism. The careful gnotobiotic and metabolic experiments demonstrated the clear metabolic relay to produce essential amino acids from glucose via lactate. The in vitro data of bacterial culture using liquid holidic medium (prepared based on the fly diet) was convincing. Isotopic tracer experiments have beautifully demonstrated lactate is a product of glucose in Lp, and amino acids are produced from lactate in Ap. Lp is therefore required for the provision of lactate. The authors successfully showed that adding lactate is sufficient for Ap to influence the host behaviour. Amino acid production in Ap might also help Lp to grow under detrimental medium that lacks isoleucine. Overall, the manuscript is written well, addressing the essential question of "community-wise" bacterial metabolism and its effect on host physiology. However, it has still too many questions unanswered. Major concerns: 1, Lack of identification of the factor regulating host appetite, as it is mentioned in the manuscript, is critical especially when the authors' take-home message seemed to be "host behaviour" altered by the microbial metabolic interaction shown in the title as well as in the abstract. What is only revealed in the study is that lactate mediates Lp's effect on the behaviour. This reviewer assumes that the rescue of increased appetite with Ile(-) diet by Ap/Lp would be because it promotes the growth of Ap. If it is the case, high abundance of Ap (~10^9 CFU) but not Lp, during gnotobiotic preparation may rescue the behaviour. 2, Strikingly, the study did not reveal the mechanism of synergistic growth of the two bacterial species in vivo (fly gut/vial). Liquid culture nicely showed the unidirectional effect of Ap on Lp growth on Ile(-) medium. However, the authors did not demonstrate that this was because of the provision of Ile by Ap, especially given that the level of Ile production by Ap is very low. If possible, they need to use the loss of function mutant of Ap, which cannot utilise lactate or synthesise Ile. Besides, if lactate from Lp can also "feed" Ap, the growth of Ap in the liquid culture should be faster, as it was evident in the fly gut. Is it possible that lactate can increase Ap's growth in the fly gut, as well as in the liquid? The growth measurement in Fig.2 needs more time course, especially for the first 24 hours; one day of bacterial culture is too long to observe the upregulation of the growth speed. If this is not the case, then the bacterial growth in liquid is irrelevant to that in solid diet or in fly gut. 3, The major finding of the present study would be the production of Ile by Ap from Lp-derived lactate. However, this is expecting, given that Acetobactereceae is proved to produce amino acids from lactate by using stable isotope labelling technique (Adler et al., Appl Environ, Microbiol 80, 4702, 2014). Considering the study did not identify the Ap-derived factor controlling the host behaviour, the present may not significantly advance our understanding of microbial metabolism in the context of the host-microbe interaction. It should also be explained why Ap does not use glucose directly to produce Ile, while it produces (secretes) Ile by exogenous lactate. Other concerns: Fig.1a, It is possible that the quantity of bacteria, rather than quality, is important for the appetite control. It might be better to test the various amount of bacterial feeding on behaviour. Fig.1c, Lp in Ile(-) is greater in amount (~10^5) compared to that in complete medium (Lp alone conditions). What is the reason for this? There should be no growth at all under Ile (-) medium for this bacterial species, but the data suggests the opposite. Fig.2b, Can the authors deny the possibility that the rescue of Lp growth by Ap is due to the provision of Ile by dead Ap? Fig.3, How much did the total Ile content increased by co-culture? Fig.3, Is the production of Ile abolished by co-culture of Ap with the ldh mutant Lp? 1	TPR	NC_BS_BIOTEC_005_RR1.txt	{ "criticism": 15, "example": 14, "importance_and_relevance": 5, "materials_and_methods": 41, "praise": 5, "presentation_and_reporting": 12, "results_and_discussion": 9, "suggestion_and_solution": 18, "total": 98 }	{ "criticism": 0.15306122448979592, "example": 0.14285714285714285, "importance_and_relevance": 0.05102040816326531, "materials_and_methods": 0.41836734693877553, "praise": 0.05102040816326531, "presentation_and_reporting": 0.12244897959183673, "results_and_discussion": 0.09183673469387756, "suggestion_and_solution": 0.1836734693877551 }	1.214286	[ { "content": "Please review this paper.", "role": "user" }, { "content": "Reviewers' comments: \n\n \n\nReviewer #1 (Remarks to the Author): \n\n \n\nSilivia et al. utilized isotope tracing based metabolomics to discover a syntrophic relationship \nbetween Acetobacter pomorum (Ap) and Lactobacillus plantarum (Lp). In general, it is found that Lp \nprovided lactate for Ap as a precursor to synthesize and secrete AAs. These AAs could allow Lp to \novercome the imbalanced diary. The story is potentially interesting. I was asked to carefully look at \nthe metabolomics technical part. And I found there are severe technical issues with isotope tracing \nbased metabolomics. \n\n \n\n1. The authors did not mention in Methods whether they have corrected nature isotope abundances \nin the final isotopologue (not isotopomer) results. Some of the unlabeled results are not easy to \nexplain. For example, in Figure 3 of 12C-glucose labeled samples, isoleucine has an observable M+4 \npeak but no M+3/M+5 peaks. If nature isotope abundances were corrected, M+1 to M+6 peaks of \nisoleucine should be close to 0. I would suspect the M+4 peak is a contamination of other \nmetabolites. The authors should provide the raw MS peaks and details on this. The similar issues \nwere also observed in Figure 5 and other figures. \n\n \n\n2. For each metabolite, M+0 to M+n (n=6 for isoleucine) isotopologues should be demonstrated. But \nthe authors only demonstrated M+3 to M+5 without explanations. I would like to see all \nisotopologue results with the correction of nature isotope abundances. It is common practice in \nisotope-tracing experiments. \n\n \n\n3. The metabolic pathways from glucose or lactate to isoleucine are not explained or proposed. \nSince lactate has only 3 carbons, while isoleucine has 6 carbons. In general, in U-13C-lactate labeling \nexperiment, it is reasonable that isoleucine has 3 carbons labelled from lactate. But in Figure 5, M+4 \nis the major isotopologue peak. How is lactate used for synthesize isoleucine (M+4)? It might be \npossible, but a metabolic pathway and explanation is required. \n\n \n\n4. Similar question for U-13C-glucose labeling experiment, glucose has 6 carbons, but the authors \nonly demonstrated isoleucine (M+5)? How is glucose used for synthesize isoleucine? Converted to \nlactate first? Then the same question repeated as comment 3. \n\n \n\n\f5. In Figure 3, for the third and the fourth bars, the summed peak areas should be equal to each \nother (since they have the same experimental condition), but with different distributions on \nisotopologues. How to explain the differences on summed peak areas??? \n\n \n\n6. In addition, the MID pattern of isoleucine in figure 3 (second bar) and figure 5 (third bar) are quite \ndifferent. How to explain the difference? \n\n \n\n7. Page 11, line 269, “the LpWCFS1∆ldh mutant strain failed to suppress yeast appetite (Fig. 4a)”. But \nactually, as the result showed, compared with the LpWCFS result, the LpWCFS1∆ldh did not have \nsignificant changes for increased yeast appetite. \n\n \n\n8. Figure 5, the authors should provide a scheme to show how M+3, M+4 and M+5 isoleucine were \nproduced. \n\n \n\n9. In the isotope-tracer experiments, could author use mole percentage enrichment (MPE) to \nquantify the new synthesized metabolites (isoleucine and other AAs)? It would help to clearly \ndemonstrate the percentage of new synthesized metabolites. \n\n \n\n10. It is very challenging to read supplementary Figure S4, please change to a color scheme. I will re-\nevaluate this figure in revision. \n\n \n\n11. The details for identification of metabolite should be given. \n\n \n\n \n\n \n\nReviewer #2 (Remarks to the Author): \n\n \n\nThis well-written manuscript by Henriques et all examines uses a multiprong approach to \nunderstand how two bacteria, Acetobacter Pomorum (Ap) and Lactobacillus plantarum (Lp) \nmetabolically interact within the context of Drosophila melanogaster host diet. By using an liquid \nchemically defined diet that mimics the Drosophila holidic diet, the authors examined how these \ntwo bacterial species grow in the absence of two essential amino acids – isoleucine and histidine. \nThese experiments revealed that although Lp is an isoleucine auxotroph, coculture of Lp with Ap \n\n\fallows Lp to grow in the absence of isoleucine. Using stable isotope tracers, the authors then \ndemonstrate that in the presence of Lp, Ap synthesized and secreted labeled isoleucine. Moreover, \nthe authors demonstrate the Lp synthesizes isoleucine using lactate provided by Ap, thus revealing \nan elegant symbiosis between the two bacteria. Finally, the authors provide evidence that this \nsymbiotic relationship is likely not responsible for the ability of these two bacteria to jointly regulate \nDrosophila feeding behavior. \n\n \n\nOverall, I found this a clear, logical, and important study. While the mechanism by which these two \nbacteria jointly regulate feeding behavior remains elusive, it doesn’t diminish my interest in this \nstory. In fact, I found it refreshing that the authors pointedly demonstrated that isoleucine \nproduction is essential for Lp growth but not fly behavior – a breath of fresh air in a field where \nthese sort of metabolic interactions are usually the basis for wild speculation. While I can think of \nadditional experiments that could be conducted by the authors, none would improve the \nmanuscript. \n\n \n\n \n\n \n\nReviewer #3 (Remarks to the Author): \n\n \n\nIn this manuscript, Henriques et al. showed an intriguing metabolic connection between two \ndominant commensal bacterial species in Drosophila gut, Acetobacter pomorum (Ap) and \nLactobacillus plantarum (Lp). Given their previous finding that these two gut microbes are required \nfor controlling the appetite of the host, the authors attempted to identify the underlying \nmechanism. The careful gnotobiotic and metabolic experiments demonstrated the clear metabolic \nrelay to produce essential amino acids from glucose via lactate. The in vitro data of bacterial culture \nusing liquid holidic medium (prepared based on the fly diet) was convincing. Isotopic tracer \nexperiments have beautifully demonstrated lactate is a product of glucose in Lp, and amino acids are \nproduced from lactate in Ap. Lp is therefore required for the provision of lactate. The authors \nsuccessfully showed that adding lactate is sufficient for Ap to influence the host behaviour. Amino \nacid production in Ap might also help Lp to grow under detrimental medium that lacks isoleucine. \nOverall, the manuscript is written well, addressing the essential question of \"community-wise\" \nbacterial metabolism and its effect on host physiology. However, it has still too many questions \nunanswered. \n\n \n\n \n\nMajor concerns: \n\n \n\n\f1, Lack of identification of the factor regulating host appetite, as it is mentioned in the manuscript, is \ncritical especially when the authors' take-home message seemed to be \"host behaviour\" altered by \nthe microbial metabolic interaction shown in the title as well as in the abstract. What is only \nrevealed in the study is that lactate mediates Lp's effect on the behaviour. This reviewer assumes \nthat the rescue of increased appetite with Ile(-) diet by Ap/Lp would be because it promotes the \ngrowth of Ap. If it is the case, high abundance of Ap (~10^9 CFU) but not Lp, during gnotobiotic \npreparation may rescue the behaviour. \n\n \n\n2, Strikingly, the study did not reveal the mechanism of synergistic growth of the two bacterial \nspecies in vivo (fly gut/vial). Liquid culture nicely showed the unidirectional effect of Ap on Lp \ngrowth on Ile(-) medium. However, the authors did not demonstrate that this was because of the \nprovision of Ile by Ap, especially given that the level of Ile production by Ap is very low. If possible, \nthey need to use the loss of function mutant of Ap, which cannot utilise lactate or synthesise Ile. \nBesides, if lactate from Lp can also \"feed\" Ap, the growth of Ap in the liquid culture should be faster, \nas it was evident in the fly gut. Is it possible that lactate can increase Ap's growth in the fly gut, as \nwell as in the liquid? The growth measurement in Fig.2 needs more time course, especially for the \nfirst 24 hours; one day of bacterial culture is too long to observe the upregulation of the growth \nspeed. If this is not the case, then the bacterial growth in liquid is irrelevant to that in solid diet or in \nfly gut. \n\n \n\n3, The major finding of the present study would be the production of Ile by Ap from Lp-derived \nlactate. However, this is expecting, given that Acetobactereceae is proved to produce amino acids \nfrom lactate by using stable isotope labelling technique (Adler et al., Appl Environ, Microbiol 80, \n4702, 2014). Considering the study did not identify the Ap-derived factor controlling the host \nbehaviour, the present may not significantly advance our understanding of microbial metabolism in \nthe context of the host-microbe interaction. It should also be explained why Ap does not use glucose \ndirectly to produce Ile, while it produces (secretes) Ile by exogenous lactate. \n\n \n\nOther concerns: \n\nFig.1a, It is possible that the quantity of bacteria, rather than quality, is important for the appetite \ncontrol. It might be better to test the various amount of bacterial feeding on behaviour. \n\nFig.1c, Lp in Ile(-) is greater in amount (~10^5) compared to that in complete medium (Lp alone \nconditions). What is the reason for this? There should be no growth at all under Ile (-) medium for \nthis bacterial species, but the data suggests the opposite. \n\nFig.2b, Can the authors deny the possibility that the rescue of Lp growth by Ap is due to the \nprovision of Ile by dead Ap? \n\nFig.3, How much did the total Ile content increased by co-culture? \n\nFig.3, Is the production of Ile abolished by co-culture of Ap with the ldh mutant Lp? \n\n\f1 \n\n \n\n\n", "role": "assistant" } ]	2,020
Single cell transcriptomics of human epidermis identifies basal stem cell transition states	Reviewers' Comments: Reviewer #1: Remarks to the Author: In this manuscript by Wang et al. the authors utilize Single-Cell-RNA-Sequencing by 10X genomics to explore cellular heterogeneity of human neonatal foreskin epidermis. To my knowledge this is one of the first single-cell-RNA-seq analysis of the epidermis of neonatal human skin. Neonatal human skin is difficult to acquire with foreskin epidermis being the most practical to gather. Furthermore, neonatal mammalian skin is very different than adult tissue. This is because neonatal skin is at a point of generating new skin to grow in conjunction with the organism compared to resting adult skin, making the analysis performed by the authors critically important. The authors analyzed their single cell data sets with multiple computational pipelines, including; Seurat, Monocle, and SoptSC. These analyses have identified multiple populations of basal- progenitor cells in neonatal epidermis in addition to the differentiated layers such as the spinous keratinocytes. These clusters are interesting but expected to be found in Single-Cell-RNA-seq experiments. Importantly, the authors validate their initial analysis by using novel SoptSC software packages that produces investigate cell-cell interactions in addition to the Cellular Entropy Estimator. This result was highly interesting because it adds an additional dimension to RNA-velocity analysis, which was also performed, to reveal the potential for epidermal progenitor differentiation along a trajectory. The authors used the multiple biocomputational pipelines to identify new genes (epigenetic- UHRF1/Hells and Pttg1, Rrm2, Ass1, Col17a1 etc) to generate a novel molecular map of the basal layer of the skin. In addition, the authors tested whether the identified molecular factors were critical for the formation of the establishment of skin equivalents in culture, which revealed their importance in the formation of healthy epidermis. This manuscript is well written, and the data is novel with multiple detailed computational analysis performed in a statistically rigorous manner for Single Cell RNA seq data. The manuscript is of very high quality and is ready for publication. Reviewer #2: Remarks to the Author: The paper by Shuxiong Wang et al. describes a comprehensive single-cell transcriptomics and computational analysis of cellular heterogeneity in human interfollicular epidermis. The authors find stable functional heterogeneity that correlates with spatial heterogeneity and is further corroborated by functional analyses. The authors identify HELLS and PPTG as novel factors involved in epidermal homeostasis and differentiation. While the paper has great potential, I propose that the following issues are addressed before publication: Major • The paper describes droplet-based scRNA-seq of primary cells from five donors. However, many of the presented analyses are based on a single donor. Supplementary Figure 4 indicates some variability in the cell clustering from different donors. For example, the GRN cluster is not detected in all of the donors. In my opinion, it would be important to show that the key findings from analyses of Library 3 are reproducible across donors. These include the expression of key marker genes (highlighted in Figure 2A) and the results of the pseudotime analysis. • On page 9 the authors state that “SoptSC unbiasedly reconstructed a putative BAS-SPN-GRN keratinocyte differentiation trajectory (Figure 4B)”. However, the figure does not clearly show GRN being the end point of the trajectory. Instead, based on 4A-B it seems that some of the GRN cells in fact precede SPN cells in pseudotime. In my opinion, the figure could be improved by addition of a panel visualizing pseudotime vs. cell identities (using the colour scheme from panel A). Minor • The cell-cell network inference analysis is rather difficult to interpret from Figure 3. From the heatmaps it would seem that the Basal cell cohort expresses on average the highest levels of genes of each pathway, yet based on the circos plots they are not among the active clusters. Could the authors please explain this discrepancy? Would it be make sense to show the expression of ligands and receptors separately? Finally, in the methods section it is suggested that the circos plots have directionality indicated by arrows. However, the figure is much too small to see such detail. • Combining entropy and RNA velocity is an elegant and innovative approach for delineating differentiation trajectories. Unfortunately, panel 4E is too small for seeing the direction of the vectors. Please improve this as it is central information. (The same applies also to Fig. 5E) • P. 18 methods: “. For downstream analyses, we kept cells which met the following filtering criteria per biological replicate per condition: <6000 UMI/cell, and <10% mitochondrial gene expression. “ Is this UMI filter correct? From Supplementary Figure 2 it is evident that a significant fraction of cells has more than 6000 UMIs. • Figure 1C and 1G legends are identical – in the interest of clarity it would be good to point out the different clustering methods used for defining the cohorts • Figure 1H does not seem to be cited in the text • In the results section supervised clustering in Seurat is mentioned. But the methods section describes unsupervised clustering. Which one was used in fact? • Typos: P.20 “circus plots” –> circos plots Reviewer #3: Remarks to the Author: The authors of the submitted manuscript perform single cell RNAseq analysis of neonatal foreskin epidermal keratinocytes. Based on the analysis the authors propose the existence of four distinct populations of basal cells, and provide some evidence for heterogeneity within the basal layer of the epidermis based on expression of markers. However, the authors provide no quantification of their staining nor coevaluation of the proposed markers. The authors go on to infer based simply on RNA expression analysis that paracrine and autocrine signaling are essential for homeostasis without any validating data. They subsequently propose that cells are organized along discrete cellular trajectories with key transition states. Functional analysis of a couple of markers for different markers associated with the basal population forms the basis for proposing that this further support that cells are organized in a cellular hierarchy and that specific epigenetic regulators control progenitor behavior. Although the dataset could have some interest for the skin community, the manuscript appears somewhat premature, lacking general controls and quantification of the different described phenotypes and expression patterns. Several concerns with both the analyses and the interpretation of the resulting data are outlined below. Major comments First and foremost, the authors clearly state that they produced 5 libraries for single cell analyses, however, they have chosen to focus on one library from a single individual. Human samples are heterogenous by nature and focusing the analysis on only one sample raises major concerns whether the detected cell populations and differentially expressed genes are indeed representative and would stand up to further rigorous analyses of the additional samples analyzed. There are efficient algorithms that allow batch corrections between different samples thereby providing the basis for analysis of single cell samples isolated and analyzed from multiple individuals. It is simply not sufficient to base the entire study essentially on a single sample. The decision to proceed using SoptSC analyses is not entirely clear. It does identify an additional population of cells that are classified by the authors as basal, however, fails to distinguish melanocytes and Langerhans and erythrocytes as separate populations, while Seurat successfully clusters them separately, and so is arguably giving a better performance for this specific task. Regardless of the choice between clustering with Seurat or SoptSC Figure 1 is difficult to interpret. For example, the plots provided illustrating the expression profiles of markers associated with differentiation and progenitor status do not align between figure 1 D and H. It seems that there are many more cells expressing INV and K10 at high levels, as well as a more pronounced population of cells that overlap for K14 (a basal cell marker) and K10 (a differentiation marker) in plots presented in Figure 1H than in plots presented in Figure 1D. The same raw data on differential expression and levels of gene expression is presumably used to generate both plots, so it is unclear why such a discrepancy should be present. Naturally this could reflect the fact that the authors chose different color schemes to present the data. Figure 1 D displays expression data ranging from light yellow to red, while in figure H data is presented from light yellow through red to dark brown (with no scale bar for both). Therefore, as the data is presented at the moment it is difficult to interpret it in a meaningful way. The authors then performed differential expression analyses on the individual populations that are assigned the basal identity in SoptSC and demonstrate via IHC in Figure 2 E-H that markers are expressed in subset of cells within the foreskin epidermis. This analysis needs to be supported larger images showing a bigger area of the epidermis and also quantification of the analysis. Instead of zooming in on regions like the authors have done in E-H, it would be much more informative to display the staining similar to that for K14/K10 as in Figure 2D. Importantly, from Figure 2B it is evident that some of the markers identified are not exclusive to just one population of basal cells. Co-staining for the markers in figure 2 (E-H) is a requirement alongside quantification of the analysis. The resulted for the RNA analysis needs further support. Firstly, the proposed signaling analysis (Figure 3) does not provide evidence for autocrine or paracrine signaling within the epithelium. This provides the basis for hypothesizes that can then subsequently be tested functionally. Secondly, the data presented for pseudotime trajectories and velocity is not very strong and the provided supporting data based on knock down is currently not analyzed in a manner whereby the authors can relate this to the proposed trajectories. Importantly, the analyses include no quantification of phenotypes, expression analysis for markers identified in the in vivo sample, as well as data that support that the same population dynamics observed in the one in vivo samples is recapitulated in the in vitro skin reconstitution model system included here (Figure 4). In Figure 5 the authors use PTTG1 and RRM2 KD to assess their role in skin reconstitution assays and conclude that PTTG1 as a marker of BAS1 population is essential for epidermal homeostasis unlike the other BAS populations identified. Much more sophisticated analyses are required for such statements. Again, further characterization is a requirement and importantly this should be supported by analysis of replicate samples (currently n=1). Minor comments Supplementary figure 1: Labels for dyes are missing All figures - the figures are not display consistently e.g. order of cell populations, labels are missing, information needs to be obtained from previous figures.	Reviewers' Comments: Reviewer #1: Remarks to the Author: In this manuscript by Wang et al. the authors utilize Single-Cell-RNA-Sequencing by 10X genomics to explore cellular heterogeneity of human neonatal foreskin epidermis. To my knowledge this is one of the first single-cell-RNA-seq analysis of the epidermis of neonatal human skin. Neonatal human skin is difficult to acquire with foreskin epidermis being the most practical to gather. Furthermore, neonatal mammalian skin is very different than adult tissue. This is because neonatal skin is at a point of generating new skin to grow in conjunction with the organism compared to resting adult skin, making the analysis performed by the authors critically important. The authors analyzed their single cell data sets with multiple computational pipelines, including; Seurat, Monocle, and SoptSC. These analyses have identified multiple populations of basal- progenitor cells in neonatal epidermis in addition to the differentiated layers such as the spinous keratinocytes. These clusters are interesting but expected to be found in Single-Cell-RNA-seq experiments. Importantly, the authors validate their initial analysis by using novel SoptSC software packages that produces investigate cell-cell interactions in addition to the Cellular Entropy Estimator. This result was highly interesting because it adds an additional dimension to RNA-velocity analysis, which was also performed, to reveal the potential for epidermal progenitor differentiation along a trajectory. The authors used the multiple biocomputational pipelines to identify new genes (epigenetic- UHRF1/Hells and Pttg1, Rrm2, Ass1, Col17a1 etc) to generate a novel molecular map of the basal layer of the skin. In addition, the authors tested whether the identified molecular factors were critical for the formation of the establishment of skin equivalents in culture, which revealed their importance in the formation of healthy epidermis. This manuscript is well written, and the data is novel with multiple detailed computational analysis performed in a statistically rigorous manner for Single Cell RNA seq data. The manuscript is of very high quality and is ready for publication. Reviewer #2: Remarks to the Author: The paper by Shuxiong Wang et al. describes a comprehensive single-cell transcriptomics and computational analysis of cellular heterogeneity in human interfollicular epidermis. The authors find stable functional heterogeneity that correlates with spatial heterogeneity and is further corroborated by functional analyses. The authors identify HELLS and PPTG as novel factors involved in epidermal homeostasis and differentiation. While the paper has great potential, I propose that the following issues are addressed before publication: Major • The paper describes droplet-based scRNA-seq of primary cells from five donors. However, many of the presented analyses are based on a single donor. Supplementary Figure 4 indicates some variability in the cell clustering from different donors. For example, the GRN cluster is not detected in all of the donors. In my opinion, it would be important to show that the key findings from analyses of Library 3 are reproducible across donors. These include the expression of key marker genes (highlighted in Figure 2A) and the results of the pseudotime analysis. • On page 9 the authors state that “SoptSC unbiasedly reconstructed a putative BAS-SPN-GRN keratinocyte differentiation trajectory (Figure 4B)”. However, the figure does not clearly show GRN being the end point of the trajectory. Instead, based on 4A-B it seems that some of the GRN cells in fact precede SPN cells in pseudotime. In my opinion, the figure could be improved by addition of a panel visualizing pseudotime vs. cell identities (using the colour scheme from panel A). Minor • The cell-cell network inference analysis is rather difficult to interpret from Figure 3. From the heatmaps it would seem that the Basal cell cohort expresses on average the highest levels of genes of each pathway, yet based on the circos plots they are not among the active clusters. Could the authors please explain this discrepancy? Would it be make sense to show the expression of ligands and receptors separately? Finally, in the methods section it is suggested that the circos plots have directionality indicated by arrows. However, the figure is much too small to see such detail. • Combining entropy and RNA velocity is an elegant and innovative approach for delineating differentiation trajectories. Unfortunately, panel 4E is too small for seeing the direction of the vectors. Please improve this as it is central information. (The same applies also to Fig. 5E) • P. 18 methods: “. For downstream analyses, we kept cells which met the following filtering criteria per biological replicate per condition: <6000 UMI/cell, and <10% mitochondrial gene expression. “ Is this UMI filter correct? From Supplementary Figure 2 it is evident that a significant fraction of cells has more than 6000 UMIs. • Figure 1C and 1G legends are identical – in the interest of clarity it would be good to point out the different clustering methods used for defining the cohorts • Figure 1H does not seem to be cited in the text • In the results section supervised clustering in Seurat is mentioned. But the methods section describes unsupervised clustering. Which one was used in fact? • Typos: P.20 “circus plots” –> circos plots Reviewer #3: Remarks to the Author: The authors of the submitted manuscript perform single cell RNAseq analysis of neonatal foreskin epidermal keratinocytes. Based on the analysis the authors propose the existence of four distinct populations of basal cells, and provide some evidence for heterogeneity within the basal layer of the epidermis based on expression of markers. However, the authors provide no quantification of their staining nor coevaluation of the proposed markers. The authors go on to infer based simply on RNA expression analysis that paracrine and autocrine signaling are essential for homeostasis without any validating data. They subsequently propose that cells are organized along discrete cellular trajectories with key transition states. Functional analysis of a couple of markers for different markers associated with the basal population forms the basis for proposing that this further support that cells are organized in a cellular hierarchy and that specific epigenetic regulators control progenitor behavior. Although the dataset could have some interest for the skin community, the manuscript appears somewhat premature, lacking general controls and quantification of the different described phenotypes and expression patterns. Several concerns with both the analyses and the interpretation of the resulting data are outlined below. Major comments First and foremost, the authors clearly state that they produced 5 libraries for single cell analyses, however, they have chosen to focus on one library from a single individual. Human samples are heterogenous by nature and focusing the analysis on only one sample raises major concerns whether the detected cell populations and differentially expressed genes are indeed representative and would stand up to further rigorous analyses of the additional samples analyzed. There are efficient algorithms that allow batch corrections between different samples thereby providing the basis for analysis of single cell samples isolated and analyzed from multiple individuals. It is simply not sufficient to base the entire study essentially on a single sample. The decision to proceed using SoptSC analyses is not entirely clear. It does identify an additional population of cells that are classified by the authors as basal, however, fails to distinguish melanocytes and Langerhans and erythrocytes as separate populations, while Seurat successfully clusters them separately, and so is arguably giving a better performance for this specific task. Regardless of the choice between clustering with Seurat or SoptSC Figure 1 is difficult to interpret. For example, the plots provided illustrating the expression profiles of markers associated with differentiation and progenitor status do not align between figure 1 D and H. It seems that there are many more cells expressing INV and K10 at high levels, as well as a more pronounced population of cells that overlap for K14 (a basal cell marker) and K10 (a differentiation marker) in plots presented in Figure 1H than in plots presented in Figure 1D. The same raw data on differential expression and levels of gene expression is presumably used to generate both plots, so it is unclear why such a discrepancy should be present. Naturally this could reflect the fact that the authors chose different color schemes to present the data. Figure 1 D displays expression data ranging from light yellow to red, while in figure H data is presented from light yellow through red to dark brown (with no scale bar for both). Therefore, as the data is presented at the moment it is difficult to interpret it in a meaningful way. The authors then performed differential expression analyses on the individual populations that are assigned the basal identity in SoptSC and demonstrate via IHC in Figure 2 E-H that markers are expressed in subset of cells within the foreskin epidermis. This analysis needs to be supported larger images showing a bigger area of the epidermis and also quantification of the analysis. Instead of zooming in on regions like the authors have done in E-H, it would be much more informative to display the staining similar to that for K14/K10 as in Figure 2D. Importantly, from Figure 2B it is evident that some of the markers identified are not exclusive to just one population of basal cells. Co-staining for the markers in figure 2 (E-H) is a requirement alongside quantification of the analysis. The resulted for the RNA analysis needs further support. Firstly, the proposed signaling analysis (Figure 3) does not provide evidence for autocrine or paracrine signaling within the epithelium. This provides the basis for hypothesizes that can then subsequently be tested functionally. Secondly, the data presented for pseudotime trajectories and velocity is not very strong and the provided supporting data based on knock down is currently not analyzed in a manner whereby the authors can relate this to the proposed trajectories. Importantly, the analyses include no quantification of phenotypes, expression analysis for markers identified in the in vivo sample, as well as data that support that the same population dynamics observed in the one in vivo samples is recapitulated in the in vitro skin reconstitution model system included here (Figure 4). In Figure 5 the authors use PTTG1 and RRM2 KD to assess their role in skin reconstitution assays and conclude that PTTG1 as a marker of BAS1 population is essential for epidermal homeostasis unlike the other BAS populations identified. Much more sophisticated analyses are required for such statements. Again, further characterization is a requirement and importantly this should be supported by analysis of replicate samples (currently n=1). Minor comments Supplementary figure 1: Labels for dyes are missing All figures - the figures are not display consistently e.g. order of cell populations, labels are missing, information needs to be obtained from previous figures.	TPR	NC_BS_BIOTEC_006_RR1.txt	{ "criticism": 19, "example": 17, "importance_and_relevance": 6, "materials_and_methods": 62, "praise": 6, "presentation_and_reporting": 17, "results_and_discussion": 5, "suggestion_and_solution": 13, "total": 81 }	{ "criticism": 0.2345679012345679, "example": 0.20987654320987653, "importance_and_relevance": 0.07407407407407407, "materials_and_methods": 0.7654320987654321, "praise": 0.07407407407407407, "presentation_and_reporting": 0.20987654320987653, "results_and_discussion": 0.06172839506172839, "suggestion_and_solution": 0.16049382716049382 }	1.790123	[ { "content": "Please review this paper.", "role": "user" }, { "content": "Reviewers' Comments: \n \nReviewer #1: \nRemarks to the Author: \nIn this manuscript by Wang et al. the authors utilize Single-Cell-RNA-Sequencing by 10X genomics \nto explore cellular heterogeneity of human neonatal foreskin epidermis. To my knowledge this is \none of the first single-cell-RNA-seq analysis of the epidermis of neonatal human skin. Neonatal \nhuman skin is difficult to acquire with foreskin epidermis being the most practical to gather. \nFurthermore, neonatal mammalian skin is very different than adult tissue. This is because neonatal \nskin is at a point of generating new skin to grow in conjunction with the organism compared to \nresting adult skin, making the analysis performed by the authors critically important. \nThe authors analyzed their single cell data sets with multiple computational pipelines, including; \nSeurat, Monocle, and SoptSC. These analyses have identified multiple populations of basal-\nprogenitor cells in neonatal epidermis in addition to the differentiated layers such as the spinous \nkeratinocytes. These clusters are interesting but expected to be found in Single-Cell-RNA-seq \nexperiments. Importantly, the authors validate their initial analysis by using novel SoptSC \nsoftware packages that produces investigate cell-cell interactions in addition to the Cellular \nEntropy Estimator. This result was highly interesting because it adds an additional dimension to \nRNA-velocity analysis, which was also performed, to reveal the potential for epidermal progenitor \ndifferentiation along a trajectory. \nThe authors used the multiple biocomputational pipelines to identify new genes (epigenetic- \nUHRF1/Hells and Pttg1, Rrm2, Ass1, Col17a1 etc) to generate a novel molecular map of the basal \nlayer of the skin. In addition, the authors tested whether the identified molecular factors were \ncritical for the formation of the establishment of skin equivalents in culture, which revealed their \nimportance in the formation of healthy epidermis. \nThis manuscript is well written, and the data is novel with multiple detailed computational analysis \nperformed in a statistically rigorous manner for Single Cell RNA seq data. The manuscript is of very \nhigh quality and is ready for publication. \n \n \n \nReviewer #2: \nRemarks to the Author: \nThe paper by Shuxiong Wang et al. describes a comprehensive single-cell transcriptomics and \ncomputational analysis of cellular heterogeneity in human interfollicular epidermis. The authors \nfind stable functional heterogeneity that correlates with spatial heterogeneity and is further \ncorroborated by functional analyses. The authors identify HELLS and PPTG as novel factors \ninvolved in epidermal homeostasis and differentiation. \nWhile the paper has great potential, I propose that the following issues are addressed before \npublication: \n \nMajor \n• The paper describes droplet-based scRNA-seq of primary cells from five donors. However, many \nof the presented analyses are based on a single donor. Supplementary Figure 4 indicates some \nvariability in the cell clustering from different donors. For example, the GRN cluster is not detected \nin all of the donors. In my opinion, it would be important to show that the key findings from \nanalyses of Library 3 are reproducible across donors. These include the expression of key marker \ngenes (highlighted in Figure 2A) and the results of the pseudotime analysis. \n• On page 9 the authors state that “SoptSC unbiasedly reconstructed a putative BAS-SPN-GRN \nkeratinocyte differentiation trajectory (Figure 4B)”. However, the figure does not clearly show GRN \nbeing the end point of the trajectory. Instead, based on 4A-B it seems that some of the GRN cells \nin fact precede SPN cells in pseudotime. In my opinion, the figure could be improved by addition of \na panel visualizing pseudotime vs. cell identities (using the colour scheme from panel A). \n \nMinor \n\n\f• The cell-cell network inference analysis is rather difficult to interpret from Figure 3. From the \nheatmaps it would seem that the Basal cell cohort expresses on average the highest levels of \ngenes of each pathway, yet based on the circos plots they are not among the active clusters. \nCould the authors please explain this discrepancy? Would it be make sense to show the expression \nof ligands and receptors separately? Finally, in the methods section it is suggested that the circos \nplots have directionality indicated by arrows. However, the figure is much too small to see such \ndetail. \n• Combining entropy and RNA velocity is an elegant and innovative approach for delineating \ndifferentiation trajectories. Unfortunately, panel 4E is too small for seeing the direction of the \nvectors. Please improve this as it is central information. (The same applies also to Fig. 5E) \n• P. 18 methods: “. For downstream analyses, we kept cells which met the following filtering \ncriteria per biological replicate per condition: <6000 UMI/cell, and <10% mitochondrial gene \nexpression. “ Is this UMI filter correct? From Supplementary Figure 2 it is evident that a significant \nfraction of cells has more than 6000 UMIs. \n• Figure 1C and 1G legends are identical – in the interest of clarity it would be good to point out \nthe different clustering methods used for defining the cohorts \n• Figure 1H does not seem to be cited in the text \n• In the results section supervised clustering in Seurat is mentioned. But the methods section \ndescribes unsupervised clustering. Which one was used in fact? \n• Typos: P.20 “circus plots” –> circos plots \n \n \n \nReviewer #3: \nRemarks to the Author: \nThe authors of the submitted manuscript perform single cell RNAseq analysis of neonatal foreskin \nepidermal keratinocytes. Based on the analysis the authors propose the existence of four distinct \npopulations of basal cells, and provide some evidence for heterogeneity within the basal layer of \nthe epidermis based on expression of markers. However, the authors provide no quantification of \ntheir staining nor coevaluation of the proposed markers. The authors go on to infer based simply \non RNA expression analysis that paracrine and autocrine signaling are essential for homeostasis \nwithout any validating data. They subsequently propose that cells are organized along discrete \ncellular trajectories with key transition states. Functional analysis of a couple of markers for \ndifferent markers associated with the basal population forms the basis for proposing that this \nfurther support that cells are organized in a cellular hierarchy and that specific epigenetic \nregulators control progenitor behavior. \n \nAlthough the dataset could have some interest for the skin community, the manuscript appears \nsomewhat premature, lacking general controls and quantification of the different described \nphenotypes and expression patterns. Several concerns with both the analyses and the \ninterpretation of the resulting data are outlined below. \n \nMajor comments \nFirst and foremost, the authors clearly state that they produced 5 libraries for single cell analyses, \nhowever, they have chosen to focus on one library from a single individual. Human samples are \nheterogenous by nature and focusing the analysis on only one sample raises major concerns \nwhether the detected cell populations and differentially expressed genes are indeed representative \nand would stand up to further rigorous analyses of the additional samples analyzed. There are \nefficient algorithms that allow batch corrections between different samples thereby providing the \nbasis for analysis of single cell samples isolated and analyzed from multiple individuals. It is simply \nnot sufficient to base the entire study essentially on a single sample. \n \nThe decision to proceed using SoptSC analyses is not entirely clear. It does identify an additional \npopulation of cells that are classified by the authors as basal, however, fails to distinguish \nmelanocytes and Langerhans and erythrocytes as separate populations, while Seurat successfully \n\n\fclusters them separately, and so is arguably giving a better performance for this specific task. \nRegardless of the choice between clustering with Seurat or SoptSC Figure 1 is difficult to interpret. \nFor example, the plots provided illustrating the expression profiles of markers associated with \ndifferentiation and progenitor status do not align between figure 1 D and H. It seems that there \nare many more cells expressing INV and K10 at high levels, as well as a more pronounced \npopulation of cells that overlap for K14 (a basal cell marker) and K10 (a differentiation marker) in \nplots presented in Figure 1H than in plots presented in Figure 1D. The same raw data on \ndifferential expression and levels of gene expression is presumably used to generate both plots, so \nit is unclear why such a discrepancy should be present. Naturally this could reflect the fact that the \nauthors chose different color schemes to present the data. Figure 1 D displays expression data \nranging from light yellow to red, while in figure H data is presented from light yellow through red \nto dark brown (with no scale bar for both). Therefore, as the data is presented at the moment it is \ndifficult to interpret it in a meaningful way. \n \nThe authors then performed differential expression analyses on the individual populations that are \nassigned the basal identity in SoptSC and demonstrate via IHC in Figure 2 E-H that markers are \nexpressed in subset of cells within the foreskin epidermis. This analysis needs to be supported \nlarger images showing a bigger area of the epidermis and also quantification of the analysis. \nInstead of zooming in on regions like the authors have done in E-H, it would be much more \ninformative to display the staining similar to that for K14/K10 as in Figure 2D. Importantly, from \nFigure 2B it is evident that some of the markers identified are not exclusive to just one population \nof basal cells. Co-staining for the markers in figure 2 (E-H) is a requirement alongside \nquantification of the analysis. \n \nThe resulted for the RNA analysis needs further support. Firstly, the proposed signaling analysis \n(Figure 3) does not provide evidence for autocrine or paracrine signaling within the epithelium. \nThis provides the basis for hypothesizes that can then subsequently be tested functionally. \nSecondly, the data presented for pseudotime trajectories and velocity is not very strong and the \nprovided supporting data based on knock down is currently not analyzed in a manner whereby the \nauthors can relate this to the proposed trajectories. Importantly, the analyses include no \nquantification of phenotypes, expression analysis for markers identified in the in vivo sample, as \nwell as data that support that the same population dynamics observed in the one in vivo samples \nis recapitulated in the in vitro skin reconstitution model system included here (Figure 4). \n \nIn Figure 5 the authors use PTTG1 and RRM2 KD to assess their role in skin reconstitution assays \nand conclude that PTTG1 as a marker of BAS1 population is essential for epidermal homeostasis \nunlike the other BAS populations identified. Much more sophisticated analyses are required for \nsuch statements. Again, further characterization is a requirement and importantly this should be \nsupported by analysis of replicate samples (currently n=1). \n \n \nMinor comments \nSupplementary figure 1: Labels for dyes are missing \n \nAll figures - the figures are not display consistently e.g. order of cell populations, labels are \nmissing, information needs to be obtained from previous figures. \n \n\n\f", "role": "assistant" } ]	2,020
A machine learning-based chemoproteomic approach to identify drug targets and binding	Reviewers' Comments: Reviewer #2: Remarks to the Author: In this work, Piazza and Beaton et al. improve their previous limited proteolysis approach to identify targets of small molecules in cell lysates. Compared to previous studies, the authors used a dose- response approach that allows the use of a more stringent statistical method. I had previously reviewed this manuscript for another journal and most of my comments have been addressed. This has significantly improved the manuscript, but there are still some points that I think the authors need to address. 1. The authors should address the limitations of their intact cell experiment, since limited proteolysis needs to be performed after lysis (even if the treatment is performed in living cells). The authors should include in the discussion that the lysis procedure can lead to reequilibration of compound binding and protein-protein interactions. This can lead to the compound no longer preventing proteolysis of some proteins, or the compound being able to access proteins that were not available before (e.g., due to subcellular compartmentalization) – leading to possible false negatives and positives. 2. While I appreciate that the authors now cite more recent and sensitive chemical biology approaches, the comparison of their method to these is still not entirely fair. a. First, only one replicate from the Savitski 2014 dataset was used. It is clear from the publication from which this dataset was retrieved that in this mode of TPP (TPP-TR), at least two replicates of each condition should be used. b. Second, the authors did not properly annotate the TPP dataset, probably due to problems matching IPI access numbers to Uniprot IDs – a quick look reveals that the third hit and some hits further down (LIMK2, STK4 and STK11) are not annotated as kinases despite being present in kinhub.org (and the authors Table S6). This leads to these proteins being annotated as false positives. c. This led me to repeat the benchmark, by using the same datasets (Savitski 2014, LiP-Quant and Deep-LiP_Quant), but with a more sensitive way of calling hits that was used to analyze this staurosporine TPP dataset and published last year (Childs et al. Mol. Cell. Prot. 2019) and carefully annotating all the datasets. I attach a markdown file of this analysis that I am happy that the authors use directly in their manuscript. This highlights that, as pointed out in my first comments, LiP (and also Deep-LiP) has lower coverage than other chemical biology approaches and is therefore less sensitive. There is no need for the authors to try to hide that LiP-Quant is less sensitive than other approaches, since it has other strengths, such as the ability to pinpoint the binding site of the compound.	Reviewers' Comments: Reviewer #2: Remarks to the Author: In this work, Piazza and Beaton et al. improve their previous limited proteolysis approach to identify targets of small molecules in cell lysates. Compared to previous studies, the authors used a dose- response approach that allows the use of a more stringent statistical method. I had previously reviewed this manuscript for another journal and most of my comments have been addressed. This has significantly improved the manuscript, but there are still some points that I think the authors need to address. 1. The authors should address the limitations of their intact cell experiment, since limited proteolysis needs to be performed after lysis (even if the treatment is performed in living cells). The authors should include in the discussion that the lysis procedure can lead to reequilibration of compound binding and protein-protein interactions. This can lead to the compound no longer preventing proteolysis of some proteins, or the compound being able to access proteins that were not available before (e.g., due to subcellular compartmentalization) – leading to possible false negatives and positives. 2. While I appreciate that the authors now cite more recent and sensitive chemical biology approaches, the comparison of their method to these is still not entirely fair. a. First, only one replicate from the Savitski 2014 dataset was used. It is clear from the publication from which this dataset was retrieved that in this mode of TPP (TPP-TR), at least two replicates of each condition should be used. b. Second, the authors did not properly annotate the TPP dataset, probably due to problems matching IPI access numbers to Uniprot IDs – a quick look reveals that the third hit and some hits further down (LIMK2, STK4 and STK11) are not annotated as kinases despite being present in kinhub.org (and the authors Table S6). This leads to these proteins being annotated as false positives. c. This led me to repeat the benchmark, by using the same datasets (Savitski 2014, LiP-Quant and Deep-LiP_Quant), but with a more sensitive way of calling hits that was used to analyze this staurosporine TPP dataset and published last year (Childs et al. Mol. Cell. Prot. 2019) and carefully annotating all the datasets. I attach a markdown file of this analysis that I am happy that the authors use directly in their manuscript. This highlights that, as pointed out in my first comments, LiP (and also Deep-LiP) has lower coverage than other chemical biology approaches and is therefore less sensitive. There is no need for the authors to try to hide that LiP-Quant is less sensitive than other approaches, since it has other strengths, such as the ability to pinpoint the binding site of the compound.	TPR	NC_BS_BIOTEC_008_RR1.txt	{ "criticism": 2, "example": 1, "importance_and_relevance": 1, "materials_and_methods": 12, "praise": 3, "presentation_and_reporting": 0, "results_and_discussion": 1, "suggestion_and_solution": 5, "total": 25 }	{ "criticism": 0.08, "example": 0.04, "importance_and_relevance": 0.04, "materials_and_methods": 0.48, "praise": 0.12, "presentation_and_reporting": 0, "results_and_discussion": 0.04, "suggestion_and_solution": 0.2 }	1	[ { "content": "Please review this paper.", "role": "user" }, { "content": "\n \n\nReviewers' Comments: \n \nReviewer #2: \nRemarks to the Author: \nIn this work, Piazza and Beaton et al. improve their previous limited proteolysis approach to identify \ntargets of small molecules in cell lysates. Compared to previous studies, the authors used a dose-\nresponse approach that allows the use of a more stringent statistical method. I had previously \nreviewed this manuscript for another journal and most of my comments have been addressed. This \nhas significantly improved the manuscript, but there are still some points that I think the authors need \nto address. \n1. The authors should address the limitations of their intact cell experiment, since limited proteolysis \nneeds to be performed after lysis (even if the treatment is performed in living cells). The authors \nshould include in the discussion that the lysis procedure can lead to reequilibration of compound \nbinding and protein-protein interactions. This can lead to the compound no longer preventing \nproteolysis of some proteins, or the compound being able to access proteins that were not available \nbefore (e.g., due to subcellular compartmentalization) – leading to possible false negatives and \npositives. \n2. While I appreciate that the authors now cite more recent and sensitive chemical biology \napproaches, the comparison of their method to these is still not entirely fair. \na. First, only one replicate from the Savitski 2014 dataset was used. It is clear from the publication \nfrom which this dataset was retrieved that in this mode of TPP (TPP-TR), at least two replicates of \neach condition should be used. \nb. Second, the authors did not properly annotate the TPP dataset, probably due to problems matching \nIPI access numbers to Uniprot IDs – a quick look reveals that the third hit and some hits further down \n(LIMK2, STK4 and STK11) are not annotated as kinases despite being present in kinhub.org (and the \nauthors Table S6). This leads to these proteins being annotated as false positives. \nc. This led me to repeat the benchmark, by using the same datasets (Savitski 2014, LiP-Quant and \nDeep-LiP_Quant), but with a more sensitive way of calling hits that was used to analyze this \nstaurosporine TPP dataset and published last year (Childs et al. Mol. Cell. Prot. 2019) and carefully \nannotating all the datasets. I attach a markdown file of this analysis that I am happy that the authors \nuse directly in their manuscript. This highlights that, as pointed out in my first comments, LiP (and \nalso Deep-LiP) has lower coverage than other chemical biology approaches and is therefore less \nsensitive. There is no need for the authors to try to hide that LiP-Quant is less sensitive than other \napproaches, since it has other strengths, such as the ability to pinpoint the binding site of the \ncompound. \n \n\n\f \n\n\n\f\n", "role": "assistant" } ]	2,020
Widespread protein lysine acetylation in gut microbiome and its alterations in patients with	Reviewers' Comments: Reviewer #1: Remarks to the Author: The authors have addressed the critiques raised. Reviewer #2: Remarks to the Author: In general, the authors have done a careful and diligent job in considering and responding to the various review comments. This included some new work to address concerns, along with very detailed comments of new text/figures added. While not perfect, the revisions reflect most of the major comments of the three reviewers - as such, this manuscript is greatly improved in terms of both content and clarity. I have only two remaining concerns: 1. P.2, top - the authors' statements of CD in the Abstract are still too over-reaching. In response to reviewer comments, they have now toned down the tenor in the manuscript, but this abstract needs to be revised to soften the results of the relatively few CD samples, which bring up some issues of general conclusions. 2. p. 10, lines 295-302 - the addition of the temporal variability (or lack thereof) of the protein acetylation is good, but this occurs too late in the manuscript, and in fact too late even in the CD discussion. The reader will want to know this information far earlier in the text - the authors should see how they can either move this section or allude to the key message much earlier. This would give the reader some assurance that the acetylation variability is not stoichastic. Reviewer #3: Remarks to the Author: The authors have addressed the concerns raised. They have attempted to showcase the value of this resource by better describing which proteins are being modified and providing at least one example of structural mapping. This could have been extended but it already gives others an idea that there will be many examples of novel regulatory acetylation sites potentially found in these data. The authors also improved other aspects of the paper and are more cautious about the messages that can be derived from a relative small cohort size. I have not further concerns.	Reviewers' Comments: Reviewer #1: Remarks to the Author: The authors have addressed the critiques raised. Reviewer #2: Remarks to the Author: In general, the authors have done a careful and diligent job in considering and responding to the various review comments. This included some new work to address concerns, along with very detailed comments of new text/figures added. While not perfect, the revisions reflect most of the major comments of the three reviewers - as such, this manuscript is greatly improved in terms of both content and clarity. I have only two remaining concerns: 1. P.2, top - the authors' statements of CD in the Abstract are still too over-reaching. In response to reviewer comments, they have now toned down the tenor in the manuscript, but this abstract needs to be revised to soften the results of the relatively few CD samples, which bring up some issues of general conclusions. 2. p. 10, lines 295-302 - the addition of the temporal variability (or lack thereof) of the protein acetylation is good, but this occurs too late in the manuscript, and in fact too late even in the CD discussion. The reader will want to know this information far earlier in the text - the authors should see how they can either move this section or allude to the key message much earlier. This would give the reader some assurance that the acetylation variability is not stoichastic. Reviewer #3: Remarks to the Author: The authors have addressed the concerns raised. They have attempted to showcase the value of this resource by better describing which proteins are being modified and providing at least one example of structural mapping. This could have been extended but it already gives others an idea that there will be many examples of novel regulatory acetylation sites potentially found in these data. The authors also improved other aspects of the paper and are more cautious about the messages that can be derived from a relative small cohort size. I have not further concerns.	TPR	NC_BS_BIOTEC_010_RR2.txt	{ "criticism": 4, "example": 1, "importance_and_relevance": 1, "materials_and_methods": 4, "praise": 6, "presentation_and_reporting": 3, "results_and_discussion": 2, "suggestion_and_solution": 4, "total": 15 }	{ "criticism": 0.26666666666666666, "example": 0.06666666666666667, "importance_and_relevance": 0.06666666666666667, "materials_and_methods": 0.26666666666666666, "praise": 0.4, "presentation_and_reporting": 0.2, "results_and_discussion": 0.13333333333333333, "suggestion_and_solution": 0.26666666666666666 }	1.666667	[ { "content": "Please review this paper.", "role": "user" }, { "content": "\n\n\fReviewers' Comments: \n \nReviewer #1: \nRemarks to the Author: \nThe authors have addressed the critiques raised. \n \n \n \nReviewer #2: \nRemarks to the Author: \nIn general, the authors have done a careful and diligent job in considering and responding to the \nvarious review comments. This included some new work to address concerns, along with very \ndetailed comments of new text/figures added. While not perfect, the revisions reflect most of the \nmajor comments of the three reviewers - as such, this manuscript is greatly improved in terms of \nboth content and clarity. \n \nI have only two remaining concerns: \n \n1. P.2, top - the authors' statements of CD in the Abstract are still too over-reaching. In response \nto reviewer comments, they have now toned down the tenor in the manuscript, but this abstract \nneeds to be revised to soften the results of the relatively few CD samples, which bring up some \nissues of general conclusions. \n \n2. p. 10, lines 295-302 - the addition of the temporal variability (or lack thereof) of the protein \nacetylation is good, but this occurs too late in the manuscript, and in fact too late even in the CD \ndiscussion. The reader will want to know this information far earlier in the text - the authors \nshould see how they can either move this section or allude to the key message much earlier. This \nwould give the reader some assurance that the acetylation variability is not stoichastic. \n \n \n \nReviewer #3: \nRemarks to the Author: \nThe authors have addressed the concerns raised. They have attempted to showcase the value of \nthis resource by better describing which proteins are being modified and providing at least one \nexample of structural mapping. This could have been extended but it already gives others an idea \nthat there will be many examples of novel regulatory acetylation sites potentially found in these \ndata. The authors also improved other aspects of the paper and are more cautious about the \nmessages that can be derived from a relative small cohort size. I have not further concerns. \n\n\f", "role": "assistant" } ]	2,020
Widespread protein lysine acetylation in gut microbiome and its alterations in patients with	Reviewers' Comments: Reviewer #1: Remarks to the Author: This paper is novel for presenting a large-scale look at human intestinal microbiome (in situ colonic mucosal wash) protein modification (K-Ac), which is an important frontier issue in microbiome ecology. The data is analyzed to describe microbial composition and taxon-associated gene state in this microbiome compartment for a set of control adults, and in a comparison of pediatric endoscopic patients nominally Crohn's disease or control. The technical pipeline is state of art, and overall it provides a number of interesting observations. There are a few limitations of this study. Major 1. The paper implies that Kac-modification is highlighting representation of organisms and their functions that cannot be observed by straight metaproteomics. However, it is possible that these features are concordant with whole metaproteomics, with just additional incremental change when the Kac subset is assessed. That is, the Kac changes are mainly reflecting simply the underlying shift in the microbiome composition and function, rather than something special about the Kac subset. This possibility is supported by the observation that many of the same differential organisms and genes reported in the literature by metagenomics and metatranscriptomics in the normal microbiome, or in healthy vs. IBD, are also reported here using the Kac subset. This really should be addressed and clarified, to provide context on what is being gained by Kac analysis. (a) A way to analytically do this is to provide description and comparisons first using the all-peptide metaproteomics data for taxon and gene representation. This analysis would provide context for the present dataset. (b) And/or, in the discussion the authors could systemically tabulate the Kac subset findings compare for how these compare to the metaproteomics (or metagenomics and metatranscriptomics )literature for controls and disease/healthy comparisons. As presented, it is possible that the differential Kac features may simply can't tell whether Kac fraction is telling us Technical 2. The statistics are can be better presented. The paper presents many types of data processing and statistical methods, but these aren't justified (why the particular method is most suitable or even appropriate?): paired T-test (line 184 for F/B ratio- is this normally distributed data?), hypergeometric test (line 191, for COG distribution among Kac vs. unmodified data); PLS-DA (line 227, for patient/control comparisons. And in some cases, the particular analytic question and how it was analyzed was unclear (e.g., Mann-Whitney in line 292, referring to a list of features and hence uncertain comparisons of microbial taxa, gene sets, and subject subsets). A section in methods or supplemental information can be used to provide rationale/justification for the analytics, and more clarity where needed in the results-findings regarding statistics used. 3. Down-regulated genes in CD (p. 10, para. starting with 238). It refers to "upregulated" genes, but it seems that these should be termed "down-regulated" Reviewer #2: Remarks to the Author: This manuscript outlines a detailed study of the level and nature of protein lysine acetylation in the human microbiome. The authors utilize an anti-acetyl-lysine (Kac) antibody enrichment strategy and tandem mass spectrometry approach to identify 35,200 Kac peptides corresponding to 31,821 Kac sites from the microbial or host 27 proteins in human gut microbiome samples. They conduct a detailed functional analysis on the protein types, and then apply this approach to the analysis of Editorial Note: Parts of this Peer Review File have been redacted as indicated to maintain patient confidentiality. pediatric Crohn’s 31 disease (CD) patient microbiome, where they identified 52 host and 136 microbial protein Kac sites that 32 were differentially abundant in CD versus controls. In general, this manuscript is well-written and logically designed. The technical approach is sound and systematic. However, there are some fairly serious issues that reduce enthusiasm, as listed below: 1. The title strongly suggests that this manuscript is primarily a “methods paper,” as there is really no description of why acetylation could be important in disease? While this is okay, the authors might consider revising to better define a “science message” if one is available. 2. Abstract (line 25) and Introduction - of all the possible PTMs, why focus on acetylation? The rationale is very weak and vague, which diminishes the overall impact of the study. Are there expectations that more or less acetylation is associated with disease somehow? 3. Line 98 – the authors should discuss the rationale for using the integrated gut microbial gene catalog vs. a matched assembled metagenome of the exact samples should be discussed and defended. How do the authors know that the gene catalog adequately reflects the exact nature and identify of the microbiota in their patients, especially the diseased cohort? 4. Line 114 – the observation of only 6 Kac peptides common in the enrichment compared to the 117 observed in the metaproteome is confusing and concerning. This issue is revisited on line 363, where the authors suggest that Kac peptides in the unenriched samples are likely false identifications. This is confusing and raises serious concerns with respect to false positive control in the other samples, and whether many of those might be false as well? 5. Line 117 – it seems a bit surprising that 80% of all protein groups have acetylation – was this expected? In light of the point above, this raises issue about false positives here again? 6. Line 156 – Do the authors suspect that the highest number of acetylation sites simply reflect the fact that Firmicutes and Bacteroidetes are simply the most abundant? In that case, acetylation might not be related to microbiome condition. 7. Line 216 – is there a basis to expect protein acetylation to be important in CD patients? Why this particular focus group? What will the “differences” mean? Line 383 provides some speculation, but even this linkage is quite vague and suspect. 8. The general outcome of this work is a broad and descriptive summary of protein acetylation, but delivers minimal biological value about what these differences mean? 9. Line 417 – One of the most concerning and unsettling aspects of this study is the sample cohort. The number of patients are low but not unacceptable for such a study. The most serious concern is that protein acetylation is known to be dynamic and thus temporal aspects are critical here. The use of only one sample per patient makes this study without context – how do the authors know that collection of samples from the same patients at another time-point wouldn’t produce dramatically different results. This could be especially true in the diseased patients. This study would have been more meaningful if the authors had conducted multiple time-point measurements on a more limited number of the samples. Reviewer #3: Remarks to the Author: In this manuscript, Zhang and colleagues have performed a characterization of the acetylation sites present in metaproteomic samples, including changes occurring in IBD patients. They have identified on the order of 30,000 acetylation sites in bacterial proteins, which they have mapped to bacterial taxonomic diversity and protein functional annotations. As expected, most acetylation maps to proteins that relate to translation and metabolism. Interestingly, by performing also a separate proteomic analysis the authors could determinate the ratio of acetylated vs. non-modified peptides for different taxonomic groups. Finally, they analysed a small number of IBD patients and controls and performed a proof-of-principle analysis of how acetylation changes in the microbial samples of these patients. I think the number of samples is too low but serves as a useful test case application. The large number of acetylation sites identified in the study serves as potential exciting resource to study protein regulation by acetylation. However, the authors don’t really make an effort of exploring this resource. The taxonomic differences in ratio of proteome acetylation is an interesting aspect that is mentioned mostly in discussion but it seems to be the most interesting result of the manuscript and could be better analysed. As it currently stands, the manuscript is mostly a quick description of acetylation patterns and does not go very far in novel discoveries. Major concerns The authors have identified what it is likely to be the largest collection of microbial acetylation sites to date. However, very little is done with this resource beyond a brief description of the taxonomic diversity and biological processes connected with these peptides. Understanding how most of the acetylation regulates protein activity is an important open question and this resource could be a fantastic opportunity to study this problem. The authors should provide some examples of how this could be achieved to motivate others to explore this great resource they have generated. For acetylated sites that match proteins that exist across many bacterial genomes the authors could align the protein sequences and map the acetylation sites to the alignment in order to study regions that are often acetylated across a large number of proteomes/species. For a number of these, there will be structural models that can serve as homology templates to study the structural impact of mapped acetylation sites. Since many of these will be in metabolic enzymes there should be many enzymes that are present in many different species and a number of structural models available to represent these enzymes. For example, the authors could ask if acetylation sites are often near catalytic residues for example. In my opinion, the most interesting aspect of the results presented were the potential differences in ratio of acetylated vs non-modified peptides along the taxonomic diversity studied. The authors speculated that this differences in ratios could be due to metabolic differences with some species being known producers of acetate. Acetate can be converted to acetyl-phosphate in some species which can in turn directly chemically acetylate proteins (Weinert et al. Mol Cell 2013). The authors could try to further analyse this association using isolate genomes or very complete metagenomic assembled genomes. Can the authors find a relationship between the enzymes encoded in the genomes of some of the taxonomic groups and their ratios of acetylated vs non-modified peptides ? Minor The authors need to make available in supplementary material the list of identified peptides with MS relevant associated statistics in a way that can facilitate their further analysis. The motif analysis of acetylated peptides could be relegated to supplementary information or shown briefly as an additional panel in Figure 1. The IBD results serve as an interesting proof-of-principle but the sample sizes are not large enough to identify variations that may still be meaningful. This should be addressed in the discussion. Figure 6 could as well be moved to supplementary and significant differences indicated in Figure 5.	Reviewers' Comments: Reviewer #1: Remarks to the Author: This paper is novel for presenting a large-scale look at human intestinal microbiome (in situ colonic mucosal wash) protein modification (K-Ac), which is an important frontier issue in microbiome ecology. The data is analyzed to describe microbial composition and taxon-associated gene state in this microbiome compartment for a set of control adults, and in a comparison of pediatric endoscopic patients nominally Crohn's disease or control. The technical pipeline is state of art, and overall it provides a number of interesting observations. There are a few limitations of this study. Major 1. The paper implies that Kac-modification is highlighting representation of organisms and their functions that cannot be observed by straight metaproteomics. However, it is possible that these features are concordant with whole metaproteomics, with just additional incremental change when the Kac subset is assessed. That is, the Kac changes are mainly reflecting simply the underlying shift in the microbiome composition and function, rather than something special about the Kac subset. This possibility is supported by the observation that many of the same differential organisms and genes reported in the literature by metagenomics and metatranscriptomics in the normal microbiome, or in healthy vs. IBD, are also reported here using the Kac subset. This really should be addressed and clarified, to provide context on what is being gained by Kac analysis. (a) A way to analytically do this is to provide description and comparisons first using the all-peptide metaproteomics data for taxon and gene representation. This analysis would provide context for the present dataset. (b) And/or, in the discussion the authors could systemically tabulate the Kac subset findings compare for how these compare to the metaproteomics (or metagenomics and metatranscriptomics )literature for controls and disease/healthy comparisons. As presented, it is possible that the differential Kac features may simply can't tell whether Kac fraction is telling us Technical 2. The statistics are can be better presented. The paper presents many types of data processing and statistical methods, but these aren't justified (why the particular method is most suitable or even appropriate?): paired T-test (line 184 for F/B ratio- is this normally distributed data?), hypergeometric test (line 191, for COG distribution among Kac vs. unmodified data); PLS-DA (line 227, for patient/control comparisons. And in some cases, the particular analytic question and how it was analyzed was unclear (e.g., Mann-Whitney in line 292, referring to a list of features and hence uncertain comparisons of microbial taxa, gene sets, and subject subsets). A section in methods or supplemental information can be used to provide rationale/justification for the analytics, and more clarity where needed in the results-findings regarding statistics used. 3. Down-regulated genes in CD (p. 10, para. starting with 238). It refers to "upregulated" genes, but it seems that these should be termed "down-regulated" Reviewer #2: Remarks to the Author: This manuscript outlines a detailed study of the level and nature of protein lysine acetylation in the human microbiome. The authors utilize an anti-acetyl-lysine (Kac) antibody enrichment strategy and tandem mass spectrometry approach to identify 35,200 Kac peptides corresponding to 31,821 Kac sites from the microbial or host 27 proteins in human gut microbiome samples. They conduct a detailed functional analysis on the protein types, and then apply this approach to the analysis of Editorial Note: Parts of this Peer Review File have been redacted as indicated to maintain patient confidentiality. pediatric Crohn’s 31 disease (CD) patient microbiome, where they identified 52 host and 136 microbial protein Kac sites that 32 were differentially abundant in CD versus controls. In general, this manuscript is well-written and logically designed. The technical approach is sound and systematic. However, there are some fairly serious issues that reduce enthusiasm, as listed below: 1. The title strongly suggests that this manuscript is primarily a “methods paper,” as there is really no description of why acetylation could be important in disease? While this is okay, the authors might consider revising to better define a “science message” if one is available. 2. Abstract (line 25) and Introduction - of all the possible PTMs, why focus on acetylation? The rationale is very weak and vague, which diminishes the overall impact of the study. Are there expectations that more or less acetylation is associated with disease somehow? 3. Line 98 – the authors should discuss the rationale for using the integrated gut microbial gene catalog vs. a matched assembled metagenome of the exact samples should be discussed and defended. How do the authors know that the gene catalog adequately reflects the exact nature and identify of the microbiota in their patients, especially the diseased cohort? 4. Line 114 – the observation of only 6 Kac peptides common in the enrichment compared to the 117 observed in the metaproteome is confusing and concerning. This issue is revisited on line 363, where the authors suggest that Kac peptides in the unenriched samples are likely false identifications. This is confusing and raises serious concerns with respect to false positive control in the other samples, and whether many of those might be false as well? 5. Line 117 – it seems a bit surprising that 80% of all protein groups have acetylation – was this expected? In light of the point above, this raises issue about false positives here again? 6. Line 156 – Do the authors suspect that the highest number of acetylation sites simply reflect the fact that Firmicutes and Bacteroidetes are simply the most abundant? In that case, acetylation might not be related to microbiome condition. 7. Line 216 – is there a basis to expect protein acetylation to be important in CD patients? Why this particular focus group? What will the “differences” mean? Line 383 provides some speculation, but even this linkage is quite vague and suspect. 8. The general outcome of this work is a broad and descriptive summary of protein acetylation, but delivers minimal biological value about what these differences mean? 9. Line 417 – One of the most concerning and unsettling aspects of this study is the sample cohort. The number of patients are low but not unacceptable for such a study. The most serious concern is that protein acetylation is known to be dynamic and thus temporal aspects are critical here. The use of only one sample per patient makes this study without context – how do the authors know that collection of samples from the same patients at another time-point wouldn’t produce dramatically different results. This could be especially true in the diseased patients. This study would have been more meaningful if the authors had conducted multiple time-point measurements on a more limited number of the samples. Reviewer #3: Remarks to the Author: In this manuscript, Zhang and colleagues have performed a characterization of the acetylation sites present in metaproteomic samples, including changes occurring in IBD patients. They have identified on the order of 30,000 acetylation sites in bacterial proteins, which they have mapped to bacterial taxonomic diversity and protein functional annotations. As expected, most acetylation maps to proteins that relate to translation and metabolism. Interestingly, by performing also a separate proteomic analysis the authors could determinate the ratio of acetylated vs. non-modified peptides for different taxonomic groups. Finally, they analysed a small number of IBD patients and controls and performed a proof-of-principle analysis of how acetylation changes in the microbial samples of these patients. I think the number of samples is too low but serves as a useful test case application. The large number of acetylation sites identified in the study serves as potential exciting resource to study protein regulation by acetylation. However, the authors don’t really make an effort of exploring this resource. The taxonomic differences in ratio of proteome acetylation is an interesting aspect that is mentioned mostly in discussion but it seems to be the most interesting result of the manuscript and could be better analysed. As it currently stands, the manuscript is mostly a quick description of acetylation patterns and does not go very far in novel discoveries. Major concerns The authors have identified what it is likely to be the largest collection of microbial acetylation sites to date. However, very little is done with this resource beyond a brief description of the taxonomic diversity and biological processes connected with these peptides. Understanding how most of the acetylation regulates protein activity is an important open question and this resource could be a fantastic opportunity to study this problem. The authors should provide some examples of how this could be achieved to motivate others to explore this great resource they have generated. For acetylated sites that match proteins that exist across many bacterial genomes the authors could align the protein sequences and map the acetylation sites to the alignment in order to study regions that are often acetylated across a large number of proteomes/species. For a number of these, there will be structural models that can serve as homology templates to study the structural impact of mapped acetylation sites. Since many of these will be in metabolic enzymes there should be many enzymes that are present in many different species and a number of structural models available to represent these enzymes. For example, the authors could ask if acetylation sites are often near catalytic residues for example. In my opinion, the most interesting aspect of the results presented were the potential differences in ratio of acetylated vs non-modified peptides along the taxonomic diversity studied. The authors speculated that this differences in ratios could be due to metabolic differences with some species being known producers of acetate. Acetate can be converted to acetyl-phosphate in some species which can in turn directly chemically acetylate proteins (Weinert et al. Mol Cell 2013). The authors could try to further analyse this association using isolate genomes or very complete metagenomic assembled genomes. Can the authors find a relationship between the enzymes encoded in the genomes of some of the taxonomic groups and their ratios of acetylated vs non-modified peptides ? Minor The authors need to make available in supplementary material the list of identified peptides with MS relevant associated statistics in a way that can facilitate their further analysis. The motif analysis of acetylated peptides could be relegated to supplementary information or shown briefly as an additional panel in Figure 1. The IBD results serve as an interesting proof-of-principle but the sample sizes are not large enough to identify variations that may still be meaningful. This should be addressed in the discussion. Figure 6 could as well be moved to supplementary and significant differences indicated in Figure 5.	TPR	NC_BS_BIOTEC_010_RR1.txt	{ "criticism": 15, "example": 14, "importance_and_relevance": 13, "materials_and_methods": 53, "praise": 10, "presentation_and_reporting": 9, "results_and_discussion": 10, "suggestion_and_solution": 19, "total": 92 }	{ "criticism": 0.16304347826086957, "example": 0.15217391304347827, "importance_and_relevance": 0.14130434782608695, "materials_and_methods": 0.5760869565217391, "praise": 0.10869565217391304, "presentation_and_reporting": 0.09782608695652174, "results_and_discussion": 0.10869565217391304, "suggestion_and_solution": 0.20652173913043478 }	1.554348	[ { "content": "Please review this paper.", "role": "user" }, { "content": "Reviewers' Comments: \n \nReviewer #1: \nRemarks to the Author: \nThis paper is novel for presenting a large-scale look at human intestinal microbiome (in situ colonic \nmucosal wash) protein modification (K-Ac), which is an important frontier issue in microbiome \necology. The data is analyzed to describe microbial composition and taxon-associated gene state in \nthis microbiome compartment for a set of control adults, and in a comparison of pediatric \nendoscopic patients nominally Crohn's disease or control. The technical pipeline is state of art, and \noverall it provides a number of interesting observations. \n \nThere are a few limitations of this study. \n \nMajor \n1. The paper implies that Kac-modification is highlighting representation of organisms and their \nfunctions that cannot be observed by straight metaproteomics. However, it is possible that these \nfeatures are concordant with whole metaproteomics, with just additional incremental change when \nthe Kac subset is assessed. That is, the Kac changes are mainly reflecting simply the underlying \nshift in the microbiome composition and function, rather than something special about the Kac \nsubset. This possibility is supported by the observation that many of the same differential \norganisms and genes reported in the literature by metagenomics and metatranscriptomics in the \nnormal microbiome, or in healthy vs. IBD, are also reported here using the Kac subset. \nThis really should be addressed and clarified, to provide context on what is being gained by Kac \nanalysis. (a) A way to analytically do this is to provide description and comparisons first using the \nall-peptide metaproteomics data for taxon and gene representation. This analysis would provide \ncontext for the present dataset. (b) And/or, in the discussion the authors could systemically \ntabulate the Kac subset findings compare for how these compare to the metaproteomics (or \nmetagenomics and metatranscriptomics )literature for controls and disease/healthy comparisons. \nAs presented, it is possible that the differential Kac features may simply can't tell whether Kac \nfraction is telling us \n \n \nTechnical \n2. The statistics are can be better presented. The paper presents many types of data processing \nand statistical methods, but these aren't justified (why the particular method is most suitable or \neven appropriate?): paired T-test (line 184 for F/B ratio- is this normally distributed data?), \nhypergeometric test (line 191, for COG distribution among Kac vs. unmodified data); PLS-DA (line \n227, for patient/control comparisons. And in some cases, the particular analytic question and how \nit was analyzed was unclear (e.g., Mann-Whitney in line 292, referring to a list of features and \nhence uncertain comparisons of microbial taxa, gene sets, and subject subsets). A section in \nmethods or supplemental information can be used to provide rationale/justification for the \nanalytics, and more clarity where needed in the results-findings regarding statistics used. \n \n3. Down-regulated genes in CD (p. 10, para. starting with 238). It refers to \"upregulated\" genes, \nbut it seems that these should be termed \"down-regulated\" \n \n \n \nReviewer #2: \nRemarks to the Author: \nThis manuscript outlines a detailed study of the level and nature of protein lysine acetylation in the \nhuman microbiome. The authors utilize an anti-acetyl-lysine (Kac) antibody enrichment strategy \nand tandem mass spectrometry approach to identify 35,200 Kac peptides corresponding to 31,821 \nKac sites from the microbial or host 27 proteins in human gut microbiome samples. They conduct \na detailed functional analysis on the protein types, and then apply this approach to the analysis of \n\nEditorial Note: Parts of this Peer Review File have been redacted as indicated to maintain patient confidentiality. \fpediatric Crohn’s 31 disease (CD) patient microbiome, where they identified 52 host and 136 \nmicrobial protein Kac sites that 32 were differentially abundant in CD versus controls. \n \nIn general, this manuscript is well-written and logically designed. The technical approach is sound \nand systematic. However, there are some fairly serious issues that reduce enthusiasm, as listed \nbelow: \n \n1. The title strongly suggests that this manuscript is primarily a “methods paper,” as there is really \nno description of why acetylation could be important in disease? While this is okay, the authors \nmight consider revising to better define a “science message” if one is available. \n \n2. Abstract (line 25) and Introduction - of all the possible PTMs, why focus on acetylation? The \nrationale is very weak and vague, which diminishes the overall impact of the study. Are there \nexpectations that more or less acetylation is associated with disease somehow? \n \n3. Line 98 – the authors should discuss the rationale for using the integrated gut microbial gene \ncatalog vs. a matched assembled metagenome of the exact samples should be discussed and \ndefended. How do the authors know that the gene catalog adequately reflects the exact nature \nand identify of the microbiota in their patients, especially the diseased cohort? \n \n4. Line 114 – the observation of only 6 Kac peptides common in the enrichment compared to the \n117 observed in the metaproteome is confusing and concerning. This issue is revisited on line 363, \nwhere the authors suggest that Kac peptides in the unenriched samples are likely false \nidentifications. This is confusing and raises serious concerns with respect to false positive control in \nthe other samples, and whether many of those might be false as well? \n \n5. Line 117 – it seems a bit surprising that 80% of all protein groups have acetylation – was this \nexpected? In light of the point above, this raises issue about false positives here again? \n \n6. Line 156 – Do the authors suspect that the highest number of acetylation sites simply reflect \nthe fact that Firmicutes and Bacteroidetes are simply the most abundant? In that case, acetylation \nmight not be related to microbiome condition. \n \n7. Line 216 – is there a basis to expect protein acetylation to be important in CD patients? Why \nthis particular focus group? What will the “differences” mean? Line 383 provides some speculation, \nbut even this linkage is quite vague and suspect. \n \n8. The general outcome of this work is a broad and descriptive summary of protein acetylation, but \ndelivers minimal biological value about what these differences mean? \n \n9. Line 417 – One of the most concerning and unsettling aspects of this study is the sample \ncohort. The number of patients are low but not unacceptable for such a study. The most serious \nconcern is that protein acetylation is known to be dynamic and thus temporal aspects are critical \nhere. The use of only one sample per patient makes this study without context – how do the \nauthors know that collection of samples from the same patients at another time-point wouldn’t \nproduce dramatically different results. This could be especially true in the diseased patients. This \nstudy would have been more meaningful if the authors had conducted multiple time-point \nmeasurements on a more limited number of the samples. \n \n \n \nReviewer #3: \nRemarks to the Author: \nIn this manuscript, Zhang and colleagues have performed a characterization of the acetylation \nsites present in metaproteomic samples, including changes occurring in IBD patients. They have \n\n\fidentified on the order of 30,000 acetylation sites in bacterial proteins, which they have mapped to \nbacterial taxonomic diversity and protein functional annotations. As expected, most acetylation \nmaps to proteins that relate to translation and metabolism. Interestingly, by performing also a \nseparate proteomic analysis the authors could determinate the ratio of acetylated vs. non-modified \npeptides for different taxonomic groups. Finally, they analysed a small number of IBD patients and \ncontrols and performed a proof-of-principle analysis of how acetylation changes in the microbial \nsamples of these patients. I think the number of samples is too low but serves as a useful test \ncase application. The large number of acetylation sites identified in the study serves as potential \nexciting resource to study protein regulation by acetylation. However, the authors don’t really \nmake an effort of exploring this resource. The taxonomic differences in ratio of proteome \nacetylation is an interesting aspect that is mentioned mostly in discussion but it seems to be the \nmost interesting result of the manuscript and could be better analysed. As it currently stands, the \nmanuscript is mostly a quick description of acetylation patterns and does not go very far in novel \ndiscoveries. \n \nMajor concerns \n \nThe authors have identified what it is likely to be the largest collection of microbial acetylation \nsites to date. However, very little is done with this resource beyond a brief description of the \ntaxonomic diversity and biological processes connected with these peptides. Understanding how \nmost of the acetylation regulates protein activity is an important open question and this resource \ncould be a fantastic opportunity to study this problem. The authors should provide some examples \nof how this could be achieved to motivate others to explore this great resource they have \ngenerated. For acetylated sites that match proteins that exist across many bacterial genomes the \nauthors could align the protein sequences and map the acetylation sites to the alignment in order \nto study regions that are often acetylated across a large number of proteomes/species. For a \nnumber of these, there will be structural models that can serve as homology templates to study \nthe structural impact of mapped acetylation sites. Since many of these will be in metabolic \nenzymes there should be many enzymes that are present in many different species and a number \nof structural models available to represent these enzymes. For example, the authors could ask if \nacetylation sites are often near catalytic residues for example. \n \nIn my opinion, the most interesting aspect of the results presented were the potential differences \nin ratio of acetylated vs non-modified peptides along the taxonomic diversity studied. The authors \nspeculated that this differences in ratios could be due to metabolic differences with some species \nbeing known producers of acetate. Acetate can be converted to acetyl-phosphate in some species \nwhich can in turn directly chemically acetylate proteins (Weinert et al. Mol Cell 2013). The authors \ncould try to further analyse this association using isolate genomes or very complete metagenomic \nassembled genomes. Can the authors find a relationship between the enzymes encoded in the \ngenomes of some of the taxonomic groups and their ratios of acetylated vs non-modified peptides \n? \n \nMinor \nThe authors need to make available in supplementary material the list of identified peptides with \nMS relevant associated statistics in a way that can facilitate their further analysis. \n \nThe motif analysis of acetylated peptides could be relegated to supplementary information or \nshown briefly as an additional panel in Figure 1. \n \nThe IBD results serve as an interesting proof-of-principle but the sample sizes are not large \nenough to identify variations that may still be meaningful. This should be addressed in the \ndiscussion. Figure 6 could as well be moved to supplementary and significant differences indicated \nin Figure 5. \n\n\f\f\n", "role": "assistant" } ]	2,020
A scalable CRISPR/Cas9-based fluorescent reporter assay to study DNA double-strand break	REVIEWERS' COMMENTS: Reviewer #2 (Remarks to the Author): The authors have overall addressed my queries.	REVIEWERS' COMMENTS: Reviewer #2 (Remarks to the Author): The authors have overall addressed my queries.	TPR	NC_BS_BIOTEC_011_RR3.txt	{ "criticism": 0, "example": 0, "importance_and_relevance": 0, "materials_and_methods": 0, "praise": 0, "presentation_and_reporting": 0, "results_and_discussion": 0, "suggestion_and_solution": 0, "total": 1 }	{ "criticism": 0, "example": 0, "importance_and_relevance": 0, "materials_and_methods": 0, "praise": 0, "presentation_and_reporting": 0, "results_and_discussion": 0, "suggestion_and_solution": 0 }	0	[ { "content": "Please review this paper.", "role": "user" }, { "content": " \n\n \n\n\fREVIEWERS' COMMENTS: \n \nReviewer #2 (Remarks to the Author): \n \nThe authors have overall addressed my queries. \n\n\f\n", "role": "assistant" } ]	2,020
A scalable CRISPR/Cas9-based fluorescent reporter assay to study DNA double-strand break	Reviewers' comments: Reviewer #1 (Remarks to the Author): The manuscript from Dr. Mardin’s group describes a colorimetric assay system (color assay tracing repair (CAT-R)) to simultaneously monitor outcomes of DSB repair via end-protection and end- resection pathways. This method relies on CRISPR/Cas9 to introduce DSBs in a tandem fluorescent reporter, which can distinguish small insertions/deletions from large deletions. Further, they validated the utility of the CAT-R system using 1) clinically relevant small molecule pharmacological inhibitors and 2) in a custom-designed genetic screen targeting gene involved in DDR to evaluate the contribution of these genes in DNA DSB repair choice. Notably, they found through the assay system that impairing nucleotide excision repair (NER) favor error-free repair, providing a possible way to increase the rate of CRISPR-based knock-ins by homology-directed repair. The concept of measuring DSB repair choices in cells has been widely studied in DNA damage repair field, and many similar assay system has been developed Certo, M. T. et al. Nat. Methods 8, 671–676 (2011), Gunn, A. & Stark, J. M. Methods Mol. Biol. 920, 379–91 (2012), Kuhar, R. et al., Nucleic Acids Res. 42, e4 (2016), Moynahan et. Al. Molecular Cell, 263-272 (2001). The unique feature of the assay system described in this manuscript is the usage of a CRISPR/ Cas9 to introduce specific lesions in the eGFP gene. They generated a HEK293CAT-R cell-line and tested the system by targeting individual DSB repair genes or chemical inhibitors. Most experiments are well designed, and data are organized in a way to support the main conclusion. However, there are some concerns regarding the novelty of the study and result interpretations. The most interesting observation that the Authors just touched upon was the importance of modulating NER to increase CRISPR-based knock-in efficiency. Developing a story based on that observation would have been more exciting and will be of general interest. Comments: 1) The Authors need to clearly mention superiority of the proposed assay system from other DSB repair assays already present. Additionally, multiple cell lines should be tested as clearly HEK293 and RPE1 showed differences in basal level. 2) The double-positive (mCherry and eGFP) population shows both un-transfected and error-free repair active cells. This might affect the interpretation of the overall repair capacity and repair choice of the cells. The author needs to state how they have overcome this issue clearly. 3) Between the large deletions (double negative population) Vs. Indels (eGFP negative) there will be a range of other events, is the assay sensitive enough to measure the array of fluorescence change? This might be helpful in understanding the extent of DDR protein's influence on pathway choice. 4) Using a CRISPR library to asses the role of different repair proteins on DSB repair choices is a robust way to test the system and identify new dependencies on pathway choice. However, the authors need to address the transfection efficiency of both gRNA to eGFP and CRISPR library to cells. From that screen, authors found that the NER pathway plays an active role in CRISPR-knock in ability (although clear data is missing), this data has a potential and should be expanded and studied with more positive and negative controls to make a claim. Reviewer #2 (Remarks to the Author): The manuscript by Roidos et al describes a scalable system to analyze events that can happen at the site of a double-strand break. The system is based on generating a cell line stably expressing (1) two proteins with different fluorescent properties from the same promoter and (2) inducible Cas9 endonuclease. The authors then claim that transfection of a gRNA targeting one of the two fluorescent proteins (mCherry and GFP) would lead to one of three events after induction of Cas9: (1) error-free repair and expression of both proteins, (2) “small” deletions or insertions leading to loss of that protein or (3) “large” deletions leading to loss of both proteins. The three populations of cells are then detected by flow cytometry. The authors provide evidence that the balance of the three outcomes in this two-color assay could distinguish the choice of the DNA double-strand repair system used to repair the lesion. They go on to show potential applicability of this assay in drug and genetic screen to search for chemical or genetic entities that could affect certain DNA repair pathways by measuring shifts in the three populations of cells. The method is interesting and potentially applicable for high-throughput screens, however, the manuscript has some major issues related to interpretation, validation and reproducibility of the method that need to be addresses before it can be suitable for publication as described below. 1. The approach has one fundamental flaw – the DNA molecules digested by Cas9 and repaired error-free can also be cleaved again until they are repaired with errors such that they can longer be substrates for the Cas9 enzyme. Therefore, this is not a very good system to estimate the fraction of error-free repair. 2. Related to the point above, it is impossible to distinguish error-free repair and uncleaved DNA, further arguing that error-free repair can not be estimated reliably based on this system. 3. The authors should validate their system by sorting the different populations of cells (mCherry+/GFP-, mCherry-/GFP-, mCherry+/GFP+) and performing NGS on each one of them to make sure that the profiles of indels and deletions actually follow their flow analysis. Furthermore, the authors should actually measure the sizes of their large deletions using some long read NGS techniques on the sorted populations of cells. These are very important issues not addressed in the manuscript. 4. The authors designed the targeting gRNA such that deletions >250bp would inactivate both proteins. Such deletions are called “large”. The ratio between “small” indels and “large” deletions serves as the basis for distinguishing the contributions of different repair pathways. However, the reason for choosing the 250 bp threshold to distinguish different repair pathways is not described in the paper. 5. I was very surprised to not find a single p-value estimate of significance of the many effects described in the manuscript. For example, there are no estimates of significance of the observed effects in the knockouts of specific genes (Figure 2), in the genetic screen (Figure 4) or in the treatments with specific drugs (Figure 3). The authors must provide these estimates, and they have to be based on analyses of these various knockouts and treatments in different biological replicas, ie different batches of cells. 6. Related to the above, the authors do not provide any information on the stability of their method in different biological replicas. They only mention 3 technical replicas in Methods. The information on biological replication is critical to understanding the robustness of their method and validity of the conclusions. 7. I am somewhat surprised that the authors did not develop cell lines stably expressing the gRNA. Transfection of gRNA complexes can be quite variable and also lead to non-physiologically high amounts of the synthetic RNAs in cells that in turn can lead to all sorts of artifacts (PMID 26697058). For example, there is a big difference between the two WT samples in Figure 2A. This is another reason to show how reproducible the system is in different biological replicas. Also, a stable cell line would be an improvement since it is more likely to stabilize performance of this method across multiple laboratories if it is to be adopted. 8. Related to the above, the authors chose immortalized non-cancerous cell line HEK293. Would it not be more appropriate to test this system in cancerous cell lines since they are more likely to be used as drug targets? I would imagine that the reason HEK293 cells are used here is because they are easy to transfect, but this significantly limits applicability of this method. This is another reason to generate cell lines stably expressing the gRNA(s). 9. The attempt to use this method to build a predictor based on machine-learning approach random forest to classify different drugs is interesting. However, I see a number of issues with it: 9.1. I can not follow the “Random forest model – bioinformatics pipeline” section in the Methods. It definitely needs to be improved for clarity and logic. Among others, the authors need to explain why they remove data points similar to DSB controls and what are those DSB controls. What is “N” in the Table 1? Number of wells? And so on… 9.2. The performance of this model in the text needs to be described better. Did the authors validate it only on drugs they tested? What if they used drugs not used to build the model? If this method is to be applied broadly, this question has to be addressed. Some additional comments: 1. In Material and Methods section, the authors wrote: "Single-cell suspensions were analyzed, and cells with strong eGFP (488-440 530/30) and mCherry (561-610/20) signals were sorted using FACSAria I cell sorter (BD 441 Biosciences) to enrich cells harboring the reporter.” Is the reporter cell line represented by clonal or mixed population? Do all the cells have the same single genomic locus insertion? This point is not very clear. 2. Inhibitor studies in Figure 3. Do the authors have evidence that these drugs work as expected at these concentrations in this cell line? 3. In the Discussion, the authors state that “Our results are consistent with a recent study describing large deletions ranging from 250 bps to 6 kbs to occur more than 20% of the cases upon Cas9 induced DSBs in mouse embryonic stem cells and in RPE-1 cells (ref 32).” The reference for this sentence seems to be incorrect. Reviewer #3 (Remarks to the Author): In this manuscript the authors assembled a fluorescence-based reporter that allows simultaneous monitoring of end-protection and end-resection DSB repair processes, and showed that it enabled classification of a panel of DDR-targeting small molecules / DDR genes in choice of DSB repair pathways. The authors went on to demonstrate that inhibiting NER led to increased efficiency in SSTR. Overall the manuscript is clearly written and conclusions were supported by experimental findings. I have a few comments that the authors may consider when revising the manuscript - 1) While it appears that the eGFP-targeting gRNAs are efficient in inducing DSBs at the integrated cassette (Figure 1b; From the scatter plot one would estimate that DSBs were induced in at least 92% of cells) the genomic cleavage assay suggests otherwise (Supp Figure 1b). The authors should explain this apparent discrepancy. 2) It remains unclear how "error-free" repair is effected upon DSB induction in CAT-R. The authors described HR as one possible means but is there a homologous template to allow HR to take place? What is the contribution of small InDels that took place in the multiplicity of 3 nucleotides that would presumably preserve eGFP? 3) The statement that "These results agree well with the idea that in the presence of PARP, the DSBs are repaired by alternative NHEJ pathway, which may contribute to the formation of large deletions due to extensive end-resection" is confusing given that "PARP inhibitors led to a reduction of small InDels and an increase of large deletions on average by 4% (+/- 1.6) at 50nM (Figure 3f). 4) CAT-R does not appear to be very sensitive in detecting change in "error-free" repair when established HR factors are inactivated. Perhaps the authors can discuss the limitation of the DSB repair reporter. 5) The authors showed that pre-treatment with the DDR gene-targeting small molecules affect cell cycle distribution. (How) Does change in cell cycle distribution affect DSB induction per se? 6) I see that there are multiple "WT" plots in Figure 2a, 2c and 2e. Are they "WT" representative plots?	Reviewers' comments: Reviewer #1 (Remarks to the Author): The manuscript from Dr. Mardin’s group describes a colorimetric assay system (color assay tracing repair (CAT-R)) to simultaneously monitor outcomes of DSB repair via end-protection and end- resection pathways. This method relies on CRISPR/Cas9 to introduce DSBs in a tandem fluorescent reporter, which can distinguish small insertions/deletions from large deletions. Further, they validated the utility of the CAT-R system using 1) clinically relevant small molecule pharmacological inhibitors and 2) in a custom-designed genetic screen targeting gene involved in DDR to evaluate the contribution of these genes in DNA DSB repair choice. Notably, they found through the assay system that impairing nucleotide excision repair (NER) favor error-free repair, providing a possible way to increase the rate of CRISPR-based knock-ins by homology-directed repair. The concept of measuring DSB repair choices in cells has been widely studied in DNA damage repair field, and many similar assay system has been developed Certo, M. T. et al. Nat. Methods 8, 671–676 (2011), Gunn, A. & Stark, J. M. Methods Mol. Biol. 920, 379–91 (2012), Kuhar, R. et al., Nucleic Acids Res. 42, e4 (2016), Moynahan et. Al. Molecular Cell, 263-272 (2001). The unique feature of the assay system described in this manuscript is the usage of a CRISPR/ Cas9 to introduce specific lesions in the eGFP gene. They generated a HEK293CAT-R cell-line and tested the system by targeting individual DSB repair genes or chemical inhibitors. Most experiments are well designed, and data are organized in a way to support the main conclusion. However, there are some concerns regarding the novelty of the study and result interpretations. The most interesting observation that the Authors just touched upon was the importance of modulating NER to increase CRISPR-based knock-in efficiency. Developing a story based on that observation would have been more exciting and will be of general interest. Comments: 1) The Authors need to clearly mention superiority of the proposed assay system from other DSB repair assays already present. Additionally, multiple cell lines should be tested as clearly HEK293 and RPE1 showed differences in basal level. 2) The double-positive (mCherry and eGFP) population shows both un-transfected and error-free repair active cells. This might affect the interpretation of the overall repair capacity and repair choice of the cells. The author needs to state how they have overcome this issue clearly. 3) Between the large deletions (double negative population) Vs. Indels (eGFP negative) there will be a range of other events, is the assay sensitive enough to measure the array of fluorescence change? This might be helpful in understanding the extent of DDR protein's influence on pathway choice. 4) Using a CRISPR library to asses the role of different repair proteins on DSB repair choices is a robust way to test the system and identify new dependencies on pathway choice. However, the authors need to address the transfection efficiency of both gRNA to eGFP and CRISPR library to cells. From that screen, authors found that the NER pathway plays an active role in CRISPR-knock in ability (although clear data is missing), this data has a potential and should be expanded and studied with more positive and negative controls to make a claim. Reviewer #2 (Remarks to the Author): The manuscript by Roidos et al describes a scalable system to analyze events that can happen at the site of a double-strand break. The system is based on generating a cell line stably expressing (1) two proteins with different fluorescent properties from the same promoter and (2) inducible Cas9 endonuclease. The authors then claim that transfection of a gRNA targeting one of the two fluorescent proteins (mCherry and GFP) would lead to one of three events after induction of Cas9: (1) error-free repair and expression of both proteins, (2) “small” deletions or insertions leading to loss of that protein or (3) “large” deletions leading to loss of both proteins. The three populations of cells are then detected by flow cytometry. The authors provide evidence that the balance of the three outcomes in this two-color assay could distinguish the choice of the DNA double-strand repair system used to repair the lesion. They go on to show potential applicability of this assay in drug and genetic screen to search for chemical or genetic entities that could affect certain DNA repair pathways by measuring shifts in the three populations of cells. The method is interesting and potentially applicable for high-throughput screens, however, the manuscript has some major issues related to interpretation, validation and reproducibility of the method that need to be addresses before it can be suitable for publication as described below. 1. The approach has one fundamental flaw – the DNA molecules digested by Cas9 and repaired error-free can also be cleaved again until they are repaired with errors such that they can longer be substrates for the Cas9 enzyme. Therefore, this is not a very good system to estimate the fraction of error-free repair. 2. Related to the point above, it is impossible to distinguish error-free repair and uncleaved DNA, further arguing that error-free repair can not be estimated reliably based on this system. 3. The authors should validate their system by sorting the different populations of cells (mCherry+/GFP-, mCherry-/GFP-, mCherry+/GFP+) and performing NGS on each one of them to make sure that the profiles of indels and deletions actually follow their flow analysis. Furthermore, the authors should actually measure the sizes of their large deletions using some long read NGS techniques on the sorted populations of cells. These are very important issues not addressed in the manuscript. 4. The authors designed the targeting gRNA such that deletions >250bp would inactivate both proteins. Such deletions are called “large”. The ratio between “small” indels and “large” deletions serves as the basis for distinguishing the contributions of different repair pathways. However, the reason for choosing the 250 bp threshold to distinguish different repair pathways is not described in the paper. 5. I was very surprised to not find a single p-value estimate of significance of the many effects described in the manuscript. For example, there are no estimates of significance of the observed effects in the knockouts of specific genes (Figure 2), in the genetic screen (Figure 4) or in the treatments with specific drugs (Figure 3). The authors must provide these estimates, and they have to be based on analyses of these various knockouts and treatments in different biological replicas, ie different batches of cells. 6. Related to the above, the authors do not provide any information on the stability of their method in different biological replicas. They only mention 3 technical replicas in Methods. The information on biological replication is critical to understanding the robustness of their method and validity of the conclusions. 7. I am somewhat surprised that the authors did not develop cell lines stably expressing the gRNA. Transfection of gRNA complexes can be quite variable and also lead to non-physiologically high amounts of the synthetic RNAs in cells that in turn can lead to all sorts of artifacts (PMID 26697058). For example, there is a big difference between the two WT samples in Figure 2A. This is another reason to show how reproducible the system is in different biological replicas. Also, a stable cell line would be an improvement since it is more likely to stabilize performance of this method across multiple laboratories if it is to be adopted. 8. Related to the above, the authors chose immortalized non-cancerous cell line HEK293. Would it not be more appropriate to test this system in cancerous cell lines since they are more likely to be used as drug targets? I would imagine that the reason HEK293 cells are used here is because they are easy to transfect, but this significantly limits applicability of this method. This is another reason to generate cell lines stably expressing the gRNA(s). 9. The attempt to use this method to build a predictor based on machine-learning approach random forest to classify different drugs is interesting. However, I see a number of issues with it: 9.1. I can not follow the “Random forest model – bioinformatics pipeline” section in the Methods. It definitely needs to be improved for clarity and logic. Among others, the authors need to explain why they remove data points similar to DSB controls and what are those DSB controls. What is “N” in the Table 1? Number of wells? And so on… 9.2. The performance of this model in the text needs to be described better. Did the authors validate it only on drugs they tested? What if they used drugs not used to build the model? If this method is to be applied broadly, this question has to be addressed. Some additional comments: 1. In Material and Methods section, the authors wrote: "Single-cell suspensions were analyzed, and cells with strong eGFP (488-440 530/30) and mCherry (561-610/20) signals were sorted using FACSAria I cell sorter (BD 441 Biosciences) to enrich cells harboring the reporter.” Is the reporter cell line represented by clonal or mixed population? Do all the cells have the same single genomic locus insertion? This point is not very clear. 2. Inhibitor studies in Figure 3. Do the authors have evidence that these drugs work as expected at these concentrations in this cell line? 3. In the Discussion, the authors state that “Our results are consistent with a recent study describing large deletions ranging from 250 bps to 6 kbs to occur more than 20% of the cases upon Cas9 induced DSBs in mouse embryonic stem cells and in RPE-1 cells (ref 32).” The reference for this sentence seems to be incorrect. Reviewer #3 (Remarks to the Author): In this manuscript the authors assembled a fluorescence-based reporter that allows simultaneous monitoring of end-protection and end-resection DSB repair processes, and showed that it enabled classification of a panel of DDR-targeting small molecules / DDR genes in choice of DSB repair pathways. The authors went on to demonstrate that inhibiting NER led to increased efficiency in SSTR. Overall the manuscript is clearly written and conclusions were supported by experimental findings. I have a few comments that the authors may consider when revising the manuscript - 1) While it appears that the eGFP-targeting gRNAs are efficient in inducing DSBs at the integrated cassette (Figure 1b; From the scatter plot one would estimate that DSBs were induced in at least 92% of cells) the genomic cleavage assay suggests otherwise (Supp Figure 1b). The authors should explain this apparent discrepancy. 2) It remains unclear how "error-free" repair is effected upon DSB induction in CAT-R. The authors described HR as one possible means but is there a homologous template to allow HR to take place? What is the contribution of small InDels that took place in the multiplicity of 3 nucleotides that would presumably preserve eGFP? 3) The statement that "These results agree well with the idea that in the presence of PARP, the DSBs are repaired by alternative NHEJ pathway, which may contribute to the formation of large deletions due to extensive end-resection" is confusing given that "PARP inhibitors led to a reduction of small InDels and an increase of large deletions on average by 4% (+/- 1.6) at 50nM (Figure 3f). 4) CAT-R does not appear to be very sensitive in detecting change in "error-free" repair when established HR factors are inactivated. Perhaps the authors can discuss the limitation of the DSB repair reporter. 5) The authors showed that pre-treatment with the DDR gene-targeting small molecules affect cell cycle distribution. (How) Does change in cell cycle distribution affect DSB induction per se? 6) I see that there are multiple "WT" plots in Figure 2a, 2c and 2e. Are they "WT" representative plots?	TPR	NC_BS_BIOTEC_011_RR1.txt	{ "criticism": 17, "example": 5, "importance_and_relevance": 5, "materials_and_methods": 64, "praise": 7, "presentation_and_reporting": 12, "results_and_discussion": 12, "suggestion_and_solution": 20, "total": 106 }	{ "criticism": 0.16037735849056603, "example": 0.04716981132075472, "importance_and_relevance": 0.04716981132075472, "materials_and_methods": 0.6037735849056604, "praise": 0.0660377358490566, "presentation_and_reporting": 0.11320754716981132, "results_and_discussion": 0.11320754716981132, "suggestion_and_solution": 0.18867924528301888 }	1.339623	[ { "content": "Please review this paper.", "role": "user" }, { "content": "Reviewers' comments: \n \nReviewer #1 (Remarks to the Author): \n \nThe manuscript from Dr. Mardin’s group describes a colorimetric assay system (color assay tracing \nrepair (CAT-R)) to simultaneously monitor outcomes of DSB repair via end-protection and end-\nresection pathways. This method relies on CRISPR/Cas9 to introduce DSBs in a tandem fluorescent \nreporter, which can distinguish small insertions/deletions from large deletions. Further, they \nvalidated the utility of the CAT-R system using 1) clinically relevant small molecule \npharmacological inhibitors and 2) in a custom-designed genetic screen targeting gene involved in \nDDR to evaluate the contribution of these genes in DNA DSB repair choice. Notably, they found \nthrough the assay system that impairing nucleotide excision repair (NER) favor error-free repair, \nproviding a possible way to increase the rate of CRISPR-based knock-ins by homology-directed \nrepair. \nThe concept of measuring DSB repair choices in cells has been widely studied in DNA damage \nrepair field, and many similar assay system has been developed Certo, M. T. et al. Nat. Methods 8, \n671–676 (2011), Gunn, A. & Stark, J. M. Methods Mol. Biol. 920, 379–91 (2012), Kuhar, R. et al., \nNucleic Acids Res. 42, e4 (2016), Moynahan et. Al. Molecular Cell, 263-272 (2001). The unique \nfeature of the assay system described in this manuscript is the usage of a CRISPR/ Cas9 to \nintroduce specific lesions in the eGFP gene. They generated a HEK293CAT-R cell-line and tested \nthe system by targeting individual DSB repair genes or chemical inhibitors. Most experiments are \nwell designed, and data are organized in a way to support the main conclusion. However, there are \nsome concerns regarding the novelty of the study and result interpretations. The most interesting \nobservation that the Authors just touched upon was the importance of modulating NER to increase \nCRISPR-based knock-in efficiency. Developing a story based on that observation would have been \nmore exciting and will be of general interest. \nComments: \n1) The Authors need to clearly mention superiority of the proposed assay system from other DSB \nrepair assays already present. Additionally, multiple cell lines should be tested as clearly HEK293 \nand RPE1 showed differences in basal level. \n2) The double-positive (mCherry and eGFP) population shows both un-transfected and error-free \nrepair active cells. This might affect the interpretation of the overall repair capacity and repair \nchoice of the cells. The author needs to state how they have overcome this issue clearly. \n3) Between the large deletions (double negative population) Vs. Indels (eGFP negative) there will \nbe a range of other events, is the assay sensitive enough to measure the array of fluorescence \nchange? This might be helpful in understanding the extent of DDR protein's influence on pathway \nchoice. \n4) Using a CRISPR library to asses the role of different repair proteins on DSB repair choices is a \nrobust way to test the system and identify new dependencies on pathway choice. However, the \nauthors need to address the transfection efficiency of both gRNA to eGFP and CRISPR library to \ncells. From that screen, authors found that the NER pathway plays an active role in CRISPR-knock \nin ability (although clear data is missing), this data has a potential and should be expanded and \nstudied with more positive and negative controls to make a claim. \n \n \n \n \nReviewer #2 (Remarks to the Author): \n \nThe manuscript by Roidos et al describes a scalable system to analyze events that can happen at \nthe site of a double-strand break. The system is based on generating a cell line stably expressing \n(1) two proteins with different fluorescent properties from the same promoter and (2) inducible \nCas9 endonuclease. The authors then claim that transfection of a gRNA targeting one of the two \nfluorescent proteins (mCherry and GFP) would lead to one of three events after induction of Cas9: \n(1) error-free repair and expression of both proteins, (2) “small” deletions or insertions leading to \n\n\floss of that protein or (3) “large” deletions leading to loss of both proteins. The three populations \nof cells are then detected by flow cytometry. The authors provide evidence that the balance of the \nthree outcomes in this two-color assay could distinguish the choice of the DNA double-strand \nrepair system used to repair the lesion. They go on to show potential applicability of this assay in \ndrug and genetic screen to search for chemical or genetic entities that could affect certain DNA \nrepair pathways by measuring shifts in the three populations of cells. \n \nThe method is interesting and potentially applicable for high-throughput screens, however, the \nmanuscript has some major issues related to interpretation, validation and reproducibility of the \nmethod that need to be addresses before it can be suitable for publication as described below. \n \n1. The approach has one fundamental flaw – the DNA molecules digested by Cas9 and repaired \nerror-free can also be cleaved again until they are repaired with errors such that they can longer \nbe substrates for the Cas9 enzyme. Therefore, this is not a very good system to estimate the \nfraction of error-free repair. \n \n2. Related to the point above, it is impossible to distinguish error-free repair and uncleaved DNA, \nfurther arguing that error-free repair can not be estimated reliably based on this system. \n \n3. The authors should validate their system by sorting the different populations of cells \n(mCherry+/GFP-, mCherry-/GFP-, mCherry+/GFP+) and performing NGS on each one of them to \nmake sure that the profiles of indels and deletions actually follow their flow analysis. Furthermore, \nthe authors should actually measure the sizes of their large deletions using some long read NGS \ntechniques on the sorted populations of cells. These are very important issues not addressed in the \nmanuscript. \n \n4. The authors designed the targeting gRNA such that deletions >250bp would inactivate both \nproteins. Such deletions are called “large”. The ratio between “small” indels and “large” deletions \nserves as the basis for distinguishing the contributions of different repair pathways. However, the \nreason for choosing the 250 bp threshold to distinguish different repair pathways is not described \nin the paper. \n \n5. I was very surprised to not find a single p-value estimate of significance of the many effects \ndescribed in the manuscript. For example, there are no estimates of significance of the observed \neffects in the knockouts of specific genes (Figure 2), in the genetic screen (Figure 4) or in the \ntreatments with specific drugs (Figure 3). The authors must provide these estimates, and they \nhave to be based on analyses of these various knockouts and treatments in different biological \nreplicas, ie different batches of cells. \n \n6. Related to the above, the authors do not provide any information on the stability of their \nmethod in different biological replicas. They only mention 3 technical replicas in Methods. The \ninformation on biological replication is critical to understanding the robustness of their method and \nvalidity of the conclusions. \n \n7. I am somewhat surprised that the authors did not develop cell lines stably expressing the gRNA. \nTransfection of gRNA complexes can be quite variable and also lead to non-physiologically high \namounts of the synthetic RNAs in cells that in turn can lead to all sorts of artifacts (PMID \n26697058). For example, there is a big difference between the two WT samples in Figure 2A. This \nis another reason to show how reproducible the system is in different biological replicas. Also, a \nstable cell line would be an improvement since it is more likely to stabilize performance of this \nmethod across multiple laboratories if it is to be adopted. \n \n8. Related to the above, the authors chose immortalized non-cancerous cell line HEK293. Would it \nnot be more appropriate to test this system in cancerous cell lines since they are more likely to be \nused as drug targets? I would imagine that the reason HEK293 cells are used here is because they \n\n\fare easy to transfect, but this significantly limits applicability of this method. This is another \nreason to generate cell lines stably expressing the gRNA(s). \n \n9. The attempt to use this method to build a predictor based on machine-learning approach \nrandom forest to classify different drugs is interesting. However, I see a number of issues with it: \n \n9.1. I can not follow the “Random forest model – bioinformatics pipeline” section in the Methods. It \ndefinitely needs to be improved for clarity and logic. Among others, the authors need to explain \nwhy they remove data points similar to DSB controls and what are those DSB controls. What is “N” \nin the Table 1? Number of wells? And so on… \n \n9.2. The performance of this model in the text needs to be described better. Did the authors \nvalidate it only on drugs they tested? What if they used drugs not used to build the model? If this \nmethod is to be applied broadly, this question has to be addressed. \n \nSome additional comments: \n \n1. In Material and Methods section, the authors wrote: \"Single-cell suspensions were analyzed, \nand cells with strong eGFP (488-440 530/30) and mCherry (561-610/20) signals were sorted using \nFACSAria I cell sorter (BD 441 Biosciences) to enrich cells harboring the reporter.” Is the reporter \ncell line represented by clonal or mixed population? Do all the cells have the same single genomic \nlocus insertion? This point is not very clear. \n \n2. Inhibitor studies in Figure 3. Do the authors have evidence that these drugs work as expected \nat these concentrations in this cell line? \n \n3. In the Discussion, the authors state that “Our results are consistent with a recent study \ndescribing large deletions ranging from 250 bps to 6 kbs to occur more than 20% of the cases \nupon Cas9 induced DSBs in mouse embryonic stem cells and in RPE-1 cells (ref 32).” The \nreference for this sentence seems to be incorrect. \n \n \n \n \n \nReviewer #3 (Remarks to the Author): \n \nIn this manuscript the authors assembled a fluorescence-based reporter that allows simultaneous \nmonitoring of end-protection and end-resection DSB repair processes, and showed that it enabled \nclassification of a panel of DDR-targeting small molecules / DDR genes in choice of DSB repair \npathways. The authors went on to demonstrate that inhibiting NER led to increased efficiency in \nSSTR. Overall the manuscript is clearly written and conclusions were supported by experimental \nfindings. \n \nI have a few comments that the authors may consider when revising the manuscript - \n \n1) While it appears that the eGFP-targeting gRNAs are efficient in inducing DSBs at the integrated \ncassette (Figure 1b; From the scatter plot one would estimate that DSBs were induced in at least \n92% of cells) the genomic cleavage assay suggests otherwise (Supp Figure 1b). The authors \nshould explain this apparent discrepancy. \n \n2) It remains unclear how \"error-free\" repair is effected upon DSB induction in CAT-R. The authors \ndescribed HR as one possible means but is there a homologous template to allow HR to take place? \nWhat is the contribution of small InDels that took place in the multiplicity of 3 nucleotides that \nwould presumably preserve eGFP? \n\n\f \n3) The statement that \"These results agree well with the idea that in the presence of PARP, the \nDSBs are repaired by alternative NHEJ pathway, which may contribute to the formation of large \ndeletions due to extensive end-resection\" is confusing given that \"PARP inhibitors led to a \nreduction of small InDels and an increase of large deletions on average by 4% (+/- 1.6) at 50nM \n(Figure 3f). \n \n4) CAT-R does not appear to be very sensitive in detecting change in \"error-free\" repair when \nestablished HR factors are inactivated. Perhaps the authors can discuss the limitation of the DSB \nrepair reporter. \n \n5) The authors showed that pre-treatment with the DDR gene-targeting small molecules affect cell \ncycle distribution. (How) Does change in cell cycle distribution affect DSB induction per se? \n \n6) I see that there are multiple \"WT\" plots in Figure 2a, 2c and 2e. Are they \"WT\" representative \nplots? \n\n\f\f\n", "role": "assistant" } ]	2,020
A scalable CRISPR/Cas9-based fluorescent reporter assay to study DNA double-strand break	REVIEWER COMMENTS Reviewer #1 (Remarks to the Author): The authors have addressed most of my comments. Reviewer #2 (Remarks to the Author): The manuscript has improved significantly and the authors have addressed a significant fraction of the queries. However, I still have the following issues. 1. The authors have presumably sorted the mCherry+/GFP- and mCherry-/GFP- populations and directly measured the distributions of the lengths of breaks in each sub-population using Illumina and Oxford Nanopore platforms. However, the authors do not explicitly state what populations were used for the sequencing neither in the main text nor the Methods part. They only use word “sorted” in the Figure 1 and Supplementary Figure 1 and therefore it is not exactly clear what was done. This analysis is at the core of validation of their high-throughput assay. As such, it should be emphasized and described in much more detail in the main text and methods. It would be also good to see a panel in a main figure that shows exact gating parameters used in this sorting- sequencing experiment. 2. In terms of the inducible Cas9 cell lines, the authors state that they cannot detect Cas9 expression without Doxycycline induction, while they still observe the unwanted DSBs events in the absence of induction and claim that those are due to the leakiness of the inducible promoters. I think the authors should provide more evidence to support this conclusion, and/or they should provide a more detailed clarification of the observed results. Furthermore, the Cas9-mediated site- specific DNA cleavage is at the core of their method and the authors should show that the increase in the breaks and increase in the fraction of the mCherry+/GFP- and mCherry-/GFP- populations is Dox-dependent. 3. The authors now include estimates of the statistical significance, however, they only use them in the context of yes or no answers as to the significance of their observations. The readers might want to look at the actual values to make their own determination about relative significance of various treatments. Therefore, I think the author should make a Supplementary Dataset with the actual p-values and %-ages for every sorted fraction for every drug treatment and gene knockout (eg, ATM, PARP etc) tested in this work. 4. The authors provide 3 supplementary tables, but they only mention one of them in the text. Reviewer #3 (Remarks to the Author): The authors have adequately addressed my critiques.	REVIEWER COMMENTS Reviewer #1 (Remarks to the Author): The authors have addressed most of my comments. Reviewer #2 (Remarks to the Author): The manuscript has improved significantly and the authors have addressed a significant fraction of the queries. However, I still have the following issues. 1. The authors have presumably sorted the mCherry+/GFP- and mCherry-/GFP- populations and directly measured the distributions of the lengths of breaks in each sub-population using Illumina and Oxford Nanopore platforms. However, the authors do not explicitly state what populations were used for the sequencing neither in the main text nor the Methods part. They only use word “sorted” in the Figure 1 and Supplementary Figure 1 and therefore it is not exactly clear what was done. This analysis is at the core of validation of their high-throughput assay. As such, it should be emphasized and described in much more detail in the main text and methods. It would be also good to see a panel in a main figure that shows exact gating parameters used in this sorting- sequencing experiment. 2. In terms of the inducible Cas9 cell lines, the authors state that they cannot detect Cas9 expression without Doxycycline induction, while they still observe the unwanted DSBs events in the absence of induction and claim that those are due to the leakiness of the inducible promoters. I think the authors should provide more evidence to support this conclusion, and/or they should provide a more detailed clarification of the observed results. Furthermore, the Cas9-mediated site- specific DNA cleavage is at the core of their method and the authors should show that the increase in the breaks and increase in the fraction of the mCherry+/GFP- and mCherry-/GFP- populations is Dox-dependent. 3. The authors now include estimates of the statistical significance, however, they only use them in the context of yes or no answers as to the significance of their observations. The readers might want to look at the actual values to make their own determination about relative significance of various treatments. Therefore, I think the author should make a Supplementary Dataset with the actual p-values and %-ages for every sorted fraction for every drug treatment and gene knockout (eg, ATM, PARP etc) tested in this work. 4. The authors provide 3 supplementary tables, but they only mention one of them in the text. Reviewer #3 (Remarks to the Author): The authors have adequately addressed my critiques.	TPR	NC_BS_BIOTEC_011_RR2.txt	{ "criticism": 3, "example": 1, "importance_and_relevance": 0, "materials_and_methods": 10, "praise": 2, "presentation_and_reporting": 7, "results_and_discussion": 2, "suggestion_and_solution": 6, "total": 21 }	{ "criticism": 0.14285714285714285, "example": 0.047619047619047616, "importance_and_relevance": 0, "materials_and_methods": 0.47619047619047616, "praise": 0.09523809523809523, "presentation_and_reporting": 0.3333333333333333, "results_and_discussion": 0.09523809523809523, "suggestion_and_solution": 0.2857142857142857 }	1.47619	[ { "content": "Please review this paper.", "role": "user" }, { "content": "\n\n\fREVIEWER COMMENTS \n \n\n \nReviewer #1 (Remarks to the Author): \n \nThe authors have addressed most of my comments. \n \n \n\n \nReviewer #2 (Remarks to the Author): \n \nThe manuscript has improved significantly and the authors have addressed a significant fraction of \nthe queries. However, I still have the following issues. \n \n1. The authors have presumably sorted the mCherry+/GFP- and mCherry-/GFP- populations and \ndirectly measured the distributions of the lengths of breaks in each sub-population using Illumina \nand Oxford Nanopore platforms. However, the authors do not explicitly state what populations \nwere used for the sequencing neither in the main text nor the Methods part. They only use word \n“sorted” in the Figure 1 and Supplementary Figure 1 and therefore it is not exactly clear what was \ndone. This analysis is at the core of validation of their high-throughput assay. As such, it should be \nemphasized and described in much more detail in the main text and methods. It would be also \ngood to see a panel in a main figure that shows exact gating parameters used in this sorting-\nsequencing experiment. \n \n2. In terms of the inducible Cas9 cell lines, the authors state that they cannot detect Cas9 \nexpression without Doxycycline induction, while they still observe the unwanted DSBs events in \nthe absence of induction and claim that those are due to the leakiness of the inducible promoters. \nI think the authors should provide more evidence to support this conclusion, and/or they should \nprovide a more detailed clarification of the observed results. Furthermore, the Cas9-mediated site-\nspecific DNA cleavage is at the core of their method and the authors should show that the increase \nin the breaks and increase in the fraction of the mCherry+/GFP- and mCherry-/GFP- populations is \nDox-dependent. \n \n3. The authors now include estimates of the statistical significance, however, they only use them \nin the context of yes or no answers as to the significance of their observations. The readers might \nwant to look at the actual values to make their own determination about relative significance of \nvarious treatments. Therefore, I think the author should make a Supplementary Dataset with the \nactual p-values and %-ages for every sorted fraction for every drug treatment and gene knockout \n(eg, ATM, PARP etc) tested in this work. \n \n4. The authors provide 3 supplementary tables, but they only mention one of them in the text. \n \n \n \nReviewer #3 (Remarks to the Author): \n \nThe authors have adequately addressed my critiques. \n\n\f", "role": "assistant" } ]	2,020
High throughput error corrected Nanopore single cell transcriptome sequencing	Reviewers' Comments: Reviewer #1: Remarks to the Author: The authors have addressed my concerns on the description of the transcript novelty. It was nice to see that novel transcripts are detected at least by 2 cells. However, I would strongly suggest they use SQANTI categories to describe the novelty as this is the standard in the field and most of their definitions of novel transcript classes correspond to an already defined SQANTI category. The authors should be sure none of the novel transcripts have non-canonical splice sites. I do not expect this to be a major problem because of the 2 cells filter, but just in case.	Reviewers' Comments: Reviewer #1: Remarks to the Author: The authors have addressed my concerns on the description of the transcript novelty. It was nice to see that novel transcripts are detected at least by 2 cells. However, I would strongly suggest they use SQANTI categories to describe the novelty as this is the standard in the field and most of their definitions of novel transcript classes correspond to an already defined SQANTI category. The authors should be sure none of the novel transcripts have non-canonical splice sites. I do not expect this to be a major problem because of the 2 cells filter, but just in case.	TPR	NC_BS_BIOTEC_014_RR1.txt	{ "criticism": 0, "example": 0, "importance_and_relevance": 0, "materials_and_methods": 4, "praise": 1, "presentation_and_reporting": 1, "results_and_discussion": 0, "suggestion_and_solution": 2, "total": 5 }	{ "criticism": 0, "example": 0, "importance_and_relevance": 0, "materials_and_methods": 0.8, "praise": 0.2, "presentation_and_reporting": 0.2, "results_and_discussion": 0, "suggestion_and_solution": 0.4 }	1.6	[ { "content": "Please review this paper.", "role": "user" }, { "content": "\n \n\nReviewers' Comments: \n \nReviewer #1: \nRemarks to the Author: \nThe authors have addressed my concerns on the description of the transcript novelty. It was nice to \nsee that novel transcripts are detected at least by 2 cells. \n \nHowever, I would strongly suggest they use SQANTI categories to describe the novelty as this is the \nstandard in the field and most of their definitions of novel transcript classes correspond to an already \ndefined SQANTI category. \n \nThe authors should be sure none of the novel transcripts have non-canonical splice sites. I do not \nexpect this to be a major problem because of the 2 cells filter, but just in case. \n\n \n\n\n", "role": "assistant" } ]	2,019
Faecal microbiota transplantation for the treatment of diarrhoea induced by tyrosine-kinase	REVIEWERS' COMMENTS: Reviewer #1 (Remarks to the Author): I have no further comments following a review of the authors’ response to reviewers’ comments Reviewer #2 (Remarks to the Author): 1. One of my primary concerns was that all patients underwent a bowel prep which puts the sham group at a disadvantage as it will decrease the diversity of an already disrupted community. That is likely why we see nearly no improvement in the sham group. The authors have responded to this by providing a separate explanation for each of these using two arguments that 1) bowel prep does not affect gut microbiota by referencing a study that used PCR-DGGE method and 2) other groups have used water as a control. There is sufficient data that bowel prep does indeed alter gut microbiome in studies that use next generation sequencing platforms as the one used by the authors (https://pubmed.ncbi.nlm.nih.gov/27015015/). The authors did not measure the microbiome before and after bowel prep in their cohort so they cannot definitively establish this. The authors then described several large trials that have used water as a sham placebo but fail to mention that none of them gave a bowel prep prior to treatment and my concern was not based on the fact that water cannot be a placebo but rather that the authors could have further disrupted an already vulnerable community with bowel prep and that can explain the lack of any response in the sham group. The authors then argue against using autologous stool as a comparator group and I am not sure if they got confused with my comment “sham FMT refers to patients own stool”. I had suggested that they should have used the patient’s own stool (allogenic FMT) as a control group and all the examples provided by the authors do exactly that- use the patient’s own stool as the comparator and not autologous stool as the authors mention in the rebuttal. I am not sure what the authors mean by lack of protocol for collecting stool from the patients with diarrhea. A sham/allogenic FMT would follow the exact same protocol for collection and processing as the donor and all the examples outlined by the authors in fact do this. 2. The authors did not respond to my comment “The data is too preliminary to discuss the role of engraftment in driving improvement as the study lacks an appropriate control group. Moreover it would be helpful if the authors consider tracking the microbes longitudinally (given they have the samples) to determine if changes in relative abundance of microbes such as Akkermansia correspond to recurrence of diarrhea.” Further the authors did not perform longitudinal sampling on the sham group and compare the active FMT to time point 0 of the sham group so we do not know the level of variability/changes in the microbiome in the sham group following intervention. Overall my enthusiasm is in fact further dampened following author’s response as it highlights additional drawbacks in the study design and the interpretation of the results. Reviewer #3 (Remarks to the Author): Authors have satisfactory assessed all my queries. Reviewer #4 (Remarks to the Author): All my comments have been addressed appropriately. Looks good!	REVIEWERS' COMMENTS: Reviewer #1 (Remarks to the Author): I have no further comments following a review of the authors’ response to reviewers’ comments Reviewer #2 (Remarks to the Author): 1. One of my primary concerns was that all patients underwent a bowel prep which puts the sham group at a disadvantage as it will decrease the diversity of an already disrupted community. That is likely why we see nearly no improvement in the sham group. The authors have responded to this by providing a separate explanation for each of these using two arguments that 1) bowel prep does not affect gut microbiota by referencing a study that used PCR-DGGE method and 2) other groups have used water as a control. There is sufficient data that bowel prep does indeed alter gut microbiome in studies that use next generation sequencing platforms as the one used by the authors (https://pubmed.ncbi.nlm.nih.gov/27015015/). The authors did not measure the microbiome before and after bowel prep in their cohort so they cannot definitively establish this. The authors then described several large trials that have used water as a sham placebo but fail to mention that none of them gave a bowel prep prior to treatment and my concern was not based on the fact that water cannot be a placebo but rather that the authors could have further disrupted an already vulnerable community with bowel prep and that can explain the lack of any response in the sham group. The authors then argue against using autologous stool as a comparator group and I am not sure if they got confused with my comment “sham FMT refers to patients own stool”. I had suggested that they should have used the patient’s own stool (allogenic FMT) as a control group and all the examples provided by the authors do exactly that- use the patient’s own stool as the comparator and not autologous stool as the authors mention in the rebuttal. I am not sure what the authors mean by lack of protocol for collecting stool from the patients with diarrhea. A sham/allogenic FMT would follow the exact same protocol for collection and processing as the donor and all the examples outlined by the authors in fact do this. 2. The authors did not respond to my comment “The data is too preliminary to discuss the role of engraftment in driving improvement as the study lacks an appropriate control group. Moreover it would be helpful if the authors consider tracking the microbes longitudinally (given they have the samples) to determine if changes in relative abundance of microbes such as Akkermansia correspond to recurrence of diarrhea.” Further the authors did not perform longitudinal sampling on the sham group and compare the active FMT to time point 0 of the sham group so we do not know the level of variability/changes in the microbiome in the sham group following intervention. Overall my enthusiasm is in fact further dampened following author’s response as it highlights additional drawbacks in the study design and the interpretation of the results. Reviewer #3 (Remarks to the Author): Authors have satisfactory assessed all my queries. Reviewer #4 (Remarks to the Author): All my comments have been addressed appropriately. Looks good!	TPR	NC_BS_CAN_001_RR1txt.txt	{ "criticism": 5, "example": 0, "importance_and_relevance": 0, "materials_and_methods": 11, "praise": 3, "presentation_and_reporting": 0, "results_and_discussion": 1, "suggestion_and_solution": 2, "total": 18 }	{ "criticism": 0.2777777777777778, "example": 0, "importance_and_relevance": 0, "materials_and_methods": 0.6111111111111112, "praise": 0.16666666666666666, "presentation_and_reporting": 0, "results_and_discussion": 0.05555555555555555, "suggestion_and_solution": 0.1111111111111111 }	1.222222	[ { "content": "Please review this paper.", "role": "user" }, { "content": "\nREVIEWERS' COMMENTS: \n\nReviewer #1 (Remarks to the Author): \n\nI have no further comments following a review of the authors’ response to reviewers’ comments \n\nReviewer #2 (Remarks to the Author): \n\n1. One of my primary concerns was that all patients underwent a bowel prep which puts the sham \ngroup at a disadvantage as it will decrease the diversity of an already disrupted community. That \nis likely why we see nearly no improvement in the sham group. \nThe authors have responded to this by providing a separate explanation for each of these using \ntwo arguments that 1) bowel prep does not affect gut microbiota by referencing a study that used \nPCR-DGGE method and 2) other groups have used water as a control. \nThere is sufficient data that bowel prep does indeed alter gut microbiome in studies that use next \ngeneration sequencing platforms as the one used by the authors \n(https://pubmed.ncbi.nlm.nih.gov/27015015/). The authors did not measure the microbiome \nbefore and after bowel prep in their cohort so they cannot definitively establish this. The authors \nthen described several large trials that have used water as a sham placebo but fail to mention that \nnone of them gave a bowel prep prior to treatment and my concern was not based on the fact that \nwater cannot be a placebo but rather that the authors could have further disrupted an already \nvulnerable community with bowel prep and that can explain the lack of any response in the sham \ngroup. \nThe authors then argue against using autologous stool as a comparator group and I am not sure if \nthey got confused with my comment “sham FMT refers to patients own stool”. I had suggested \nthat they should have used the patient’s own stool (allogenic FMT) as a control group and all the \nexamples provided by the authors do exactly that- use the patient’s own stool as the comparator \nand not autologous stool as the authors mention in the rebuttal. I am not sure what the authors \nmean by lack of protocol for collecting stool from the patients with diarrhea. A sham/allogenic FMT \nwould follow the exact same protocol for collection and processing as the donor and all the \nexamples outlined by the authors in fact do this. \n2. The authors did not respond to my comment \n“The data is too preliminary to discuss the role of engraftment in driving improvement as the study \nlacks an appropriate control group. Moreover it would be helpful if the authors consider tracking \nthe microbes longitudinally (given they have the samples) to determine if changes in relative \nabundance of microbes such as Akkermansia correspond to recurrence of diarrhea.” Further the \nauthors did not perform longitudinal sampling on the sham group and compare the active FMT to \ntime point 0 of the sham group so we do not know the level of variability/changes in the \nmicrobiome in the sham group following intervention. \nOverall my enthusiasm is in fact further dampened following author’s response as it highlights \nadditional drawbacks in the study design and the interpretation of the results. \n\nReviewer #3 (Remarks to the Author): \n\nAuthors have satisfactory assessed all my queries. \n\n\fReviewer #4 (Remarks to the Author): \n\nAll my comments have been addressed appropriately. Looks good! \n\n\f", "role": "assistant" } ]	2,020
ITGB3-mediated uptake of small extracellular vesicles facilitates intercellular communication	Reviewers' comments: Reviewer #1 (Remarks to the Author): This manuscript from the yCajal lab describes experiments aimed at determining a role for avb3 integrin in internalisation of extracellular vesicles (EVs) . The authors have found that knocking down b3 integrin reduced the level of EV uptake by MDA-MB-231 cells. They show that activation of focal adhesion kinase (FAK) and dynamin are also required for EV uptake. Inhibition of this EV uptake process seems to reduce the clonal growth of MDA-MB-213 cells, possibly because this renders the cells less sensitive to growth promoting factors in the secretome. The authors have identified an interesting phenomenon, and the findings would be of interest to the readership of Nat. Comms. However, as the paper stands, the data do not support the conclusions drawn by the authors. Moreover, the paper needs substantially more mechanistic insight into b3 integrin –dependent EV internalisation before acceptance in Nat. Comms. is justified. Major points: 1. The FACS-based assay to measure uptake of fluorescently labelled EVs does not discriminate between EVs that are attached to the cell surface and those that are internalised into the cell. High-resolution confocal microscopy to directly visualise labelled EVs inside cells/endosomes would go a long way to support their findings. The authors should also visualise EV internalisation in cells where dynamin and ITGB3 have been inhibited/knocked down to determine whether these factors are indeed required for EV internalisation and not simply binding of EVs to the plasma membrane. 2. To ensure that the defects in EV internalisation observed in Fig 2 are not simply due to reduced levels of ITGB3 at the cell surface, the surface expression of ITGB3 following treatment with FAK14, Dyngo, PitSTOP and Heparin should be measured. Moreover, as FAK14 treatment occurs over a long time-period, the total levels of b3 integrin should also be measured for this treatment. 3. The small molecule inhibitor of dynamin Dyngo4 has reported off-target activities that inhibit fluid-phase endocytosis. There is a danger that the results in Fig2B and Fig5A linking EV internalisation/FAK activation to dynamin, are artefactual. Therefore, EV internalisation (Fig2B) and EV-induced FAK phosphorylation (Fig5a) should be repeated with knockdown of one or more dynamin isoforms and/or overexpression of dominant negative dynamin (Dynamin-2K44A). Similarly, experiments using the FAK inhibitor FAK14 should be repeated with FAK knockdown where possible. 4. Generally, there is a lack of data providing mechanistic detail on the role of HSPGs and b3 integrin during EV internalisation. This study would benefit greatly from a set of experiments using high resolution microscopy to dissect: a) The co-localisation of labelled EVs with b3 at the cell surface. And the requirement of HSPGs for this; b) The internalisation of EV-b3 complexes/foci within endosomes, for instance EEA1-posotive endosomes; c) The recruitment of FAK to EV-b3 positive foci, and the phosphorylation-status of FAK within these structure. 5. The authors show that treatment with the FAK inhibitor FAK14 reduces EV internalisation (Fig2C) and conclude that EV uptake requires FAK activity. siRNA-mediated knockdown of FAK should achieve similar results. Moreover, it would be predicted that FAK activation/constitutively- active FAK would rescue EV uptake by shITGB3 cells, is this the case? 6. From the NTA profiles shown in Fig3A it appears that shCONTROL cells release far more EVs than shITGB3 cells, however, when normalised to cell number (Fig3C) the opposite is the case. This suggests that there are far fewer shITGB3 cells in the EV preps and the authors should provide an explanation for this. Do shITGB3 cells grow more slowly than shCONTROL cells? As mentioned above, this should this should be formally tested as the results of this are crucial for the interpretation of additional experiments, e.g. the colony formation assays in (Fig1) and the IMR90 co-culture in (Supp Fig 4). 7. Throughout this study, western blots are performed on EV lysates without the relevant whole cell lysate samples. This means that all data concerning the constituents of shCONTROL vs shITGB3 EVs are difficult to interpret. Molecular weight markers are absent from all western blot images and should be included. 8. The authors show compelling evidence that ITGB3 promotes CD81/TSG101-containing exosome biogenesis. However, the paper would benefit from further characterisation of the EV populations released by shCONTROL and shITGB3 cells. For instance, sucrose or Iodixanol gradients should be used to separate vesicle populations prior to western blotting. 9. The claim that EV-induced activation of FAK is a driver of exosome biogenesis needs to be further supported. The mechanism for this is unclear. The authors should address this by investigating the role of ITGB3/FAK in regulating; a) the biogenesis of ILVs in the MVB; b) MVB dynamics (localisation, trafficking and exocytosis) 10. The authors show that dynamin activity is required for FAK activation following EV treatment. Are additional components of integrin signalling (Src, Akt, Erk) also activated following EV treatment? Moreover, does FAK signalling play a role in colony formation (Fig1A). For instance, do conditioned medium-treated cells have higher levels of phospho-FAK? Do FAK inhibitors impair colony formation in the presence of conditioned media? 11. In Fig 5B the authors use TCA precipitation of cell culture supernatant to show that the levels of secreted CD81 are reduced following FAK inhibition. From this they conclude that FAK activity is responsible for driving exosome biogenesis and secretion, but these data are somewhat preliminary. It is unclear, for instance, why they have not performed western blots on EVs isolated from FAK14-treated cells. This should also be supported with NTA of EVs isolated in the presence of the FAK14 to determine. If the authors are right, this should lead to similar changes in EV size distribution and concentration as those observed in the shITGB3 cells. 12. Generally, the conclusions outlined in the discussion text and in Fig5C are overstated with only preliminary supporting evidence. For instance, the conclusion that FAK activation following EV internalisation promotes syntenin-ALIX-ESCRT recruitment on MVBs is totally unsupported. Either the evidence should be provided, or the text should be toned down. Minor points 1. The authors show compelling evidence that the EV contents of conditioned medium contribute towards colony formation of MDA-MB-231 cells, and this is subsequently attributed to their uptake of extracellular vesicles, likely exosomes. It is unclear whether large microvesicles/ectosomes or small EVs within the conditioned media are responsible for this however. This could be addressed by assessing the colony forming capacity of CM from which only microvesicles/ectosomes have been depleted, for instance, by subjecting the conditioned media to a 10,000 x g spin. 2. In Fig 2A the authors maintain target cells at 4 oC during incubation with EVs. The aim of this experiment is to assess whether EV internalisation is an energy dependent or a passive process. The design of this experiment is problematic as passive fusion of membrane compartments is likely to be severely inhibited at 4 oC due to reduced membrane fluidity. Can the authors show that passive fusion of membranes is uninhibited at 4 oC? 3. Fibronectin is a described EV cargo and fibronectin-containing EVs have been shown to support formation of cell adhesions (Sung et al, Nat. Comms. 2015). It is a possibility that the authors may be observing ECM-induced FAK activation following EV treatment. Do the EVs isolated form their cells contain fibronectin or other ECM components? If this is the case, the involvement of these ECM components in their EV-induced FAK activation should be investigated. 4. Line number 159 describes Dyngo-4a as a “small molecule inhibitor that blocks GTPase activity”. Reference 46 is incorrect- this reference from 2009 is for the drug Dynole where as Dyngo-4a was developed in 2013. Dyngo-4a reduces GTPase activity of dynamin and is believed to act by reducing its ability to oligomerise. The reference they require is Traffic. 2013 Dec;14(12):1272-89. ‘Building a better dynasore: the dyngo compounds potently inhibit dynamin and endocytosis’ McCluskey A et al. Reviewer #2 (Remarks to the Author): In this study, Fuentes et al showed that ITGB3 knock-down inhibits extracellular vesicle- uptake ( through dynamin-dependent endocytosis), change the composition and size of vesicles present in the culture media of ITGB3 KD cells, and that ITGB3+ vesicles are required for colony formation of breast cancer cell. The authors propose that ITGB3 controls both exosome uptake and release through activation of the Focal Adhesion Kinase, and speculate that endosome could be the signalling platform. The experiments are well designed and executed, and well-presented. The manuscript is clear and well written. hoverer, the statements are very strong and not always supported by the data. I have two major concerns that needs to be addressed (especially the first one): 1) In Figure 2a and B, the authors use a 4 degrees C temperature block to inhibits energy dependent internalization. To my knowledge, ligand-receptor interaction is not abolished at 40C. a good example is the binding of LDL to its receptor that still occurs at 4 degrees C, although LDL internalization is abolished (Brown and Golstein PNAS 1974). Therefore, one would expect to detect fluorescently labelled exosomes at the surface of the acceptor cells at 4 degrees C. This is not the case in fig2 a and b. This suggests to me that ITGB3 does not contribute to exosome docking at the surface of the cell, which contradicts author's statement and conclusion. ITGB3 may be involved somehow in EV endocytosis, but does not seem to be sufficient to enable docking of exosomes to the target cell. This diminishes greatly the finding of the paper. 2) The authors concluded that ITGB3 controls also exosome release because they found that exosomes emanating from ITGB3KD cell are different in size and composition. Another interpretation is that ITGB3 is dedicated to the internalization of a specific subtype of exosomes. This population would corresponds to small EVs (50 -100nm) that would be depleted of escrt complex and other hits found in the paper. This would also explain why in fig2A and B vesivle- internalization is reduced by a factor r 2 (i.e the larger EV are still internalized). The authors might consider discussing this possibility. To conclude, the first point is an important caveat that needs to be addressed by the authors, because it change dramatically the conclusion made by the authors.	Reviewers' comments: Reviewer #1 (Remarks to the Author): This manuscript from the yCajal lab describes experiments aimed at determining a role for avb3 integrin in internalisation of extracellular vesicles (EVs) . The authors have found that knocking down b3 integrin reduced the level of EV uptake by MDA-MB-231 cells. They show that activation of focal adhesion kinase (FAK) and dynamin are also required for EV uptake. Inhibition of this EV uptake process seems to reduce the clonal growth of MDA-MB-213 cells, possibly because this renders the cells less sensitive to growth promoting factors in the secretome. The authors have identified an interesting phenomenon, and the findings would be of interest to the readership of Nat. Comms. However, as the paper stands, the data do not support the conclusions drawn by the authors. Moreover, the paper needs substantially more mechanistic insight into b3 integrin –dependent EV internalisation before acceptance in Nat. Comms. is justified. Major points: 1. The FACS-based assay to measure uptake of fluorescently labelled EVs does not discriminate between EVs that are attached to the cell surface and those that are internalised into the cell. High-resolution confocal microscopy to directly visualise labelled EVs inside cells/endosomes would go a long way to support their findings. The authors should also visualise EV internalisation in cells where dynamin and ITGB3 have been inhibited/knocked down to determine whether these factors are indeed required for EV internalisation and not simply binding of EVs to the plasma membrane. 2. To ensure that the defects in EV internalisation observed in Fig 2 are not simply due to reduced levels of ITGB3 at the cell surface, the surface expression of ITGB3 following treatment with FAK14, Dyngo, PitSTOP and Heparin should be measured. Moreover, as FAK14 treatment occurs over a long time-period, the total levels of b3 integrin should also be measured for this treatment. 3. The small molecule inhibitor of dynamin Dyngo4 has reported off-target activities that inhibit fluid-phase endocytosis. There is a danger that the results in Fig2B and Fig5A linking EV internalisation/FAK activation to dynamin, are artefactual. Therefore, EV internalisation (Fig2B) and EV-induced FAK phosphorylation (Fig5a) should be repeated with knockdown of one or more dynamin isoforms and/or overexpression of dominant negative dynamin (Dynamin-2K44A). Similarly, experiments using the FAK inhibitor FAK14 should be repeated with FAK knockdown where possible. 4. Generally, there is a lack of data providing mechanistic detail on the role of HSPGs and b3 integrin during EV internalisation. This study would benefit greatly from a set of experiments using high resolution microscopy to dissect: a) The co-localisation of labelled EVs with b3 at the cell surface. And the requirement of HSPGs for this; b) The internalisation of EV-b3 complexes/foci within endosomes, for instance EEA1-posotive endosomes; c) The recruitment of FAK to EV-b3 positive foci, and the phosphorylation-status of FAK within these structure. 5. The authors show that treatment with the FAK inhibitor FAK14 reduces EV internalisation (Fig2C) and conclude that EV uptake requires FAK activity. siRNA-mediated knockdown of FAK should achieve similar results. Moreover, it would be predicted that FAK activation/constitutively- active FAK would rescue EV uptake by shITGB3 cells, is this the case? 6. From the NTA profiles shown in Fig3A it appears that shCONTROL cells release far more EVs than shITGB3 cells, however, when normalised to cell number (Fig3C) the opposite is the case. This suggests that there are far fewer shITGB3 cells in the EV preps and the authors should provide an explanation for this. Do shITGB3 cells grow more slowly than shCONTROL cells? As mentioned above, this should this should be formally tested as the results of this are crucial for the interpretation of additional experiments, e.g. the colony formation assays in (Fig1) and the IMR90 co-culture in (Supp Fig 4). 7. Throughout this study, western blots are performed on EV lysates without the relevant whole cell lysate samples. This means that all data concerning the constituents of shCONTROL vs shITGB3 EVs are difficult to interpret. Molecular weight markers are absent from all western blot images and should be included. 8. The authors show compelling evidence that ITGB3 promotes CD81/TSG101-containing exosome biogenesis. However, the paper would benefit from further characterisation of the EV populations released by shCONTROL and shITGB3 cells. For instance, sucrose or Iodixanol gradients should be used to separate vesicle populations prior to western blotting. 9. The claim that EV-induced activation of FAK is a driver of exosome biogenesis needs to be further supported. The mechanism for this is unclear. The authors should address this by investigating the role of ITGB3/FAK in regulating; a) the biogenesis of ILVs in the MVB; b) MVB dynamics (localisation, trafficking and exocytosis) 10. The authors show that dynamin activity is required for FAK activation following EV treatment. Are additional components of integrin signalling (Src, Akt, Erk) also activated following EV treatment? Moreover, does FAK signalling play a role in colony formation (Fig1A). For instance, do conditioned medium-treated cells have higher levels of phospho-FAK? Do FAK inhibitors impair colony formation in the presence of conditioned media? 11. In Fig 5B the authors use TCA precipitation of cell culture supernatant to show that the levels of secreted CD81 are reduced following FAK inhibition. From this they conclude that FAK activity is responsible for driving exosome biogenesis and secretion, but these data are somewhat preliminary. It is unclear, for instance, why they have not performed western blots on EVs isolated from FAK14-treated cells. This should also be supported with NTA of EVs isolated in the presence of the FAK14 to determine. If the authors are right, this should lead to similar changes in EV size distribution and concentration as those observed in the shITGB3 cells. 12. Generally, the conclusions outlined in the discussion text and in Fig5C are overstated with only preliminary supporting evidence. For instance, the conclusion that FAK activation following EV internalisation promotes syntenin-ALIX-ESCRT recruitment on MVBs is totally unsupported. Either the evidence should be provided, or the text should be toned down. Minor points 1. The authors show compelling evidence that the EV contents of conditioned medium contribute towards colony formation of MDA-MB-231 cells, and this is subsequently attributed to their uptake of extracellular vesicles, likely exosomes. It is unclear whether large microvesicles/ectosomes or small EVs within the conditioned media are responsible for this however. This could be addressed by assessing the colony forming capacity of CM from which only microvesicles/ectosomes have been depleted, for instance, by subjecting the conditioned media to a 10,000 x g spin. 2. In Fig 2A the authors maintain target cells at 4 oC during incubation with EVs. The aim of this experiment is to assess whether EV internalisation is an energy dependent or a passive process. The design of this experiment is problematic as passive fusion of membrane compartments is likely to be severely inhibited at 4 oC due to reduced membrane fluidity. Can the authors show that passive fusion of membranes is uninhibited at 4 oC? 3. Fibronectin is a described EV cargo and fibronectin-containing EVs have been shown to support formation of cell adhesions (Sung et al, Nat. Comms. 2015). It is a possibility that the authors may be observing ECM-induced FAK activation following EV treatment. Do the EVs isolated form their cells contain fibronectin or other ECM components? If this is the case, the involvement of these ECM components in their EV-induced FAK activation should be investigated. 4. Line number 159 describes Dyngo-4a as a “small molecule inhibitor that blocks GTPase activity”. Reference 46 is incorrect- this reference from 2009 is for the drug Dynole where as Dyngo-4a was developed in 2013. Dyngo-4a reduces GTPase activity of dynamin and is believed to act by reducing its ability to oligomerise. The reference they require is Traffic. 2013 Dec;14(12):1272-89. ‘Building a better dynasore: the dyngo compounds potently inhibit dynamin and endocytosis’ McCluskey A et al. Reviewer #2 (Remarks to the Author): In this study, Fuentes et al showed that ITGB3 knock-down inhibits extracellular vesicle- uptake ( through dynamin-dependent endocytosis), change the composition and size of vesicles present in the culture media of ITGB3 KD cells, and that ITGB3+ vesicles are required for colony formation of breast cancer cell. The authors propose that ITGB3 controls both exosome uptake and release through activation of the Focal Adhesion Kinase, and speculate that endosome could be the signalling platform. The experiments are well designed and executed, and well-presented. The manuscript is clear and well written. hoverer, the statements are very strong and not always supported by the data. I have two major concerns that needs to be addressed (especially the first one): 1) In Figure 2a and B, the authors use a 4 degrees C temperature block to inhibits energy dependent internalization. To my knowledge, ligand-receptor interaction is not abolished at 40C. a good example is the binding of LDL to its receptor that still occurs at 4 degrees C, although LDL internalization is abolished (Brown and Golstein PNAS 1974). Therefore, one would expect to detect fluorescently labelled exosomes at the surface of the acceptor cells at 4 degrees C. This is not the case in fig2 a and b. This suggests to me that ITGB3 does not contribute to exosome docking at the surface of the cell, which contradicts author's statement and conclusion. ITGB3 may be involved somehow in EV endocytosis, but does not seem to be sufficient to enable docking of exosomes to the target cell. This diminishes greatly the finding of the paper. 2) The authors concluded that ITGB3 controls also exosome release because they found that exosomes emanating from ITGB3KD cell are different in size and composition. Another interpretation is that ITGB3 is dedicated to the internalization of a specific subtype of exosomes. This population would corresponds to small EVs (50 -100nm) that would be depleted of escrt complex and other hits found in the paper. This would also explain why in fig2A and B vesivle- internalization is reduced by a factor r 2 (i.e the larger EV are still internalized). The authors might consider discussing this possibility. To conclude, the first point is an important caveat that needs to be addressed by the authors, because it change dramatically the conclusion made by the authors.	TPR	NC_BS_CAN_002_RR1.txt	{ "criticism": 14, "example": 12, "importance_and_relevance": 1, "materials_and_methods": 53, "praise": 5, "presentation_and_reporting": 9, "results_and_discussion": 14, "suggestion_and_solution": 21, "total": 104 }	{ "criticism": 0.1346153846153846, "example": 0.11538461538461539, "importance_and_relevance": 0.009615384615384616, "materials_and_methods": 0.5096153846153846, "praise": 0.04807692307692308, "presentation_and_reporting": 0.08653846153846154, "results_and_discussion": 0.1346153846153846, "suggestion_and_solution": 0.20192307692307693 }	1.240385	[ { "content": "Please review this paper.", "role": "user" }, { "content": "Reviewers' comments: \n \nReviewer #1 (Remarks to the Author): \n \nThis manuscript from the yCajal lab describes experiments aimed at determining a role for avb3 \nintegrin in internalisation of extracellular vesicles (EVs) . The authors have found that knocking \ndown b3 integrin reduced the level of EV uptake by MDA-MB-231 cells. They show that activation \nof focal adhesion kinase (FAK) and dynamin are also required for EV uptake. Inhibition of this EV \nuptake process seems to reduce the clonal growth of MDA-MB-213 cells, possibly because this \nrenders the cells less sensitive to growth promoting factors in the secretome. \nThe authors have identified an interesting phenomenon, and the findings would be of interest to \nthe readership of Nat. Comms. However, as the paper stands, the data do not support the \nconclusions drawn by the authors. Moreover, the paper needs substantially more mechanistic \ninsight into b3 integrin –dependent EV internalisation before acceptance in Nat. Comms. is \njustified. \n \nMajor points: \n1. The FACS-based assay to measure uptake of fluorescently labelled EVs does not discriminate \nbetween EVs that are attached to the cell surface and those that are internalised into the cell. \nHigh-resolution confocal microscopy to directly visualise labelled EVs inside cells/endosomes would \ngo a long way to support their findings. The authors should also visualise EV internalisation in cells \nwhere dynamin and ITGB3 have been inhibited/knocked down to determine whether these factors \nare indeed required for EV internalisation and not simply binding of EVs to the plasma membrane. \n2. To ensure that the defects in EV internalisation observed in Fig 2 are not simply due to reduced \nlevels of ITGB3 at the cell surface, the surface expression of ITGB3 following treatment with \nFAK14, Dyngo, PitSTOP and Heparin should be measured. Moreover, as FAK14 treatment occurs \nover a long time-period, the total levels of b3 integrin should also be measured for this treatment. \n3. The small molecule inhibitor of dynamin Dyngo4 has reported off-target activities that inhibit \nfluid-phase endocytosis. There is a danger that the results in Fig2B and Fig5A linking EV \ninternalisation/FAK activation to dynamin, are artefactual. Therefore, EV internalisation (Fig2B) \nand EV-induced FAK phosphorylation (Fig5a) should be repeated with knockdown of one or more \ndynamin isoforms and/or overexpression of dominant negative dynamin (Dynamin-2K44A). \nSimilarly, experiments using the FAK inhibitor FAK14 should be repeated with FAK knockdown \nwhere possible. \n4. Generally, there is a lack of data providing mechanistic detail on the role of HSPGs and b3 \nintegrin during EV internalisation. This study would benefit greatly from a set of experiments using \nhigh resolution microscopy to dissect: a) The co-localisation of labelled EVs with b3 at the cell \nsurface. And the requirement of HSPGs for this; b) The internalisation of EV-b3 complexes/foci \nwithin endosomes, for instance EEA1-posotive endosomes; c) The recruitment of FAK to EV-b3 \npositive foci, and the phosphorylation-status of FAK within these structure. \n5. The authors show that treatment with the FAK inhibitor FAK14 reduces EV internalisation \n(Fig2C) and conclude that EV uptake requires FAK activity. siRNA-mediated knockdown of FAK \nshould achieve similar results. Moreover, it would be predicted that FAK activation/constitutively-\nactive FAK would rescue EV uptake by shITGB3 cells, is this the case? \n6. From the NTA profiles shown in Fig3A it appears that shCONTROL cells release far more EVs \nthan shITGB3 cells, however, when normalised to cell number (Fig3C) the opposite is the case. \nThis suggests that there are far fewer shITGB3 cells in the EV preps and the authors should \nprovide an explanation for this. Do shITGB3 cells grow more slowly than shCONTROL cells? As \nmentioned above, this should this should be formally tested as the results of this are crucial for the \ninterpretation of additional experiments, e.g. the colony formation assays in (Fig1) and the IMR90 \nco-culture in (Supp Fig 4). \n7. Throughout this study, western blots are performed on EV lysates without the relevant whole \ncell lysate samples. This means that all data concerning the constituents of shCONTROL vs \nshITGB3 EVs are difficult to interpret. Molecular weight markers are absent from all western blot \nimages and should be included. \n\n\f8. The authors show compelling evidence that ITGB3 promotes CD81/TSG101-containing exosome \nbiogenesis. However, the paper would benefit from further characterisation of the EV populations \nreleased by shCONTROL and shITGB3 cells. For instance, sucrose or Iodixanol gradients should be \nused to separate vesicle populations prior to western blotting. \n9. The claim that EV-induced activation of FAK is a driver of exosome biogenesis needs to be \nfurther supported. The mechanism for this is unclear. The authors should address this by \ninvestigating the role of ITGB3/FAK in regulating; a) the biogenesis of ILVs in the MVB; b) MVB \ndynamics (localisation, trafficking and exocytosis) \n10. The authors show that dynamin activity is required for FAK activation following EV treatment. \nAre additional components of integrin signalling (Src, Akt, Erk) also activated following EV \ntreatment? Moreover, does FAK signalling play a role in colony formation (Fig1A). For instance, do \nconditioned medium-treated cells have higher levels of phospho-FAK? Do FAK inhibitors impair \ncolony formation in the presence of conditioned media? \n11. In Fig 5B the authors use TCA precipitation of cell culture supernatant to show that the levels \nof secreted CD81 are reduced following FAK inhibition. From this they conclude that FAK activity is \nresponsible for driving exosome biogenesis and secretion, but these data are somewhat \npreliminary. It is unclear, for instance, why they have not performed western blots on EVs isolated \nfrom FAK14-treated cells. This should also be supported with NTA of EVs isolated in the presence \nof the FAK14 to determine. If the authors are right, this should lead to similar changes in EV size \ndistribution and concentration as those observed in the shITGB3 cells. \n12. Generally, the conclusions outlined in the discussion text and in Fig5C are overstated with only \npreliminary supporting evidence. For instance, the conclusion that FAK activation following EV \ninternalisation promotes syntenin-ALIX-ESCRT recruitment on MVBs is totally unsupported. Either \nthe evidence should be provided, or the text should be toned down. \n \n \nMinor points \n1. The authors show compelling evidence that the EV contents of conditioned medium contribute \ntowards colony formation of MDA-MB-231 cells, and this is subsequently attributed to their uptake \nof extracellular vesicles, likely exosomes. It is unclear whether large microvesicles/ectosomes or \nsmall EVs within the conditioned media are responsible for this however. This could be addressed \nby assessing the colony forming capacity of CM from which only microvesicles/ectosomes have \nbeen depleted, for instance, by subjecting the conditioned media to a 10,000 x g spin. \n2. In Fig 2A the authors maintain target cells at 4 oC during incubation with EVs. The aim of this \nexperiment is to assess whether EV internalisation is an energy dependent or a passive process. \nThe design of this experiment is problematic as passive fusion of membrane compartments is likely \nto be severely inhibited at 4 oC due to reduced membrane fluidity. Can the authors show that \npassive fusion of membranes is uninhibited at 4 oC? \n3. Fibronectin is a described EV cargo and fibronectin-containing EVs have been shown to support \nformation of cell adhesions (Sung et al, Nat. Comms. 2015). It is a possibility that the authors \nmay be observing ECM-induced FAK activation following EV treatment. Do the EVs isolated form \ntheir cells contain fibronectin or other ECM components? If this is the case, the involvement of \nthese ECM components in their EV-induced FAK activation should be investigated. \n4. Line number 159 describes Dyngo-4a as a “small molecule inhibitor that blocks GTPase \nactivity”. Reference 46 is incorrect- this reference from 2009 is for the drug Dynole where as \nDyngo-4a was developed in 2013. Dyngo-4a reduces GTPase activity of dynamin and is believed to \nact by reducing its ability to oligomerise. The reference they require is Traffic. 2013 \nDec;14(12):1272-89. ‘Building a better dynasore: the dyngo compounds potently inhibit dynamin \nand endocytosis’ McCluskey A et al. \n \n \n \n\n \n\n\fReviewer #2 (Remarks to the Author): \n \nIn this study, Fuentes et al showed that ITGB3 knock-down inhibits extracellular vesicle- uptake ( \nthrough dynamin-dependent endocytosis), change the composition and size of vesicles present in \nthe culture media of ITGB3 KD cells, and that ITGB3+ vesicles are required for colony formation of \nbreast cancer cell. \n \nThe authors propose that ITGB3 controls both exosome uptake and release through activation of \nthe Focal Adhesion Kinase, and speculate that endosome could be the signalling platform. \n \nThe experiments are well designed and executed, and well-presented. The manuscript is clear and \nwell written. hoverer, the statements are very strong and not always supported by the data. \n \nI have two major concerns that needs to be addressed (especially the first one): \n1) \nIn Figure 2a and B, the authors use a 4 degrees C temperature block to inhibits energy dependent \ninternalization. To my knowledge, ligand-receptor interaction is not abolished at 40C. a good \nexample is the binding of LDL to its receptor that still occurs at 4 degrees C, although LDL \ninternalization is abolished (Brown and Golstein PNAS 1974). Therefore, one would expect to \ndetect fluorescently labelled exosomes at the surface of the acceptor cells at 4 degrees C. This is \nnot the case in fig2 a and b. This suggests to me that ITGB3 does not contribute to exosome \ndocking at the surface of the cell, which contradicts author's statement and conclusion. ITGB3 may \nbe involved somehow in EV endocytosis, but does not seem to be sufficient to enable docking of \nexosomes to the target cell. This diminishes greatly the finding of the paper. \n2) \nThe authors concluded that ITGB3 controls also exosome release because they found that \nexosomes emanating from ITGB3KD cell are different in size and composition. Another \ninterpretation is that ITGB3 is dedicated to the internalization of a specific subtype of exosomes. \nThis population would corresponds to small EVs (50 -100nm) that would be depleted of escrt \ncomplex and other hits found in the paper. This would also explain why in fig2A and B vesivle-\ninternalization is reduced by a factor r 2 (i.e the larger EV are still internalized). The authors might \nconsider discussing this possibility. \n \nTo conclude, the first point is an important caveat that needs to be addressed by the authors, \nbecause it change dramatically the conclusion made by the authors. \n\n\f\n", "role": "assistant" } ]	2,021
ITGB3-mediated uptake of small extracellular vesicles facilitates intercellular communication	Reviewers' comments: Reviewer #1 (Remarks to the Author): This revision has significantly improved the paper and I appreciate the authors taking my suggestions on board. I feel, however, that some issues remain as the paper stands, particularly regarding the visualisation of internalised EVs in recipient cells by confocal microscopy. Main points: Imaging – 1. a) In order to overcome difficulties in distinguishing between donor and recipient ITGB3-positive structures using confocal microscopy (Supporting Figure 1), the authors could repeat these imaging experiments using EVs derived from ITGB3-knockdown cells. This could work as loss of ITGB3 in donor cells did not affect recipient cell-EV uptake (Figure 1). The experiment displayed in rebuttal letter Supporting Figure 3 could also be repeated using EVs derived from ITGB3- knockdown cells. b) In supporting Figure 4, the authors were not able to demonstrate the colocalisation between labelled EVs and p-FAK in recipient cells. Although we agree with the authors that confocal microscopy is at the limit of resolution for detection of single EVs, it is well established that EVs accumulate within the endosomal system upon internalisation in recipient cells, and EV-positive vesicular structures can be usually clearly visualised using such microscopy techniques. Are the PKH26-positive structures associated with focal adhesions? Live cell imaging of recipient cells transiently expressing GFP-FAK +/- ITGB3si in the presence of labelled EVs would help characterising the temporal events of FAK activation downstream of EV uptake, and to answer whether EVs colocalise with FAK in recipient cells. c) In the rebuttal letter the authors include images of labelled EVs with ITGB3 and these images look a lot better than the ones shown in figure 3d. I think the other points made in the rebuttal letter should also be included in the paper. Also fig 3d would benefit from showing each channel separately black on white background as this would be much clearer. 2. The authors show that depletion of EVs from CM impairs its ability to promote clonal growth in MDA-MB-231 cells (Fig. 1). However, instead of experimentally testing whether the incubation of recipient MDA-MB-231 cells with EVs stimulate clonal growth, the authors evaluated the migration of recipient IMR90 fibroblasts (Supp. Fig 9). Although these data are interesting, it is not clear why this EV-function readout (i.e. cell migration) is used at this stage. Does the incubation of MDA-MB- 231 cells with EVs stimulate their clonal growth? Is ITGB3 in recipient cells required for this? 3. The link between syndecans and the mechanism of EV uptake that the authors described is still very weak. Perhaps the MS data for the proteome of EVs might inform on whether a particular syndecan is highly abundant in MDA-MB-231-EV preps, and if so, it would be of interest to evaluate the effect that knocking-down such syndecan in donor cells has on EV uptake. 4. The TITLE is misleading, as the authors did not study the integrin recycling throughout the MS. Instead, the phenotype regulating EV uptake is coordinated with integrin internalisation, and the title should reflect this. Other points: 1. The order of presentation of the data and the organisation of the paper renders it difficult to read and understand. For example, line 181 pertains to figure 3c-d then the next figures that are mentioned are in the order of Fig2b, sup fig 8a, fig2e, fig 2f then figure 4a. Also Fig. 3b is not mentioned in text. This is similar in the supplementary data which should be in the same order in the text and linked to the same data figure. 2. The choice to use IMR90 cell exosomes in figure 1 is not explicitly explained. The text needs a sentence or two to explain the reason for using these cells in relation to cancer. 3. Supplementary figure 6 needs to be labelled as such and in a separate data file 4. Supplementary figure 5A needs labelling control and shITGB3 5. The authors should discuss work from Meisner-Kober lab showing that filopodia mediate EV uptake (https://rupress.org/jcb/article-lookup/doi/10.1083/jcb.201506084) Reviewer #2 (Remarks to the Author): In the revised manuscript, the authors invested a large effort to further characterize the EV behaviour/fate at 4 degrees C (new fig2a and 3). The data and the study has been greatly improved. They also satisfyingly answers comment raised in point 2 of the initial review. Concerning the initial point 1 (interpretation of lack of surface binding at 4 C). This reviewer agrees that itg3 participates in EV uptake and is convinced by the data (especially new fig2a and 3). The explanations provided in the response letter are also clear. There is no strong EV binding at the cell surface at 4C. Again, this suggests (to this reviewer) the lack of BONA-FIDE EV-receptor because protein-protein interaction are normally maintained at 4 C. This reviewer feels that it should be clearly stated in the discussion that ITG3 participates in EV uptake but that the lack of surface binding at 4C suggests that ITG3 is not a bona fide EV receptor. This could be emphasised through comparison with the LDL receptor, using the same language used by the authors in the response letter. To summarise, data are convincing, the model in fig 6 still stands especially the dynamin dependency, but the lack of strong statement discussing the point raised above is a high risk to mislead readers and the scientific community who may conclude that ITG3 is indeed a specific bona fide EV receptor.	Reviewers' comments: Reviewer #1 (Remarks to the Author): This revision has significantly improved the paper and I appreciate the authors taking my suggestions on board. I feel, however, that some issues remain as the paper stands, particularly regarding the visualisation of internalised EVs in recipient cells by confocal microscopy. Main points: Imaging – 1. a) In order to overcome difficulties in distinguishing between donor and recipient ITGB3-positive structures using confocal microscopy (Supporting Figure 1), the authors could repeat these imaging experiments using EVs derived from ITGB3-knockdown cells. This could work as loss of ITGB3 in donor cells did not affect recipient cell-EV uptake (Figure 1). The experiment displayed in rebuttal letter Supporting Figure 3 could also be repeated using EVs derived from ITGB3- knockdown cells. b) In supporting Figure 4, the authors were not able to demonstrate the colocalisation between labelled EVs and p-FAK in recipient cells. Although we agree with the authors that confocal microscopy is at the limit of resolution for detection of single EVs, it is well established that EVs accumulate within the endosomal system upon internalisation in recipient cells, and EV-positive vesicular structures can be usually clearly visualised using such microscopy techniques. Are the PKH26-positive structures associated with focal adhesions? Live cell imaging of recipient cells transiently expressing GFP-FAK +/- ITGB3si in the presence of labelled EVs would help characterising the temporal events of FAK activation downstream of EV uptake, and to answer whether EVs colocalise with FAK in recipient cells. c) In the rebuttal letter the authors include images of labelled EVs with ITGB3 and these images look a lot better than the ones shown in figure 3d. I think the other points made in the rebuttal letter should also be included in the paper. Also fig 3d would benefit from showing each channel separately black on white background as this would be much clearer. 2. The authors show that depletion of EVs from CM impairs its ability to promote clonal growth in MDA-MB-231 cells (Fig. 1). However, instead of experimentally testing whether the incubation of recipient MDA-MB-231 cells with EVs stimulate clonal growth, the authors evaluated the migration of recipient IMR90 fibroblasts (Supp. Fig 9). Although these data are interesting, it is not clear why this EV-function readout (i.e. cell migration) is used at this stage. Does the incubation of MDA-MB- 231 cells with EVs stimulate their clonal growth? Is ITGB3 in recipient cells required for this? 3. The link between syndecans and the mechanism of EV uptake that the authors described is still very weak. Perhaps the MS data for the proteome of EVs might inform on whether a particular syndecan is highly abundant in MDA-MB-231-EV preps, and if so, it would be of interest to evaluate the effect that knocking-down such syndecan in donor cells has on EV uptake. 4. The TITLE is misleading, as the authors did not study the integrin recycling throughout the MS. Instead, the phenotype regulating EV uptake is coordinated with integrin internalisation, and the title should reflect this. Other points: 1. The order of presentation of the data and the organisation of the paper renders it difficult to read and understand. For example, line 181 pertains to figure 3c-d then the next figures that are mentioned are in the order of Fig2b, sup fig 8a, fig2e, fig 2f then figure 4a. Also Fig. 3b is not mentioned in text. This is similar in the supplementary data which should be in the same order in the text and linked to the same data figure. 2. The choice to use IMR90 cell exosomes in figure 1 is not explicitly explained. The text needs a sentence or two to explain the reason for using these cells in relation to cancer. 3. Supplementary figure 6 needs to be labelled as such and in a separate data file 4. Supplementary figure 5A needs labelling control and shITGB3 5. The authors should discuss work from Meisner-Kober lab showing that filopodia mediate EV uptake (https://rupress.org/jcb/article-lookup/doi/10.1083/jcb.201506084) Reviewer #2 (Remarks to the Author): In the revised manuscript, the authors invested a large effort to further characterize the EV behaviour/fate at 4 degrees C (new fig2a and 3). The data and the study has been greatly improved. They also satisfyingly answers comment raised in point 2 of the initial review. Concerning the initial point 1 (interpretation of lack of surface binding at 4 C). This reviewer agrees that itg3 participates in EV uptake and is convinced by the data (especially new fig2a and 3). The explanations provided in the response letter are also clear. There is no strong EV binding at the cell surface at 4C. Again, this suggests (to this reviewer) the lack of BONA-FIDE EV-receptor because protein-protein interaction are normally maintained at 4 C. This reviewer feels that it should be clearly stated in the discussion that ITG3 participates in EV uptake but that the lack of surface binding at 4C suggests that ITG3 is not a bona fide EV receptor. This could be emphasised through comparison with the LDL receptor, using the same language used by the authors in the response letter. To summarise, data are convincing, the model in fig 6 still stands especially the dynamin dependency, but the lack of strong statement discussing the point raised above is a high risk to mislead readers and the scientific community who may conclude that ITG3 is indeed a specific bona fide EV receptor.	TPR	NC_BS_CAN_002_RR2.txt	{ "criticism": 9, "example": 14, "importance_and_relevance": 1, "materials_and_methods": 25, "praise": 5, "presentation_and_reporting": 15, "results_and_discussion": 2, "suggestion_and_solution": 14, "total": 49 }	{ "criticism": 0.1836734693877551, "example": 0.2857142857142857, "importance_and_relevance": 0.02040816326530612, "materials_and_methods": 0.5102040816326531, "praise": 0.10204081632653061, "presentation_and_reporting": 0.30612244897959184, "results_and_discussion": 0.04081632653061224, "suggestion_and_solution": 0.2857142857142857 }	1.734694	[ { "content": "Please review this paper.", "role": "user" }, { "content": "\n\n\fReviewers' comments: \n \nReviewer #1 (Remarks to the Author): \n \nThis revision has significantly improved the paper and I appreciate the authors taking my \nsuggestions on board. I feel, however, that some issues remain as the paper stands, particularly \nregarding the visualisation of internalised EVs in recipient cells by confocal microscopy. \n \nMain points: \nImaging – \n1. \na) In order to overcome difficulties in distinguishing between donor and recipient ITGB3-positive \nstructures using confocal microscopy (Supporting Figure 1), the authors could repeat these \nimaging experiments using EVs derived from ITGB3-knockdown cells. This could work as loss of \nITGB3 in donor cells did not affect recipient cell-EV uptake (Figure 1). The experiment displayed in \nrebuttal letter Supporting Figure 3 could also be repeated using EVs derived from ITGB3-\nknockdown cells. \n \nb) In supporting Figure 4, the authors were not able to demonstrate the colocalisation between \nlabelled EVs and p-FAK in recipient cells. Although we agree with the authors that confocal \nmicroscopy is at the limit of resolution for detection of single EVs, it is well established that EVs \naccumulate within the endosomal system upon internalisation in recipient cells, and EV-positive \nvesicular structures can be usually clearly visualised using such microscopy techniques. Are the \nPKH26-positive structures associated with focal adhesions? Live cell imaging of recipient cells \ntransiently expressing GFP-FAK +/- ITGB3si in the presence of labelled EVs would help \ncharacterising the temporal events of FAK activation downstream of EV uptake, and to answer \nwhether EVs colocalise with FAK in recipient cells. \n \nc) In the rebuttal letter the authors include images of labelled EVs with ITGB3 and these images \nlook a lot better than the ones shown in figure 3d. I think the other points made in the rebuttal \nletter should also be included in the paper. Also fig 3d would benefit from showing each channel \nseparately black on white background as this would be much clearer. \n \n \n2. The authors show that depletion of EVs from CM impairs its ability to promote clonal growth in \nMDA-MB-231 cells (Fig. 1). However, instead of experimentally testing whether the incubation of \nrecipient MDA-MB-231 cells with EVs stimulate clonal growth, the authors evaluated the migration \nof recipient IMR90 fibroblasts (Supp. Fig 9). Although these data are interesting, it is not clear why \nthis EV-function readout (i.e. cell migration) is used at this stage. Does the incubation of MDA-MB-\n231 cells with EVs stimulate their clonal growth? Is ITGB3 in recipient cells required for this? \n \n3. The link between syndecans and the mechanism of EV uptake that the authors described is still \nvery weak. Perhaps the MS data for the proteome of EVs might inform on whether a particular \nsyndecan is highly abundant in MDA-MB-231-EV preps, and if so, it would be of interest to \nevaluate the effect that knocking-down such syndecan in donor cells has on EV uptake. \n \n4. The TITLE is misleading, as the authors did not study the integrin recycling throughout the MS. \nInstead, the phenotype regulating EV uptake is coordinated with integrin internalisation, and the \ntitle should reflect this. \n \nOther points: \n1. The order of presentation of the data and the organisation of the paper renders it difficult to \nread and understand. For example, line 181 pertains to figure 3c-d then the next figures that are \nmentioned are in the order of Fig2b, sup fig 8a, fig2e, fig 2f then figure 4a. Also Fig. 3b is not \nmentioned in text. This is similar in the supplementary data which should be in the same order in \n\n\fthe text and linked to the same data figure. \n2. The choice to use IMR90 cell exosomes in figure 1 is not explicitly explained. The text needs a \nsentence or two to explain the reason for using these cells in relation to cancer. \n3. Supplementary figure 6 needs to be labelled as such and in a separate data file \n4. Supplementary figure 5A needs labelling control and shITGB3 \n5. The authors should discuss work from Meisner-Kober lab showing that filopodia mediate EV \nuptake (https://rupress.org/jcb/article-lookup/doi/10.1083/jcb.201506084) \n \n \n \n \n \nReviewer #2 (Remarks to the Author): \n \nIn the revised manuscript, the authors invested a large effort to further characterize the EV \nbehaviour/fate at 4 degrees C (new fig2a and 3). The data and the study has been greatly \nimproved. They also satisfyingly answers comment raised in point 2 of the initial review. \nConcerning the initial point 1 (interpretation of lack of surface binding at 4 C). This reviewer \nagrees that itg3 participates in EV uptake and is convinced by the data (especially new fig2a and \n3). The explanations provided in the response letter are also clear. There is no strong EV binding \nat the cell surface at 4C. Again, this suggests (to this reviewer) the lack of BONA-FIDE EV-receptor \nbecause protein-protein interaction are normally maintained at 4 C. This reviewer feels that it \nshould be clearly stated in the discussion that ITG3 participates in EV uptake but that the lack of \nsurface binding at 4C suggests that ITG3 is not a bona fide EV receptor. This could be emphasised \nthrough comparison with the LDL receptor, using the same language used by the authors in the \nresponse letter. \nTo summarise, data are convincing, the model in fig 6 still stands especially the dynamin \ndependency, but the lack of strong statement discussing the point raised above is a high risk to \nmislead readers and the scientific community who may conclude that ITG3 is indeed a specific \nbona fide EV receptor. \n\n\f", "role": "assistant" } ]	2,021
ITGB3-mediated uptake of small extracellular vesicles facilitates intercellular communication	Reviewer #1 (Remarks to the Author): Although I feel that this 2nd revision is an incremental improvement on the previous revision, there are still a number of questions that remain unanswered – such as the mechanism by which HSPGs affect EV uptake, or the identity of cellular compartments in which EVs accumulate in recipient cells. Furthermore, I feel that it is hard to draw any meaningful conclusions from some of the confocal microscopy data presented in the paper. For these reasons, I am still cautious to recommend this manuscript for publication in Nature Comms. as it currently stands. Specific points: - The title needs to be modified, as the paper does not address the effect of ITGB3-mediated EV internalisation on breast cancer metastasis. Even though previous work from the authors suggested that ITGB3 is required for breast cancer metastasis in the mouse lung, the title should only refer to work in the present manuscript. Instead of “breast cancer metastasis”, consider using “breast cancer clonal growth” or a similar term. - Throughout the text, there seems to be confusion regarding the term “recycling of integrins”. In the trafficking field, “receptor recycling” refers to the “return” of receptors from intracellular compartments (e.g. recycling endosomes) to the plasma membrane. In many parts of the text (e.g. abstract line 38, introduction line 95, results line 204 and 307, discussion line 379), the authors use the term when instead should be referring to endocytosis/internalisation of integrins. This should be changed accordingly. - The authors need to plot quantifications of EVs with sizes ranging 150 – 200 nm. Although the authors claim in the text that ITGB3-kd leads to a reduction of EVs within this range in the condition media, this is not shown in Fig 4C. - The quantification of colocalisation between labelled EVs and recipient cell EEA1 (Supp. Fig. 9) revealed that the loss of ITGB3 in recipient cells does not influence the accumulation of EVs in early endosomes. Is it fair to say these observations do not follow the authors conclusions nor the model presented in Fig. 6H, where ITGB3 is suggested to mediate dynamin-dependent internalisation of EVs? Would one not expect the levels of EVs in early endosomes to be affected by ITGB3-loss? The observation that ITGB3-kd does not affect accumulation of EV in early endosomes needs to be added to the results text, and we advise the authors to discuss this in the discussion. - Invert lookup table in the confocal microscopy images, so signal is displayed in black and background in white. - Add in supplemental the FACS plots (I think that these had been in a previous version of the paper) from which the bar charts for EV internalisation come from. Also display how the gating was setup. - Add citation in line 107 and line 300. - Labels in Fig. 3C are wrong? - For the story to flow in a more logical manner, we recommend that the authors move the data regarding the analysis of ITGB3-dependent EVs (Figure 4 and 5) to the end, following the requirement of FAK for EV internalisation, and change the text accordingly.	Reviewer #1 (Remarks to the Author): Although I feel that this 2nd revision is an incremental improvement on the previous revision, there are still a number of questions that remain unanswered – such as the mechanism by which HSPGs affect EV uptake, or the identity of cellular compartments in which EVs accumulate in recipient cells. Furthermore, I feel that it is hard to draw any meaningful conclusions from some of the confocal microscopy data presented in the paper. For these reasons, I am still cautious to recommend this manuscript for publication in Nature Comms. as it currently stands. Specific points: - The title needs to be modified, as the paper does not address the effect of ITGB3-mediated EV internalisation on breast cancer metastasis. Even though previous work from the authors suggested that ITGB3 is required for breast cancer metastasis in the mouse lung, the title should only refer to work in the present manuscript. Instead of “breast cancer metastasis”, consider using “breast cancer clonal growth” or a similar term. - Throughout the text, there seems to be confusion regarding the term “recycling of integrins”. In the trafficking field, “receptor recycling” refers to the “return” of receptors from intracellular compartments (e.g. recycling endosomes) to the plasma membrane. In many parts of the text (e.g. abstract line 38, introduction line 95, results line 204 and 307, discussion line 379), the authors use the term when instead should be referring to endocytosis/internalisation of integrins. This should be changed accordingly. - The authors need to plot quantifications of EVs with sizes ranging 150 – 200 nm. Although the authors claim in the text that ITGB3-kd leads to a reduction of EVs within this range in the condition media, this is not shown in Fig 4C. - The quantification of colocalisation between labelled EVs and recipient cell EEA1 (Supp. Fig. 9) revealed that the loss of ITGB3 in recipient cells does not influence the accumulation of EVs in early endosomes. Is it fair to say these observations do not follow the authors conclusions nor the model presented in Fig. 6H, where ITGB3 is suggested to mediate dynamin-dependent internalisation of EVs? Would one not expect the levels of EVs in early endosomes to be affected by ITGB3-loss? The observation that ITGB3-kd does not affect accumulation of EV in early endosomes needs to be added to the results text, and we advise the authors to discuss this in the discussion. - Invert lookup table in the confocal microscopy images, so signal is displayed in black and background in white. - Add in supplemental the FACS plots (I think that these had been in a previous version of the paper) from which the bar charts for EV internalisation come from. Also display how the gating was setup. - Add citation in line 107 and line 300. - Labels in Fig. 3C are wrong? - For the story to flow in a more logical manner, we recommend that the authors move the data regarding the analysis of ITGB3-dependent EVs (Figure 4 and 5) to the end, following the requirement of FAK for EV internalisation, and change the text accordingly.	TPR	NC_BS_CAN_002_RR3.txt	{ "criticism": 6, "example": 6, "importance_and_relevance": 0, "materials_and_methods": 7, "praise": 0, "presentation_and_reporting": 15, "results_and_discussion": 4, "suggestion_and_solution": 11, "total": 29 }	{ "criticism": 0.20689655172413793, "example": 0.20689655172413793, "importance_and_relevance": 0, "materials_and_methods": 0.2413793103448276, "praise": 0, "presentation_and_reporting": 0.5172413793103449, "results_and_discussion": 0.13793103448275862, "suggestion_and_solution": 0.3793103448275862 }	1.689655	[ { "content": "Please review this paper.", "role": "user" }, { "content": "\n \n \n\n\fReviewer #1 (Remarks to the Author): \n \nAlthough I feel that this 2nd revision is an incremental improvement on the previous revision, \nthere are still a number of questions that remain unanswered – such as the mechanism by which \nHSPGs affect EV uptake, or the identity of cellular compartments in which EVs accumulate in \nrecipient cells. Furthermore, I feel that it is hard to draw any meaningful conclusions from some of \nthe confocal microscopy data presented in the paper. For these reasons, I am still cautious to \nrecommend this manuscript for publication in Nature Comms. as it currently stands. \n \nSpecific points: \n- The title needs to be modified, as the paper does not address the effect of ITGB3-mediated EV \ninternalisation on breast cancer metastasis. Even though previous work from the authors \nsuggested that ITGB3 is required for breast cancer metastasis in the mouse lung, the title should \nonly refer to work in the present manuscript. Instead of “breast cancer metastasis”, consider using \n“breast cancer clonal growth” or a similar term. \n- Throughout the text, there seems to be confusion regarding the term “recycling of integrins”. In \nthe trafficking field, “receptor recycling” refers to the “return” of receptors from intracellular \ncompartments (e.g. recycling endosomes) to the plasma membrane. In many parts of the text \n(e.g. abstract line 38, introduction line 95, results line 204 and 307, discussion line 379), the \nauthors use the term when instead should be referring to endocytosis/internalisation of integrins. \nThis should be changed accordingly. \n- The authors need to plot quantifications of EVs with sizes ranging 150 – 200 nm. Although the \nauthors claim in the text that ITGB3-kd leads to a reduction of EVs within this range in the \ncondition media, this is not shown in Fig 4C. \n- The quantification of colocalisation between labelled EVs and recipient cell EEA1 (Supp. Fig. 9) \nrevealed that the loss of ITGB3 in recipient cells does not influence the accumulation of EVs in \nearly endosomes. Is it fair to say these observations do not follow the authors conclusions nor the \nmodel presented in Fig. 6H, where ITGB3 is suggested to mediate dynamin-dependent \ninternalisation of EVs? Would one not expect the levels of EVs in early endosomes to be affected \nby ITGB3-loss? The observation that ITGB3-kd does not affect accumulation of EV in early \nendosomes needs to be added to the results text, and we advise the authors to discuss this in the \ndiscussion. \n- Invert lookup table in the confocal microscopy images, so signal is displayed in black and \nbackground in white. \n- Add in supplemental the FACS plots (I think that these had been in a previous version of the \npaper) from which the bar charts for EV internalisation come from. Also display how the gating \nwas setup. \n- Add citation in line 107 and line 300. \n- Labels in Fig. 3C are wrong? \n- For the story to flow in a more logical manner, we recommend that the authors move the data \nregarding the analysis of ITGB3-dependent EVs (Figure 4 and 5) to the end, following the \nrequirement of FAK for EV internalisation, and change the text accordingly. \n\n\f", "role": "assistant" } ]	2,021
T cell exhaustion and a failure in antigen presentation drive resistance to the graft-versus-	Reviewers' comments: Reviewer #1 (Remarks to the Author): Zhou et al report on graft-versus-leukemia studies in a model system that relies on T cells that recognize the minor histocompatibility antigen H60. The authors show that Leukemia-derived H60 was inefficiently cross-presented and they use different approaches to enhance GVL effects including an agonistic CD40 antibody, blockade of PD1, Tim-3, TIGIT or LAG3. T cell exhaustion and leukemia burden are used a major read-outs. The main finding is that FGK45 administered at the time of transplant was that the agonistic CD40 antibody was most potent at enhancing GVL effects. Major comments: 1. H60 is an artificial antigen, in patients multiple hematopoietic antigens (including major MHC molecules such as HLA-DP1 which are not matched) are recognized by the donor T cells after allo- HSCT. The authors should reproduce their findings in a different leukemia model that relies at least on one additional antigen. This would make their finding relevant for the clinic. 2. FGK45 alone augmented the anti-H60 response, even in the absence of BC-CML, - this shows that there were sources of H60 other than the leukemia derived H60. How can the authors be sure that the TetH60+ (H60 reactive) T cells are then really leukemia reactive? 3. The authors state that the anti CD40 antibody was most effective - why did they not perform a survival study of the leukemia bearing mice? 4. The authors use a CML model (Figure 1) - CML is hardly anymore an indication for allo-HSZT. The studies should focus on AML. 5. Figure 6: anti PD1 treatment after allo-HSCT causes GVHD -did the authors analyze GVHD target organs? 6. The H60-expressing mouse that relies on bone marrow (BM) transduced with retroviruses that express cDNAs encoding the human bcr-abl and NUP98/HOXA9 translocations is artificial. Genetic models of AML are more relevant for the clinic. 7. The authors show that a-CD40 at time of transplantation augments GVL -this is very early treatment - in the clinic relapse is treated when it shows MRD positivity or hematological relapse. Such scenario should be tested by the authors (established AML treated with a-CD40). Reviewer #2 (Remarks to the Author): In this manuscript, Zhou et al. examined mechanisms of the failure in graft-versus-leukemia (GVL) using a mouse model wherein GVL is mediated by minor histocompatibility antigen H60-specific T cells that are trackable with MHC tetramer. In this model, H60 specific memory CD8 T cells generated after vaccination were adoptively transferred into irradiated recipient mice along with T cell-depleted bone marrow as well as H60 expressing leukemia cells. They found that H60-specific memory CD8 T cells failed to control the expansion of the leukemia cell population due to inefficient H60 antigen presentation. Instead, H60-antigen presented by leukemia cells contributed to the differentiation of these memory CD8 T cells into exhausted T cells. Such exhausted CD8 T cells included a PD-1+ Tcf1+ subset, which was previously observed during chronic viral infection and in solid tumors. PD-1 blockade diminished T cell exhaustion and improved GVL. Furthermore, the authors showed that treatment with anti-CD40 antibody augmented H60-specific memory CD8 T cell-mediated GVL. Similar to what was seen in the model of memory CD8 T cell transfer, naïve CD8 T cells differentiated into exhausted T cells after adoptive transfer. Overall, this work provides a comprehensive analysis of how CD8 T cells respond to leukemia cells and also shows the mechanisms of the failure in GVL as well as a potential therapeutic approach to improve GVL. This manuscript could be improved if the authors address the following issues and questions. 1. In the mouse model used in this study, the authors irradiated H60-expressing B6 mice and then donor bone marrow, H60-specific memory CD8 T cells, and H60+ blast crisis chronic myelogenous leukemia (BC-CML) were adoptively transferred into these irradiated mice. Why were H60- expressing B6 mice used as a recipient instead of H60 negative regular B6 mice? This information will be helpful for readers to understand the experiments in this manuscript. 2. In Fig. 1C, D, and E, a burden of BC-CML is shown in the presence of H60-specific CD8 T cells. Do these H60-specific CD8 T cells contribute to the control of BC-CML? Does the expansion of the BC-CML population further increase if H60-specific CD8 T cells are not transferred? 3. Experimental designs of supplemental Fig. 2D and E are not clear. In these experiments, bone marrow cells were transplanted into irradiated B6.H60 mice but the designs give the impression that bone marrow cells were isolated from B6.H60 mice. Please revise them. 4. On pages 12 and 20, the authors mentioned that they were unable to detect H60-specific CD8 T cells that were PD1+ CXCR5+, a subset shown to respond to PD-1-blockade in LCMV infection. What clone and fluorochrome were used for staining of CXCR5 (there is no information in the MM section)? CXCR5 expression in CD8 T cells is dim compared with B cells, and specific clone(s) as well as brightest fluorochrome are required to detect CXCR5 on exhausted CD8 T cells. Alternatively, the existence of CXCR5 can be transcriptionally confirmed. Do the authors have the data of CXCR5 mRNA levels in PD1 positive Tim3 low H60-specific CD8 T cells (compared with those in PD1 positive Tim3 high H60-specific CD8 T cells or naïve CD8 T cells)? Since expression patterns of other markers (Tcf-1, CD39, KLRG1, and Ly108) are very similar to the LCMV system, the Reviewer is wondering if CXCR5 staining would fail due to a technical issue. If the authors do not have conclusive data about CXCR5 expression, the Reviewer suggests removing these parts from the manuscript. 5. It is very difficult to distinguish the lines in the histograms in Fig. 7D. Please revise them. 6. On page 13 line 11, the authors mentioned: “FGK45 primarily acted on host cells as there was no reduction in GVL in recipients of CD40-/- BM (Fig. 7F).” What are the host cells? Do host cells mean recipient cells? Please clarify them. 7. It is very difficult to understand the rationale, the experimental design, and the data of Fig. 7G. Why were mixed bone marrow chimera mice made? Why were these mixed bone marrow chimera mice re-irradiated for transplantation? This part needs more information for a better understanding of the data. 8. On page 16 line 15, (Tn) should be right after “CD8 T cells from either unmanipulated C3H.SW donors”. 9. In Fig. 10A and B, it will be better to show Tox expression levels in unstimulated naïve CD8 T cells as a negative control. 10. On page 16 the last line to page 17 the first line, the authors said “TOX expression in TetH60- negative cells in Tn-recipients was bimodal, with discrete populations of PD-1 High TOX High and PD-1 low TOX low cells (Fig. 10E). FGK45 increased the frequency of TetH60 negative PD-1 high TOX high cells in spleen, perhaps by enhancing alloreactive CD8 cell expansion.” However, Fig. 10E analyzes Tcf-1 expression (no TOX expression data). So, the Reviewer thinks that Tcf1 expression in TetH60-negative cells in Tn-recipients was bimodal. Also, Fig. 10E shows that PD-1 high cells are Tcf-1 low. Conversely, PD-1 low cells are Tcf-1 high. Please revise and clarify this part.	Reviewers' comments: Reviewer #1 (Remarks to the Author): Zhou et al report on graft-versus-leukemia studies in a model system that relies on T cells that recognize the minor histocompatibility antigen H60. The authors show that Leukemia-derived H60 was inefficiently cross-presented and they use different approaches to enhance GVL effects including an agonistic CD40 antibody, blockade of PD1, Tim-3, TIGIT or LAG3. T cell exhaustion and leukemia burden are used a major read-outs. The main finding is that FGK45 administered at the time of transplant was that the agonistic CD40 antibody was most potent at enhancing GVL effects. Major comments: 1. H60 is an artificial antigen, in patients multiple hematopoietic antigens (including major MHC molecules such as HLA-DP1 which are not matched) are recognized by the donor T cells after allo- HSCT. The authors should reproduce their findings in a different leukemia model that relies at least on one additional antigen. This would make their finding relevant for the clinic. 2. FGK45 alone augmented the anti-H60 response, even in the absence of BC-CML, - this shows that there were sources of H60 other than the leukemia derived H60. How can the authors be sure that the TetH60+ (H60 reactive) T cells are then really leukemia reactive? 3. The authors state that the anti CD40 antibody was most effective - why did they not perform a survival study of the leukemia bearing mice? 4. The authors use a CML model (Figure 1) - CML is hardly anymore an indication for allo-HSZT. The studies should focus on AML. 5. Figure 6: anti PD1 treatment after allo-HSCT causes GVHD -did the authors analyze GVHD target organs? 6. The H60-expressing mouse that relies on bone marrow (BM) transduced with retroviruses that express cDNAs encoding the human bcr-abl and NUP98/HOXA9 translocations is artificial. Genetic models of AML are more relevant for the clinic. 7. The authors show that a-CD40 at time of transplantation augments GVL -this is very early treatment - in the clinic relapse is treated when it shows MRD positivity or hematological relapse. Such scenario should be tested by the authors (established AML treated with a-CD40). Reviewer #2 (Remarks to the Author): In this manuscript, Zhou et al. examined mechanisms of the failure in graft-versus-leukemia (GVL) using a mouse model wherein GVL is mediated by minor histocompatibility antigen H60-specific T cells that are trackable with MHC tetramer. In this model, H60 specific memory CD8 T cells generated after vaccination were adoptively transferred into irradiated recipient mice along with T cell-depleted bone marrow as well as H60 expressing leukemia cells. They found that H60-specific memory CD8 T cells failed to control the expansion of the leukemia cell population due to inefficient H60 antigen presentation. Instead, H60-antigen presented by leukemia cells contributed to the differentiation of these memory CD8 T cells into exhausted T cells. Such exhausted CD8 T cells included a PD-1+ Tcf1+ subset, which was previously observed during chronic viral infection and in solid tumors. PD-1 blockade diminished T cell exhaustion and improved GVL. Furthermore, the authors showed that treatment with anti-CD40 antibody augmented H60-specific memory CD8 T cell-mediated GVL. Similar to what was seen in the model of memory CD8 T cell transfer, naïve CD8 T cells differentiated into exhausted T cells after adoptive transfer. Overall, this work provides a comprehensive analysis of how CD8 T cells respond to leukemia cells and also shows the mechanisms of the failure in GVL as well as a potential therapeutic approach to improve GVL. This manuscript could be improved if the authors address the following issues and questions. 1. In the mouse model used in this study, the authors irradiated H60-expressing B6 mice and then donor bone marrow, H60-specific memory CD8 T cells, and H60+ blast crisis chronic myelogenous leukemia (BC-CML) were adoptively transferred into these irradiated mice. Why were H60- expressing B6 mice used as a recipient instead of H60 negative regular B6 mice? This information will be helpful for readers to understand the experiments in this manuscript. 2. In Fig. 1C, D, and E, a burden of BC-CML is shown in the presence of H60-specific CD8 T cells. Do these H60-specific CD8 T cells contribute to the control of BC-CML? Does the expansion of the BC-CML population further increase if H60-specific CD8 T cells are not transferred? 3. Experimental designs of supplemental Fig. 2D and E are not clear. In these experiments, bone marrow cells were transplanted into irradiated B6.H60 mice but the designs give the impression that bone marrow cells were isolated from B6.H60 mice. Please revise them. 4. On pages 12 and 20, the authors mentioned that they were unable to detect H60-specific CD8 T cells that were PD1+ CXCR5+, a subset shown to respond to PD-1-blockade in LCMV infection. What clone and fluorochrome were used for staining of CXCR5 (there is no information in the MM section)? CXCR5 expression in CD8 T cells is dim compared with B cells, and specific clone(s) as well as brightest fluorochrome are required to detect CXCR5 on exhausted CD8 T cells. Alternatively, the existence of CXCR5 can be transcriptionally confirmed. Do the authors have the data of CXCR5 mRNA levels in PD1 positive Tim3 low H60-specific CD8 T cells (compared with those in PD1 positive Tim3 high H60-specific CD8 T cells or naïve CD8 T cells)? Since expression patterns of other markers (Tcf-1, CD39, KLRG1, and Ly108) are very similar to the LCMV system, the Reviewer is wondering if CXCR5 staining would fail due to a technical issue. If the authors do not have conclusive data about CXCR5 expression, the Reviewer suggests removing these parts from the manuscript. 5. It is very difficult to distinguish the lines in the histograms in Fig. 7D. Please revise them. 6. On page 13 line 11, the authors mentioned: “FGK45 primarily acted on host cells as there was no reduction in GVL in recipients of CD40-/- BM (Fig. 7F).” What are the host cells? Do host cells mean recipient cells? Please clarify them. 7. It is very difficult to understand the rationale, the experimental design, and the data of Fig. 7G. Why were mixed bone marrow chimera mice made? Why were these mixed bone marrow chimera mice re-irradiated for transplantation? This part needs more information for a better understanding of the data. 8. On page 16 line 15, (Tn) should be right after “CD8 T cells from either unmanipulated C3H.SW donors”. 9. In Fig. 10A and B, it will be better to show Tox expression levels in unstimulated naïve CD8 T cells as a negative control. 10. On page 16 the last line to page 17 the first line, the authors said “TOX expression in TetH60- negative cells in Tn-recipients was bimodal, with discrete populations of PD-1 High TOX High and PD-1 low TOX low cells (Fig. 10E). FGK45 increased the frequency of TetH60 negative PD-1 high TOX high cells in spleen, perhaps by enhancing alloreactive CD8 cell expansion.” However, Fig. 10E analyzes Tcf-1 expression (no TOX expression data). So, the Reviewer thinks that Tcf1 expression in TetH60-negative cells in Tn-recipients was bimodal. Also, Fig. 10E shows that PD-1 high cells are Tcf-1 low. Conversely, PD-1 low cells are Tcf-1 high. Please revise and clarify this part.	TPR	NC_BS_CAN_003_RR1.txt	{ "criticism": 4, "example": 11, "importance_and_relevance": 1, "materials_and_methods": 43, "praise": 2, "presentation_and_reporting": 11, "results_and_discussion": 3, "suggestion_and_solution": 14, "total": 87 }	{ "criticism": 0.04597701149425287, "example": 0.12643678160919541, "importance_and_relevance": 0.011494252873563218, "materials_and_methods": 0.4942528735632184, "praise": 0.022988505747126436, "presentation_and_reporting": 0.12643678160919541, "results_and_discussion": 0.034482758620689655, "suggestion_and_solution": 0.16091954022988506 }	1.022989	[ { "content": "Please review this paper.", "role": "user" }, { "content": "Reviewers' comments: \n\nReviewer #1 (Remarks to the Author): \n\nZhou et al report on graft-versus-leukemia studies in a model system that relies on T cells that \nrecognize the minor histocompatibility antigen H60. The authors show that Leukemia-derived H60 \nwas inefficiently cross-presented and they use different approaches to enhance GVL effects \nincluding an agonistic CD40 antibody, blockade of PD1, Tim-3, TIGIT or LAG3. T cell exhaustion \nand leukemia burden are used a major read-outs. The main finding is that FGK45 administered at \nthe time of transplant was that the agonistic CD40 antibody was most potent at enhancing GVL \neffects. \n\nMajor comments: \n1. H60 is an artificial antigen, in patients multiple hematopoietic antigens (including major MHC \nmolecules such as HLA-DP1 which are not matched) are recognized by the donor T cells after allo-\nHSCT. The authors should reproduce their findings in a different leukemia model that relies at least \non one additional antigen. This would make their finding relevant for the clinic. \n2. FGK45 alone augmented the anti-H60 response, even in the absence of BC-CML, - this shows \nthat there were sources of H60 other than the leukemia derived H60. How can the authors be sure \nthat the TetH60+ (H60 reactive) T cells are then really leukemia reactive? \n3. The authors state that the anti CD40 antibody was most effective - why did they not perform a \nsurvival study of the leukemia bearing mice? \n4. The authors use a CML model (Figure 1) - CML is hardly anymore an indication for allo-HSZT. \nThe studies should focus on AML. \n5. Figure 6: anti PD1 treatment after allo-HSCT causes GVHD -did the authors analyze GVHD \ntarget organs? \n6. The H60-expressing mouse that relies on bone marrow (BM) transduced with retroviruses that \nexpress cDNAs encoding the human bcr-abl and NUP98/HOXA9 translocations is artificial. Genetic \nmodels of AML are more relevant for the clinic. \n7. The authors show that a-CD40 at time of transplantation augments GVL -this is very early \ntreatment - in the clinic relapse is treated when it shows MRD positivity or hematological relapse. \nSuch scenario should be tested by the authors (established AML treated with a-CD40). \n\nReviewer #2 (Remarks to the Author): \n\nIn this manuscript, Zhou et al. examined mechanisms of the failure in graft-versus-leukemia (GVL) \nusing a mouse model wherein GVL is mediated by minor histocompatibility antigen H60-specific T \ncells that are trackable with MHC tetramer. In this model, H60 specific memory CD8 T cells \ngenerated after vaccination were adoptively transferred into irradiated recipient mice along with T \ncell-depleted bone marrow as well as H60 expressing leukemia cells. They found that H60-specific \nmemory CD8 T cells failed to control the expansion of the leukemia cell population due to \ninefficient H60 antigen presentation. Instead, H60-antigen presented by leukemia cells contributed \nto the differentiation of these memory CD8 T cells into exhausted T cells. Such exhausted CD8 T \ncells included a PD-1+ Tcf1+ subset, which was previously observed during chronic viral infection \nand in solid tumors. PD-1 blockade diminished T cell exhaustion and improved GVL. Furthermore, \nthe authors showed that treatment with anti-CD40 antibody augmented H60-specific memory CD8 \nT cell-mediated GVL. Similar to what was seen in the model of memory CD8 T cell transfer, naïve \nCD8 T cells differentiated into exhausted T cells after adoptive transfer. \n\nOverall, this work provides a comprehensive analysis of how CD8 T cells respond to leukemia cells \nand also shows the mechanisms of the failure in GVL as well as a potential therapeutic approach to \nimprove GVL. This manuscript could be improved if the authors address the following issues and \nquestions. \n\n\f1. In the mouse model used in this study, the authors irradiated H60-expressing B6 mice and then \ndonor bone marrow, H60-specific memory CD8 T cells, and H60+ blast crisis chronic myelogenous \nleukemia (BC-CML) were adoptively transferred into these irradiated mice. Why were H60-\nexpressing B6 mice used as a recipient instead of H60 negative regular B6 mice? This information \nwill be helpful for readers to understand the experiments in this manuscript. \n\n2. In Fig. 1C, D, and E, a burden of BC-CML is shown in the presence of H60-specific CD8 T cells. \nDo these H60-specific CD8 T cells contribute to the control of BC-CML? Does the expansion of the \nBC-CML population further increase if H60-specific CD8 T cells are not transferred? \n\n3. Experimental designs of supplemental Fig. 2D and E are not clear. In these experiments, bone \nmarrow cells were transplanted into irradiated B6.H60 mice but the designs give the impression \nthat bone marrow cells were isolated from B6.H60 mice. Please revise them. \n\n4. On pages 12 and 20, the authors mentioned that they were unable to detect H60-specific CD8 T \ncells that were PD1+ CXCR5+, a subset shown to respond to PD-1-blockade in LCMV infection. \nWhat clone and fluorochrome were used for staining of CXCR5 (there is no information in the MM \nsection)? CXCR5 expression in CD8 T cells is dim compared with B cells, and specific clone(s) as \nwell as brightest fluorochrome are required to detect CXCR5 on exhausted CD8 T cells. \nAlternatively, the existence of CXCR5 can be transcriptionally confirmed. Do the authors have the \ndata of CXCR5 mRNA levels in PD1 positive Tim3 low H60-specific CD8 T cells (compared with \nthose in PD1 positive Tim3 high H60-specific CD8 T cells or naïve CD8 T cells)? Since expression \npatterns of other markers (Tcf-1, CD39, KLRG1, and Ly108) are very similar to the LCMV system, \nthe Reviewer is wondering if CXCR5 staining would fail due to a technical issue. If the authors do \nnot have conclusive data about CXCR5 expression, the Reviewer suggests removing these parts \nfrom the manuscript. \n\n5. It is very difficult to distinguish the lines in the histograms in Fig. 7D. Please revise them. \n\n6. On page 13 line 11, the authors mentioned: “FGK45 primarily acted on host cells as there was \nno reduction in GVL in recipients of CD40-/- BM (Fig. 7F).” What are the host cells? Do host cells \nmean recipient cells? Please clarify them. \n\n7. It is very difficult to understand the rationale, the experimental design, and the data of Fig. 7G. \nWhy were mixed bone marrow chimera mice made? Why were these mixed bone marrow chimera \nmice re-irradiated for transplantation? This part needs more information for a better understanding \nof the data. \n\n8. On page 16 line 15, (Tn) should be right after “CD8 T cells from either unmanipulated C3H.SW \ndonors”. \n\n9. In Fig. 10A and B, it will be better to show Tox expression levels in unstimulated naïve CD8 T \ncells as a negative control. \n\n10. On page 16 the last line to page 17 the first line, the authors said “TOX expression in TetH60-\nnegative cells in Tn-recipients was bimodal, with discrete populations of PD-1 High TOX High and \nPD-1 low TOX low cells (Fig. 10E). FGK45 increased the frequency of TetH60 negative PD-1 high \nTOX high cells in spleen, perhaps by enhancing alloreactive CD8 cell expansion.” However, Fig. \n10E analyzes Tcf-1 expression (no TOX expression data). So, the Reviewer thinks that Tcf1 \nexpression in TetH60-negative cells in Tn-recipients was bimodal. Also, Fig. 10E shows that PD-1 \nhigh cells are Tcf-1 low. Conversely, PD-1 low cells are Tcf-1 high. Please revise and clarify this \npart.\n\n\f", "role": "assistant" } ]	2,020
T cell exhaustion and a failure in antigen presentation drive resistance to the graft-versus-	REVIEWERS' COMMENTS: Reviewer #2 (Remarks to the Author): The authors answered all concerns/issues raised by the reviewer.	REVIEWERS' COMMENTS: Reviewer #2 (Remarks to the Author): The authors answered all concerns/issues raised by the reviewer.	TPR	NC_BS_CAN_003_RR2.txt	{ "criticism": 0, "example": 0, "importance_and_relevance": 0, "materials_and_methods": 0, "praise": 0, "presentation_and_reporting": 0, "results_and_discussion": 0, "suggestion_and_solution": 0, "total": 1 }	{ "criticism": 0, "example": 0, "importance_and_relevance": 0, "materials_and_methods": 0, "praise": 0, "presentation_and_reporting": 0, "results_and_discussion": 0, "suggestion_and_solution": 0 }	0	[ { "content": "Please review this paper.", "role": "user" }, { "content": "\n\n\fREVIEWERS' COMMENTS: \n\nReviewer #2 (Remarks to the Author): \n\nThe authors answered all concerns/issues raised by the reviewer.\n\n\f\n", "role": "assistant" } ]	2,020
A rare variant of African ancestry activates 8q24 lncRNA hub by modulating cancer	Reviewers' comments: Reviewer #1 (Remarks to the Author): This study explores the potential role of a rare SNP that is seen at highest frequency in African populations and associated with a twofold prostate cancer risk as comparted to non-SNP carriers. The authors posit that there is a functional role that enables the initiation or development of prostate cancer through alterations in non-coding regions of the genome. Using 4C capture approaches to explore to DNA interactions they nominate as set of lncRNAs that are activated and may enhance the activation of MYC, a commonly activated oncogene in all prostate cancer. Validation was performed on the TCGA cohort. 1) The authors are not entirely clear as to what prostate cancer Hi-C data was used for this study. To date there are few published cancer datasets. It would be important to specify the data set(s). There are also data sets using benign prostate cells (RWPE) with and without ERG expression, another important prostate cancer oncogene. Were these data also explored? Addition of these data would be useful to understand how universal these findings are in prostate disease progression. This is particularly of interest because SPDEF is also an ETS transcription factor that was identified. 2) Exploration of the TCGA public data is important to attempt to validate their findings. Despite the small numbers of cases, they do see higher MYC levels in cases with the risk alleles. It would be important to clarify if these cases have or do not have MYC amplifications? This is a common region of genomic amplification in prostate cancer. 3) It is not clear to this reviewer if the outcomes data are valid. The TCGA did not consistently collect outcome data. It this correct? What are the demographics of these cases used? 4) The authors might extend these validation findings to some of the larger datasets that were generated on advanced prostate cancer (e.g., SU2C-PCF, WES, RNAseq data) or ICGC whole genome data in order to compare the risk allele or other stated associations. Minor Comments 1) The abbreviation for DNase I hypersensitive sites, DHS, should be defined in the main text. Perhaps even briefly defined for the non-expert. 2) The HiC data is an important inclusion in this study. The authors cite the Wang paper as the main reference, which is providing a convenient browser but is not the original description of Hi-C data. 3) “ESMA experiments”. When possible please introduce acronyms to the non-specialist audience. Reviewer #2 (Remarks to the Author): This work by Walavalkar et al. describes the regulatory expression of a lncRNA hub modulating a prostate cancer associated enhancer. The authors asked the mechanistic function of a rare snp found in the 8q24 locus associated with greater susceptibility in prostate cancer. Using several ChIP-, ATTAC- and RNA-seq dataset published by different consortiums, they established that the rare variant is located within a region supporting enhancer features. Following cloning the different “natural” variants of this putative enhancer in front of a reporter gene, they show that it embeds transcription activation capacity. The analysis of HiC dataset and 4C single genes experiments indicate that the enhancer region is in close contact with lncRNA promoters associated with cMyc expression. TCGA clinical and expression dataset suggest that the enhancer variants are associated with poor outcome. Interestingly, the variants define a consensus binding site for the SPDEF androgen transcription factor. Chip experiments show that SPDEF can bind to these rare variants in vivo and support a model in which it further activates the lncRNA expression. Although the conclusions of the manuscript are very attractive, several major issues need to be addressed for publication in Nature communication, especially much more has to be presented concerning the clinical aspects and enhancer activities. Major issues/suggestions 1. The luciferase test is poorly explained and has major variabilities that are not explained by the authors. For instance, the comparison between allele G and T without androgen show almost a 2 fold increase in Figure 2B, but none in Figure 2D (green points) and 4 fold change in Figure 5D. These variations strongly compromise the reliability of such measurements. A lentivirus integration might help to get more robust results. In figure 2, it should be presented the vector and the internal normalization process, since it is very complicated to understand for the reader. As it is, it shadows the demonstration the the alleles have a significant differences for the enhancer activity- a key message of the paper. What about chromatin data to support further opening of the region (attac, Mnase etc..) in the different allele conditions? 2. “eRNA expression from the enhancer is not changed”. This quite strange since eRNA are supposed to be cryptic RNA. Their role as enhancer is more about the chromatin state and TF binding capacity, rather than the eRNA expression itself. In addition, the eRNA POLII might be activated but the eRNA rapidly degraded. So It would not be a surprise that whatever the enhancer is active or not, the eRNA (already very low) levels remain constant. 3. The comparison with breast tumor TAD (sup S2A) should be properly presented and the figure suppS2A placed in the main part. 4. Figure 4D and 4E are completely unbalanced and cannot be used for the model since the statistics are nowhere near any relevance! 737 cases compared with 15 and 1! The authors should find a beeter cohort or engage their own cohort if they want to draw any clinical conclusion. I suspect that the absence of the “n” number for Figure 4F would reflect also very poor numbers. If this is the case, it should be removed or placed in supplementary file to be only discussed as preliminary insights. Without good clinical data, the manuscript would miss an important message and could not be accepted in Nature communication. 5 Figure 5B missed 2 important controls: with a scramble oligo and using cellular extract of cells treated with siSPDEF. 6. The experiment in Figure 5C is also a crucial aspect of the manuscript and as it is, it fails to convince the reader if SPDEF binds differently the different alleles. The ChIP analysis has no negative or positive controls to determine whether the SPDEF signals are significant or not. The author should perform the experiment by probing SPDEF known promoters and other negative regions of the genome. Additionally, a normalization gene should be used to confirm that the IP is robust. 7. the competition experiment is very interesting but requires a more drastic dose response. The authors should perform a proper competition by increasing different doses (and measuring it in vivo) of the competitive plasmid/allele, and not just one. Minor comments/questions 1. figure 1D some tumors have no signal for ATAC seq, is there any reason? Gleasons? Clinical specificity of these tumors? 2. typo line 183 “loos” instead of loops 3. typo line 225 “(Fig. (Fig 4C)” 4. Figure 2C, the bottom of the figure with chip data of FOXA1, DHS or K27ac are not discussed at all. remove 5. In figure 5D, 3Xflag SPDEF should be at the top of the gel. 6. Figure 5G. the legend is missing: is it fold change luciferase? And on the X axis, is it ratio of the plasmids? It should be indicated and the results better described in the text.	Reviewers' comments: Reviewer #1 (Remarks to the Author): This study explores the potential role of a rare SNP that is seen at highest frequency in African populations and associated with a twofold prostate cancer risk as comparted to non-SNP carriers. The authors posit that there is a functional role that enables the initiation or development of prostate cancer through alterations in non-coding regions of the genome. Using 4C capture approaches to explore to DNA interactions they nominate as set of lncRNAs that are activated and may enhance the activation of MYC, a commonly activated oncogene in all prostate cancer. Validation was performed on the TCGA cohort. 1) The authors are not entirely clear as to what prostate cancer Hi-C data was used for this study. To date there are few published cancer datasets. It would be important to specify the data set(s). There are also data sets using benign prostate cells (RWPE) with and without ERG expression, another important prostate cancer oncogene. Were these data also explored? Addition of these data would be useful to understand how universal these findings are in prostate disease progression. This is particularly of interest because SPDEF is also an ETS transcription factor that was identified. 2) Exploration of the TCGA public data is important to attempt to validate their findings. Despite the small numbers of cases, they do see higher MYC levels in cases with the risk alleles. It would be important to clarify if these cases have or do not have MYC amplifications? This is a common region of genomic amplification in prostate cancer. 3) It is not clear to this reviewer if the outcomes data are valid. The TCGA did not consistently collect outcome data. It this correct? What are the demographics of these cases used? 4) The authors might extend these validation findings to some of the larger datasets that were generated on advanced prostate cancer (e.g., SU2C-PCF, WES, RNAseq data) or ICGC whole genome data in order to compare the risk allele or other stated associations. Minor Comments 1) The abbreviation for DNase I hypersensitive sites, DHS, should be defined in the main text. Perhaps even briefly defined for the non-expert. 2) The HiC data is an important inclusion in this study. The authors cite the Wang paper as the main reference, which is providing a convenient browser but is not the original description of Hi-C data. 3) “ESMA experiments”. When possible please introduce acronyms to the non-specialist audience. Reviewer #2 (Remarks to the Author): This work by Walavalkar et al. describes the regulatory expression of a lncRNA hub modulating a prostate cancer associated enhancer. The authors asked the mechanistic function of a rare snp found in the 8q24 locus associated with greater susceptibility in prostate cancer. Using several ChIP-, ATTAC- and RNA-seq dataset published by different consortiums, they established that the rare variant is located within a region supporting enhancer features. Following cloning the different “natural” variants of this putative enhancer in front of a reporter gene, they show that it embeds transcription activation capacity. The analysis of HiC dataset and 4C single genes experiments indicate that the enhancer region is in close contact with lncRNA promoters associated with cMyc expression. TCGA clinical and expression dataset suggest that the enhancer variants are associated with poor outcome. Interestingly, the variants define a consensus binding site for the SPDEF androgen transcription factor. Chip experiments show that SPDEF can bind to these rare variants in vivo and support a model in which it further activates the lncRNA expression. Although the conclusions of the manuscript are very attractive, several major issues need to be addressed for publication in Nature communication, especially much more has to be presented concerning the clinical aspects and enhancer activities. Major issues/suggestions 1. The luciferase test is poorly explained and has major variabilities that are not explained by the authors. For instance, the comparison between allele G and T without androgen show almost a 2 fold increase in Figure 2B, but none in Figure 2D (green points) and 4 fold change in Figure 5D. These variations strongly compromise the reliability of such measurements. A lentivirus integration might help to get more robust results. In figure 2, it should be presented the vector and the internal normalization process, since it is very complicated to understand for the reader. As it is, it shadows the demonstration the the alleles have a significant differences for the enhancer activity- a key message of the paper. What about chromatin data to support further opening of the region (attac, Mnase etc..) in the different allele conditions? 2. “eRNA expression from the enhancer is not changed”. This quite strange since eRNA are supposed to be cryptic RNA. Their role as enhancer is more about the chromatin state and TF binding capacity, rather than the eRNA expression itself. In addition, the eRNA POLII might be activated but the eRNA rapidly degraded. So It would not be a surprise that whatever the enhancer is active or not, the eRNA (already very low) levels remain constant. 3. The comparison with breast tumor TAD (sup S2A) should be properly presented and the figure suppS2A placed in the main part. 4. Figure 4D and 4E are completely unbalanced and cannot be used for the model since the statistics are nowhere near any relevance! 737 cases compared with 15 and 1! The authors should find a beeter cohort or engage their own cohort if they want to draw any clinical conclusion. I suspect that the absence of the “n” number for Figure 4F would reflect also very poor numbers. If this is the case, it should be removed or placed in supplementary file to be only discussed as preliminary insights. Without good clinical data, the manuscript would miss an important message and could not be accepted in Nature communication. 5 Figure 5B missed 2 important controls: with a scramble oligo and using cellular extract of cells treated with siSPDEF. 6. The experiment in Figure 5C is also a crucial aspect of the manuscript and as it is, it fails to convince the reader if SPDEF binds differently the different alleles. The ChIP analysis has no negative or positive controls to determine whether the SPDEF signals are significant or not. The author should perform the experiment by probing SPDEF known promoters and other negative regions of the genome. Additionally, a normalization gene should be used to confirm that the IP is robust. 7. the competition experiment is very interesting but requires a more drastic dose response. The authors should perform a proper competition by increasing different doses (and measuring it in vivo) of the competitive plasmid/allele, and not just one. Minor comments/questions 1. figure 1D some tumors have no signal for ATAC seq, is there any reason? Gleasons? Clinical specificity of these tumors? 2. typo line 183 “loos” instead of loops 3. typo line 225 “(Fig. (Fig 4C)” 4. Figure 2C, the bottom of the figure with chip data of FOXA1, DHS or K27ac are not discussed at all. remove 5. In figure 5D, 3Xflag SPDEF should be at the top of the gel. 6. Figure 5G. the legend is missing: is it fold change luciferase? And on the X axis, is it ratio of the plasmids? It should be indicated and the results better described in the text.	TPR	NC_BS_CAN_004_RR1.txt	{ "criticism": 12, "example": 12, "importance_and_relevance": 7, "materials_and_methods": 45, "praise": 3, "presentation_and_reporting": 17, "results_and_discussion": 12, "suggestion_and_solution": 19, "total": 80 }	{ "criticism": 0.15, "example": 0.15, "importance_and_relevance": 0.0875, "materials_and_methods": 0.5625, "praise": 0.0375, "presentation_and_reporting": 0.2125, "results_and_discussion": 0.15, "suggestion_and_solution": 0.2375 }	1.5875	[ { "content": "Please review this paper.", "role": "user" }, { "content": "Reviewers' comments: \n \nReviewer #1 (Remarks to the Author): \n \nThis study explores the potential role of a rare SNP that is seen at highest frequency in African \npopulations and associated with a twofold prostate cancer risk as comparted to non-SNP carriers. The \nauthors posit that there is a functional role that enables the initiation or development of prostate \ncancer through alterations in non-coding regions of the genome. Using 4C capture approaches to \nexplore to DNA interactions they nominate as set of lncRNAs that are activated and may enhance the \nactivation of MYC, a commonly activated oncogene in all prostate cancer. Validation was performed on \nthe TCGA cohort. \n \n1) The authors are not entirely clear as to what prostate cancer Hi-C data was used for this study. To \ndate there are few published cancer datasets. It would be important to specify the data set(s). There \nare also data sets using benign prostate cells (RWPE) with and without ERG expression, another \nimportant prostate cancer oncogene. Were these data also explored? Addition of these data would be \nuseful to understand how universal these findings are in prostate disease progression. This is \nparticularly of interest because SPDEF is also an ETS transcription factor that was identified. \n \n2) Exploration of the TCGA public data is important to attempt to validate their findings. Despite the \nsmall numbers of cases, they do see higher MYC levels in cases with the risk alleles. It would be \nimportant to clarify if these cases have or do not have MYC amplifications? This is a common region of \ngenomic amplification in prostate cancer. \n \n3) It is not clear to this reviewer if the outcomes data are valid. The TCGA did not consistently collect \noutcome data. It this correct? What are the demographics of these cases used? \n \n4) The authors might extend these validation findings to some of the larger datasets that were \ngenerated on advanced prostate cancer (e.g., SU2C-PCF, WES, RNAseq data) or ICGC whole genome \ndata in order to compare the risk allele or other stated associations. \n \n \n \nMinor Comments \n \n1) The abbreviation for DNase I hypersensitive sites, DHS, should be defined in the main text. Perhaps \neven briefly defined for the non-expert. \n \n2) The HiC data is an important inclusion in this study. The authors cite the Wang paper as the main \nreference, which is providing a convenient browser but is not the original description of Hi-C data. \n \n3) “ESMA experiments”. When possible please introduce acronyms to the non-specialist audience. \n \n \n \nReviewer #2 (Remarks to the Author): \n \nThis work by Walavalkar et al. describes the regulatory expression of a lncRNA hub modulating a \nprostate cancer associated enhancer. The authors asked the mechanistic function of a rare snp found \nin the 8q24 locus associated with greater susceptibility in prostate cancer. Using several ChIP-, \nATTAC- and RNA-seq dataset published by different consortiums, they established that the rare \nvariant is located within a region supporting enhancer features. Following cloning the different \n“natural” variants of this putative enhancer in front of a reporter gene, they show that it embeds \ntranscription activation capacity. The analysis of HiC dataset and 4C single genes experiments indicate \nthat the enhancer region is in close contact with lncRNA promoters associated with cMyc expression. \n\n\fTCGA clinical and expression dataset suggest that the enhancer variants are associated with poor \noutcome. Interestingly, the variants define a consensus binding site for the SPDEF androgen \ntranscription factor. Chip experiments show that SPDEF can bind to these rare variants in vivo and \nsupport a model in which it further activates the lncRNA expression. \nAlthough the conclusions of the manuscript are very attractive, several major issues need to be \naddressed for publication in Nature communication, especially much more has to be presented \nconcerning the clinical aspects and enhancer activities. \n \nMajor issues/suggestions \n1. The luciferase test is poorly explained and has major variabilities that are not explained by the \nauthors. For instance, the comparison between allele G and T without androgen show almost a 2 fold \nincrease in Figure 2B, but none in Figure 2D (green points) and 4 fold change in Figure 5D. These \nvariations strongly compromise the reliability of such measurements. A lentivirus integration might \nhelp to get more robust results. In figure 2, it should be presented the vector and the internal \nnormalization process, since it is very complicated to understand for the reader. As it is, it shadows \nthe demonstration the the alleles have a significant differences for the enhancer activity- a key \nmessage of the paper. What about chromatin data to support further opening of the region (attac, \nMnase etc..) in the different allele conditions? \n2. “eRNA expression from the enhancer is not changed”. This quite strange since eRNA are supposed \nto be cryptic RNA. Their role as enhancer is more about the chromatin state and TF binding capacity, \nrather than the eRNA expression itself. In addition, the eRNA POLII might be activated but the eRNA \nrapidly degraded. So It would not be a surprise that whatever the enhancer is active or not, the eRNA \n(already very low) levels remain constant. \n3. The comparison with breast tumor TAD (sup S2A) should be properly presented and the figure \nsuppS2A placed in the main part. \n4. Figure 4D and 4E are completely unbalanced and cannot be used for the model since the statistics \nare nowhere near any relevance! 737 cases compared with 15 and 1! The authors should find a beeter \ncohort or engage their own cohort if they want to draw any clinical conclusion. I suspect that the \nabsence of the “n” number for Figure 4F would reflect also very poor numbers. If this is the case, it \nshould be removed or placed in supplementary file to be only discussed as preliminary insights. \nWithout good clinical data, the manuscript would miss an important message and could not be \naccepted in Nature communication. \n \n5 Figure 5B missed 2 important controls: with a scramble oligo and using cellular extract of cells \ntreated with siSPDEF. \n6. The experiment in Figure 5C is also a crucial aspect of the manuscript and as it is, it fails to \nconvince the reader if SPDEF binds differently the different alleles. The ChIP analysis has no negative \nor positive controls to determine whether the SPDEF signals are significant or not. The author should \nperform the experiment by probing SPDEF known promoters and other negative regions of the \ngenome. Additionally, a normalization gene should be used to confirm that the IP is robust. \n7. the competition experiment is very interesting but requires a more drastic dose response. The \nauthors should perform a proper competition by increasing different doses (and measuring it in vivo) \nof the competitive plasmid/allele, and not just one. \nMinor comments/questions \n1. figure 1D some tumors have no signal for ATAC seq, is there any reason? Gleasons? Clinical \nspecificity of these tumors? \n2. typo line 183 “loos” instead of loops \n3. typo line 225 “(Fig. (Fig 4C)” \n4. Figure 2C, the bottom of the figure with chip data of FOXA1, DHS or K27ac are not discussed at all. \nremove \n5. In figure 5D, 3Xflag SPDEF should be at the top of the gel. \n6. Figure 5G. the legend is missing: is it fold change luciferase? And on the X axis, is it ratio of the \nplasmids? It should be indicated and the results better described in the text. \n\n\f", "role": "assistant" } ]	2,020
A rare variant of African ancestry activates 8q24 lncRNA hub by modulating cancer	REVIEWERS' COMMENTS: Reviewer #1 (Remarks to the Author): The authors have addressed my questions. Reviewer #3 (Reviewer Replacement to comment on behalf of Reviewer #2, Remarks to the Author): The authors have provided thorough responses to the reviewer comments.	REVIEWERS' COMMENTS: Reviewer #1 (Remarks to the Author): The authors have addressed my questions. Reviewer #3 (Reviewer Replacement to comment on behalf of Reviewer #2, Remarks to the Author): The authors have provided thorough responses to the reviewer comments.	TPR	NC_BS_CAN_004_RR2.txt	{ "criticism": 0, "example": 0, "importance_and_relevance": 0, "materials_and_methods": 0, "praise": 0, "presentation_and_reporting": 0, "results_and_discussion": 0, "suggestion_and_solution": 0, "total": 2 }	{ "criticism": 0, "example": 0, "importance_and_relevance": 0, "materials_and_methods": 0, "praise": 0, "presentation_and_reporting": 0, "results_and_discussion": 0, "suggestion_and_solution": 0 }	0	[ { "content": "Please review this paper.", "role": "user" }, { "content": "\n \n \n\n\fREVIEWERS' COMMENTS: \n \nReviewer #1 (Remarks to the Author): \n \nThe authors have addressed my questions. \n \n \nReviewer #3 (Reviewer Replacement to comment on behalf of Reviewer #2, Remarks to the \nAuthor): \n \nThe authors have provided thorough responses to the reviewer comments. \n\n\f\n", "role": "assistant" } ]	2,020
Engineering AvidCARs for combinatorial antigen recognition and reversible control of CAR	9 Reviewers' Comments: Reviewer #1: Remarks to the Author: Thank you - the authors have addressed my comments to my satisfaction. Very nice work. Reviewer #2: Remarks to the Author: All the issues and concerns are addressed properly. I have no any further questions.	9 Reviewers' Comments: Reviewer #1: Remarks to the Author: Thank you - the authors have addressed my comments to my satisfaction. Very nice work. Reviewer #2: Remarks to the Author: All the issues and concerns are addressed properly. I have no any further questions.	TPR	NC_BS_CAN_005_RR2.txt	{ "criticism": 1, "example": 0, "importance_and_relevance": 0, "materials_and_methods": 0, "praise": 2, "presentation_and_reporting": 0, "results_and_discussion": 0, "suggestion_and_solution": 0, "total": 4 }	{ "criticism": 0.25, "example": 0, "importance_and_relevance": 0, "materials_and_methods": 0, "praise": 0.5, "presentation_and_reporting": 0, "results_and_discussion": 0, "suggestion_and_solution": 0 }	0.75	[ { "content": "Please review this paper.", "role": "user" }, { "content": "\n9 \n\n\fReviewers' Comments: \n \nReviewer #1: \nRemarks to the Author: \nThank you - the authors have addressed my comments to my satisfaction. Very nice work. \n \n \n \nReviewer #2: \nRemarks to the Author: \nAll the issues and concerns are addressed properly. I have no any further questions. \n\n\f", "role": "assistant" } ]	2,020
Engineering AvidCARs for combinatorial antigen recognition and reversible control of CAR	Reviewers' Comments: Reviewer #1: Remarks to the Author: Salzer et al. present a new spin on the CAR platform that is based on coupling reduced-affinity targeting domains to exogenous CAR dimerization. The authors propose this as a better system than other published modalities for combinatorial targeting. Overall, the engineering work is nicely done and well thought-out. However: Major comments: 1. The concept needs to be fleshed out with more examples of targeting domains. I do not think it is sufficient to use only the monovalent E11.4.1 binder. Also, more models with scFv to show how generalizable this concept is. 2. One of the major claims here is that the AVID-CAR system can discriminate single- from double- antigen-expressing cells. This is done quite well in vitro but the in vivo work is not convincing. The authors would need to show in the same animal that single antigen tumor cells are ignored while double antigen cells are killed. The claim on page 12 line 282-283 is not actually supported by their data. 3. Page 5, lines 112-124. This claim would be greatly enhanced by actual experimental data to test their mathematical prediction. 4. Page 6, line 136. Since the data suggest that the low affinity AvidCAR confers high sensitivity only in settings where the ligands are present in dimeric form, this makes me question the relevance of this platform to real life. What two different tumor antigens are present in dimeric form? 5. Page 11 lines 262-264 and Fig 5d,e. Here the results are quite marginal and the follow-up is short. Therefore the data are not particularly convincing. 6. Page 12, line 301-302. There actually are examples of bispecific CARs that show “AND” gating logic. For example, Kloss et al, Nat Biotech 2013. 7. Page 13, line 320-322. Increasing tumor specificity by requiring a dependence on two different antigens (i.e. AND gating) opens not one but two doors for tumor escape through antigen loss (or downmodulation). Therefore I am not sure that this design is ultimately advisable. Minor comments: 1. It would be interesting to see more in-depth mechanistic studies. For example, page 8 line 185, are there differences in the activation of signaling proteins that might explain why the different designs have disparate effects on killing vs cytokine production? 2. Fig 2d. although there is no statistical significance in the high and intermediate affinity comparisons, I wonder whether with additional repeats of this experiment, these results will hold up. There is a clear trend to reduced activity in the ser-BBz constructs. Reviewer #2: Remarks to the Author: The one of major challenges of CART therapy is controlling the toxicities associated with the therapy due the lack of cancer specific targets and treatment associated side effects, such as tumor lysis syndrome and cytokine storm, as well as off target effects. Developing novel ON-switch and AND-gate systems that can not only effectively control specificity of CARs, but also minimally affect efficacy of the therapy is critically important to advance the field of CART therapy. In this report, the authors developed a novel avidity-controlled CAR (AvidCAR) platform with inducible and logic control functions that enable to control CART function via ON-switch and AND- gate mechanisms. However, the data present in current version of manuscript are far from being convincing and solid enough to show that the systems are optimized to be potentially useful. Major changes have to be made before publication. Some suggestion for revision: 1. Based on the data shown in Figure 2F and Supp. 4B, that suggest: a. Overall, AvidCAR activities are lower than high-affinity CAR; b. At lower antigen density, AvidCAR still reactive. C. At high antigen density, there is nearly no differences between AvidCAR and monomeric low affinity Ser- CAR for both lytic and cytokine production. Therefore, it seems that monomeric low affinity Ser- CAR is safer and more specific to high antigen density (equivalent to tumors), where spares low antigen density (equivalent to normal tissues). 2. The tumor lines that express EGFR or Her2/neu at different levels should be tested to demonstrate that AvidCAR can differentially recognize tumors that express antigen at high levels. 3. The authors conclude that AvidCAR cannot use scFv, but only in FV form, which might be misleading, since only one scFv (4D5-5) was tested. 4. AvidCAR works with only 4-1bb domain based CAR, which also imply that the systems can be affected by some unpredicted and uncontrollable factors and makes this platform less reliable and useful. 5. Only Her2/neu high expressing cell lines were used in Affibody mutant R10A AvidCAR work. Tumors, or cell lines that express Her2/neu at different levels should be tested, to exclude the possibility that the dimerized R10A AvidCAR has no discrimination for antigen expressions. 6. Although cytokine productions were used in some experiments, the in vitro test was mainly relayed on CTL (killing) assay. All the CTL assays were used fixed single E:T ratio, which cannot exactly tell the difference of the killing ability of T cells, especially at high E:T ratio, such as for Figure 5A&B, E:T ratio of 10:1 was used. The lytic activity at different E:T ratio should be shown for most of the experiments. 7. The Nalm6 model used in this report was much weaker than most of the reports by other people who used 1e6 Nalm6 (i.v.) and treatment started at day 5 or 7 later with 1 to 10 million of transduced CART cells. While in this report, 5e5 Nalm6 with treatment started at 3 days (Figure 5C) or 5 days (Figure 6C) with 10 million of CART cells, with such condition, 10 million FMC63 CD19 CART cells are far more than enough to completely cure the mice, while AvidCAR only showed minimum effect on controlling tumor growth. That’s might be the reasons why the experiments (for both Figure 5C and 6C) were terminated at extremely early after treatments. 8. For Nalm6 model, long time (usually two to three month) follow-up and survival cures are required. 9. Since the systems are mainly developed for the treatment of solid tumors and the targets developed in this report are EGFR and Her2/neu, the system should be tested in solid tumor models.	Reviewers' Comments: Reviewer #1: Remarks to the Author: Salzer et al. present a new spin on the CAR platform that is based on coupling reduced-affinity targeting domains to exogenous CAR dimerization. The authors propose this as a better system than other published modalities for combinatorial targeting. Overall, the engineering work is nicely done and well thought-out. However: Major comments: 1. The concept needs to be fleshed out with more examples of targeting domains. I do not think it is sufficient to use only the monovalent E11.4.1 binder. Also, more models with scFv to show how generalizable this concept is. 2. One of the major claims here is that the AVID-CAR system can discriminate single- from double- antigen-expressing cells. This is done quite well in vitro but the in vivo work is not convincing. The authors would need to show in the same animal that single antigen tumor cells are ignored while double antigen cells are killed. The claim on page 12 line 282-283 is not actually supported by their data. 3. Page 5, lines 112-124. This claim would be greatly enhanced by actual experimental data to test their mathematical prediction. 4. Page 6, line 136. Since the data suggest that the low affinity AvidCAR confers high sensitivity only in settings where the ligands are present in dimeric form, this makes me question the relevance of this platform to real life. What two different tumor antigens are present in dimeric form? 5. Page 11 lines 262-264 and Fig 5d,e. Here the results are quite marginal and the follow-up is short. Therefore the data are not particularly convincing. 6. Page 12, line 301-302. There actually are examples of bispecific CARs that show “AND” gating logic. For example, Kloss et al, Nat Biotech 2013. 7. Page 13, line 320-322. Increasing tumor specificity by requiring a dependence on two different antigens (i.e. AND gating) opens not one but two doors for tumor escape through antigen loss (or downmodulation). Therefore I am not sure that this design is ultimately advisable. Minor comments: 1. It would be interesting to see more in-depth mechanistic studies. For example, page 8 line 185, are there differences in the activation of signaling proteins that might explain why the different designs have disparate effects on killing vs cytokine production? 2. Fig 2d. although there is no statistical significance in the high and intermediate affinity comparisons, I wonder whether with additional repeats of this experiment, these results will hold up. There is a clear trend to reduced activity in the ser-BBz constructs. Reviewer #2: Remarks to the Author: The one of major challenges of CART therapy is controlling the toxicities associated with the therapy due the lack of cancer specific targets and treatment associated side effects, such as tumor lysis syndrome and cytokine storm, as well as off target effects. Developing novel ON-switch and AND-gate systems that can not only effectively control specificity of CARs, but also minimally affect efficacy of the therapy is critically important to advance the field of CART therapy. In this report, the authors developed a novel avidity-controlled CAR (AvidCAR) platform with inducible and logic control functions that enable to control CART function via ON-switch and AND- gate mechanisms. However, the data present in current version of manuscript are far from being convincing and solid enough to show that the systems are optimized to be potentially useful. Major changes have to be made before publication. Some suggestion for revision: 1. Based on the data shown in Figure 2F and Supp. 4B, that suggest: a. Overall, AvidCAR activities are lower than high-affinity CAR; b. At lower antigen density, AvidCAR still reactive. C. At high antigen density, there is nearly no differences between AvidCAR and monomeric low affinity Ser- CAR for both lytic and cytokine production. Therefore, it seems that monomeric low affinity Ser- CAR is safer and more specific to high antigen density (equivalent to tumors), where spares low antigen density (equivalent to normal tissues). 2. The tumor lines that express EGFR or Her2/neu at different levels should be tested to demonstrate that AvidCAR can differentially recognize tumors that express antigen at high levels. 3. The authors conclude that AvidCAR cannot use scFv, but only in FV form, which might be misleading, since only one scFv (4D5-5) was tested. 4. AvidCAR works with only 4-1bb domain based CAR, which also imply that the systems can be affected by some unpredicted and uncontrollable factors and makes this platform less reliable and useful. 5. Only Her2/neu high expressing cell lines were used in Affibody mutant R10A AvidCAR work. Tumors, or cell lines that express Her2/neu at different levels should be tested, to exclude the possibility that the dimerized R10A AvidCAR has no discrimination for antigen expressions. 6. Although cytokine productions were used in some experiments, the in vitro test was mainly relayed on CTL (killing) assay. All the CTL assays were used fixed single E:T ratio, which cannot exactly tell the difference of the killing ability of T cells, especially at high E:T ratio, such as for Figure 5A&B, E:T ratio of 10:1 was used. The lytic activity at different E:T ratio should be shown for most of the experiments. 7. The Nalm6 model used in this report was much weaker than most of the reports by other people who used 1e6 Nalm6 (i.v.) and treatment started at day 5 or 7 later with 1 to 10 million of transduced CART cells. While in this report, 5e5 Nalm6 with treatment started at 3 days (Figure 5C) or 5 days (Figure 6C) with 10 million of CART cells, with such condition, 10 million FMC63 CD19 CART cells are far more than enough to completely cure the mice, while AvidCAR only showed minimum effect on controlling tumor growth. That’s might be the reasons why the experiments (for both Figure 5C and 6C) were terminated at extremely early after treatments. 8. For Nalm6 model, long time (usually two to three month) follow-up and survival cures are required. 9. Since the systems are mainly developed for the treatment of solid tumors and the targets developed in this report are EGFR and Her2/neu, the system should be tested in solid tumor models.	TPR	NC_BS_CAN_005_RR1.txt	{ "criticism": 9, "example": 11, "importance_and_relevance": 4, "materials_and_methods": 38, "praise": 2, "presentation_and_reporting": 2, "results_and_discussion": 3, "suggestion_and_solution": 11, "total": 74 }	{ "criticism": 0.12162162162162163, "example": 0.14864864864864866, "importance_and_relevance": 0.05405405405405406, "materials_and_methods": 0.5135135135135135, "praise": 0.02702702702702703, "presentation_and_reporting": 0.02702702702702703, "results_and_discussion": 0.04054054054054054, "suggestion_and_solution": 0.14864864864864866 }	1.081081	[ { "content": "Please review this paper.", "role": "user" }, { "content": "Reviewers' Comments: \n \nReviewer #1: \nRemarks to the Author: \nSalzer et al. present a new spin on the CAR platform that is based on coupling reduced-affinity \ntargeting domains to exogenous CAR dimerization. The authors propose this as a better system \nthan other published modalities for combinatorial targeting. \n \nOverall, the engineering work is nicely done and well thought-out. However: \n \nMajor comments: \n1. The concept needs to be fleshed out with more examples of targeting domains. I do not think it \nis sufficient to use only the monovalent E11.4.1 binder. Also, more models with scFv to show how \ngeneralizable this concept is. \n2. One of the major claims here is that the AVID-CAR system can discriminate single- from double-\nantigen-expressing cells. This is done quite well in vitro but the in vivo work is not convincing. The \nauthors would need to show in the same animal that single antigen tumor cells are ignored while \ndouble antigen cells are killed. The claim on page 12 line 282-283 is not actually supported by \ntheir data. \n3. Page 5, lines 112-124. This claim would be greatly enhanced by actual experimental data to test \ntheir mathematical prediction. \n4. Page 6, line 136. Since the data suggest that the low affinity AvidCAR confers high sensitivity \nonly in settings where the ligands are present in dimeric form, this makes me question the \nrelevance of this platform to real life. What two different tumor antigens are present in dimeric \nform? \n5. Page 11 lines 262-264 and Fig 5d,e. Here the results are quite marginal and the follow-up is \nshort. Therefore the data are not particularly convincing. \n6. Page 12, line 301-302. There actually are examples of bispecific CARs that show “AND” gating \nlogic. For example, Kloss et al, Nat Biotech 2013. \n7. Page 13, line 320-322. Increasing tumor specificity by requiring a dependence on two different \nantigens (i.e. AND gating) opens not one but two doors for tumor escape through antigen loss (or \ndownmodulation). Therefore I am not sure that this design is ultimately advisable. \n \nMinor comments: \n1. It would be interesting to see more in-depth mechanistic studies. For example, page 8 line 185, \nare there differences in the activation of signaling proteins that might explain why the different \ndesigns have disparate effects on killing vs cytokine production? \n2. Fig 2d. although there is no statistical significance in the high and intermediate affinity \ncomparisons, I wonder whether with additional repeats of this experiment, these results will hold \nup. There is a clear trend to reduced activity in the ser-BBz constructs. \n \n \n \nReviewer #2: \nRemarks to the Author: \nThe one of major challenges of CART therapy is controlling the toxicities associated with the \ntherapy due the lack of cancer specific targets and treatment associated side effects, such as \ntumor lysis syndrome and cytokine storm, as well as off target effects. Developing novel ON-switch \nand AND-gate systems that can not only effectively control specificity of CARs, but also minimally \naffect efficacy of the therapy is critically important to advance the field of CART therapy. \nIn this report, the authors developed a novel avidity-controlled CAR (AvidCAR) platform with \ninducible and logic control functions that enable to control CART function via ON-switch and AND-\ngate mechanisms. \n \nHowever, the data present in current version of manuscript are far from being convincing and solid \n\n\fenough to show that the systems are optimized to be potentially useful. Major changes have to be \nmade before publication. \n \nSome suggestion for revision: \n \n1. Based on the data shown in Figure 2F and Supp. 4B, that suggest: a. Overall, AvidCAR activities \nare lower than high-affinity CAR; b. At lower antigen density, AvidCAR still reactive. C. At high \nantigen density, there is nearly no differences between AvidCAR and monomeric low affinity Ser-\nCAR for both lytic and cytokine production. Therefore, it seems that monomeric low affinity Ser-\nCAR is safer and more specific to high antigen density (equivalent to tumors), where spares low \nantigen density (equivalent to normal tissues). \n \n2. The tumor lines that express EGFR or Her2/neu at different levels should be tested to \ndemonstrate that AvidCAR can differentially recognize tumors that express antigen at high levels. \n \n3. The authors conclude that AvidCAR cannot use scFv, but only in FV form, which might be \nmisleading, since only one scFv (4D5-5) was tested. \n \n \n4. AvidCAR works with only 4-1bb domain based CAR, which also imply that the systems can be \naffected by some unpredicted and uncontrollable factors and makes this platform less reliable and \nuseful. \n \n5. Only Her2/neu high expressing cell lines were used in Affibody mutant R10A AvidCAR work. \nTumors, or cell lines that express Her2/neu at different levels should be tested, to exclude the \npossibility that the dimerized R10A AvidCAR has no discrimination for antigen expressions. \n \n6. Although cytokine productions were used in some experiments, the in vitro test was mainly \nrelayed on CTL (killing) assay. All the CTL assays were used fixed single E:T ratio, which cannot \nexactly tell the difference of the killing ability of T cells, especially at high E:T ratio, such as for \nFigure 5A&B, E:T ratio of 10:1 was used. The lytic activity at different E:T ratio should be shown \nfor most of the experiments. \n \n \n7. The Nalm6 model used in this report was much weaker than most of the reports by other people \nwho used 1e6 Nalm6 (i.v.) and treatment started at day 5 or 7 later with 1 to 10 million of \ntransduced CART cells. While in this report, 5e5 Nalm6 with treatment started at 3 days (Figure \n5C) or 5 days (Figure 6C) with 10 million of CART cells, with such condition, 10 million FMC63 \nCD19 CART cells are far more than enough to completely cure the mice, while AvidCAR only \nshowed minimum effect on controlling tumor growth. That’s might be the reasons why the \nexperiments (for both Figure 5C and 6C) were terminated at extremely early after treatments. \n \n \n8. For Nalm6 model, long time (usually two to three month) follow-up and survival cures are \nrequired. \n \n \n9. Since the systems are mainly developed for the treatment of solid tumors and the targets \ndeveloped in this report are EGFR and Her2/neu, the system should be tested in solid tumor \nmodels. \n\n\f", "role": "assistant" } ]	2,020
OSMR controls glioma stem cell respiration and confers resistance of glioblastoma to	Reviewers' Comments: Reviewer #1: Remarks to the Author: Sharanek and colleagues describe very convincingly that the cytokine receptor for Oncostatin M (OSMR) is localized in mitochondria of brain tumor stem cells (BTSCs) and regulates oxidative phosphorylation and thereby confers resistance to ionizing radiation. This was done by the help of western blot of subcellular fractions and by performing knockdown experiments. They identified interaction partners by IP-MS and were able to show that OSM regulates cellular respiration independent of EGFRvIII and that a knockdown of OSMR sensitizes BTSCs to irradiation in vitro and in vivo lowering sphere and tumor growth. Data are novel and impressing. However, in extended data fig 7 the authors describe a similar effect on primary CGN cells lowering the translational relevance of targeting OSMR. A few other criticisms are listed below: 1.) Extended data fig. 6: Did authors also analyze other brain cells? Glial cells would have been of particular interest. How was the purity of CGNs confirmed? What was the purity? 2.) Fig. e-f: Why are control values in f) higher as compared to e)? 3.) Fig. 3: Not only sphere numbers but also sphere sizes should be assessed. Pictures in a) and d) should show a larger area and more representative spheres. 4.) Page 8, last sentence: This is an overinterpretation. It has not been shown that especially slow-cycling BTSCs have been eliminated. 5.) For some of the western blot pictures it appears as if the samples were not run on the same gel because especially for OSMR there is often a lot of background. Please comment. 6.) Fig. 3f and g: these graphs should contain data with and without IR like in 3b, c and e). 7.) Glioblastoma are not any more called glioblastoma multiforme but simply glioblastoma. Please correct. 8.) Extended data, fig. 5a-b: Did the authors also test other time points and tissues? Reviewer #2: Remarks to the Author: The contribution by Sharanek and colleagues adds on a previous contribution (Nat. Neuroscience 2016) that showed that the suppression of the receptor for oncostatin M (OSMR) prevents proliferation and tumor growth of glioblastoma cells and human xenografts in mice. OSMR is a co- receptor of the plasma membrane EGFRvIII receptor. In this manuscript they put forward the puzzling idea that OSMR is localized in mitochondria of brain tumor stem cells (BTSC) and in mouse liver and neurons and is responsible for the activation of mitochondrial respiration by interaction with subunits of complex I of the respiratory chain. In addition, they show that knockdown of OSMR increases the response of brain xenografts to ionizing radiation (IR). My major concerns with this piece of work are the methodologies used to claim the mitochondrial localization of OSMR and details of the mechanism of action of OSMR in mitochondrial respiration as detailed in the following comments. Addressing these concerns is required before the manuscript could be considered for publication in Nat. Commun. Major points. 1. Fig. 1a-f and Extended Fig. 5: It is claimed that OSMR is present in mitochondria of BTSCs, liver and brain. However, membrane fractions are known to co-purify with mitochondria. These blots should include biomarkers of plasma and ER membranes to exclude the contamination of plasma/ER membranes in the Mito-fraction. Plasma membrane as such needs to be included as a fraction of the separation and a blot of EGFRvIII could illuminate the findings. 2. Fig. 1 g-h: The proteinase K/Triton X100 experiment supports that OSMR is protected in a membranous compartment but not necessarily mitochondria. They should address a specific import assay of OSMR in isolated mitochondria. 3. Fig. 1 k-n, Fig.1 r-s and Supp. Table 1: Blots of IPs and the “interactome” study by IP-LC- MS/MS are carried out from solubilized proteins from whole cellular lysates where the abundant mitochondrial proteins are highly represented. The IPs should be performed from solubilized proteins from isolated mitochondria. Moreover, the interactome data is really misleading they only show a set of mitochondrial proteins out of the more than thousand proteins that are obtained by this basic approach. This sort of studies need to be quantitatively carried out using SILAC. For instance, in their case, by labeling with heavy and light amino acids the OSRM and knockdown- OSRM cells, respectively. For details in the way to procedure see the example in He et al., PNAS (2018) 115, 2988-2993. 4. The mechanisms to explain the upregulation of mitochondrial respiration mediated by OSMR either alone or in response to the incubation of the cells with OSM (Figs. 2a,e and Extended Figs. 2 and 3) are not explained and far from clear. The major downstream signaling pathways that are activated in response to OSM are JAK/STAT, Ras/Raf/MAPK and PI3K pathways which could be modulating the activity and phosphorylation status of oxidative phosphorylation complexes, these facts are not taken into consideration and, in contrast, they emphasize the potential interaction of OSMR with NDUFS1, a matrix protruding subunit of complex I, which is unconvincing (see point 3) and by itself does not explain the increase in respiration. They should determine the enzymatic activities of respiratory chain complexes and eventually the differential phosphorylation of immunocaptured respiratory complexes. Minor points: 1.- The differential redox stress of the xenografts in response to IR should be documented.	Reviewers' Comments: Reviewer #1: Remarks to the Author: Sharanek and colleagues describe very convincingly that the cytokine receptor for Oncostatin M (OSMR) is localized in mitochondria of brain tumor stem cells (BTSCs) and regulates oxidative phosphorylation and thereby confers resistance to ionizing radiation. This was done by the help of western blot of subcellular fractions and by performing knockdown experiments. They identified interaction partners by IP-MS and were able to show that OSM regulates cellular respiration independent of EGFRvIII and that a knockdown of OSMR sensitizes BTSCs to irradiation in vitro and in vivo lowering sphere and tumor growth. Data are novel and impressing. However, in extended data fig 7 the authors describe a similar effect on primary CGN cells lowering the translational relevance of targeting OSMR. A few other criticisms are listed below: 1.) Extended data fig. 6: Did authors also analyze other brain cells? Glial cells would have been of particular interest. How was the purity of CGNs confirmed? What was the purity? 2.) Fig. e-f: Why are control values in f) higher as compared to e)? 3.) Fig. 3: Not only sphere numbers but also sphere sizes should be assessed. Pictures in a) and d) should show a larger area and more representative spheres. 4.) Page 8, last sentence: This is an overinterpretation. It has not been shown that especially slow-cycling BTSCs have been eliminated. 5.) For some of the western blot pictures it appears as if the samples were not run on the same gel because especially for OSMR there is often a lot of background. Please comment. 6.) Fig. 3f and g: these graphs should contain data with and without IR like in 3b, c and e). 7.) Glioblastoma are not any more called glioblastoma multiforme but simply glioblastoma. Please correct. 8.) Extended data, fig. 5a-b: Did the authors also test other time points and tissues? Reviewer #2: Remarks to the Author: The contribution by Sharanek and colleagues adds on a previous contribution (Nat. Neuroscience 2016) that showed that the suppression of the receptor for oncostatin M (OSMR) prevents proliferation and tumor growth of glioblastoma cells and human xenografts in mice. OSMR is a co- receptor of the plasma membrane EGFRvIII receptor. In this manuscript they put forward the puzzling idea that OSMR is localized in mitochondria of brain tumor stem cells (BTSC) and in mouse liver and neurons and is responsible for the activation of mitochondrial respiration by interaction with subunits of complex I of the respiratory chain. In addition, they show that knockdown of OSMR increases the response of brain xenografts to ionizing radiation (IR). My major concerns with this piece of work are the methodologies used to claim the mitochondrial localization of OSMR and details of the mechanism of action of OSMR in mitochondrial respiration as detailed in the following comments. Addressing these concerns is required before the manuscript could be considered for publication in Nat. Commun. Major points. 1. Fig. 1a-f and Extended Fig. 5: It is claimed that OSMR is present in mitochondria of BTSCs, liver and brain. However, membrane fractions are known to co-purify with mitochondria. These blots should include biomarkers of plasma and ER membranes to exclude the contamination of plasma/ER membranes in the Mito-fraction. Plasma membrane as such needs to be included as a fraction of the separation and a blot of EGFRvIII could illuminate the findings. 2. Fig. 1 g-h: The proteinase K/Triton X100 experiment supports that OSMR is protected in a membranous compartment but not necessarily mitochondria. They should address a specific import assay of OSMR in isolated mitochondria. 3. Fig. 1 k-n, Fig.1 r-s and Supp. Table 1: Blots of IPs and the “interactome” study by IP-LC- MS/MS are carried out from solubilized proteins from whole cellular lysates where the abundant mitochondrial proteins are highly represented. The IPs should be performed from solubilized proteins from isolated mitochondria. Moreover, the interactome data is really misleading they only show a set of mitochondrial proteins out of the more than thousand proteins that are obtained by this basic approach. This sort of studies need to be quantitatively carried out using SILAC. For instance, in their case, by labeling with heavy and light amino acids the OSRM and knockdown- OSRM cells, respectively. For details in the way to procedure see the example in He et al., PNAS (2018) 115, 2988-2993. 4. The mechanisms to explain the upregulation of mitochondrial respiration mediated by OSMR either alone or in response to the incubation of the cells with OSM (Figs. 2a,e and Extended Figs. 2 and 3) are not explained and far from clear. The major downstream signaling pathways that are activated in response to OSM are JAK/STAT, Ras/Raf/MAPK and PI3K pathways which could be modulating the activity and phosphorylation status of oxidative phosphorylation complexes, these facts are not taken into consideration and, in contrast, they emphasize the potential interaction of OSMR with NDUFS1, a matrix protruding subunit of complex I, which is unconvincing (see point 3) and by itself does not explain the increase in respiration. They should determine the enzymatic activities of respiratory chain complexes and eventually the differential phosphorylation of immunocaptured respiratory complexes. Minor points: 1.- The differential redox stress of the xenografts in response to IR should be documented.	TPR	NC_BS_CAN_006_RR1.txt	{ "criticism": 6, "example": 16, "importance_and_relevance": 3, "materials_and_methods": 30, "praise": 3, "presentation_and_reporting": 17, "results_and_discussion": 1, "suggestion_and_solution": 14, "total": 69 }	{ "criticism": 0.08695652173913043, "example": 0.2318840579710145, "importance_and_relevance": 0.043478260869565216, "materials_and_methods": 0.43478260869565216, "praise": 0.043478260869565216, "presentation_and_reporting": 0.2463768115942029, "results_and_discussion": 0.014492753623188406, "suggestion_and_solution": 0.2028985507246377 }	1.304348	[ { "content": "Please review this paper.", "role": "user" }, { "content": "Reviewers' Comments: \n \nReviewer #1: \nRemarks to the Author: \nSharanek and colleagues describe very convincingly that the cytokine receptor for Oncostatin M \n(OSMR) is localized in mitochondria of brain tumor stem cells (BTSCs) and regulates oxidative \nphosphorylation and thereby confers resistance to ionizing radiation. This was done by the help of \nwestern blot of subcellular fractions and by performing knockdown experiments. They identified \ninteraction partners by IP-MS and were able to show that OSM regulates cellular respiration \nindependent of EGFRvIII and that a knockdown of OSMR sensitizes BTSCs to irradiation in vitro \nand in vivo lowering sphere and tumor growth. Data are novel and impressing. However, in \nextended data fig 7 the authors describe a similar effect on primary CGN cells lowering the \ntranslational relevance of targeting OSMR. A few other criticisms are listed below: \n1.) Extended data fig. 6: Did authors also analyze other brain cells? Glial cells would have been of \nparticular interest. How was the purity of CGNs confirmed? What was the purity? \n2.) Fig. e-f: Why are control values in f) higher as compared to e)? \n3.) Fig. 3: Not only sphere numbers but also sphere sizes should be assessed. Pictures in a) and d) \nshould show a larger area and more representative spheres. \n4.) Page 8, last sentence: This is an overinterpretation. It has not been shown that especially \nslow-cycling BTSCs have been eliminated. \n5.) For some of the western blot pictures it appears as if the samples were not run on the same \ngel because especially for OSMR there is often a lot of background. Please comment. \n6.) Fig. 3f and g: these graphs should contain data with and without IR like in 3b, c and e). \n7.) Glioblastoma are not any more called glioblastoma multiforme but simply glioblastoma. Please \ncorrect. \n8.) Extended data, fig. 5a-b: Did the authors also test other time points and tissues? \n \n \n \nReviewer #2: \nRemarks to the Author: \nThe contribution by Sharanek and colleagues adds on a previous contribution (Nat. Neuroscience \n2016) that showed that the suppression of the receptor for oncostatin M (OSMR) prevents \nproliferation and tumor growth of glioblastoma cells and human xenografts in mice. OSMR is a co-\nreceptor of the plasma membrane EGFRvIII receptor. In this manuscript they put forward the \npuzzling idea that OSMR is localized in mitochondria of brain tumor stem cells (BTSC) and in \nmouse liver and neurons and is responsible for the activation of mitochondrial respiration by \ninteraction with subunits of complex I of the respiratory chain. In addition, they show that \nknockdown of OSMR increases the response of brain xenografts to ionizing radiation (IR). My \nmajor concerns with this piece of work are the methodologies used to claim the mitochondrial \nlocalization of OSMR and details of the mechanism of action of OSMR in mitochondrial respiration \nas detailed in the following comments. Addressing these concerns is required before the \nmanuscript could be considered for publication in Nat. Commun. \nMajor points. \n1. Fig. 1a-f and Extended Fig. 5: It is claimed that OSMR is present in mitochondria of BTSCs, liver \nand brain. However, membrane fractions are known to co-purify with mitochondria. These blots \nshould include biomarkers of plasma and ER membranes to exclude the contamination of \nplasma/ER membranes in the Mito-fraction. Plasma membrane as such needs to be included as a \nfraction of the separation and a blot of EGFRvIII could illuminate the findings. \n2. Fig. 1 g-h: The proteinase K/Triton X100 experiment supports that OSMR is protected in a \nmembranous compartment but not necessarily mitochondria. They should address a specific \nimport assay of OSMR in isolated mitochondria. \n3. Fig. 1 k-n, Fig.1 r-s and Supp. Table 1: Blots of IPs and the “interactome” study by IP-LC-\nMS/MS are carried out from solubilized proteins from whole cellular lysates where the abundant \nmitochondrial proteins are highly represented. The IPs should be performed from solubilized \n\n\fproteins from isolated mitochondria. Moreover, the interactome data is really misleading they only \nshow a set of mitochondrial proteins out of the more than thousand proteins that are obtained by \nthis basic approach. This sort of studies need to be quantitatively carried out using SILAC. For \ninstance, in their case, by labeling with heavy and light amino acids the OSRM and knockdown-\nOSRM cells, respectively. For details in the way to procedure see the example in He et al., PNAS \n(2018) 115, 2988-2993. \n4. The mechanisms to explain the upregulation of mitochondrial respiration mediated by OSMR \neither alone or in response to the incubation of the cells with OSM (Figs. 2a,e and Extended Figs. 2 \nand 3) are not explained and far from clear. The major downstream signaling pathways that are \nactivated in response to OSM are JAK/STAT, Ras/Raf/MAPK and PI3K pathways which could be \nmodulating the activity and phosphorylation status of oxidative phosphorylation complexes, these \nfacts are not taken into consideration and, in contrast, they emphasize the potential interaction of \nOSMR with NDUFS1, a matrix protruding subunit of complex I, which is unconvincing (see point 3) \nand by itself does not explain the increase in respiration. They should determine the enzymatic \nactivities of respiratory chain complexes and eventually the differential phosphorylation of \nimmunocaptured respiratory complexes. \nMinor points: \n1.- The differential redox stress of the xenografts in response to IR should be documented. \n \n \n\n\f \n\n\f\n", "role": "assistant" } ]	2,020
OSMR controls glioma stem cell respiration and confers resistance of glioblastoma to	Reviewers' Comments: Reviewer #1: Remarks to the Author: My concerns have been sufficiently expressed. signed: Christel Herold-Mende Reviewer #2: Remarks to the Author: The revised paper has been much improved showing the localization of OSMR in mitochondria of different cellular types. However, this reviewer does not find mechanistic explanation, or sufficiently convincing arguments, to explain the OSMR-mediated activation of mitochondrial respiration. Likewise, there is no mechanistic explanation to address the inhibition of the activities of the four respiratory complexes studied when OSMR is silenced.	Reviewers' Comments: Reviewer #1: Remarks to the Author: My concerns have been sufficiently expressed. signed: Christel Herold-Mende Reviewer #2: Remarks to the Author: The revised paper has been much improved showing the localization of OSMR in mitochondria of different cellular types. However, this reviewer does not find mechanistic explanation, or sufficiently convincing arguments, to explain the OSMR-mediated activation of mitochondrial respiration. Likewise, there is no mechanistic explanation to address the inhibition of the activities of the four respiratory complexes studied when OSMR is silenced.	TPR	NC_BS_CAN_006_RR2.txt	{ "criticism": 2, "example": 0, "importance_and_relevance": 0, "materials_and_methods": 1, "praise": 2, "presentation_and_reporting": 0, "results_and_discussion": 0, "suggestion_and_solution": 1, "total": 4 }	{ "criticism": 0.5, "example": 0, "importance_and_relevance": 0, "materials_and_methods": 0.25, "praise": 0.5, "presentation_and_reporting": 0, "results_and_discussion": 0, "suggestion_and_solution": 0.25 }	1.5	[ { "content": "Please review this paper.", "role": "user" }, { "content": "\n\n\fReviewers' Comments: \n \nReviewer #1: \nRemarks to the Author: \nMy concerns have been sufficiently expressed. \nsigned: Christel Herold-Mende \n \n \n \nReviewer #2: \nRemarks to the Author: \nThe revised paper has been much improved showing the localization of OSMR in mitochondria of \ndifferent cellular types. However, this reviewer does not find mechanistic explanation, or \nsufficiently convincing arguments, to explain the OSMR-mediated activation of mitochondrial \nrespiration. Likewise, there is no mechanistic explanation to address the inhibition of the activities \nof the four respiratory complexes studied when OSMR is silenced. \n\n\f", "role": "assistant" } ]	2,020
Genomic characterization of malignant progression in neoplastic pancreatic cysts	REVIEWERS' COMMENTS: Reviewer#1: the authors addressed all my concerns adequately. Reviewer#2: I appreciate the extensive comments made by the authors made on exactly delineating why this study demonstrates novelty compared to prior data. Also the clarification on ATM is accepted.	REVIEWERS' COMMENTS: Reviewer#1: the authors addressed all my concerns adequately. Reviewer#2: I appreciate the extensive comments made by the authors made on exactly delineating why this study demonstrates novelty compared to prior data. Also the clarification on ATM is accepted.	TPR	NC_BS_CAN_007_RR2.txt	{ "criticism": 0, "example": 0, "importance_and_relevance": 0, "materials_and_methods": 1, "praise": 2, "presentation_and_reporting": 1, "results_and_discussion": 0, "suggestion_and_solution": 0, "total": 3 }	{ "criticism": 0, "example": 0, "importance_and_relevance": 0, "materials_and_methods": 0.3333333333333333, "praise": 0.6666666666666666, "presentation_and_reporting": 0.3333333333333333, "results_and_discussion": 0, "suggestion_and_solution": 0 }	1.333333	[ { "content": "Please review this paper.", "role": "user" }, { "content": "\n\fREVIEWERS' COMMENTS: \n\n \n\n \n\nReviewer#1: \n\nthe authors addressed all my concerns adequately. \n\n \n\nReviewer#2: \n\nI appreciate the extensive comments made by the authors made on exactly delineating why this \nstudy demonstrates novelty compared to prior data. Also the clarification on ATM is accepted. \n\n\f\n", "role": "assistant" } ]	2,020
Genomic characterization of malignant progression in neoplastic pancreatic cysts	Reviewers' comments: Reviewer #1 (Remarks to the Author): The authors claim that their data “established that both IPMNs and MCNs are direct precursors of invasive pancreatic cancer“. This would be the most important conclusion of this manuscript and, if true, it is a significant observation. An important limitation as mentioned by the authors is the small number of patients. However, as the authors point out, this is the largest study of that short published to date. The authors’ central conclusion that IPMNs/MCNs are in lineage with invasive pancreatic cancer was argued only by point mutations and not rearrangements. Point mutation based lineage analysis suffers from two issues: 1. Some point mutations (such as commonly seen driver mutations i.e. KRAS_G12V) are not clonally specific as they can be shared among inter-patient samples and can appear in intra-patient samples by chance. 2. Even if LCM was used in some cases, there is a possibility from contamination from the invasive component as invasive cells migrate. Therefor, in this reviewer’s opinion, it would be helpful to alleviate these two concerns by some additional analysis and clarification. For example, what would happen if all specific point mutations that are found in cancers (or maybe pancreatic cancers) across individuals such as the KRAS (G12V) would be eliminated from the analysis? And what if all mutations that are found in more that one patient in the current study are not used in the analysis? Statistical analysis that shows the likelihood that a patient-specific mutation can be seen in random in intra-patient samples could augment their conclusion. Allelic frequency analysis of shared mutations can also be used to eliminate the possibility of contamination. The detection of many driver mutations in only pre-invasive neoplasms is also curious. It is true that subclones evolve on their own and could accumulate mutations that are different that the invasive component, but this is observed here in commonly seen driver genes in invasive pancreatic cancer. Furthermore, the authors found IPMN regions that shared no mutations with the invasive cancer, suggesting that they represented genetically independent clones. The authors also point out that there is notably higher degree of intra-patient similarity between invasive and non-invasive samples in copy-number space that the mutational space but did not give any biological explanation. Taken together, the information presented by the authors makes their claim that “IPMNs and MCNs are direct precursors of invasive pancreatic cancer” somewhat questionable but not indisputable. The manuscript might improve if some of the points mentioned above are clarified.   The SMAD4 story is also perplexing and should be further clarified because, as presented in this manuscript, it will likely confuse the scientific community. SMAD4 alterations in both IPMNs and invasive cancers are found in this study. We know that in invasive carcinoma biallelic inactivation of SMAD4 is very common. This can happen by combinations of mutations and rearrangements. However, this study is not comprehensive as it lucks the SMAD4 alterations that are due to junctions or deletions. Therefor, the question of the possibility of SMAD4 biallelic inactivation in IPMN/MCNs has not been answered. Could it be possible that only monoallelic SMAD4 alterations are present in IPMN/MCN and it is the second hit on the other allele that turns them to invasive? Also there is a possibility that the detected SMAD4 mutations in pre-invasive neoplasia samples are due to contamination. This point should be mentioned as a limitation or further be argued using allelic frequencies. Reviewer #2 (Remarks to the Author): In this manuscript, Noe and colleagues have conducted whole exome sequencing on paired IPMNs/MCNs and associated invasive cancers from 18 patients. In addition they have conducted targeted multi region sequencing from additional non invasive foci of IPMNs/MCNs. Using this data, they have constructed an evolutionary cascade of IPMN/MCN progression to invasive cancer, including putative timelines of progression. The major findings of the study are as follows: A: IPMNs and MCNs are precursors of the associated pancreatic cancer. B: Non invasive IPMNs and MCNs in a single patient may contain heteregeneous driver mutations such as in KRAS or RNF43. C: Some IPMNs represent cancerization of the ducts by retrograde extension of the cancer. D: Some mutations in IPMNs/MCNs are early (KRAS, GNAS) and others are late (SMAD4, TGFBR2) E: IPMNs and MCNs do not progress overnight to pancreatic cancers and there is a timeline of progression in terms of several years. This represents a window of opportunity for early detection. Each and every one of these findings has been demonstrated previously by work from Johns Hopkins and Memorial Sloan Kettering investigators in the field. While I have no quibbles with the volume and rigor of work, I am unable to grasp the novelty of the findings in this manuscript. Minor point: I would be very cautious about 17% somatic mutation rate in ATM especially from archival material. ATM (a large gene) has a lot of dubious "mutations" in literature due to spurious variant calling.	Reviewers' comments: Reviewer #1 (Remarks to the Author): The authors claim that their data “established that both IPMNs and MCNs are direct precursors of invasive pancreatic cancer“. This would be the most important conclusion of this manuscript and, if true, it is a significant observation. An important limitation as mentioned by the authors is the small number of patients. However, as the authors point out, this is the largest study of that short published to date. The authors’ central conclusion that IPMNs/MCNs are in lineage with invasive pancreatic cancer was argued only by point mutations and not rearrangements. Point mutation based lineage analysis suffers from two issues: 1. Some point mutations (such as commonly seen driver mutations i.e. KRAS_G12V) are not clonally specific as they can be shared among inter-patient samples and can appear in intra-patient samples by chance. 2. Even if LCM was used in some cases, there is a possibility from contamination from the invasive component as invasive cells migrate. Therefor, in this reviewer’s opinion, it would be helpful to alleviate these two concerns by some additional analysis and clarification. For example, what would happen if all specific point mutations that are found in cancers (or maybe pancreatic cancers) across individuals such as the KRAS (G12V) would be eliminated from the analysis? And what if all mutations that are found in more that one patient in the current study are not used in the analysis? Statistical analysis that shows the likelihood that a patient-specific mutation can be seen in random in intra-patient samples could augment their conclusion. Allelic frequency analysis of shared mutations can also be used to eliminate the possibility of contamination. The detection of many driver mutations in only pre-invasive neoplasms is also curious. It is true that subclones evolve on their own and could accumulate mutations that are different that the invasive component, but this is observed here in commonly seen driver genes in invasive pancreatic cancer. Furthermore, the authors found IPMN regions that shared no mutations with the invasive cancer, suggesting that they represented genetically independent clones. The authors also point out that there is notably higher degree of intra-patient similarity between invasive and non-invasive samples in copy-number space that the mutational space but did not give any biological explanation. Taken together, the information presented by the authors makes their claim that “IPMNs and MCNs are direct precursors of invasive pancreatic cancer” somewhat questionable but not indisputable. The manuscript might improve if some of the points mentioned above are clarified.   The SMAD4 story is also perplexing and should be further clarified because, as presented in this manuscript, it will likely confuse the scientific community. SMAD4 alterations in both IPMNs and invasive cancers are found in this study. We know that in invasive carcinoma biallelic inactivation of SMAD4 is very common. This can happen by combinations of mutations and rearrangements. However, this study is not comprehensive as it lucks the SMAD4 alterations that are due to junctions or deletions. Therefor, the question of the possibility of SMAD4 biallelic inactivation in IPMN/MCNs has not been answered. Could it be possible that only monoallelic SMAD4 alterations are present in IPMN/MCN and it is the second hit on the other allele that turns them to invasive? Also there is a possibility that the detected SMAD4 mutations in pre-invasive neoplasia samples are due to contamination. This point should be mentioned as a limitation or further be argued using allelic frequencies. Reviewer #2 (Remarks to the Author): In this manuscript, Noe and colleagues have conducted whole exome sequencing on paired IPMNs/MCNs and associated invasive cancers from 18 patients. In addition they have conducted targeted multi region sequencing from additional non invasive foci of IPMNs/MCNs. Using this data, they have constructed an evolutionary cascade of IPMN/MCN progression to invasive cancer, including putative timelines of progression. The major findings of the study are as follows: A: IPMNs and MCNs are precursors of the associated pancreatic cancer. B: Non invasive IPMNs and MCNs in a single patient may contain heteregeneous driver mutations such as in KRAS or RNF43. C: Some IPMNs represent cancerization of the ducts by retrograde extension of the cancer. D: Some mutations in IPMNs/MCNs are early (KRAS, GNAS) and others are late (SMAD4, TGFBR2) E: IPMNs and MCNs do not progress overnight to pancreatic cancers and there is a timeline of progression in terms of several years. This represents a window of opportunity for early detection. Each and every one of these findings has been demonstrated previously by work from Johns Hopkins and Memorial Sloan Kettering investigators in the field. While I have no quibbles with the volume and rigor of work, I am unable to grasp the novelty of the findings in this manuscript. Minor point: I would be very cautious about 17% somatic mutation rate in ATM especially from archival material. ATM (a large gene) has a lot of dubious "mutations" in literature due to spurious variant calling.	TPR	NC_BS_CAN_007_RR1.txt	{ "criticism": 9, "example": 1, "importance_and_relevance": 3, "materials_and_methods": 22, "praise": 1, "presentation_and_reporting": 4, "results_and_discussion": 7, "suggestion_and_solution": 6, "total": 42 }	{ "criticism": 0.21428571428571427, "example": 0.023809523809523808, "importance_and_relevance": 0.07142857142857142, "materials_and_methods": 0.5238095238095238, "praise": 0.023809523809523808, "presentation_and_reporting": 0.09523809523809523, "results_and_discussion": 0.16666666666666666, "suggestion_and_solution": 0.14285714285714285 }	1.261905	[ { "content": "Please review this paper.", "role": "user" }, { "content": "Reviewers' comments: \n\n \n\nReviewer #1 (Remarks to the Author): \n\n \n\nThe authors claim that their data “established that both IPMNs and MCNs are direct precursors of \ninvasive pancreatic cancer“. This would be the most important conclusion of this manuscript and, if \ntrue, it is a significant observation. An important limitation as mentioned by the authors is the small \nnumber of patients. However, as the authors point out, this is the largest study of that short \npublished to date. \n\nThe authors’ central conclusion that IPMNs/MCNs are in lineage with invasive pancreatic cancer was \nargued only by point mutations and not rearrangements. Point mutation based lineage analysis \nsuffers from two issues: 1. Some point mutations (such as commonly seen driver mutations i.e. \nKRAS_G12V) are not clonally specific as they can be shared among inter-patient samples and can \nappear in intra-patient samples by chance. 2. Even if LCM was used in some cases, there is a \npossibility from contamination from the invasive component as invasive cells migrate. Therefor, in \nthis reviewer’s opinion, it would be helpful to alleviate these two concerns by some additional \nanalysis and clarification. For example, what would happen if all specific point mutations that are \nfound in cancers (or maybe pancreatic cancers) across individuals such as the KRAS (G12V) would be \neliminated from the analysis? And what if all mutations that are found in more that one patient in \nthe current study are not used in the analysis? Statistical analysis that shows the likelihood that a \npatient-specific mutation can be seen in random in intra-patient samples could augment their \nconclusion. Allelic frequency analysis of shared mutations can also be used to eliminate the \npossibility of contamination. \n\nThe detection of many driver mutations in only pre-invasive neoplasms is also curious. It is true that \nsubclones evolve on their own and could accumulate mutations that are different that the invasive \ncomponent, but this is observed here in commonly seen driver genes in invasive pancreatic cancer. \nFurthermore, the authors found IPMN regions that shared no mutations with the invasive cancer, \nsuggesting that they represented genetically independent clones. The authors also point out that \nthere is notably higher degree of intra-patient similarity between invasive and non-invasive samples \nin copy-number space that the mutational space but did not give any biological explanation. Taken \ntogether, the information presented by the authors makes their claim that “IPMNs and MCNs are \ndirect precursors of invasive pancreatic cancer” somewhat questionable but not indisputable. The \nmanuscript might improve if some of the points mentioned above are clarified.   \n\nThe SMAD4 story is also perplexing and should be further clarified because, as presented in this \nmanuscript, it will likely confuse the scientific community. SMAD4 alterations in both IPMNs and \ninvasive cancers are found in this study. We know that in invasive carcinoma biallelic inactivation of \nSMAD4 is very common. This can happen by combinations of mutations and rearrangements. \nHowever, this study is not comprehensive as it lucks the SMAD4 alterations that are due to junctions \nor deletions. Therefor, the question of the possibility of SMAD4 biallelic inactivation in IPMN/MCNs \nhas not been answered. Could it be possible that only monoallelic SMAD4 alterations are present in \n\n\fIPMN/MCN and it is the second hit on the other allele that turns them to invasive? Also there is a \npossibility that the detected SMAD4 mutations in pre-invasive neoplasia samples are due to \ncontamination. This point should be mentioned as a limitation or further be argued using allelic \nfrequencies. \n\n \n\n \n\n \n\nReviewer #2 (Remarks to the Author): \n\n \n\nIn this manuscript, Noe and colleagues have conducted whole exome sequencing on paired \nIPMNs/MCNs and associated invasive cancers from 18 patients. In addition they have conducted \ntargeted multi region sequencing from additional non invasive foci of IPMNs/MCNs. Using this data, \nthey have constructed an evolutionary cascade of IPMN/MCN progression to invasive cancer, \nincluding putative timelines of progression. \n\n \n\nThe major findings of the study are as follows: \n\nA: IPMNs and MCNs are precursors of the associated pancreatic cancer. \n\nB: Non invasive IPMNs and MCNs in a single patient may contain heteregeneous driver mutations \nsuch as in KRAS or RNF43. \n\nC: Some IPMNs represent cancerization of the ducts by retrograde extension of the cancer. \n\nD: Some mutations in IPMNs/MCNs are early (KRAS, GNAS) and others are late (SMAD4, TGFBR2) \n\nE: IPMNs and MCNs do not progress overnight to pancreatic cancers and there is a timeline of \nprogression in terms of several years. This represents a window of opportunity for early detection. \n\n \n\nEach and every one of these findings has been demonstrated previously by work from Johns Hopkins \nand Memorial Sloan Kettering investigators in the field. While I have no quibbles with the volume \nand rigor of work, I am unable to grasp the novelty of the findings in this manuscript. \n\n \n\nMinor point: I would be very cautious about 17% somatic mutation rate in ATM especially from \narchival material. ATM (a large gene) has a lot of dubious \"mutations\" in literature due to spurious \nvariant calling. \n\n\f\n\f", "role": "assistant" } ]	2,020
Pancreatic cancers suppress negative feedback of glucose transport to reprogram chromatin	Reviewers' Comments: Reviewer #1: Remarks to the Author: This study by Bechard et al builds on the authors’ previous findings that cells derived from distant metastases of pancreatic ductal adenocarcinoma (PDAC) are specifically dependent on the oxidative pentose phosphate pathway (ox-PPP) enzyme, phosphogluconate dehydrogenase (PGD), for growth in vitro. This work suggested that a distinct metabolic program supports metastatic spread and is linked to extensive changes in chromatin regulation. Since multiple studies show that primary PDAC and metastasis share the same genetic profiles, the exploration of the non-genetic basis of metastatic outgrowth and dependency is an important area of study and the role of altered metabolic pathways in the survival of distant metastatic PDACs is a new paradigm in metastasis research. In the present study the authors have examined the impact of PDG on the regulation of glucose uptake. They find that the presence of high levels of PDG in cells derived from distant metastasis generates a positive feedback pathway to maintain high glucose uptake. They report that thioredoxin- interacting protein (TXNIP), an inhibitor of cellular glucose uptake, is reduced in these cells compared to cells from peritoneal metastasis, and that PGD knockdown increases TXNIP levels and suppresses glucose u ptake. They relate these effects to the inhibition of the MondoA transcription factor by accumulation of the PDG substrate, 6PG. They also show that TNXIP overexpression specifically reduces the growth of 3D organoids from distant metastasis. Finally, the authors show in patient tumor samples, that TXNIP staining is reduced distant metastases compared to peritoneal metastases. These studies are novel and of interest and the data showing the relationship between PDG, TXNIP and glucose uptake are clear-cut. The primary concern is that limitations in the model systems restrict the conclusions that can be drawn. The study is primarily framed around the difference between peritoneal and distant metastasis. Moreover, most experiments compare a single pair of peritoneal vs distant metastasis models. It is not clear whether the differences noted relate to features particular to the peritoneal metastasis used as a reference rather than properties acquired as primary tumors undergo distant metastatic spread. Also, the generality of the observations is not established. Finally, the cell lines used in this manuscript do not form tumors in xenograft assays, and so their relevance to the in vivo situation is uncertain. While it is understood that these models are difficult to obtain, some additional support would be needed to confirm the conclusions of the paper. 1) Figure 1B should include additional primary tumor-derived lines in the comparison. The present analysis includes long- established cell lines from repositories and a single peritoneal met sample as references. A more relevant and complete comparison would utilize matched early passage primary and mets (preferably) or a number of early passage primary PDA lines. 2) The functional studies are based on a single pair of cell lines (38Per vs 38Lg). The phenotypes should be tested in multiple cell lines. 3) Please include immunoblots for TXNIP (Fig 1D) and the histone marks (Fig 3a). 4) The rationale for excluding primary tumors in the IHC analysis in Figure 4 is not clear. As noted above, it is not certain whether TXNIP is decreased in metastasis compared to primary PDAC or whether TXNIP levels are specifically elevated in peritoneal metastases. 5) The regulation of nucl-cyto translocation of MondoA by 6PG is an important observation (Extended data 3). The authors should clarify how these metabolites are treated in this experiment. Since phosphorylated compounds like 6PG and G6P are poorly cell-permeable, it is important to show the intracellular concentration of the 3PG after exogenous addition. 6) Are steady-state 6PG levels reduced in the PGD-high cells? Also, in 7) The impact of the study would be enhanced by some insight into how inhibition of PGD regulates mondoA -TXNIP axis? 8) In Figure 4 the authors used siRNA to knockdown TXNIP and performed 3D growth assays over 7-21 days. Since siRNA causes transient knockdown the authors may want to consider additional controls. Reviewer #2: Remarks to the Author: This is an interesting and well done manuscript that describes a role for altered metabolism in distant pancreatic cancer metastatic lesions. This manuscripts follows others from the same group and should be of high general interest. The data is convincing and clearly shows that glucose uptake is upregulate in distant lesions that are dependent on PGD and that high PDG levels result in suppression of TXNIP. This pathway dissection is followed up by a reasonable set of experiments to validate the biological significance of their working model. Particularly interesting is the finding that in PDG tumors the levels of H3K27ac are upregulated, perhaps suggesting a broad rewiring of gene expression in this subgroup of lesions. On the downside, the manuscript is largely descriptive and lacks mechanistic depth. This is a major concern. 1. The proposal is that TXNIP loss leads to an increase in glucose uptake, which is consistent with a body of literature and clearly shown here. They propose that the increase in glucose uptake leads to an increase in Acetyl-CoA resulting in an ncrease levels of histone acetylation. which is not explicitly tested here. This is important given TXNIP’s many roles that seem to be unrelated to functions in glucose uptake and/or metabolism. For example, TXNIP loss in MEFs leads to activation of PI3-Kinase, which could easily account for their observations. 2. The staining for H3K27Ac is convincing as is the ChIP profiling for H3K27Ac at CDH2 shown in Figure 3b. At least a few more TXNIP regulated targets shown in figure 3b, should be evaluated. It may be just as easy to perform a ChIP-seq experiment for H3K27ac. 3. PGD loss increases TXNIP expression and the authors’ have some data that induction is MondoA-dependent. They should test this directly and determine whether TXNIP induction depends on MondoA as they data as it now stands is only correlative. It would also be valuable to determine if PGD loss leads to an increase in G6P as this is the primary regulator of MondoA transcriptional activity. Likewise measuring G6P levels in PGDhigh lesions would also be informative. Determining how G6P regulates MondoA activity is a big ask and is beyond the scope of this manuscript. 4. This is more of a minor concern. The author’s present their data as consistent with both positive and negative feedback loops. Clearly this is complicated, but the description was cumbersome in my opinion and they don’t really test the regulatory loops rigorously. It may be more straight forward just to describe the metabolic wiring of cells in the PDGhigh and the PDGlow states. Reviewer #3: Remarks to the Author: Bechard et al study distant vs peritoneal metastasis in PDAC. They find that distant metastasis with high PGD (identified in a previous paper) have low expression of TXNIP. Reduced expression of TXNIP is known to lead to increased GLUT1 expression and therefore glucose uptake. Accordingly, they find that PGD regulated TXNIP expression impacts on glucose uptake. In line, exogenous TXNIP decreased 3D growth of distant metastasis but not peritoneal metastasis (while TXNIP silencing impair peritoneal metastasis) and TXNIP was less expressed in distant metastasis from patients. The main finding of this paper is that PGD regulates (by an unknown mechanism) TXNIP expression and that this decreased expression is important for the previously described change in histone acetylation. Major comments: 1) The focus on a switch from a negative to a positive feedback loop is confusing and not supported by data. In my opinion, the data simply suggest that PGD is a negative regulator of TXNIP. 2) Accordingly, it is required that title and abstract reflect the actual data presented in the manuscript rather than a speculative interpretation of the data. 3) The authors mention several times the context of evolution and non-genetic driver. These terms should be removed from the manuscript or experiments should be provided to support the claims. For example, is PGD overexpression or TXNIP deletion sufficient to drive a non-metastatic PDAC cell to lung or liver? 4) Evidence should be provided that altering PGD or TXNIP prevents in vivo metastasis formation.	Reviewers' Comments: Reviewer #1: Remarks to the Author: This study by Bechard et al builds on the authors’ previous findings that cells derived from distant metastases of pancreatic ductal adenocarcinoma (PDAC) are specifically dependent on the oxidative pentose phosphate pathway (ox-PPP) enzyme, phosphogluconate dehydrogenase (PGD), for growth in vitro. This work suggested that a distinct metabolic program supports metastatic spread and is linked to extensive changes in chromatin regulation. Since multiple studies show that primary PDAC and metastasis share the same genetic profiles, the exploration of the non-genetic basis of metastatic outgrowth and dependency is an important area of study and the role of altered metabolic pathways in the survival of distant metastatic PDACs is a new paradigm in metastasis research. In the present study the authors have examined the impact of PDG on the regulation of glucose uptake. They find that the presence of high levels of PDG in cells derived from distant metastasis generates a positive feedback pathway to maintain high glucose uptake. They report that thioredoxin- interacting protein (TXNIP), an inhibitor of cellular glucose uptake, is reduced in these cells compared to cells from peritoneal metastasis, and that PGD knockdown increases TXNIP levels and suppresses glucose u ptake. They relate these effects to the inhibition of the MondoA transcription factor by accumulation of the PDG substrate, 6PG. They also show that TNXIP overexpression specifically reduces the growth of 3D organoids from distant metastasis. Finally, the authors show in patient tumor samples, that TXNIP staining is reduced distant metastases compared to peritoneal metastases. These studies are novel and of interest and the data showing the relationship between PDG, TXNIP and glucose uptake are clear-cut. The primary concern is that limitations in the model systems restrict the conclusions that can be drawn. The study is primarily framed around the difference between peritoneal and distant metastasis. Moreover, most experiments compare a single pair of peritoneal vs distant metastasis models. It is not clear whether the differences noted relate to features particular to the peritoneal metastasis used as a reference rather than properties acquired as primary tumors undergo distant metastatic spread. Also, the generality of the observations is not established. Finally, the cell lines used in this manuscript do not form tumors in xenograft assays, and so their relevance to the in vivo situation is uncertain. While it is understood that these models are difficult to obtain, some additional support would be needed to confirm the conclusions of the paper. 1) Figure 1B should include additional primary tumor-derived lines in the comparison. The present analysis includes long- established cell lines from repositories and a single peritoneal met sample as references. A more relevant and complete comparison would utilize matched early passage primary and mets (preferably) or a number of early passage primary PDA lines. 2) The functional studies are based on a single pair of cell lines (38Per vs 38Lg). The phenotypes should be tested in multiple cell lines. 3) Please include immunoblots for TXNIP (Fig 1D) and the histone marks (Fig 3a). 4) The rationale for excluding primary tumors in the IHC analysis in Figure 4 is not clear. As noted above, it is not certain whether TXNIP is decreased in metastasis compared to primary PDAC or whether TXNIP levels are specifically elevated in peritoneal metastases. 5) The regulation of nucl-cyto translocation of MondoA by 6PG is an important observation (Extended data 3). The authors should clarify how these metabolites are treated in this experiment. Since phosphorylated compounds like 6PG and G6P are poorly cell-permeable, it is important to show the intracellular concentration of the 3PG after exogenous addition. 6) Are steady-state 6PG levels reduced in the PGD-high cells? Also, in 7) The impact of the study would be enhanced by some insight into how inhibition of PGD regulates mondoA -TXNIP axis? 8) In Figure 4 the authors used siRNA to knockdown TXNIP and performed 3D growth assays over 7-21 days. Since siRNA causes transient knockdown the authors may want to consider additional controls. Reviewer #2: Remarks to the Author: This is an interesting and well done manuscript that describes a role for altered metabolism in distant pancreatic cancer metastatic lesions. This manuscripts follows others from the same group and should be of high general interest. The data is convincing and clearly shows that glucose uptake is upregulate in distant lesions that are dependent on PGD and that high PDG levels result in suppression of TXNIP. This pathway dissection is followed up by a reasonable set of experiments to validate the biological significance of their working model. Particularly interesting is the finding that in PDG tumors the levels of H3K27ac are upregulated, perhaps suggesting a broad rewiring of gene expression in this subgroup of lesions. On the downside, the manuscript is largely descriptive and lacks mechanistic depth. This is a major concern. 1. The proposal is that TXNIP loss leads to an increase in glucose uptake, which is consistent with a body of literature and clearly shown here. They propose that the increase in glucose uptake leads to an increase in Acetyl-CoA resulting in an ncrease levels of histone acetylation. which is not explicitly tested here. This is important given TXNIP’s many roles that seem to be unrelated to functions in glucose uptake and/or metabolism. For example, TXNIP loss in MEFs leads to activation of PI3-Kinase, which could easily account for their observations. 2. The staining for H3K27Ac is convincing as is the ChIP profiling for H3K27Ac at CDH2 shown in Figure 3b. At least a few more TXNIP regulated targets shown in figure 3b, should be evaluated. It may be just as easy to perform a ChIP-seq experiment for H3K27ac. 3. PGD loss increases TXNIP expression and the authors’ have some data that induction is MondoA-dependent. They should test this directly and determine whether TXNIP induction depends on MondoA as they data as it now stands is only correlative. It would also be valuable to determine if PGD loss leads to an increase in G6P as this is the primary regulator of MondoA transcriptional activity. Likewise measuring G6P levels in PGDhigh lesions would also be informative. Determining how G6P regulates MondoA activity is a big ask and is beyond the scope of this manuscript. 4. This is more of a minor concern. The author’s present their data as consistent with both positive and negative feedback loops. Clearly this is complicated, but the description was cumbersome in my opinion and they don’t really test the regulatory loops rigorously. It may be more straight forward just to describe the metabolic wiring of cells in the PDGhigh and the PDGlow states. Reviewer #3: Remarks to the Author: Bechard et al study distant vs peritoneal metastasis in PDAC. They find that distant metastasis with high PGD (identified in a previous paper) have low expression of TXNIP. Reduced expression of TXNIP is known to lead to increased GLUT1 expression and therefore glucose uptake. Accordingly, they find that PGD regulated TXNIP expression impacts on glucose uptake. In line, exogenous TXNIP decreased 3D growth of distant metastasis but not peritoneal metastasis (while TXNIP silencing impair peritoneal metastasis) and TXNIP was less expressed in distant metastasis from patients. The main finding of this paper is that PGD regulates (by an unknown mechanism) TXNIP expression and that this decreased expression is important for the previously described change in histone acetylation. Major comments: 1) The focus on a switch from a negative to a positive feedback loop is confusing and not supported by data. In my opinion, the data simply suggest that PGD is a negative regulator of TXNIP. 2) Accordingly, it is required that title and abstract reflect the actual data presented in the manuscript rather than a speculative interpretation of the data. 3) The authors mention several times the context of evolution and non-genetic driver. These terms should be removed from the manuscript or experiments should be provided to support the claims. For example, is PGD overexpression or TXNIP deletion sufficient to drive a non-metastatic PDAC cell to lung or liver? 4) Evidence should be provided that altering PGD or TXNIP prevents in vivo metastasis formation.	TPR	NC_BS_CAN_008_RR1.txt	{ "criticism": 13, "example": 5, "importance_and_relevance": 8, "materials_and_methods": 41, "praise": 5, "presentation_and_reporting": 6, "results_and_discussion": 6, "suggestion_and_solution": 17, "total": 73 }	{ "criticism": 0.1780821917808219, "example": 0.0684931506849315, "importance_and_relevance": 0.1095890410958904, "materials_and_methods": 0.5616438356164384, "praise": 0.0684931506849315, "presentation_and_reporting": 0.0821917808219178, "results_and_discussion": 0.0821917808219178, "suggestion_and_solution": 0.2328767123287671 }	1.383562	[ { "content": "Please review this paper.", "role": "user" }, { "content": "Reviewers' Comments: \n \nReviewer #1: \nRemarks to the Author: \nThis study by Bechard et al builds on the authors’ previous findings that cells derived from distant metastases of pancreatic \nductal adenocarcinoma (PDAC) are specifically dependent on the oxidative pentose phosphate pathway (ox-PPP) enzyme, \nphosphogluconate dehydrogenase (PGD), for growth in vitro. This work suggested that a distinct metabolic program \nsupports metastatic spread and is linked to extensive changes in chromatin regulation. Since multiple studies show that \nprimary PDAC and metastasis share the same genetic profiles, the exploration of the non-genetic basis of metastatic \noutgrowth and dependency is an important area of study and the role of altered metabolic pathways in the survival of \ndistant metastatic PDACs is a new paradigm in metastasis research. In the present study the authors have examined the \nimpact of PDG on the regulation of glucose uptake. They find that the presence of high levels of PDG in cells derived from \ndistant metastasis generates a positive feedback pathway to maintain high glucose uptake. They report that thioredoxin-\ninteracting protein (TXNIP), an inhibitor of cellular glucose uptake, is reduced in these cells compared to cells from \nperitoneal metastasis, and that PGD knockdown increases TXNIP levels and suppresses glucose u ptake. They relate these \neffects to the inhibition of the MondoA transcription factor by accumulation of the PDG substrate, 6PG. They also show that \nTNXIP overexpression specifically reduces the growth of 3D organoids from distant metastasis. Finally, the authors show in \npatient tumor samples, that TXNIP staining is reduced distant metastases compared to peritoneal metastases. \n \nThese studies are novel and of interest and the data showing the relationship between PDG, TXNIP and glucose uptake are \nclear-cut. The primary concern is that limitations in the model systems restrict the conclusions that can be drawn. The \nstudy is primarily framed around the difference between peritoneal and distant metastasis. Moreover, most experiments \ncompare a single pair of peritoneal vs distant metastasis models. It is not clear whether the differences noted relate to \nfeatures particular to the peritoneal metastasis used as a reference rather than properties acquired as primary tumors \nundergo distant metastatic spread. Also, the generality of the observations is not established. Finally, the cell lines used in \nthis manuscript do not form tumors in xenograft assays, and so their relevance to the in vivo situation is uncertain. While \nit is understood that these models are difficult to obtain, some additional support would be needed to confirm the \nconclusions of the paper. \n \n1) Figure 1B should include additional primary tumor-derived lines in the comparison. The present analysis includes long-\nestablished cell lines from repositories and a single peritoneal met sample as references. A more relevant and complete \ncomparison would utilize matched early passage primary and mets (preferably) or a number of early passage primary PDA \nlines. \n2) The functional studies are based on a single pair of cell lines (38Per vs 38Lg). The phenotypes should be tested in \nmultiple cell lines. \n3) Please include immunoblots for TXNIP (Fig 1D) and the histone marks (Fig 3a). \n4) The rationale for excluding primary tumors in the IHC analysis in Figure 4 is not clear. As noted above, it is not certain \nwhether TXNIP is decreased in metastasis compared to primary PDAC or whether TXNIP levels are specifically elevated in \nperitoneal metastases. \n5) The regulation of nucl-cyto translocation of MondoA by 6PG is an important observation (Extended data 3). The authors \nshould clarify how these metabolites are treated in this experiment. Since phosphorylated compounds like 6PG and G6P \nare poorly cell-permeable, it is important to show the intracellular concentration of the 3PG after exogenous addition. \n6) Are steady-state 6PG levels reduced in the PGD-high cells? Also, in \n7) The impact of the study would be enhanced by some insight into how inhibition of PGD regulates mondoA -TXNIP axis? \n8) In Figure 4 the authors used siRNA to knockdown TXNIP and performed 3D growth assays over 7-21 days. Since siRNA \ncauses transient knockdown the authors may want to consider additional controls. \n \n \n \nReviewer #2: \nRemarks to the Author: \nThis is an interesting and well done manuscript that describes a role for altered metabolism in distant pancreatic cancer \nmetastatic lesions. This manuscripts follows others from the same group and should be of high general interest. The data \nis convincing and clearly shows that glucose uptake is upregulate in distant lesions that are dependent on PGD and that \nhigh PDG levels result in suppression of TXNIP. This pathway dissection is followed up by a reasonable set of experiments \nto validate the biological significance of their working model. Particularly interesting is the finding that in PDG tumors the \nlevels of H3K27ac are upregulated, perhaps suggesting a broad rewiring of gene expression in this subgroup of lesions. On \nthe downside, the manuscript is largely descriptive and lacks mechanistic depth. This is a major concern. \n1. The proposal is that TXNIP loss leads to an increase in glucose uptake, which is consistent with a body of literature and \nclearly shown here. They propose that the increase in glucose uptake leads to an increase in Acetyl-CoA resulting in an \nncrease levels of histone acetylation. which is not explicitly tested here. This is important given TXNIP’s many roles that \n\n\fseem to be unrelated to functions in glucose uptake and/or metabolism. For example, TXNIP loss in MEFs leads to \nactivation of PI3-Kinase, which could easily account for their observations. \n2. The staining for H3K27Ac is convincing as is the ChIP profiling for H3K27Ac at CDH2 shown in Figure 3b. At least a few \nmore TXNIP regulated targets shown in figure 3b, should be evaluated. It may be just as easy to perform a ChIP-seq \nexperiment for H3K27ac. \n3. PGD loss increases TXNIP expression and the authors’ have some data that induction is MondoA-dependent. They should \ntest this directly and determine whether TXNIP induction depends on MondoA as they data as it now stands is only \ncorrelative. It would also be valuable to determine if PGD loss leads to an increase in G6P as this is the primary regulator \nof MondoA transcriptional activity. Likewise measuring G6P levels in PGDhigh lesions would also be informative. \nDetermining how G6P regulates MondoA activity is a big ask and is beyond the scope of this manuscript. \n4. This is more of a minor concern. The author’s present their data as consistent with both positive and negative feedback \nloops. Clearly this is complicated, but the description was cumbersome in my opinion and they don’t really test the \nregulatory loops rigorously. It may be more straight forward just to describe the metabolic wiring of cells in the PDGhigh \nand the PDGlow states. \n \n \n \nReviewer #3: \nRemarks to the Author: \nBechard et al study distant vs peritoneal metastasis in PDAC. They find that distant metastasis with high PGD (identified in \na previous paper) have low expression of TXNIP. Reduced expression of TXNIP is known to lead to increased GLUT1 \nexpression and therefore glucose uptake. Accordingly, they find that PGD regulated TXNIP expression impacts on glucose \nuptake. In line, exogenous TXNIP decreased 3D growth of distant metastasis but not peritoneal metastasis (while TXNIP \nsilencing impair peritoneal metastasis) and TXNIP was less expressed in distant metastasis from patients. The main finding \nof this paper is that PGD regulates (by an unknown mechanism) TXNIP expression and that this decreased expression is \nimportant for the previously described change in histone acetylation. \n \nMajor comments: \n1) The focus on a switch from a negative to a positive feedback loop is confusing and not supported by data. In my \nopinion, the data simply suggest that PGD is a negative regulator of TXNIP. \n2) Accordingly, it is required that title and abstract reflect the actual data presented in the manuscript rather than a \nspeculative interpretation of the data. \n3) The authors mention several times the context of evolution and non-genetic driver. These terms should be removed \nfrom the manuscript or experiments should be provided to support the claims. For example, is PGD overexpression or \nTXNIP deletion sufficient to drive a non-metastatic PDAC cell to lung or liver? \n4) Evidence should be provided that altering PGD or TXNIP prevents in vivo metastasis formation. \n\n\f \n\n", "role": "assistant" } ]	2,020
Pancreatic cancers suppress negative feedback of glucose transport to reprogram chromatin	Reviewers' Comments: Reviewer #3: Remarks to the Author: The authors sufficiently addressed my concerns. Reviewer #4: Remarks to the Author: The authors have answered most of the questions raised during the first course of reviewing. The manuscript has improved. No more comments.	Reviewers' Comments: Reviewer #3: Remarks to the Author: The authors sufficiently addressed my concerns. Reviewer #4: Remarks to the Author: The authors have answered most of the questions raised during the first course of reviewing. The manuscript has improved. No more comments.	TPR	NC_BS_CAN_008_RR2.txt	{ "criticism": 0, "example": 0, "importance_and_relevance": 0, "materials_and_methods": 0, "praise": 2, "presentation_and_reporting": 0, "results_and_discussion": 0, "suggestion_and_solution": 0, "total": 4 }	{ "criticism": 0, "example": 0, "importance_and_relevance": 0, "materials_and_methods": 0, "praise": 0.5, "presentation_and_reporting": 0, "results_and_discussion": 0, "suggestion_and_solution": 0 }	0.5	[ { "content": "Please review this paper.", "role": "user" }, { "content": "\n\n\fReviewers' Comments: \n \nReviewer #3: \nRemarks to the Author: \nThe authors sufficiently addressed my concerns. \n \n \n \nReviewer #4: \nRemarks to the Author: \nThe authors have answered most of the questions raised during the first course of reviewing. The manuscript has \nimproved. No more comments. \n\n\f\n\f\n", "role": "assistant" } ]	2,020
FGF2 alters macrophage polarization, tumour immunity and growth and can be targeted	Reviewers' Comments: Reviewer #1: Remarks to the Author: The manuscript by Buzzelli at al. focuses on understanding the contributions of host-derived FGF2LMW to tumor growth and macrophage polarization using various tumor models. Initial studies demonstrate a significant decrease in tumor growth following subcutaneous injection into FGF2LMW-/- mice using various models (MC38, KPC and LLC). Further studies demonstrate alterations in immune profiles favoring an increased anti-tumor response in the FGF2-/-LMW mice driven by T cells. In addition, macrophages isolated from control and FGF2-/-LMW mice show altered expression of various cytokines and chemokines. To further define the contributions of macrophages to this phenotype, tumor cells were co-injected with control or FGF2-/-LMW bone marrow derived macrophages, which demonstrates reduced tumor growth although in a T cell- independent manner. Finally, irradiation was found to promote increased levels of FGF2, and treatment with anti-FGF2 led to better response to irradiation, which correlated with an increased iNOS:CD206 ratio in macrophages. Based on these findings, the authors conclude that FGF2 regulates tumor immunity and growth through altering macrophage polarization. While the link between FGFs and immune modulation within the tumor microenvironment is not necessarily novel, the findings that macrophage-derived FGF2 impacts tumor growth and that anti-FGF2 enhances efficacy of irradiation are of interest. While these studies demonstrate a role for host- derived FGF2LMW in promoting tumor growth, and effectively demonstrate efficacy of anti-FGF2 in the context of irradiation, some concerns are noted as described below. The authors should provide information in the Materials and Methods regarding the specific substrains of C57BL/6 mice that were used for these studies. As written, it appears that the FGF2LMW-/- mice were purchased from Jackson Laboratories on the C57BL/6J background and maintained in-house, and the control mice were purchased from Charles River, which provides the C57BL/6N substrain. If this is correct, this should be noted in the methods section. Furthermore, in light of recent observations that differences in substrain can impact immune phenotypes (for example, PMID 27210752), if different substrains were indeed used in these studies, this should be discussed as a potential caveat particularly of the immune phenotype analysis. Ideally, key experiments involving immune profiling, including the macrophage polarization studies, should be repeated using mice of the same substrain. Figure 3: While basal levels of a number of the factors shown are significantly different than control, the magnitude of induction seems to be similar. Does restoration of FGF2 to the FGF2LMW-/- cells impact basal expression levels of these factors? Figure 5C: The conclusion that the majority of FGF2-producing cells appear to be TAMs is not supported by the images provided as there are substantial areas of green FGF2 staining that do not appear to be associated with F4/80 staining. Quantification of dual-staining or flow cytometry- based quantification should be provided to support these conclusions. The authors conclude several times (in the title, lines 215-216 and 352) that “macrophage polarization by FGF2LMW mediates tumor growth”. However, this conclusion seems premature given the studies provided. There are several other possible explanations for the phenotypes that have not been ruled out, such as reduced tumor growth due to loss of macrophage-derived FGF2- mediated effects on tumor cells, or other non-immune host cells, which could ultimately impact both tumor growth rates and the immune environment. Thus, additional potential explanations for the phenotypes observed should be thoroughly considered. The authors demonstrate that expression levels of FGFR1 and FGFR2 are higher on TAMs compared with other immune cell populations. Does FGF2 directly activate FGFR signaling in TAMs? In Figure 1, the authors demonstrate that the reduction in tumor growth observed in the FGF2LMW-/- mice is T cell-dependent. However, in Figure 4, T cells are found to be indispensible for the reduction in tumor growth induced by the FGF2LMW-/- macrophages. Potential reasons for these differences (such as the possibility that non-macrophage FGF2LMW-expressing host cells are impacting the T cell population in the FGF2LMW-/- mice) should be discussed. Because the flow plots are presented in the context of FSC or SSC, the authors should provide the gating schemes used to identify the different immune cell populations (as Supplementary data). Please provide the number of mice used for each in vivo experiment. Figure 1: Figures J and K are switched when referred to in the text. Figure 2: Please clarify whether this antibody recognizes only the secreted form of FGF2, or whether it also recognizes the intracellular form of FGF2, and whether this impacts interpretation of the data. Figure 3A: The shifts are difficult to see in the panels provided and may be more easily interpreted as histograms. Also, please provide the iNOS:CD206 ratios along with statistical analysis. Figure 4E: The authors conclude that there is a reduction in MDSCs in tumors with FGF2LMW-/- macrophages, although this does not appear to be supported by the data presented in the graph. Figure 6: For clarity, please provide the cell number count for each plot. Although a general statement regarding statistical analysis is included in the materials and methods section, it is not clear which test was used for each graph, which could be provided in the Figure Legends. Reviewer #3: Remarks to the Author: Buzzelli et al present experimental evidence that suggests a role of FGF2 as a key modulator of macrophage polarization in tumors, as part of the "wound response" associated with localized radio-therapy. They identify FGF2 as a potential target of translational interest, since antibodies against FGF2 could remove one of radio-therapy immune-suppressive effects, while exploiting other pro-immunogenic ones. Several concerns: 1. The mechanism of radio-therapy induction of FGF2 is not clear. In vitro experiments are needed to specifically define the nature of signaling after irradiation, whether it is different in different cells types and whether it is irradiation-does dependent. 2. Is FGF2 induction part of DDR? A heat map for FGF2 null versus parental lines could help understanding the pathway. It could also provide important hints regarding the component of cross talk with innate immunity 3. The previous work of the authors in reference 25 shows an effect of FGF2 on angiogenesis, independent of irradiation. Understanding the mechanism of FGF2 induction by irradiation is key, since it may also have clinical impact in cancer metastasis. 4. A main concern is the use of only one radiotherapy regimen (9GyX2), without clear justification for this selection. Particularly in the in vivo experiments a different dose could have overcome the progression of tumors and/or the type of TAMs infiltration. Since no dose response curves were provided in the model used, one can not exclude that re-population because of inadequate dose may have also recruited M2 and/or induced polarization of TAMs. The choice of this radio-therapy dose and number of fractions needs a much more explicit justification, since others have shown very different results for instance when low doses are used. (Klug et al Cancer Cell. 2013 Nov 11;24(5):589-602) 5. Figure 5 Shows modest difference in HT29 model and the delay in MC38 is only transient. From the images it is hard to conclude a sustained superior effect by the combination 6. References on irradiation and macrophages are incomplete (missing key papers of Deng L et al, Xu J et al etc.). Reviewer #4: Remarks to the Author: In this manuscript, Buzzelli et al. show slower growth of primary tumors and decreased liver metastasis in FGF2lmw deleted mice. Mechanistic studies showed that the deletion induces T cell recruitment into the tumors and that this is critical for the decreased tumor growth. Although the authors were not able to detect FGFR1 or 2 in the T cells, they found that the majority of tumor- associated macrophages expressed FGFR1 and 2. Both FGF2lmw deletion and anti-FGF2 antibody induced TAM polarization towards a more inflammatory phenotype. Decreased tumor xenograft growth was seen in nude mice when anti-FGF2 blocking antibody was used in combination with irradiation. The manuscript reports many interesting, but somewhat dissociated results that raise several questions. In the previous paper by the same authors, metastasis-associated neutrophils expressed substantially more fibroblast growth factor 2 (FGF2) in comparison to naïve neutrophils, indicating neutrophil polarization by the tumor microenvironment. Administration of FGF2 neutralizing antibody to mice bearing experimental liver metastases phenocopied neutrophil depletion by reducing liver metastatic colony growth, vascular density, and branching (Hepatology 2017;65:1920-1935). It has also been shown that FGFR1 inhibition by gene silencing in tumor cells or administration of a compound (SSR128129E at nanomolar concentrations) inhibiting FGF-induced signaling in tumor xenografts increases the efficiency of radiotherapy (pubmed/26896280). Thus, an alternative possibility to explain the authors’ results is that FGF2lmw depletion directly causes initial TC death in the tumors in vivo, and that the resulting danger signals invite inflammatory/immune cells to the tumors. This possibility is consistent with the “T-cell independent alteration” that “resulted in less rapid tumor growth in FGF2lmw-/- mice than in WT mice”. Although the authors show that anti-FGF2 antibody treatment after irradiation did not alter clonogenic survival of SW1222 and HT29 cells, it would be essential to also analyze the effect of FGF2 silencing. In the manuscript, a limited analysis is reported on the role of TC produced FGF, and the expression of FGF receptors. A comparison of the amounts of FGF2 produced by TCs and macrophages should be shown. A more careful mapping of the autocrine pathway in the TCs is also needed. It would be essential to compare the survival of the control, FGF2 and FGFR1 silenced as well as FGF2 antibody treated tumor cell cultures after irradiation. On pages 5 and 6 the authors show that T-cells are required for tumor regression in the FGF2lmw- /- mice. However, later, in Figure 5 F-H and Supplementary Figure 4, they show that T-cells are not required for the effects of the anti-FGF2 antibody on human SW1222 and HT29 tumor xenografts in immunodeficient mice. The authors should analyze the mouse tumor cell lines in using irradiation with and without the anti-FGF2 antibody in an immunocompromised background. Alternatively, they could use another inhibitor of the FGF2-FGFR pathway, such as SSR128129E, a FGFR1 blocking antibody, soluble receptor or FGFR1 gene silencing in the tumor cells. In addition, the authors should discuss why anti-FGF2 as a monotherapy does not decrease xenograft growth (Fig 5 F-H), although FGF2lmw deletion decreases tumor growth in the isogenic mice (Fig. 1). As the authors indicate in the introduction, FGF2lmw is a highly potent angiogenic molecule. However, here the authors don`t provide any tumor vessel analysis. This should be analyzed to exclude the possible role of decreased vascular density to the decreased tumor growth. If anti- FGF2 antibody treatment decreases vascular density as shown previously in the case of liver metastasis, it could explain why anti-FGF2 antibody treatment decreases SW1222 and HT29 tumor growth in immunodeficient mice. Specific comments: Another question concerns the specificity of the effects reported here. The authors should perhaps mention that the other 21 members of FGF family are not expressed or of concern here. Similarly, they should describe which of the four receptors are present in the tumor cells they use. When reporting immune infiltration to the metastases/tumors, the authors should also report if this concerned normal tissues or circulating immune cells. The authors should re-isolate the escape clones that resumed growth and test if they regrow in corresponding experiments in FGF2lmw-/- mice. The LLC growth curves should be added to Supplementary Figure 1. Did TAMs express FGFR3 or FGFR4? Was phosphorylation of FGFR1 correlated with FGF2lmw expression or irradiation in TCs or TAMs? The anti-FGF2 antibody was used for the MC38 clonogenic assay in Supplementary Figure 5. However, only human tumor cells were used in the in vivo experiments. Why were the mouse tumor cells not used in the in vivo tumor experiments? As the authors explain, FGF2 expression has been reported to be induced in irradiated human tumors. Thus, Figs 5A and B are of little value. In Fig. 5E, much of the macrophage and FGF2 staining appears to be eradicated by the irradiation. The authors should show the specificity of anti-FGF2 by staining macrophages of the corresponding tumors grown in FGF2lmw-/- mice. Minor comments • Fig. 1 J and K are mixed on page 6? • How was liver tumor burden measured? • How was tumor volume calculated? The authors should study M1/M2 ratio in tumors beyond the (very small) volume of 500mm3. • Fig. 1 B: Please show the invasion front in high magnification, with the identification of the tumor cells (by IHC). • In all experiments, the number of mice should be mentioned in each group. The individual tumor growth curves are difficult to interpret, for example in Fig. 5F.	Reviewers' Comments: Reviewer #1: Remarks to the Author: The manuscript by Buzzelli at al. focuses on understanding the contributions of host-derived FGF2LMW to tumor growth and macrophage polarization using various tumor models. Initial studies demonstrate a significant decrease in tumor growth following subcutaneous injection into FGF2LMW-/- mice using various models (MC38, KPC and LLC). Further studies demonstrate alterations in immune profiles favoring an increased anti-tumor response in the FGF2-/-LMW mice driven by T cells. In addition, macrophages isolated from control and FGF2-/-LMW mice show altered expression of various cytokines and chemokines. To further define the contributions of macrophages to this phenotype, tumor cells were co-injected with control or FGF2-/-LMW bone marrow derived macrophages, which demonstrates reduced tumor growth although in a T cell- independent manner. Finally, irradiation was found to promote increased levels of FGF2, and treatment with anti-FGF2 led to better response to irradiation, which correlated with an increased iNOS:CD206 ratio in macrophages. Based on these findings, the authors conclude that FGF2 regulates tumor immunity and growth through altering macrophage polarization. While the link between FGFs and immune modulation within the tumor microenvironment is not necessarily novel, the findings that macrophage-derived FGF2 impacts tumor growth and that anti-FGF2 enhances efficacy of irradiation are of interest. While these studies demonstrate a role for host- derived FGF2LMW in promoting tumor growth, and effectively demonstrate efficacy of anti-FGF2 in the context of irradiation, some concerns are noted as described below. The authors should provide information in the Materials and Methods regarding the specific substrains of C57BL/6 mice that were used for these studies. As written, it appears that the FGF2LMW-/- mice were purchased from Jackson Laboratories on the C57BL/6J background and maintained in-house, and the control mice were purchased from Charles River, which provides the C57BL/6N substrain. If this is correct, this should be noted in the methods section. Furthermore, in light of recent observations that differences in substrain can impact immune phenotypes (for example, PMID 27210752), if different substrains were indeed used in these studies, this should be discussed as a potential caveat particularly of the immune phenotype analysis. Ideally, key experiments involving immune profiling, including the macrophage polarization studies, should be repeated using mice of the same substrain. Figure 3: While basal levels of a number of the factors shown are significantly different than control, the magnitude of induction seems to be similar. Does restoration of FGF2 to the FGF2LMW-/- cells impact basal expression levels of these factors? Figure 5C: The conclusion that the majority of FGF2-producing cells appear to be TAMs is not supported by the images provided as there are substantial areas of green FGF2 staining that do not appear to be associated with F4/80 staining. Quantification of dual-staining or flow cytometry- based quantification should be provided to support these conclusions. The authors conclude several times (in the title, lines 215-216 and 352) that “macrophage polarization by FGF2LMW mediates tumor growth”. However, this conclusion seems premature given the studies provided. There are several other possible explanations for the phenotypes that have not been ruled out, such as reduced tumor growth due to loss of macrophage-derived FGF2- mediated effects on tumor cells, or other non-immune host cells, which could ultimately impact both tumor growth rates and the immune environment. Thus, additional potential explanations for the phenotypes observed should be thoroughly considered. The authors demonstrate that expression levels of FGFR1 and FGFR2 are higher on TAMs compared with other immune cell populations. Does FGF2 directly activate FGFR signaling in TAMs? In Figure 1, the authors demonstrate that the reduction in tumor growth observed in the FGF2LMW-/- mice is T cell-dependent. However, in Figure 4, T cells are found to be indispensible for the reduction in tumor growth induced by the FGF2LMW-/- macrophages. Potential reasons for these differences (such as the possibility that non-macrophage FGF2LMW-expressing host cells are impacting the T cell population in the FGF2LMW-/- mice) should be discussed. Because the flow plots are presented in the context of FSC or SSC, the authors should provide the gating schemes used to identify the different immune cell populations (as Supplementary data). Please provide the number of mice used for each in vivo experiment. Figure 1: Figures J and K are switched when referred to in the text. Figure 2: Please clarify whether this antibody recognizes only the secreted form of FGF2, or whether it also recognizes the intracellular form of FGF2, and whether this impacts interpretation of the data. Figure 3A: The shifts are difficult to see in the panels provided and may be more easily interpreted as histograms. Also, please provide the iNOS:CD206 ratios along with statistical analysis. Figure 4E: The authors conclude that there is a reduction in MDSCs in tumors with FGF2LMW-/- macrophages, although this does not appear to be supported by the data presented in the graph. Figure 6: For clarity, please provide the cell number count for each plot. Although a general statement regarding statistical analysis is included in the materials and methods section, it is not clear which test was used for each graph, which could be provided in the Figure Legends. Reviewer #3: Remarks to the Author: Buzzelli et al present experimental evidence that suggests a role of FGF2 as a key modulator of macrophage polarization in tumors, as part of the "wound response" associated with localized radio-therapy. They identify FGF2 as a potential target of translational interest, since antibodies against FGF2 could remove one of radio-therapy immune-suppressive effects, while exploiting other pro-immunogenic ones. Several concerns: 1. The mechanism of radio-therapy induction of FGF2 is not clear. In vitro experiments are needed to specifically define the nature of signaling after irradiation, whether it is different in different cells types and whether it is irradiation-does dependent. 2. Is FGF2 induction part of DDR? A heat map for FGF2 null versus parental lines could help understanding the pathway. It could also provide important hints regarding the component of cross talk with innate immunity 3. The previous work of the authors in reference 25 shows an effect of FGF2 on angiogenesis, independent of irradiation. Understanding the mechanism of FGF2 induction by irradiation is key, since it may also have clinical impact in cancer metastasis. 4. A main concern is the use of only one radiotherapy regimen (9GyX2), without clear justification for this selection. Particularly in the in vivo experiments a different dose could have overcome the progression of tumors and/or the type of TAMs infiltration. Since no dose response curves were provided in the model used, one can not exclude that re-population because of inadequate dose may have also recruited M2 and/or induced polarization of TAMs. The choice of this radio-therapy dose and number of fractions needs a much more explicit justification, since others have shown very different results for instance when low doses are used. (Klug et al Cancer Cell. 2013 Nov 11;24(5):589-602) 5. Figure 5 Shows modest difference in HT29 model and the delay in MC38 is only transient. From the images it is hard to conclude a sustained superior effect by the combination 6. References on irradiation and macrophages are incomplete (missing key papers of Deng L et al, Xu J et al etc.). Reviewer #4: Remarks to the Author: In this manuscript, Buzzelli et al. show slower growth of primary tumors and decreased liver metastasis in FGF2lmw deleted mice. Mechanistic studies showed that the deletion induces T cell recruitment into the tumors and that this is critical for the decreased tumor growth. Although the authors were not able to detect FGFR1 or 2 in the T cells, they found that the majority of tumor- associated macrophages expressed FGFR1 and 2. Both FGF2lmw deletion and anti-FGF2 antibody induced TAM polarization towards a more inflammatory phenotype. Decreased tumor xenograft growth was seen in nude mice when anti-FGF2 blocking antibody was used in combination with irradiation. The manuscript reports many interesting, but somewhat dissociated results that raise several questions. In the previous paper by the same authors, metastasis-associated neutrophils expressed substantially more fibroblast growth factor 2 (FGF2) in comparison to naïve neutrophils, indicating neutrophil polarization by the tumor microenvironment. Administration of FGF2 neutralizing antibody to mice bearing experimental liver metastases phenocopied neutrophil depletion by reducing liver metastatic colony growth, vascular density, and branching (Hepatology 2017;65:1920-1935). It has also been shown that FGFR1 inhibition by gene silencing in tumor cells or administration of a compound (SSR128129E at nanomolar concentrations) inhibiting FGF-induced signaling in tumor xenografts increases the efficiency of radiotherapy (pubmed/26896280). Thus, an alternative possibility to explain the authors’ results is that FGF2lmw depletion directly causes initial TC death in the tumors in vivo, and that the resulting danger signals invite inflammatory/immune cells to the tumors. This possibility is consistent with the “T-cell independent alteration” that “resulted in less rapid tumor growth in FGF2lmw-/- mice than in WT mice”. Although the authors show that anti-FGF2 antibody treatment after irradiation did not alter clonogenic survival of SW1222 and HT29 cells, it would be essential to also analyze the effect of FGF2 silencing. In the manuscript, a limited analysis is reported on the role of TC produced FGF, and the expression of FGF receptors. A comparison of the amounts of FGF2 produced by TCs and macrophages should be shown. A more careful mapping of the autocrine pathway in the TCs is also needed. It would be essential to compare the survival of the control, FGF2 and FGFR1 silenced as well as FGF2 antibody treated tumor cell cultures after irradiation. On pages 5 and 6 the authors show that T-cells are required for tumor regression in the FGF2lmw- /- mice. However, later, in Figure 5 F-H and Supplementary Figure 4, they show that T-cells are not required for the effects of the anti-FGF2 antibody on human SW1222 and HT29 tumor xenografts in immunodeficient mice. The authors should analyze the mouse tumor cell lines in using irradiation with and without the anti-FGF2 antibody in an immunocompromised background. Alternatively, they could use another inhibitor of the FGF2-FGFR pathway, such as SSR128129E, a FGFR1 blocking antibody, soluble receptor or FGFR1 gene silencing in the tumor cells. In addition, the authors should discuss why anti-FGF2 as a monotherapy does not decrease xenograft growth (Fig 5 F-H), although FGF2lmw deletion decreases tumor growth in the isogenic mice (Fig. 1). As the authors indicate in the introduction, FGF2lmw is a highly potent angiogenic molecule. However, here the authors don`t provide any tumor vessel analysis. This should be analyzed to exclude the possible role of decreased vascular density to the decreased tumor growth. If anti- FGF2 antibody treatment decreases vascular density as shown previously in the case of liver metastasis, it could explain why anti-FGF2 antibody treatment decreases SW1222 and HT29 tumor growth in immunodeficient mice. Specific comments: Another question concerns the specificity of the effects reported here. The authors should perhaps mention that the other 21 members of FGF family are not expressed or of concern here. Similarly, they should describe which of the four receptors are present in the tumor cells they use. When reporting immune infiltration to the metastases/tumors, the authors should also report if this concerned normal tissues or circulating immune cells. The authors should re-isolate the escape clones that resumed growth and test if they regrow in corresponding experiments in FGF2lmw-/- mice. The LLC growth curves should be added to Supplementary Figure 1. Did TAMs express FGFR3 or FGFR4? Was phosphorylation of FGFR1 correlated with FGF2lmw expression or irradiation in TCs or TAMs? The anti-FGF2 antibody was used for the MC38 clonogenic assay in Supplementary Figure 5. However, only human tumor cells were used in the in vivo experiments. Why were the mouse tumor cells not used in the in vivo tumor experiments? As the authors explain, FGF2 expression has been reported to be induced in irradiated human tumors. Thus, Figs 5A and B are of little value. In Fig. 5E, much of the macrophage and FGF2 staining appears to be eradicated by the irradiation. The authors should show the specificity of anti-FGF2 by staining macrophages of the corresponding tumors grown in FGF2lmw-/- mice. Minor comments • Fig. 1 J and K are mixed on page 6? • How was liver tumor burden measured? • How was tumor volume calculated? The authors should study M1/M2 ratio in tumors beyond the (very small) volume of 500mm3. • Fig. 1 B: Please show the invasion front in high magnification, with the identification of the tumor cells (by IHC). • In all experiments, the number of mice should be mentioned in each group. The individual tumor growth curves are difficult to interpret, for example in Fig. 5F.	TPR	NC_BS_CAN_009_RR1.txt	{ "criticism": 12, "example": 20, "importance_and_relevance": 3, "materials_and_methods": 66, "praise": 1, "presentation_and_reporting": 26, "results_and_discussion": 17, "suggestion_and_solution": 33, "total": 111 }	{ "criticism": 0.10810810810810811, "example": 0.18018018018018017, "importance_and_relevance": 0.02702702702702703, "materials_and_methods": 0.5945945945945946, "praise": 0.009009009009009009, "presentation_and_reporting": 0.23423423423423423, "results_and_discussion": 0.15315315315315314, "suggestion_and_solution": 0.2972972972972973 }	1.603604	[ { "content": "Please review this paper.", "role": "user" }, { "content": "Reviewers' Comments: \n \nReviewer #1: \nRemarks to the Author: \nThe manuscript by Buzzelli at al. focuses on understanding the contributions of host-derived \nFGF2LMW to tumor growth and macrophage polarization using various tumor models. Initial \nstudies demonstrate a significant decrease in tumor growth following subcutaneous injection into \nFGF2LMW-/- mice using various models (MC38, KPC and LLC). Further studies demonstrate \nalterations in immune profiles favoring an increased anti-tumor response in the FGF2-/-LMW mice \ndriven by T cells. In addition, macrophages isolated from control and FGF2-/-LMW mice show \naltered expression of various cytokines and chemokines. To further define the contributions of \nmacrophages to this phenotype, tumor cells were co-injected with control or FGF2-/-LMW bone \nmarrow derived macrophages, which demonstrates reduced tumor growth although in a T cell-\nindependent manner. Finally, irradiation was found to promote increased levels of FGF2, and \ntreatment with anti-FGF2 led to better response to irradiation, which correlated with an increased \niNOS:CD206 ratio in macrophages. Based on these findings, the authors conclude that FGF2 \nregulates tumor immunity and growth through altering macrophage polarization. While the link \nbetween FGFs and immune modulation within the tumor microenvironment is not necessarily \nnovel, the findings that macrophage-derived FGF2 impacts tumor growth and that anti-FGF2 \nenhances efficacy of irradiation are of interest. While these studies demonstrate a role for host-\nderived FGF2LMW in promoting tumor growth, and effectively demonstrate efficacy of anti-FGF2 in \nthe context of irradiation, some concerns are noted as described below. \n \nThe authors should provide information in the Materials and Methods regarding the specific \nsubstrains of C57BL/6 mice that were used for these studies. As written, it appears that the \nFGF2LMW-/- mice were purchased from Jackson Laboratories on the C57BL/6J background and \nmaintained in-house, and the control mice were purchased from Charles River, which provides the \nC57BL/6N substrain. If this is correct, this should be noted in the methods section. Furthermore, in \nlight of recent observations that differences in substrain can impact immune phenotypes (for \nexample, PMID 27210752), if different substrains were indeed used in these studies, this should \nbe discussed as a potential caveat particularly of the immune phenotype analysis. Ideally, key \nexperiments involving immune profiling, including the macrophage polarization studies, should be \nrepeated using mice of the same substrain. \n \nFigure 3: While basal levels of a number of the factors shown are significantly different than \ncontrol, the magnitude of induction seems to be similar. Does restoration of FGF2 to the \nFGF2LMW-/- cells impact basal expression levels of these factors? \n \nFigure 5C: The conclusion that the majority of FGF2-producing cells appear to be TAMs is not \nsupported by the images provided as there are substantial areas of green FGF2 staining that do \nnot appear to be associated with F4/80 staining. Quantification of dual-staining or flow cytometry-\nbased quantification should be provided to support these conclusions. \n \nThe authors conclude several times (in the title, lines 215-216 and 352) that “macrophage \npolarization by FGF2LMW mediates tumor growth”. However, this conclusion seems premature \ngiven the studies provided. There are several other possible explanations for the phenotypes that \nhave not been ruled out, such as reduced tumor growth due to loss of macrophage-derived FGF2-\nmediated effects on tumor cells, or other non-immune host cells, which could ultimately impact \nboth tumor growth rates and the immune environment. Thus, additional potential explanations for \nthe phenotypes observed should be thoroughly considered. \n \nThe authors demonstrate that expression levels of FGFR1 and FGFR2 are higher on TAMs \ncompared with other immune cell populations. Does FGF2 directly activate FGFR signaling in TAMs? \n \nIn Figure 1, the authors demonstrate that the reduction in tumor growth observed in the \n\n\fFGF2LMW-/- mice is T cell-dependent. However, in Figure 4, T cells are found to be indispensible \nfor the reduction in tumor growth induced by the FGF2LMW-/- macrophages. Potential reasons for \nthese differences (such as the possibility that non-macrophage FGF2LMW-expressing host cells are \nimpacting the T cell population in the FGF2LMW-/- mice) should be discussed. \n \nBecause the flow plots are presented in the context of FSC or SSC, the authors should provide the \ngating schemes used to identify the different immune cell populations (as Supplementary data). \n \nPlease provide the number of mice used for each in vivo experiment. \n \nFigure 1: Figures J and K are switched when referred to in the text. \n \nFigure 2: Please clarify whether this antibody recognizes only the secreted form of FGF2, or \nwhether it also recognizes the intracellular form of FGF2, and whether this impacts interpretation \nof the data. \n \nFigure 3A: The shifts are difficult to see in the panels provided and may be more easily interpreted \nas histograms. Also, please provide the iNOS:CD206 ratios along with statistical analysis. \n \nFigure 4E: The authors conclude that there is a reduction in MDSCs in tumors with FGF2LMW-/- \nmacrophages, although this does not appear to be supported by the data presented in the graph. \n \nFigure 6: For clarity, please provide the cell number count for each plot. \n \nAlthough a general statement regarding statistical analysis is included in the materials and \nmethods section, it is not clear which test was used for each graph, which could be provided in the \nFigure Legends. \n \n \n \nReviewer #3: \nRemarks to the Author: \nBuzzelli et al present experimental evidence that suggests a role of FGF2 as a key modulator of \nmacrophage polarization in tumors, as part of the \"wound response\" associated with localized \nradio-therapy. They identify FGF2 as a potential target of translational interest, since antibodies \nagainst FGF2 could remove one of radio-therapy immune-suppressive effects, while exploiting \nother pro-immunogenic ones. \n \nSeveral concerns: \n \n1. The mechanism of radio-therapy induction of FGF2 is not clear. In vitro experiments are needed \nto specifically define the nature of signaling after irradiation, whether it is different in different cells \ntypes and whether it is irradiation-does dependent. \n2. Is FGF2 induction part of DDR? A heat map for FGF2 null versus parental lines could help \nunderstanding the pathway. It could also provide important hints regarding the component of \ncross talk with innate immunity \n3. The previous work of the authors in reference 25 shows an effect of FGF2 on angiogenesis, \nindependent of irradiation. Understanding the mechanism of FGF2 induction by irradiation is key, \nsince it may also have clinical impact in cancer metastasis. \n \n4. A main concern is the use of only one radiotherapy regimen (9GyX2), without clear justification \nfor this selection. Particularly in the in vivo experiments a different dose could have overcome the \nprogression of tumors and/or the type of TAMs infiltration. Since no dose response curves were \nprovided in the model used, one can not exclude that re-population because of inadequate dose \nmay have also recruited M2 and/or induced polarization of TAMs. The choice of this radio-therapy \n\n\fdose and number of fractions needs a much more explicit justification, since others have shown \nvery different results for instance when low doses are used. (Klug et al Cancer Cell. 2013 Nov \n11;24(5):589-602) \n5. Figure 5 Shows modest difference in HT29 model and the delay in MC38 is only transient. From \nthe images it is hard to conclude a sustained superior effect by the combination \n6. References on irradiation and macrophages are incomplete (missing key papers of Deng L et al, \nXu J et al etc.). \n \n \n \nReviewer #4: \nRemarks to the Author: \nIn this manuscript, Buzzelli et al. show slower growth of primary tumors and decreased liver \nmetastasis in FGF2lmw deleted mice. Mechanistic studies showed that the deletion induces T cell \nrecruitment into the tumors and that this is critical for the decreased tumor growth. Although the \nauthors were not able to detect FGFR1 or 2 in the T cells, they found that the majority of tumor-\nassociated macrophages expressed FGFR1 and 2. Both FGF2lmw deletion and anti-FGF2 antibody \ninduced TAM polarization towards a more inflammatory phenotype. Decreased tumor xenograft \ngrowth was seen in nude mice when anti-FGF2 blocking antibody was used in combination with \nirradiation. The manuscript reports many interesting, but somewhat dissociated results that raise \nseveral questions. \n \nIn the previous paper by the same authors, metastasis-associated neutrophils expressed \nsubstantially more fibroblast growth factor 2 (FGF2) in comparison to naïve neutrophils, indicating \nneutrophil polarization by the tumor microenvironment. Administration of FGF2 neutralizing \nantibody to mice bearing experimental liver metastases phenocopied neutrophil depletion by \nreducing liver metastatic colony growth, vascular density, and branching (Hepatology \n2017;65:1920-1935). \n \nIt has also been shown that FGFR1 inhibition by gene silencing in tumor cells or administration of a \ncompound (SSR128129E at nanomolar concentrations) inhibiting FGF-induced signaling in tumor \nxenografts increases the efficiency of radiotherapy (pubmed/26896280). \n \nThus, an alternative possibility to explain the authors’ results is that FGF2lmw depletion directly \ncauses initial TC death in the tumors in vivo, and that the resulting danger signals invite \ninflammatory/immune cells to the tumors. This possibility is consistent with the “T-cell \nindependent alteration” that “resulted in less rapid tumor growth in FGF2lmw-/- mice than in WT \nmice”. Although the authors show that anti-FGF2 antibody treatment after irradiation did not alter \nclonogenic survival of SW1222 and HT29 cells, it would be essential to also analyze the effect of \nFGF2 silencing. \n \nIn the manuscript, a limited analysis is reported on the role of TC produced FGF, and the \nexpression of FGF receptors. A comparison of the amounts of FGF2 produced by TCs and \nmacrophages should be shown. A more careful mapping of the autocrine pathway in the TCs is \nalso needed. It would be essential to compare the survival of the control, FGF2 and FGFR1 silenced \nas well as FGF2 antibody treated tumor cell cultures after irradiation. \n \nOn pages 5 and 6 the authors show that T-cells are required for tumor regression in the FGF2lmw-\n/- mice. However, later, in Figure 5 F-H and Supplementary Figure 4, they show that T-cells are \nnot required for the effects of the anti-FGF2 antibody on human SW1222 and HT29 tumor \nxenografts in immunodeficient mice. The authors should analyze the mouse tumor cell lines in \nusing irradiation with and without the anti-FGF2 antibody in an immunocompromised background. \nAlternatively, they could use another inhibitor of the FGF2-FGFR pathway, such as SSR128129E, a \nFGFR1 blocking antibody, soluble receptor or FGFR1 gene silencing in the tumor cells. In addition, \nthe authors should discuss why anti-FGF2 as a monotherapy does not decrease xenograft growth \n\n\f(Fig 5 F-H), although FGF2lmw deletion decreases tumor growth in the isogenic mice (Fig. 1). \n \nAs the authors indicate in the introduction, FGF2lmw is a highly potent angiogenic molecule. \nHowever, here the authors don`t provide any tumor vessel analysis. This should be analyzed to \nexclude the possible role of decreased vascular density to the decreased tumor growth. If anti-\nFGF2 antibody treatment decreases vascular density as shown previously in the case of liver \nmetastasis, it could explain why anti-FGF2 antibody treatment decreases SW1222 and HT29 tumor \ngrowth in immunodeficient mice. \n \nSpecific comments: \n \nAnother question concerns the specificity of the effects reported here. The authors should perhaps \nmention that the other 21 members of FGF family are not expressed or of concern here. Similarly, \nthey should describe which of the four receptors are present in the tumor cells they use. \n \nWhen reporting immune infiltration to the metastases/tumors, the authors should also report if \nthis concerned normal tissues or circulating immune cells. \n \nThe authors should re-isolate the escape clones that resumed growth and test if they regrow in \ncorresponding experiments in FGF2lmw-/- mice. \n \nThe LLC growth curves should be added to Supplementary Figure 1. \n \nDid TAMs express FGFR3 or FGFR4? Was phosphorylation of FGFR1 correlated with FGF2lmw \nexpression or irradiation in TCs or TAMs? \n \nThe anti-FGF2 antibody was used for the MC38 clonogenic assay in Supplementary Figure 5. \nHowever, only human tumor cells were used in the in vivo experiments. Why were the mouse \ntumor cells not used in the in vivo tumor experiments? \n \nAs the authors explain, FGF2 expression has been reported to be induced in irradiated human \ntumors. Thus, Figs 5A and B are of little value. In Fig. 5E, much of the macrophage and FGF2 \nstaining appears to be eradicated by the irradiation. The authors should show the specificity of \nanti-FGF2 by staining macrophages of the corresponding tumors grown in FGF2lmw-/- mice. \n \nMinor comments \n \n• Fig. 1 J and K are mixed on page 6? \n• How was liver tumor burden measured? \n• How was tumor volume calculated? The authors should study M1/M2 ratio in tumors beyond the \n(very small) volume of 500mm3. \n• Fig. 1 B: Please show the invasion front in high magnification, with the identification of the tumor \ncells (by IHC). \n• In all experiments, the number of mice should be mentioned in each group. The individual tumor \ngrowth curves are difficult to interpret, for example in Fig. 5F. \n \n\n\f \n\n", "role": "assistant" } ]	2,020
FGF2 alters macrophage polarization, tumour immunity and growth and can be targeted	3 Reviewers' Comments: Reviewer #1: Remarks to the Author: The authors have addressed reviewer concerns.	3 Reviewers' Comments: Reviewer #1: Remarks to the Author: The authors have addressed reviewer concerns.	TPR	NC_BS_CAN_009_RR3.txt	{ "criticism": 0, "example": 0, "importance_and_relevance": 0, "materials_and_methods": 0, "praise": 0, "presentation_and_reporting": 0, "results_and_discussion": 0, "suggestion_and_solution": 0, "total": 1 }	{ "criticism": 0, "example": 0, "importance_and_relevance": 0, "materials_and_methods": 0, "praise": 0, "presentation_and_reporting": 0, "results_and_discussion": 0, "suggestion_and_solution": 0 }	0	[ { "content": "Please review this paper.", "role": "user" }, { "content": "\n \n \n \n \n \n \n\n \n\n3 \n\n\fReviewers' Comments: \n \nReviewer #1: \nRemarks to the Author: \nThe authors have addressed reviewer concerns. \n\n\f\n", "role": "assistant" } ]	2,020
FGF2 alters macrophage polarization, tumour immunity and growth and can be targeted	Reviewers' Comments: Reviewer #1: Remarks to the Author: Overall, the authors have incorporated many changes and the manuscript has been improved. One concern is regarding the experiments provided in response to the suggestion to demonstrate that FGF2 directly activates FGFR signaling in TAMs. The authors have provided data in Supplemental Figure 3 in which they treated macrophages with recombinant FGF2 and examined expression of inflammatory genes. They also treated cells with tumor cell conditioned medium followed by an FGF2 antibody. It is not clear whether appropriate controls were included. The source of recombinant FGF2 was not found. BMDMs are highly sensitive to even low levels of endotoxin, commonly found in most recombinant protein preparations, which can impact expression of inflammatory genes. In addition, it does not appear that an isotype control antibody was used in the antibody blocking experiments. Therefore, these data are difficult to interpret and do not provide convincing evidence for direct regulation of macrophage polarization by FGF2 via FGFR signaling. There is also concern regarding the experiment provided in Supplemental Figure 7C. These experiments are ideally performed following serum starvation for short timepoints (5-30 minutes). Given the transient nature of FGF2-induced signaling it is unclear that analysis of signaling pathways 48 hours after treatment (in an undefined serum setting) provides meaningful/interpretable information. Furthermore, an appropriate positive control should also be included (such as a cell known known to respond to FGF2 stimulation). While it is agreed that FGFR phosphorylation can be difficult to detect, phosphorylation of FRS-2 is generally considered to be a relevant surrogate and is more specific than Akt and ERK. Reviewer #3: Remarks to the Author: The Authors have provided adequate responses and revisions after the first round of reviews Reviewer #5: Remarks to the Author: This is an interesting and clearly written paper. this reviewer has been asked to comment on whether the issues raised by reviewer 4 have been adequately addressed. In the most part, the comments from reviewer 4 have been addressed. However for point 3 the authors couldnt get the antibodies to work, and the autocrine pathway requested was not possible to provide. Point 5 was responded to by 'beyond the scope of this study', which is a little disappointing but probably does change the message of the paper Reviewer 4 point 7 - for angiogenic responses an n=3 is not sufficient for statistical analysis and this n-number should be increased The point abput regrowing escape cells in FGF2lmw-/- was not done. Again this is a shame, because although it may not change the paper it would have been of interest.	Reviewers' Comments: Reviewer #1: Remarks to the Author: Overall, the authors have incorporated many changes and the manuscript has been improved. One concern is regarding the experiments provided in response to the suggestion to demonstrate that FGF2 directly activates FGFR signaling in TAMs. The authors have provided data in Supplemental Figure 3 in which they treated macrophages with recombinant FGF2 and examined expression of inflammatory genes. They also treated cells with tumor cell conditioned medium followed by an FGF2 antibody. It is not clear whether appropriate controls were included. The source of recombinant FGF2 was not found. BMDMs are highly sensitive to even low levels of endotoxin, commonly found in most recombinant protein preparations, which can impact expression of inflammatory genes. In addition, it does not appear that an isotype control antibody was used in the antibody blocking experiments. Therefore, these data are difficult to interpret and do not provide convincing evidence for direct regulation of macrophage polarization by FGF2 via FGFR signaling. There is also concern regarding the experiment provided in Supplemental Figure 7C. These experiments are ideally performed following serum starvation for short timepoints (5-30 minutes). Given the transient nature of FGF2-induced signaling it is unclear that analysis of signaling pathways 48 hours after treatment (in an undefined serum setting) provides meaningful/interpretable information. Furthermore, an appropriate positive control should also be included (such as a cell known known to respond to FGF2 stimulation). While it is agreed that FGFR phosphorylation can be difficult to detect, phosphorylation of FRS-2 is generally considered to be a relevant surrogate and is more specific than Akt and ERK. Reviewer #3: Remarks to the Author: The Authors have provided adequate responses and revisions after the first round of reviews Reviewer #5: Remarks to the Author: This is an interesting and clearly written paper. this reviewer has been asked to comment on whether the issues raised by reviewer 4 have been adequately addressed. In the most part, the comments from reviewer 4 have been addressed. However for point 3 the authors couldnt get the antibodies to work, and the autocrine pathway requested was not possible to provide. Point 5 was responded to by 'beyond the scope of this study', which is a little disappointing but probably does change the message of the paper Reviewer 4 point 7 - for angiogenic responses an n=3 is not sufficient for statistical analysis and this n-number should be increased The point abput regrowing escape cells in FGF2lmw-/- was not done. Again this is a shame, because although it may not change the paper it would have been of interest.	TPR	NC_BS_CAN_009_RR2.txt	{ "criticism": 7, "example": 2, "importance_and_relevance": 2, "materials_and_methods": 15, "praise": 2, "presentation_and_reporting": 1, "results_and_discussion": 0, "suggestion_and_solution": 2, "total": 20 }	{ "criticism": 0.35, "example": 0.1, "importance_and_relevance": 0.1, "materials_and_methods": 0.75, "praise": 0.1, "presentation_and_reporting": 0.05, "results_and_discussion": 0, "suggestion_and_solution": 0.1 }	1.55	[ { "content": "Please review this paper.", "role": "user" }, { "content": "\n\n\fReviewers' Comments: \n \nReviewer #1: \nRemarks to the Author: \nOverall, the authors have incorporated many changes and the manuscript has been improved. One \nconcern is regarding the experiments provided in response to the suggestion to demonstrate that \nFGF2 directly activates FGFR signaling in TAMs. The authors have provided data in Supplemental \nFigure 3 in which they treated macrophages with recombinant FGF2 and examined expression of \ninflammatory genes. They also treated cells with tumor cell conditioned medium followed by an \nFGF2 antibody. It is not clear whether appropriate controls were included. The source of \nrecombinant FGF2 was not found. BMDMs are highly sensitive to even low levels of endotoxin, \ncommonly found in most recombinant protein preparations, which can impact expression of \ninflammatory genes. In addition, it does not appear that an isotype control antibody was used in \nthe antibody blocking experiments. Therefore, these data are difficult to interpret and do not \nprovide convincing evidence for direct regulation of macrophage polarization by FGF2 via FGFR \nsignaling. \n \nThere is also concern regarding the experiment provided in Supplemental Figure 7C. These \nexperiments are ideally performed following serum starvation for short timepoints (5-30 minutes). \nGiven the transient nature of FGF2-induced signaling it is unclear that analysis of signaling \npathways 48 hours after treatment (in an undefined serum setting) provides \nmeaningful/interpretable information. Furthermore, an appropriate positive control should also be \nincluded (such as a cell known known to respond to FGF2 stimulation). While it is agreed that \nFGFR phosphorylation can be difficult to detect, phosphorylation of FRS-2 is generally considered \nto be a relevant surrogate and is more specific than Akt and ERK. \n \n \n \nReviewer #3: \nRemarks to the Author: \nThe Authors have provided adequate responses and revisions after the first round of reviews \n \n \n \nReviewer #5: \nRemarks to the Author: \nThis is an interesting and clearly written paper. this reviewer has been asked to comment on \nwhether the issues raised by reviewer 4 have been adequately addressed. \n \nIn the most part, the comments from reviewer 4 have been addressed. However for point 3 the \nauthors couldnt get the antibodies to work, and the autocrine pathway requested was not possible \nto provide. \n \nPoint 5 was responded to by 'beyond the scope of this study', which is a little disappointing but \nprobably does change the message of the paper \n \nReviewer 4 point 7 - for angiogenic responses an n=3 is not sufficient for statistical analysis and \nthis n-number should be increased \n \nThe point abput regrowing escape cells in FGF2lmw-/- was not done. Again this is a shame, \nbecause although it may not change the paper it would have been of interest. \n \n \n\n\f", "role": "assistant" } ]	2,020
Reversal of pre-existing NGFR-driven tumor and immune therapy resistance	Reviewers' comments: Reviewer #1 (Remarks to the Author); expert on mouse models for melanoma and signaling: This paper by Boshuizen and colleagues demonstrates a convincing correlation of high CD271/NGFR expression in melanoma cells with (MART-1) T cell resistance and resistance to other melanoma drugs. High CD271 expression has been associated with resistance to BRAF/MEK inhibition before. Likewise, a correlation of high CD271 expression with reduced T cell infiltration has previously been shown. Of note, the point crucial for the present study – inhibition of the NGFRhigh state by blocking specifically CD271 (or e.g. by engineering CD271-specific T-cells) – has not been addressed and it remains open why CD271/NGFRhigh cells are resistant to T cells. Specific points: 1) For the key functional experiments, the authors use HSP90i, which is not at all a specific NGFR inhibitor. Downregulation of NGFR upon HSP90i might be merely correlative. Specific NGFR inhibition, at (or during acquisition of) the TR state in vitro and in vivo, has to be performed to assess whether inhibition of the CD271high state indeed prevents relapse, i.e. whether resistance is indeed NFGR driven, as claimed in the title. 2) In Figure 1, the authors present a nice method for how to generate TR melanoma cells. However, the TR state is ill defined and it is unclear why the authors picked CD271/NGFR for their further studies (as it stands, the authors selected NGFR mainly based on literature; for this, the work shown in Fig1a-e would not have been necessary). What is the molecular signature of the TR state obtained in the experiments of Figure 1 and how does it compare to the signatures shown in Figure 4? 3) As shown by others before, resistance formation involves various cellular states rather than just one particular, homogeneous cell state. TR state analysis should optimally be performed at the single cell level. Rather than showing overall NGFR expression (Fig. 1 f), the numbers of CD271/NGFRhigh cells as compared to CD271/NGFRlow cells have to be determined upon T cell resistance. Likewise, how many cells within the clones shown in Figure 3c are CD271 positive? 4) The choice of the cell lines used for a particular set of experiments is not consistent (e.g. Figure 1c vs 1h; different cell lines used for the different panels of Figure 2). Each key experiment should be repeated with at least two cell lines. 5) Why are CD271/NGFRhigh cells resistant to T cells? The mechanisms underlying resistance formation have to be addressed. Do CD271/NGFRlow cells play a role in this as well (e.g. by interaction with CD271/NGFRhigh cells)? 6) In Figure 6b, does shNGFR already exhibit an effect in non-TR cells? This control experiment needs to be included. 7) Figure 5: what about NGFR expression in biopsies from resistant patients? 8) Minor point: in general, the labelling of figure axes could be improved; as an example, in Figure 3b, it's not clear immediately that NGFR MFI is measured, etc. Reviewer #2 (Remarks to the Author); expert on PDX models, melanoma and immunotherapy: This is a nice manuscript with some interesting findings. I have only three major concerns: 1. The immunotherapy experiments are not conducted with autologous tumor and TILs. Although HLA matched it is therefore risky that some of the effects are not general but only pertaining to allogeneic responses or regulation of antigen processing. The authors should consider repeating some of the data using paired autologous TILs and tumor cells. 2. The extended data 5 suggest that there might be an off target problem with the shRNA targeting NGFR. Please repeat the experiment with CRISPR/Cas9 deletion of NGFR. 3. Although the data are suggestive of the NGFR is necessary, which can be strengthened by CRISPR, is NGFR sufficient to drive resistance? If NGFR is transgenically overexpressed, will this result in resistance?	Reviewers' comments: Reviewer #1 (Remarks to the Author); expert on mouse models for melanoma and signaling: This paper by Boshuizen and colleagues demonstrates a convincing correlation of high CD271/NGFR expression in melanoma cells with (MART-1) T cell resistance and resistance to other melanoma drugs. High CD271 expression has been associated with resistance to BRAF/MEK inhibition before. Likewise, a correlation of high CD271 expression with reduced T cell infiltration has previously been shown. Of note, the point crucial for the present study – inhibition of the NGFRhigh state by blocking specifically CD271 (or e.g. by engineering CD271-specific T-cells) – has not been addressed and it remains open why CD271/NGFRhigh cells are resistant to T cells. Specific points: 1) For the key functional experiments, the authors use HSP90i, which is not at all a specific NGFR inhibitor. Downregulation of NGFR upon HSP90i might be merely correlative. Specific NGFR inhibition, at (or during acquisition of) the TR state in vitro and in vivo, has to be performed to assess whether inhibition of the CD271high state indeed prevents relapse, i.e. whether resistance is indeed NFGR driven, as claimed in the title. 2) In Figure 1, the authors present a nice method for how to generate TR melanoma cells. However, the TR state is ill defined and it is unclear why the authors picked CD271/NGFR for their further studies (as it stands, the authors selected NGFR mainly based on literature; for this, the work shown in Fig1a-e would not have been necessary). What is the molecular signature of the TR state obtained in the experiments of Figure 1 and how does it compare to the signatures shown in Figure 4? 3) As shown by others before, resistance formation involves various cellular states rather than just one particular, homogeneous cell state. TR state analysis should optimally be performed at the single cell level. Rather than showing overall NGFR expression (Fig. 1 f), the numbers of CD271/NGFRhigh cells as compared to CD271/NGFRlow cells have to be determined upon T cell resistance. Likewise, how many cells within the clones shown in Figure 3c are CD271 positive? 4) The choice of the cell lines used for a particular set of experiments is not consistent (e.g. Figure 1c vs 1h; different cell lines used for the different panels of Figure 2). Each key experiment should be repeated with at least two cell lines. 5) Why are CD271/NGFRhigh cells resistant to T cells? The mechanisms underlying resistance formation have to be addressed. Do CD271/NGFRlow cells play a role in this as well (e.g. by interaction with CD271/NGFRhigh cells)? 6) In Figure 6b, does shNGFR already exhibit an effect in non-TR cells? This control experiment needs to be included. 7) Figure 5: what about NGFR expression in biopsies from resistant patients? 8) Minor point: in general, the labelling of figure axes could be improved; as an example, in Figure 3b, it's not clear immediately that NGFR MFI is measured, etc. Reviewer #2 (Remarks to the Author); expert on PDX models, melanoma and immunotherapy: This is a nice manuscript with some interesting findings. I have only three major concerns: 1. The immunotherapy experiments are not conducted with autologous tumor and TILs. Although HLA matched it is therefore risky that some of the effects are not general but only pertaining to allogeneic responses or regulation of antigen processing. The authors should consider repeating some of the data using paired autologous TILs and tumor cells. 2. The extended data 5 suggest that there might be an off target problem with the shRNA targeting NGFR. Please repeat the experiment with CRISPR/Cas9 deletion of NGFR. 3. Although the data are suggestive of the NGFR is necessary, which can be strengthened by CRISPR, is NGFR sufficient to drive resistance? If NGFR is transgenically overexpressed, will this result in resistance?	TPR	NC_BS_CAN_010_RR1.txt	{ "criticism": 4, "example": 8, "importance_and_relevance": 1, "materials_and_methods": 24, "praise": 3, "presentation_and_reporting": 4, "results_and_discussion": 3, "suggestion_and_solution": 9, "total": 39 }	{ "criticism": 0.10256410256410256, "example": 0.20512820512820512, "importance_and_relevance": 0.02564102564102564, "materials_and_methods": 0.6153846153846154, "praise": 0.07692307692307693, "presentation_and_reporting": 0.10256410256410256, "results_and_discussion": 0.07692307692307693, "suggestion_and_solution": 0.23076923076923078 }	1.435897	[ { "content": "Please review this paper.", "role": "user" }, { "content": "Reviewers' comments: \n\nReviewer #1 (Remarks to the Author); expert on mouse models for melanoma and signaling: \n\nThis paper by Boshuizen and colleagues demonstrates a convincing correlation of high \nCD271/NGFR expression in melanoma cells with (MART-1) T cell resistance and resistance to other \nmelanoma drugs. High CD271 expression has been associated with resistance to BRAF/MEK \ninhibition before. Likewise, a correlation of high CD271 expression with reduced T cell infiltration \nhas previously been shown. Of note, the point crucial for the present study – inhibition of the \nNGFRhigh state by blocking specifically CD271 (or e.g. by engineering CD271-specific T-cells) – \nhas not been addressed and it remains open why CD271/NGFRhigh cells are resistant to T cells. \n\nSpecific points: \n\n1) For the key functional experiments, the authors use HSP90i, which is not at all a specific NGFR \ninhibitor. Downregulation of NGFR upon HSP90i might be merely correlative. Specific NGFR \ninhibition, at (or during acquisition of) the TR state in vitro and in vivo, has to be performed to \nassess whether inhibition of the CD271high state indeed prevents relapse, i.e. whether resistance \nis indeed NFGR driven, as claimed in the title. \n\n2) In Figure 1, the authors present a nice method for how to generate TR melanoma cells. \nHowever, the TR state is ill defined and it is unclear why the authors picked CD271/NGFR for their \nfurther studies (as it stands, the authors selected NGFR mainly based on literature; for this, the \nwork shown in Fig1a-e would not have been necessary). What is the molecular signature of the TR \nstate obtained in the experiments of Figure 1 and how does it compare to the signatures shown in \nFigure 4? \n\n3) As shown by others before, resistance formation involves various cellular states rather than just \none particular, homogeneous cell state. TR state analysis should optimally be performed at the \nsingle cell level. Rather than showing overall NGFR expression (Fig. 1 f), the numbers of \nCD271/NGFRhigh cells as compared to CD271/NGFRlow cells have to be determined upon T cell \nresistance. Likewise, how many cells within the clones shown in Figure 3c are CD271 positive? \n\n4) The choice of the cell lines used for a particular set of experiments is not consistent (e.g. Figure \n1c vs 1h; different cell lines used for the different panels of Figure 2). Each key experiment should \nbe repeated with at least two cell lines. \n\n5) Why are CD271/NGFRhigh cells resistant to T cells? The mechanisms underlying resistance \nformation have to be addressed. \nDo CD271/NGFRlow cells play a role in this as well (e.g. by interaction with CD271/NGFRhigh \ncells)? \n\n6) In Figure 6b, does shNGFR already exhibit an effect in non-TR cells? This control experiment \nneeds to be included. \n\n7) Figure 5: what about NGFR expression in biopsies from resistant patients? \n\n8) Minor point: in general, the labelling of figure axes could be improved; as an example, in Figure \n3b, it's not clear immediately that NGFR MFI is measured, etc. \n\nReviewer #2 (Remarks to the Author); expert on PDX models, melanoma and immunotherapy: \n\nThis is a nice manuscript with some interesting findings. I have only three major concerns: \n\n\f1. The immunotherapy experiments are not conducted with autologous tumor and TILs. Although \nHLA matched it is therefore risky that some of the effects are not general but only pertaining to \nallogeneic responses or regulation of antigen processing. The authors should consider repeating \nsome of the data using paired autologous TILs and tumor cells. \n2. The extended data 5 suggest that there might be an off target problem with the shRNA \ntargeting NGFR. Please repeat the experiment with CRISPR/Cas9 deletion of NGFR. \n3. Although the data are suggestive of the NGFR is necessary, which can be strengthened by \nCRISPR, is NGFR sufficient to drive resistance? If NGFR is transgenically overexpressed, will this \nresult in resistance? \n\n\f\f\n", "role": "assistant" } ]	2,020
Reversal of pre-existing NGFR-driven tumor and immune therapy resistance	REVIEWER COMMENTS Reviewer #1 (Remarks to the Author): In the revised version of their manuscript, Boshuizen and colleagues have incorporated an extensive set of additional experiments, including CRISPR/Cas9-mediated NGFR KOs and ectopic NGFR expression, which all in all considerably strengthened their study. Together with some new mechanistic assays, the experiments nicely support their finding that NGFR confers T cell resistance to melanoma cells in vitro. It would have even further increased the impact of the paper if the authors had addressed whether NGFR inhibition can prevent melanoma resistance formation in vivo or, vice versa, whether NGFR overexpression in parental melanoma cells is sufficient to drive resistance in vivo. In any case, I feel that the study is now acceptable for publication. However, I suggest to clarify, for instance in the Discussion, line 376, that “ectopic expression of NGFR conferred T cell resistance to parental cells IN VITRO” or: “…to parental cells, at least in cell culture”. Furthermore, I feel that the title is not very clear; the authors might want to consider rephrasing it. Reviewer #2 (Remarks to the Author): The authors did an excellent job in answering my comments. /Jonas Nilsson Reviewer #3 (Remarks to the Author): While I congratulate the authors for the present manuscript in which they use very complete in the array of techniques and analyses, the part of the study regarding the RNAseq experiment and analysis is quite lacking. While I understand that this manuscript is late in the review process, there are several issues regarding the RNAseq part that need to be addressed. For reasons there is this common line of thinking that in all cases PCR duplicates need not be removed for RNAseq data, which is not true at all. It all depends on the scenario: 1) Amount of starting RNA: if there is enough starting RNA in your samples (~200ng or more), removing PCR duplicates does not have a big effect, and can even be detrimental depending on the read length. But if the starting material is scarce, not removing the PCR duplicates will incur into artificially inflated gene expression measures due to the low diversity of the starting library depending on the target number of reads sequenced. 2) Length of reads and Paired End (PE) vs Single End (SE): in SE, the ability to discern PCR duplicates from reads that just look like a PCR duplicate but is not is seriously hampered. Similarly, the shorter the reads, the more difficult to discern true PCR duplicates is. The authors have performed SE 65bp read sequencing which is one read and very short (common practice for RNAseq is 75x2 and optimal is 150x2). Since the authors are in the low end (and only SE) they need to have really good starting material quality and amount to justify not removing PCR duplicates. 3) Related to the last point, the length of read al so critically affects multimapping rate of reads, which when high, then also artificially inflate the expression levels of certain genes with high homology between them. Hence the authors need to provide a supplementary table with the RNA extraction metrics to discern whether it is ok to leave the PCR duplicates in or they need to be removed. This needs to be state din the M&M section as a sentence, specifying the average amount of RNA extracted from samples and also specifying whether PCR duplicates need to be removed or not. Furthermore, a supplementary table with STAR alignment metrics need to be provided. STAR outputs all these metrics as a normal part of its output. Critical metrics to show would be total raw reads, mapped reads, Total splices and break down stats for those splices, mismatch, deletion and insertion rates, multimaps and multimaps percentages as well as percentage of high multimap reads, percentage of reads that are too short, and finally total reads which are PCR duplicates and the percentage of those (whether they are removed or not, as stated above). Regarding the multimapping reads they also need to provide explanation as of they were handled (i.e. what threshold of percentage of multimaps was used to discard reads as “high multimap reads”) Lastly the authors mention that they have used the “default” settings of STAR. The default is single pass mode, which is far less accurate than the two pass mode of STAR. This needs to be specified in the M&M section, and if it is indeed the case that they have done single pass alignment with STAR, they need to provide and explanation as of why they have instead of using the two pass mode. All this information is critical to the reproducibility of the study and the evaluation of the quality of the analysis done in the RNAseq part of it.	REVIEWER COMMENTS Reviewer #1 (Remarks to the Author): In the revised version of their manuscript, Boshuizen and colleagues have incorporated an extensive set of additional experiments, including CRISPR/Cas9-mediated NGFR KOs and ectopic NGFR expression, which all in all considerably strengthened their study. Together with some new mechanistic assays, the experiments nicely support their finding that NGFR confers T cell resistance to melanoma cells in vitro. It would have even further increased the impact of the paper if the authors had addressed whether NGFR inhibition can prevent melanoma resistance formation in vivo or, vice versa, whether NGFR overexpression in parental melanoma cells is sufficient to drive resistance in vivo. In any case, I feel that the study is now acceptable for publication. However, I suggest to clarify, for instance in the Discussion, line 376, that “ectopic expression of NGFR conferred T cell resistance to parental cells IN VITRO” or: “…to parental cells, at least in cell culture”. Furthermore, I feel that the title is not very clear; the authors might want to consider rephrasing it. Reviewer #2 (Remarks to the Author): The authors did an excellent job in answering my comments. /Jonas Nilsson Reviewer #3 (Remarks to the Author): While I congratulate the authors for the present manuscript in which they use very complete in the array of techniques and analyses, the part of the study regarding the RNAseq experiment and analysis is quite lacking. While I understand that this manuscript is late in the review process, there are several issues regarding the RNAseq part that need to be addressed. For reasons there is this common line of thinking that in all cases PCR duplicates need not be removed for RNAseq data, which is not true at all. It all depends on the scenario: 1) Amount of starting RNA: if there is enough starting RNA in your samples (~200ng or more), removing PCR duplicates does not have a big effect, and can even be detrimental depending on the read length. But if the starting material is scarce, not removing the PCR duplicates will incur into artificially inflated gene expression measures due to the low diversity of the starting library depending on the target number of reads sequenced. 2) Length of reads and Paired End (PE) vs Single End (SE): in SE, the ability to discern PCR duplicates from reads that just look like a PCR duplicate but is not is seriously hampered. Similarly, the shorter the reads, the more difficult to discern true PCR duplicates is. The authors have performed SE 65bp read sequencing which is one read and very short (common practice for RNAseq is 75x2 and optimal is 150x2). Since the authors are in the low end (and only SE) they need to have really good starting material quality and amount to justify not removing PCR duplicates. 3) Related to the last point, the length of read al so critically affects multimapping rate of reads, which when high, then also artificially inflate the expression levels of certain genes with high homology between them. Hence the authors need to provide a supplementary table with the RNA extraction metrics to discern whether it is ok to leave the PCR duplicates in or they need to be removed. This needs to be state din the M&M section as a sentence, specifying the average amount of RNA extracted from samples and also specifying whether PCR duplicates need to be removed or not. Furthermore, a supplementary table with STAR alignment metrics need to be provided. STAR outputs all these metrics as a normal part of its output. Critical metrics to show would be total raw reads, mapped reads, Total splices and break down stats for those splices, mismatch, deletion and insertion rates, multimaps and multimaps percentages as well as percentage of high multimap reads, percentage of reads that are too short, and finally total reads which are PCR duplicates and the percentage of those (whether they are removed or not, as stated above). Regarding the multimapping reads they also need to provide explanation as of they were handled (i.e. what threshold of percentage of multimaps was used to discard reads as “high multimap reads”) Lastly the authors mention that they have used the “default” settings of STAR. The default is single pass mode, which is far less accurate than the two pass mode of STAR. This needs to be specified in the M&M section, and if it is indeed the case that they have done single pass alignment with STAR, they need to provide and explanation as of why they have instead of using the two pass mode. All this information is critical to the reproducibility of the study and the evaluation of the quality of the analysis done in the RNAseq part of it.	TPR	NC_BS_CAN_010_RR2.txt	{ "criticism": 5, "example": 1, "importance_and_relevance": 0, "materials_and_methods": 17, "praise": 4, "presentation_and_reporting": 7, "results_and_discussion": 3, "suggestion_and_solution": 12, "total": 27 }	{ "criticism": 0.18518518518518517, "example": 0.037037037037037035, "importance_and_relevance": 0, "materials_and_methods": 0.6296296296296297, "praise": 0.14814814814814814, "presentation_and_reporting": 0.25925925925925924, "results_and_discussion": 0.1111111111111111, "suggestion_and_solution": 0.4444444444444444 }	1.814815	[ { "content": "Please review this paper.", "role": "user" }, { "content": "\n \n \n \n\n\fREVIEWER COMMENTS \n\nReviewer #1 (Remarks to the Author): \n\nIn the revised version of their manuscript, Boshuizen and colleagues have incorporated an \nextensive set of additional experiments, including CRISPR/Cas9-mediated NGFR KOs and ectopic \nNGFR expression, which all in all considerably strengthened their study. Together with some new \nmechanistic assays, the experiments nicely support their finding that NGFR confers T cell \nresistance to melanoma cells in vitro. \nIt would have even further increased the impact of the paper if the authors had addressed whether \nNGFR inhibition can prevent melanoma resistance formation in vivo or, vice versa, whether NGFR \noverexpression in parental melanoma cells is sufficient to drive resistance in vivo. In any case, I \nfeel that the study is now acceptable for publication. However, I suggest to clarify, for instance in \nthe Discussion, line 376, that “ectopic expression of NGFR conferred T cell resistance to parental \ncells IN VITRO” or: “…to parental cells, at least in cell culture”. \nFurthermore, I feel that the title is not very clear; the authors might want to consider rephrasing \nit. \n\nReviewer #2 (Remarks to the Author): \n\nThe authors did an excellent job in answering my comments. /Jonas Nilsson \n\nReviewer #3 (Remarks to the Author): \n\nWhile I congratulate the authors for the present manuscript in which they use very complete in the \narray of techniques and analyses, the part of the study regarding the RNAseq experiment and \nanalysis is quite lacking. While I understand that this manuscript is late in the review process, \nthere are several issues regarding the RNAseq part that need to be addressed. \nFor reasons there is this common line of thinking that in all cases PCR duplicates need not be \nremoved for RNAseq data, which is not true at all. It all depends on the scenario: \n1) Amount of starting RNA: if there is enough starting RNA in your samples (~200ng or more), \nremoving PCR duplicates does not have a big effect, and can even be detrimental depending on \nthe read length. But if the starting material is scarce, not removing the PCR duplicates will incur \ninto artificially inflated gene expression measures due to the low diversity of the starting library \ndepending on the target number of reads sequenced. \n2) Length of reads and Paired End (PE) vs Single End (SE): in SE, the ability to discern PCR \nduplicates from reads that just look like a PCR duplicate but is not is seriously hampered. \nSimilarly, the shorter the reads, the more difficult to discern true PCR duplicates is. The authors \nhave performed SE 65bp read sequencing which is one read and very short (common practice for \nRNAseq is 75x2 and optimal is 150x2). Since the authors are in the low end (and only SE) they \nneed to have really good starting material quality and amount to justify not removing PCR \nduplicates. \n3) Related to the last point, the length of read al so critically affects multimapping rate of reads, \nwhich when high, then also artificially inflate the expression levels of certain genes with high \nhomology between them. \nHence the authors need to provide a supplementary table with the RNA extraction metrics to \ndiscern whether it is ok to leave the PCR duplicates in or they need to be removed. This needs to \nbe state din the M&M section as a sentence, specifying the average amount of RNA extracted from \nsamples and also specifying whether PCR duplicates need to be removed or not. \nFurthermore, a supplementary table with STAR alignment metrics need to be provided. STAR \noutputs all these metrics as a normal part of its output. Critical metrics to show would be total raw \nreads, mapped reads, Total splices and break down stats for those splices, mismatch, deletion and \n\n\finsertion rates, multimaps and multimaps percentages as well as percentage of high multimap \nreads, percentage of reads that are too short, and finally total reads which are PCR duplicates and \nthe percentage of those (whether they are removed or not, as stated above). \nRegarding the multimapping reads they also need to provide explanation as of they were handled \n(i.e. what threshold of percentage of multimaps was used to discard reads as “high multimap \nreads”) \nLastly the authors mention that they have used the “default” settings of STAR. The default is single \npass mode, which is far less accurate than the two pass mode of STAR. This needs to be specified \nin the M&M section, and if it is indeed the case that they have done single pass alignment with \nSTAR, they need to provide and explanation as of why they have instead of using the two pass \nmode. \nAll this information is critical to the reproducibility of the study and the evaluation of the quality of \nthe analysis done in the RNAseq part of it. \n\n\f", "role": "assistant" } ]	2,020
Reversal of pre-existing NGFR-driven tumor and immune therapy resistance	REVIEWERS' COMMENTS: Reviewer #3 (Remarks to the Author): The authors have satisfactorily answered all queries from this reviewer.	REVIEWERS' COMMENTS: Reviewer #3 (Remarks to the Author): The authors have satisfactorily answered all queries from this reviewer.	TPR	NC_BS_CAN_010_RR3.txt	{ "criticism": 0, "example": 0, "importance_and_relevance": 0, "materials_and_methods": 0, "praise": 0, "presentation_and_reporting": 0, "results_and_discussion": 0, "suggestion_and_solution": 0, "total": 1 }	{ "criticism": 0, "example": 0, "importance_and_relevance": 0, "materials_and_methods": 0, "praise": 0, "presentation_and_reporting": 0, "results_and_discussion": 0, "suggestion_and_solution": 0 }	0	[ { "content": "Please review this paper.", "role": "user" }, { "content": " \n \n\n\fREVIEWERS' COMMENTS: \n\nReviewer #3 (Remarks to the Author): \n\nThe authors have satisfactorily answered all queries from this reviewer.\n\n\f\n", "role": "assistant" } ]	2,020
ABL1-dependent OTULIN phosphorylation promotes genotoxic Wnt/β-catenin activation to	Reviewers' Comments: Reviewer #1: Remarks to the Author: The authors have addressed most of the major concerns raised by the reviewers. The paper is now appropriate for publication. Reviewer #2: Remarks to the Author: According to our criticisms, the authors dramatically improved the manuscript. The findings are novel and the experiments are carefully done. Although I am almost satisfied with the authors’ responses, I would like to ask one question again. As already commented in (1) and (4), the authors should show direct results indicating whether or not the state of M1-ubiquitination of β- catenin is affected by Wnt3a stimulation. The authors suggested that LUBAC-promoted linear ubiquitination has independent but collaborative roles of GSK3-dependent phosphorylation of β- catenin in controlling its stability. If Wnt simulation directly affects linear ubiquitination of β- catenin, it would give a strong impact on the Wnt research field.	Reviewers' Comments: Reviewer #1: Remarks to the Author: The authors have addressed most of the major concerns raised by the reviewers. The paper is now appropriate for publication. Reviewer #2: Remarks to the Author: According to our criticisms, the authors dramatically improved the manuscript. The findings are novel and the experiments are carefully done. Although I am almost satisfied with the authors’ responses, I would like to ask one question again. As already commented in (1) and (4), the authors should show direct results indicating whether or not the state of M1-ubiquitination of β- catenin is affected by Wnt3a stimulation. The authors suggested that LUBAC-promoted linear ubiquitination has independent but collaborative roles of GSK3-dependent phosphorylation of β- catenin in controlling its stability. If Wnt simulation directly affects linear ubiquitination of β- catenin, it would give a strong impact on the Wnt research field.	TPR	NC_BS_CAN_011_RR2.txt	{ "criticism": 0, "example": 0, "importance_and_relevance": 0, "materials_and_methods": 4, "praise": 4, "presentation_and_reporting": 1, "results_and_discussion": 2, "suggestion_and_solution": 2, "total": 8 }	{ "criticism": 0, "example": 0, "importance_and_relevance": 0, "materials_and_methods": 0.5, "praise": 0.5, "presentation_and_reporting": 0.125, "results_and_discussion": 0.25, "suggestion_and_solution": 0.25 }	1.625	[ { "content": "Please review this paper.", "role": "user" }, { "content": "\n\n\fReviewers' Comments: \n \nReviewer #1: \nRemarks to the Author: \nThe authors have addressed most of the major concerns raised by the reviewers. The paper is now \nappropriate for publication. \n \n \n \nReviewer #2: \nRemarks to the Author: \nAccording to our criticisms, the authors dramatically improved the manuscript. The findings are \nnovel and the experiments are carefully done. Although I am almost satisfied with the authors’ \nresponses, I would like to ask one question again. As already commented in (1) and (4), the \nauthors should show direct results indicating whether or not the state of M1-ubiquitination of β-\ncatenin is affected by Wnt3a stimulation. The authors suggested that LUBAC-promoted linear \nubiquitination has independent but collaborative roles of GSK3-dependent phosphorylation of β-\ncatenin in controlling its stability. If Wnt simulation directly affects linear ubiquitination of β-\ncatenin, it would give a strong impact on the Wnt research field. \n \n\n\f", "role": "assistant" } ]	2,020
ABL1-dependent OTULIN phosphorylation promotes genotoxic Wnt/β-catenin activation to	Reviewers' Comments: Reviewer #2: Remarks to the Author: The new data supports the important finding that Wnt reduces linear ubiquitination of β-catenin, thereby controlling Wnt signaling. This reviewer has no further question.	Reviewers' Comments: Reviewer #2: Remarks to the Author: The new data supports the important finding that Wnt reduces linear ubiquitination of β-catenin, thereby controlling Wnt signaling. This reviewer has no further question.	TPR	NC_BS_CAN_011_RR3.txt	{ "criticism": 1, "example": 0, "importance_and_relevance": 1, "materials_and_methods": 1, "praise": 1, "presentation_and_reporting": 0, "results_and_discussion": 0, "suggestion_and_solution": 0, "total": 2 }	{ "criticism": 0.5, "example": 0, "importance_and_relevance": 0.5, "materials_and_methods": 0.5, "praise": 0.5, "presentation_and_reporting": 0, "results_and_discussion": 0, "suggestion_and_solution": 0 }	2	[ { "content": "Please review this paper.", "role": "user" }, { "content": "\n \n\n\fReviewers' Comments: \n \nReviewer #2: \nRemarks to the Author: \nThe new data supports the important finding that Wnt reduces linear ubiquitination of β-catenin, \nthereby controlling Wnt signaling. This reviewer has no further question. \n\n\f\n", "role": "assistant" } ]	2,020
ABL1-dependent OTULIN phosphorylation promotes genotoxic Wnt/β-catenin activation to	Reviewers' Comments: Reviewer #1: Remarks to the Author: In the manuscript submitted by Wang et al., the authors provide evidence for the DNA damage- induced activation of Wnt/-catenin signaling through a mechanism which is independent of canonical signaling through FZD/LRP heterodimeric receptors and Wnt ligands. They show that genotoxic stress induces the phosphorylation of OTULIN at Tyr56, which in turn allows for its engagement with-catenin. This inhibits linear ubiquitination of -catenin by LUBAC and subsequent Lys48-linked polyubiquitination by -TRCP, thereby attenuating the proteasomal degradation of -catenin upon DNA damage. Furthermore, through a limited siRNA screen with a tyrosine kinase sub-library, they identified ABL1 as the kinase responsible for phosphorylating OTULIN upon genotoxic stress. OTULIN-mediated Wnt/-catenin activation upon genotoxic stress promotes drug resistance in TNBC tumors, thus suggesting that targeting OTULIN may be an appropriate therapeutic strategy to reduce metastasis and drug resistance in TNBC patients. While most experiments are thoroughly executed, the following additional experiments should be considered to warrant publication: Major points: 1. The finding that Dox treatment induces the association between endogenous OTULIN and catenin is very subtle (i.e. Fig 4A), although this is a major part of the mechanism for OTULIN- mediated activation of Wnt signaling. It would be helpful to have a single figure comparing relative binding of -catenin to OTULIN, HOIP, and -TRCP with and without Dox treatment to better appreciate how these differences in binding contribute to the mechanism for -catenin stabilization. For example, in the model in Figure 6J, the authors suggest that LUBAC is bound to -catenin regardless of whether Wnt signaling is turned on and OTULIN is bound. Would increasing the concentration of LUBAC in a dose-dependent manner outcompete OUTLIN for access to - catenin for linear ubiquitination? Also, the model indicates that -TRCP is bound to -catenin in both the ON and OFF states, yet the data in Figure 3 showed that -TrCP binding was reduced in response to genotoxic stress. The model should be re-drawn to better emphasize this point. 2. The model in Figure 6J also indicates that c-Abl is translocated from the nucleus to phosphorylate OTULIN in the cytoplasm, although no direct evidence was provided for this. The authors should include immunofluorescence data showing the nuclear localization of c-Abl and OTULIN with or without Dox treatment to better clarify this point. 3. In addition to c-Abl, ITK was also found in the tyrosine kinase screen (Fig. 5A) to induce the phosphorylation of OTULIN, yet it was not followed up on. The authors should provide a supplemental figure to demonstrate that ITK phosphorylates OTULIN but does not impact Wnt- signaling in combination with c-Abl, as they suggest. 4. In Figure 6, the authors show a correlation between OTULIN levels and drug resistance in TNBC cells. It would also be interesting to observe whether there is also any correlation with LUBAC expression levels in the TCGA dataset. Does modulation of LUBAC attenuate the effects of OTULIN on drug resistance in the TNBC cells? Minor points: 1. Figure S3E has a misplaced K48-Ub data label. 2. The following sentence should reference Figure S5B instead of S5C: “In contrast, inhibiting c-Abl did not affect Wnt3a-induced b-catenin stabilization.” (pg. 13) 3. Methods section is not complete, no info on where the pY56-Otulin antibody came from and no info on sources of chemicals and antibodies, cell lines, etc Reviewer #2: Remarks to the Author: This manuscript describes the underlying mechanism that Wnt/β-catenin signaling is activated by genotoxic agents in breast cancer cells. The authors showed that OTULIN, a linear linkage-specific deubiquitinase, is required for the DNA damage-induced β-catenin activation. OTULIN was phosphorylated at Tyr56 by ABL1 and activated, resulting in mediating deubiquitination of the M1 ubiquitin chain of β-catenin at lysine 133. M1-linked linear ubiquitination of β-catenin by LUBAC promoted the interaction of β-TrCP with β-catenin, thereby inducing K48-linked polyubiquitination and degradation of β-catenin. In addition, the authors showed that OTULIN depletion decreased TNBC cells resistance to chemotherapeutics in vivo. From these results, the authors proposed that OTULIN activates Wnt/β-catenin signaling by inhibiting linear ubiquitination, K48-linked polyubiquitination, and proteasomal degradation of β- catenin, upon DNA damage. Although this paper is potentially interesting, there are several issues that the authors need to address. Major comments; (1) Is OTULIN-mediated deubiquitination of β-catenin involved in Wnt ligand-dependent activation of β-catenin signaling? Wnt3a did not induce OTULIN phosphorylation (sFig. 4G) and ABL1 or DNA- PK inhibitor that attenuates OTULIN phosphorylation and functions did not affect β-catenin stabilization by Wnt3a (sFig. 5B, J). In contrast, OTULIN KO inhibited Wnt3a-induced stabilization of β-catenin and target gene expression (Fig. 2A-C). These results are controversial, therefore, the authors should interpret carefully them. (2) CYLD, a USP family deubiquitinase, binds the PUB domain of HOIP and cleaves M1- and K63- ubiquitin linkage. It is crucial to test whether CYLD regulates β-catenin levels in the presence of genotoxic agents and Wnt3a. (3) In Fig. 3B, polyubiquitination of β-catenin seems to be greatly reduced by OTULIN or OTUB1, whereas K48-linked and M1-linked polyubiquitin chains, respectively, should not be affected by these deubiquitinases. The authors need to confirm by immunoblotting using chain type specific antibodies. (4) In Fig. 3E, the authors should show the effect of genotoxic agents and Wnt3a on the enhanced M1-ubiquitination of β-catenin in OTULIN-KO cells. The authors suggested that linear ubiquitination of β-catenin may occur prior to K48 ubiquitination. Then, it is necessary to examine K48-linked ubiquitination and proteasomal degradation of β-catenin in HOIP-KO cells. (5) The authors identified that K133 is a linear ubiquitination site in β-catenin. This reviewer is wondering what is the physiological roles of M1-linked linear ubiquitination of β-catenin at steady state? It has been reported that β-catenin Lys133 is methylated by SMYD2 and this modification is critical for the binding to FOXM1, nuclear translocation, and activation of Wnt signaling [PMID:28915556]. The authors need to test the relationship or interdependence between M1- linked linear ubiquitination and methylation at Lys133 (6) In sFig. 3E, some mutants of β-catenin other than K133R reduced M1-ubiquitination. The authors should examine the ubiquitination sites of β-catenin by mass spectrometry. Otherwise, in vitro ubiquitination assay is necessary to confirm that K113 is the M1-ubiquitination site. (7) In related to Fig. 3K, the authors should show the stability of β-catenin K133R that has defect of binding to β-TrCP compared with β-catenin K19/49R and β-catenin SA mutant (S33A/S37A/T41A/S45A, which is known to be constitutively activated). In addition, does OTULIN affect β-catenin stabilization by GSK3 inhibitors such as CHIR99021 that inhibits β-catenin phosphorylation? (8) In Fig. 3A and I, the authors showed that LUBAC expression induced M1-linked linear ubiquitination and degradation of β-catenin. To clarify that K19/K49 are K48-ubiqitination sites of β-catenin, whether LUBAC degrades β-catenin K19/K49R mutant should be tested. (9) As the molecular anatomy, the ABL1-binding region of OTULIN should be determined. Furthermore, does the region overlap with β-catenin-binding region of OUTLIN? (10) The authors should clarify the relationship between phosphorylation and M1-linked linear ubiquitination of β-catenin. In Fig. 3K, M1-ubiqitination defect (K133R) did not affect β-catenin phosphorylation at T41/S45 which is required for β-TrCP binding. Does the state of β-catenin phosphorylation at S33/S37 affect M1-ubiquitination of β-catenin? M1-ubiquitination state of β- catenin SA mutant (S33A/S37A/T41A/S45A) with or without DOX should be analyzed. (11) The authors identified that ABL1 is responsible for phosphorylation of OTULIN, which regulates the Wnt/β-catenin pathway in Fig. 5. However, many studies reported that OUTLIN is involved in NF-κB signaling and cell death. Therefore, the authors need to test whether the NF-κB, MAPK, and cell death pathways are activated by genotoxic agents and inflammatory cytokines in ABL1-KO cells. (12) In Fig. 5E, ABL1 KO phenotypes should be rescued by expression of WT ABL1 but not ABL1 kinase dead mutant. (13) In Fig. 5I, the total β-catenin level was unchanged by DOX treatment. The result is not consistent with other results such as Fig. 1A. (14) In Fig. 6, direct evidence that in vivo phenotypes induced by OTULIN KO cells are mediated by Wnt/β-catenin signaling was not shown The authors should show that phenotypes induced by OTULIN KO are rescued by Wnt signaling activation such as the expression of β-catenin SA mutant.	Reviewers' Comments: Reviewer #1: Remarks to the Author: In the manuscript submitted by Wang et al., the authors provide evidence for the DNA damage- induced activation of Wnt/-catenin signaling through a mechanism which is independent of canonical signaling through FZD/LRP heterodimeric receptors and Wnt ligands. They show that genotoxic stress induces the phosphorylation of OTULIN at Tyr56, which in turn allows for its engagement with-catenin. This inhibits linear ubiquitination of -catenin by LUBAC and subsequent Lys48-linked polyubiquitination by -TRCP, thereby attenuating the proteasomal degradation of -catenin upon DNA damage. Furthermore, through a limited siRNA screen with a tyrosine kinase sub-library, they identified ABL1 as the kinase responsible for phosphorylating OTULIN upon genotoxic stress. OTULIN-mediated Wnt/-catenin activation upon genotoxic stress promotes drug resistance in TNBC tumors, thus suggesting that targeting OTULIN may be an appropriate therapeutic strategy to reduce metastasis and drug resistance in TNBC patients. While most experiments are thoroughly executed, the following additional experiments should be considered to warrant publication: Major points: 1. The finding that Dox treatment induces the association between endogenous OTULIN and catenin is very subtle (i.e. Fig 4A), although this is a major part of the mechanism for OTULIN- mediated activation of Wnt signaling. It would be helpful to have a single figure comparing relative binding of -catenin to OTULIN, HOIP, and -TRCP with and without Dox treatment to better appreciate how these differences in binding contribute to the mechanism for -catenin stabilization. For example, in the model in Figure 6J, the authors suggest that LUBAC is bound to -catenin regardless of whether Wnt signaling is turned on and OTULIN is bound. Would increasing the concentration of LUBAC in a dose-dependent manner outcompete OUTLIN for access to - catenin for linear ubiquitination? Also, the model indicates that -TRCP is bound to -catenin in both the ON and OFF states, yet the data in Figure 3 showed that -TrCP binding was reduced in response to genotoxic stress. The model should be re-drawn to better emphasize this point. 2. The model in Figure 6J also indicates that c-Abl is translocated from the nucleus to phosphorylate OTULIN in the cytoplasm, although no direct evidence was provided for this. The authors should include immunofluorescence data showing the nuclear localization of c-Abl and OTULIN with or without Dox treatment to better clarify this point. 3. In addition to c-Abl, ITK was also found in the tyrosine kinase screen (Fig. 5A) to induce the phosphorylation of OTULIN, yet it was not followed up on. The authors should provide a supplemental figure to demonstrate that ITK phosphorylates OTULIN but does not impact Wnt- signaling in combination with c-Abl, as they suggest. 4. In Figure 6, the authors show a correlation between OTULIN levels and drug resistance in TNBC cells. It would also be interesting to observe whether there is also any correlation with LUBAC expression levels in the TCGA dataset. Does modulation of LUBAC attenuate the effects of OTULIN on drug resistance in the TNBC cells? Minor points: 1. Figure S3E has a misplaced K48-Ub data label. 2. The following sentence should reference Figure S5B instead of S5C: “In contrast, inhibiting c-Abl did not affect Wnt3a-induced b-catenin stabilization.” (pg. 13) 3. Methods section is not complete, no info on where the pY56-Otulin antibody came from and no info on sources of chemicals and antibodies, cell lines, etc Reviewer #2: Remarks to the Author: This manuscript describes the underlying mechanism that Wnt/β-catenin signaling is activated by genotoxic agents in breast cancer cells. The authors showed that OTULIN, a linear linkage-specific deubiquitinase, is required for the DNA damage-induced β-catenin activation. OTULIN was phosphorylated at Tyr56 by ABL1 and activated, resulting in mediating deubiquitination of the M1 ubiquitin chain of β-catenin at lysine 133. M1-linked linear ubiquitination of β-catenin by LUBAC promoted the interaction of β-TrCP with β-catenin, thereby inducing K48-linked polyubiquitination and degradation of β-catenin. In addition, the authors showed that OTULIN depletion decreased TNBC cells resistance to chemotherapeutics in vivo. From these results, the authors proposed that OTULIN activates Wnt/β-catenin signaling by inhibiting linear ubiquitination, K48-linked polyubiquitination, and proteasomal degradation of β- catenin, upon DNA damage. Although this paper is potentially interesting, there are several issues that the authors need to address. Major comments; (1) Is OTULIN-mediated deubiquitination of β-catenin involved in Wnt ligand-dependent activation of β-catenin signaling? Wnt3a did not induce OTULIN phosphorylation (sFig. 4G) and ABL1 or DNA- PK inhibitor that attenuates OTULIN phosphorylation and functions did not affect β-catenin stabilization by Wnt3a (sFig. 5B, J). In contrast, OTULIN KO inhibited Wnt3a-induced stabilization of β-catenin and target gene expression (Fig. 2A-C). These results are controversial, therefore, the authors should interpret carefully them. (2) CYLD, a USP family deubiquitinase, binds the PUB domain of HOIP and cleaves M1- and K63- ubiquitin linkage. It is crucial to test whether CYLD regulates β-catenin levels in the presence of genotoxic agents and Wnt3a. (3) In Fig. 3B, polyubiquitination of β-catenin seems to be greatly reduced by OTULIN or OTUB1, whereas K48-linked and M1-linked polyubiquitin chains, respectively, should not be affected by these deubiquitinases. The authors need to confirm by immunoblotting using chain type specific antibodies. (4) In Fig. 3E, the authors should show the effect of genotoxic agents and Wnt3a on the enhanced M1-ubiquitination of β-catenin in OTULIN-KO cells. The authors suggested that linear ubiquitination of β-catenin may occur prior to K48 ubiquitination. Then, it is necessary to examine K48-linked ubiquitination and proteasomal degradation of β-catenin in HOIP-KO cells. (5) The authors identified that K133 is a linear ubiquitination site in β-catenin. This reviewer is wondering what is the physiological roles of M1-linked linear ubiquitination of β-catenin at steady state? It has been reported that β-catenin Lys133 is methylated by SMYD2 and this modification is critical for the binding to FOXM1, nuclear translocation, and activation of Wnt signaling [PMID:28915556]. The authors need to test the relationship or interdependence between M1- linked linear ubiquitination and methylation at Lys133 (6) In sFig. 3E, some mutants of β-catenin other than K133R reduced M1-ubiquitination. The authors should examine the ubiquitination sites of β-catenin by mass spectrometry. Otherwise, in vitro ubiquitination assay is necessary to confirm that K113 is the M1-ubiquitination site. (7) In related to Fig. 3K, the authors should show the stability of β-catenin K133R that has defect of binding to β-TrCP compared with β-catenin K19/49R and β-catenin SA mutant (S33A/S37A/T41A/S45A, which is known to be constitutively activated). In addition, does OTULIN affect β-catenin stabilization by GSK3 inhibitors such as CHIR99021 that inhibits β-catenin phosphorylation? (8) In Fig. 3A and I, the authors showed that LUBAC expression induced M1-linked linear ubiquitination and degradation of β-catenin. To clarify that K19/K49 are K48-ubiqitination sites of β-catenin, whether LUBAC degrades β-catenin K19/K49R mutant should be tested. (9) As the molecular anatomy, the ABL1-binding region of OTULIN should be determined. Furthermore, does the region overlap with β-catenin-binding region of OUTLIN? (10) The authors should clarify the relationship between phosphorylation and M1-linked linear ubiquitination of β-catenin. In Fig. 3K, M1-ubiqitination defect (K133R) did not affect β-catenin phosphorylation at T41/S45 which is required for β-TrCP binding. Does the state of β-catenin phosphorylation at S33/S37 affect M1-ubiquitination of β-catenin? M1-ubiquitination state of β- catenin SA mutant (S33A/S37A/T41A/S45A) with or without DOX should be analyzed. (11) The authors identified that ABL1 is responsible for phosphorylation of OTULIN, which regulates the Wnt/β-catenin pathway in Fig. 5. However, many studies reported that OUTLIN is involved in NF-κB signaling and cell death. Therefore, the authors need to test whether the NF-κB, MAPK, and cell death pathways are activated by genotoxic agents and inflammatory cytokines in ABL1-KO cells. (12) In Fig. 5E, ABL1 KO phenotypes should be rescued by expression of WT ABL1 but not ABL1 kinase dead mutant. (13) In Fig. 5I, the total β-catenin level was unchanged by DOX treatment. The result is not consistent with other results such as Fig. 1A. (14) In Fig. 6, direct evidence that in vivo phenotypes induced by OTULIN KO cells are mediated by Wnt/β-catenin signaling was not shown The authors should show that phenotypes induced by OTULIN KO are rescued by Wnt signaling activation such as the expression of β-catenin SA mutant.	TPR	NC_BS_CAN_011_RR1.txt	{ "criticism": 5, "example": 17, "importance_and_relevance": 1, "materials_and_methods": 43, "praise": 2, "presentation_and_reporting": 17, "results_and_discussion": 6, "suggestion_and_solution": 23, "total": 84 }	{ "criticism": 0.05952380952380952, "example": 0.20238095238095238, "importance_and_relevance": 0.011904761904761904, "materials_and_methods": 0.5119047619047619, "praise": 0.023809523809523808, "presentation_and_reporting": 0.20238095238095238, "results_and_discussion": 0.07142857142857142, "suggestion_and_solution": 0.27380952380952384 }	1.357143	[ { "content": "Please review this paper.", "role": "user" }, { "content": "Reviewers' Comments: \n \nReviewer #1: \nRemarks to the Author: \nIn the manuscript submitted by Wang et al., the authors provide evidence for the DNA damage-\ninduced activation of Wnt/-catenin signaling through a mechanism which is independent of \ncanonical signaling through FZD/LRP heterodimeric receptors and Wnt ligands. They show that \ngenotoxic stress induces the phosphorylation of OTULIN at Tyr56, which in turn allows for its \nengagement with-catenin. This inhibits linear ubiquitination of -catenin by LUBAC and \nsubsequent Lys48-linked polyubiquitination by -TRCP, thereby attenuating the proteasomal \ndegradation of -catenin upon DNA damage. Furthermore, through a limited siRNA screen with a \ntyrosine kinase sub-library, they identified ABL1 as the kinase responsible for phosphorylating \nOTULIN upon genotoxic stress. OTULIN-mediated Wnt/-catenin activation upon genotoxic stress \npromotes drug resistance in TNBC tumors, thus suggesting that targeting OTULIN may be an \nappropriate therapeutic strategy to reduce metastasis and drug resistance in TNBC patients. While \nmost experiments are thoroughly executed, the following additional experiments should be \nconsidered to warrant publication: \n \nMajor points: \n \n1. The finding that Dox treatment induces the association between endogenous OTULIN and \ncatenin is very subtle (i.e. Fig 4A), although this is a major part of the mechanism for OTULIN-\nmediated activation of Wnt signaling. It would be helpful to have a single figure comparing relative \nbinding of -catenin to OTULIN, HOIP, and -TRCP with and without Dox treatment to better \nappreciate how these differences in binding contribute to the mechanism for -catenin \nstabilization. For example, in the model in Figure 6J, the authors suggest that LUBAC is bound to \n-catenin regardless of whether Wnt signaling is turned on and OTULIN is bound. Would increasing \nthe concentration of LUBAC in a dose-dependent manner outcompete OUTLIN for access to -\ncatenin for linear ubiquitination? Also, the model indicates that -TRCP is bound to -catenin in \nboth the ON and OFF states, yet the data in Figure 3 showed that -TrCP binding was reduced in \nresponse to genotoxic stress. The model should be re-drawn to better emphasize this point. \n2. The model in Figure 6J also indicates that c-Abl is translocated from the nucleus to \nphosphorylate OTULIN in the cytoplasm, although no direct evidence was provided for this. The \nauthors should include immunofluorescence data showing the nuclear localization of c-Abl and \nOTULIN with or without Dox treatment to better clarify this point. \n3. In addition to c-Abl, ITK was also found in the tyrosine kinase screen (Fig. 5A) to induce the \nphosphorylation of OTULIN, yet it was not followed up on. The authors should provide a \nsupplemental figure to demonstrate that ITK phosphorylates OTULIN but does not impact Wnt-\nsignaling in combination with c-Abl, as they suggest. \n4. In Figure 6, the authors show a correlation between OTULIN levels and drug resistance in TNBC \ncells. It would also be interesting to observe whether there is also any correlation with LUBAC \nexpression levels in the TCGA dataset. Does modulation of LUBAC attenuate the effects of OTULIN \non drug resistance in the TNBC cells? \n \n \nMinor points: \n \n1. Figure S3E has a misplaced K48-Ub data label. \n2. The following sentence should reference Figure S5B instead of S5C: “In contrast, inhibiting c-Abl \ndid not affect Wnt3a-induced b-catenin stabilization.” (pg. 13) \n3. Methods section is not complete, no info on where the pY56-Otulin antibody came from and no \ninfo on sources of chemicals and antibodies, cell lines, etc \n \n \n \n\n\fReviewer #2: \nRemarks to the Author: \nThis manuscript describes the underlying mechanism that Wnt/β-catenin signaling is activated by \ngenotoxic agents in breast cancer cells. The authors showed that OTULIN, a linear linkage-specific \ndeubiquitinase, is required for the DNA damage-induced β-catenin activation. OTULIN was \nphosphorylated at Tyr56 by ABL1 and activated, resulting in mediating deubiquitination of the M1 \nubiquitin chain of β-catenin at lysine 133. M1-linked linear ubiquitination of β-catenin by LUBAC \npromoted the interaction of β-TrCP with β-catenin, thereby inducing K48-linked polyubiquitination \nand degradation of β-catenin. In addition, the authors showed that OTULIN depletion decreased \nTNBC cells resistance to chemotherapeutics in vivo. \n \nFrom these results, the authors proposed that OTULIN activates Wnt/β-catenin signaling by \ninhibiting linear ubiquitination, K48-linked polyubiquitination, and proteasomal degradation of β-\ncatenin, upon DNA damage. Although this paper is potentially interesting, there are several issues \nthat the authors need to address. \n \nMajor comments; \n(1) Is OTULIN-mediated deubiquitination of β-catenin involved in Wnt ligand-dependent activation \nof β-catenin signaling? Wnt3a did not induce OTULIN phosphorylation (sFig. 4G) and ABL1 or DNA-\nPK inhibitor that attenuates OTULIN phosphorylation and functions did not affect β-catenin \nstabilization by Wnt3a (sFig. 5B, J). In contrast, OTULIN KO inhibited Wnt3a-induced stabilization \nof β-catenin and target gene expression (Fig. 2A-C). These results are controversial, therefore, the \nauthors should interpret carefully them. \n \n(2) CYLD, a USP family deubiquitinase, binds the PUB domain of HOIP and cleaves M1- and K63-\nubiquitin linkage. It is crucial to test whether CYLD regulates β-catenin levels in the presence of \ngenotoxic agents and Wnt3a. \n \n(3) In Fig. 3B, polyubiquitination of β-catenin seems to be greatly reduced by OTULIN or OTUB1, \nwhereas K48-linked and M1-linked polyubiquitin chains, respectively, should not be affected by \nthese deubiquitinases. The authors need to confirm by immunoblotting using chain type specific \nantibodies. \n \n(4) In Fig. 3E, the authors should show the effect of genotoxic agents and Wnt3a on the enhanced \nM1-ubiquitination of β-catenin in OTULIN-KO cells. The authors suggested that linear ubiquitination \nof β-catenin may occur prior to K48 ubiquitination. Then, it is necessary to examine K48-linked \nubiquitination and proteasomal degradation of β-catenin in HOIP-KO cells. \n \n(5) The authors identified that K133 is a linear ubiquitination site in β-catenin. This reviewer is \nwondering what is the physiological roles of M1-linked linear ubiquitination of β-catenin at steady \nstate? It has been reported that β-catenin Lys133 is methylated by SMYD2 and this modification is \ncritical for the binding to FOXM1, nuclear translocation, and activation of Wnt signaling \n[PMID:28915556]. The authors need to test the relationship or interdependence between M1-\nlinked linear ubiquitination and methylation at Lys133 \n \n(6) In sFig. 3E, some mutants of β-catenin other than K133R reduced M1-ubiquitination. The \nauthors should examine the ubiquitination sites of β-catenin by mass spectrometry. Otherwise, in \nvitro ubiquitination assay is necessary to confirm that K113 is the M1-ubiquitination site. \n \n(7) In related to Fig. 3K, the authors should show the stability of β-catenin K133R that has defect \nof binding to β-TrCP compared with β-catenin K19/49R and β-catenin SA mutant \n(S33A/S37A/T41A/S45A, which is known to be constitutively activated). In addition, does OTULIN \naffect β-catenin stabilization by GSK3 inhibitors such as CHIR99021 that inhibits β-catenin \nphosphorylation? \n \n\n\f(8) In Fig. 3A and I, the authors showed that LUBAC expression induced M1-linked linear \nubiquitination and degradation of β-catenin. To clarify that K19/K49 are K48-ubiqitination sites of \nβ-catenin, whether LUBAC degrades β-catenin K19/K49R mutant should be tested. \n \n(9) As the molecular anatomy, the ABL1-binding region of OTULIN should be determined. \nFurthermore, does the region overlap with β-catenin-binding region of OUTLIN? \n \n(10) The authors should clarify the relationship between phosphorylation and M1-linked linear \nubiquitination of β-catenin. In Fig. 3K, M1-ubiqitination defect (K133R) did not affect β-catenin \nphosphorylation at T41/S45 which is required for β-TrCP binding. Does the state of β-catenin \nphosphorylation at S33/S37 affect M1-ubiquitination of β-catenin? M1-ubiquitination state of β-\ncatenin SA mutant (S33A/S37A/T41A/S45A) with or without DOX should be analyzed. \n \n(11) The authors identified that ABL1 is responsible for phosphorylation of OTULIN, which \nregulates the Wnt/β-catenin pathway in Fig. 5. However, many studies reported that OUTLIN is \ninvolved in NF-κB signaling and cell death. Therefore, the authors need to test whether the NF-κB, \nMAPK, and cell death pathways are activated by genotoxic agents and inflammatory cytokines in \nABL1-KO cells. \n \n(12) In Fig. 5E, ABL1 KO phenotypes should be rescued by expression of WT ABL1 but not ABL1 \nkinase dead mutant. \n \n(13) In Fig. 5I, the total β-catenin level was unchanged by DOX treatment. The result is not \nconsistent with other results such as Fig. 1A. \n \n(14) In Fig. 6, direct evidence that in vivo phenotypes induced by OTULIN KO cells are mediated \nby Wnt/β-catenin signaling was not shown The authors should show that phenotypes induced by \nOTULIN KO are rescued by Wnt signaling activation such as the expression of β-catenin SA \nmutant. \n\n\f\f\n", "role": "assistant" } ]	2,020
The REGγ inhibitor NIP30 increases sensitivity to chemotherapy in p53-deficient tumor cells	REVIEWERS' COMMENTS: Reviewer #1 (Remarks to the Author): This manuscript has been revised in response to my previous comments. Some responses are satisfactory, but others, in my opinion, remain unsatisfactory. I continue to believe that acknowledgment of earlier work (ref 32) is inadequate with respect to the findings therein and their relationship to data in the current manuscript (e.g. insight to and identification of the phosphorylation sites, effects of phosphorylation, etc, etc). I also consider the description, conduct and analysis of the in vitro degradation experiment I questioned previously to be inadequate - including weight-amounts of some proteins is a bare minimum of new information. Finally, although somewhat improved, the presentation still includes a number of odd word choices and syntax throughout and there many typos and inappropriate formatting, especially in the Reference section.	REVIEWERS' COMMENTS: Reviewer #1 (Remarks to the Author): This manuscript has been revised in response to my previous comments. Some responses are satisfactory, but others, in my opinion, remain unsatisfactory. I continue to believe that acknowledgment of earlier work (ref 32) is inadequate with respect to the findings therein and their relationship to data in the current manuscript (e.g. insight to and identification of the phosphorylation sites, effects of phosphorylation, etc, etc). I also consider the description, conduct and analysis of the in vitro degradation experiment I questioned previously to be inadequate - including weight-amounts of some proteins is a bare minimum of new information. Finally, although somewhat improved, the presentation still includes a number of odd word choices and syntax throughout and there many typos and inappropriate formatting, especially in the Reference section.	TPR	NC_BS_CAN_012_RR3.txt	{ "criticism": 3, "example": 0, "importance_and_relevance": 0, "materials_and_methods": 2, "praise": 1, "presentation_and_reporting": 1, "results_and_discussion": 0, "suggestion_and_solution": 0, "total": 6 }	{ "criticism": 0.5, "example": 0, "importance_and_relevance": 0, "materials_and_methods": 0.3333333333333333, "praise": 0.16666666666666666, "presentation_and_reporting": 0.16666666666666666, "results_and_discussion": 0, "suggestion_and_solution": 0 }	1.166667	[ { "content": "Please review this paper.", "role": "user" }, { "content": "\n \n\n\fREVIEWERS' COMMENTS: \n\nReviewer #1 (Remarks to the Author): \n\nThis manuscript has been revised in response to my previous comments. Some responses are \nsatisfactory, but others, in my opinion, remain unsatisfactory. I continue to believe that \nacknowledgment of earlier work (ref 32) is inadequate with respect to the findings therein and \ntheir relationship to data in the current manuscript (e.g. insight to and identification of the \nphosphorylation sites, effects of phosphorylation, etc, etc). I also consider the description, conduct \nand analysis of the in vitro degradation experiment I questioned previously to be inadequate - \nincluding weight-amounts of some proteins is a bare minimum of new information. Finally, \nalthough somewhat improved, the presentation still includes a number of odd word choices and \nsyntax throughout and there many typos and inappropriate formatting, especially in the Reference \nsection.\n\n\n\f\n", "role": "assistant" } ]	2,020
The REGγ inhibitor NIP30 increases sensitivity to chemotherapy in p53-deficient tumor cells	Reviewers' comments: Reviewer #1 (Remarks to the Author); expert in proteasome regulation: This is an interesting and potentially important report that describes regulation of REG gamma function via the action of two proteins: NIP30 and CDC25. The authors present data showing that NIP30 directly binds to REG gamma and decrease levels of p21, presumably by enhancing REGgamma activation of proteasome activity. They further show that NIP30 is phosphorylated and that phosphorylation attenuates NIP30/ REGgamma activated p21 degradation. They identify CDC25 as a NIP30 phosphatase and provide evidence that this activity may regulate NIP30/ REG - mediated levels of p21. In general, the data are internally consistent and offer a model for proteasome-dependent control of p21 levels and consequent effects on cell growth. A main advance of this manuscript is the identification of CDC25 as a regulator of the NIP30/ REGgamma/proteasome axis and the data that support this model. However, many aspects of the presentation are premature, descriptive and problematic. 1. The first section of the results (Figures 1 and 2) describes identification of NIP30 as a REGgamma interacting protein and show that the C-terminus of NIP has a conserved site that is phosphorylated and required for the REGgamma interaction. Similar analysis and conclusions were obtained by Jonik-Nowak et al (ref 32) but are not described in an appropriate context here. This is somewhat misleading. Describing those results more completely in the set-up to the Results section would allow the authors to streamline the paper, focus on what is new, and perhaps go deeper into the mechanisms of the effects reported here. 2. The role of NIP30 phosphorylation is a key mechanistic part of the manuscript. The authors’ data a generally consistent with this but there needs to be more direct evidence. I did not find a description of the production of phospho-specific antibodies used in the study – a complete description of the production and validation of these antibodies is essential. The use of these antibodies is introduced in experiments of Figure 3, but should be used earlier. It would be useful to have supporting evidence in the form of direct demonstration of phosphorylation with either mass-spec data, or radioactive phosphate incorporation. There must be reference to (32) where NIP30 phosphorylation of NIP30 by CKII was demonstrated directly. 3. The data in Figure 3 are not very completing. The discrimination of co-localization on the scale used here is not fine enough to demonstrate physical interactions between proteins or assessments of the strengths of these interactions, as claimed in the text. Also, there are many proteins whose highest level of expression occurs in the testis – these results are correlative and not convincing for the claims. 4. As the authors note, the molecular mechanisms by which REG gammamight mediate ubiquitin- independent protein breakdown are not very well understood. In my opinion, this is one of the major problems in understanding how REG functions in the degradation of intact proteins. Although this manuscript fills in some of this understanding, the results still lack mechanistic detail. Steady-state levels of p21 may indeed reflect degradation, but that is not shown directly. As noted above, the data are consistent with the authors’ models, but it is possible and likely that NIP30/ REGgamma could affect p21 levels indirectly by other mechanisms. In this regard, the authors should discuss the results of ref 32 that offer either conflicting or alternative possibilities and at least show the complexity of the system. Curiously, others showed that NIP30 inhibited REG proteasome activation (although phos-NIP30 increased the degree of inhibition). In any case, the authors seem to be in a position to test a direct mechanism with a reconstituted in vitro systems. They either have or should be able to prepare pure proteins involved in the proposed regulatory axis and directly test the degradation of non-ubiquitinated p21 by it. As noted early in the Introduction, but then ignored for the rest of the paper, there is controversy about the relative roles of ubiquitin-dependent and ubiquitin-independent degradation of p21. REG gammacould be part of a hybrid proteasome that degrades p21 in an ubiquitin dependent mechanism. The suggested in vitro constitution experiments would help to sort this out. 5. The entire manuscript is in need of extensive editing to correct typographical errors and errors in English usage and grammar that include the lack of necessary articles, incorrect subject-verb agreement, fractured sentences, and odd or inappropriate word choices. The bibliography contains many incomplete or incorrectly formatted citations. There are too many instances of each of these issues to cite individually here, but in total they make the manuscript distracting and annoying to read. I am sensitive to and appreciate English-as-a-second-language challenges, but it is difficult to believe that all of the authors read and approved this version of the manuscript or that other editing services are not available at Baylor. Reviewer #2 (Remarks to the Author); expert in cell cycle, DNA damage: The authors build on the work of another group (Beata Jonik-Nowak et al, Proc Natl Acad Sci, 2018, 115:E6477; PMC6048556) who showed that PIP30/FAM192A (here called NIP30) is phosphorylated by CKII, and that phosphorylated PIP30/NIP30 binds and regulates the function of REGgamma. Here Gao et al confirm those findings and further: 1) Show that overexpression of wild-type and phosphomimetic (4A) PIP30/NIP30 posttranscriptionally affects the levels of p21, which is a substrate of REGgamma. 2) Show that overexpression of nonphosphorylatable (4D) PIP30/NIP30 reverses the antiproliferative effects of REGgamma knockout 3) Claim that CDC25A dephosphorylates PIP30/NIP30, which in turn regulates p21 levels in a p53- independent manner. 4) Identify that PIP30/NIP30 is altered in a low percentage of tumors 5) Show that overexpression of wild-type and phosphophomimetic PIP30/NIP30 increases tumor sensitivity to etoposide, cisplatin, 5-fluorouracil, and doxorubicin 6) Claim that PIP30/NIP30 is a tumor suppressor 7) Present a model in which CDC25A regulates NIP30/PIP30, which in turn regulates REGgamma that then regulates p21 levels and cell cycle progression. This is an interesting study. However, there are several serious concerns. 1) Although the results showing that altering CDC25A levels affects PIP30/NIP30 phosphorylation in cells is solid, the data showing that CDC25A directly de-phosphorylates PIP30/NIP30 is very weak (Fig. 3E); this analysis used immunopurified PIP30/NIP30 and immunopurified CDC25A. It is very possible that a contaminating phosphatase is responsible for the in vitro dephosphorylation that is observed. 2) It is not clear that CDC25A’s effects on p21 levels are driven by CDC25A-mediated dephosphorylation of PIP30/NIP30. They could be through another unknown pathway. Does expression of phosphomimetic PIP30/NIP30 prevent CDC25A-induced changes in p21? 3) Most of the experiments rely on PIP30/NIP30 overexpression. This is a serious concern because overexpression studies are prone to artefacts. Additionally, given that the authors propose that PIP30/NIP30 is a tumor suppressor, a serious analysis of the effects of losing PIP30/NIP30 would be important to understanding its functions. 4) PIP30/NIP30 mutations found in tumors are very infrequent and the effects mutations that are shown in the supplemental figure are not known. Accordingly, there are no data to support the contention that PIP30/NIP30 is a tumor suppressor. 5) Although p21 levels are affected by CDC25A and PIP30/NIP30, it is not clear that changes in p21 levels are actually driving cell cycle changes induced by CDC25A and PIP30/NIP30 manipulation. Minor comments: 1) There are many grammatical errors. 2) It is not always clear how many times an experiment was done. 3) Abbreviations are not always defined. 4) Figures are too small to read, especially annotations and symbols on graphs, even when zoomed in on. 5) Some figures are not appropriately labeled. For example, in Fig. 6B, the x-axes are not labeled and the y-axes are labeled as “survival rate.” If an MTT assay was used, this is not a survival rate. Reviewer #3 (Remarks to the Author); expert in p53, mouse models: The manuscript by Gao, et al. focuses on NIP30 as an inhibitor of the REG proteasome, which is oncogenic and distinct from the 26S proteasome. The authors uncovered a tumor suppressor role for NIP30 by its interactions with REGmapping the amino acids of interaction and uncovering a phosphorylation switch for this interaction. CDC25A was identified as the major regulatory phosphatase for this inhibitory switch during cell cycle and in response to DNA damage. Cell cycle inhibitor p21 is a major target of the CDC25A-NIP30- REG axis, which induces regulation in a p53-dependent manner. Multiple approaches were employed in these studies: IP/mass spec; yeast two-hybrid; co-IP; domain and amino acid interaction mapping; xenograft tumor analyses of expressing, mutated and KD of key players; screening of a phosphatase library; cell cycle and DNA damage studies and UV-response of the skin taken from newborn REG WT, KO, Tpr53 KO and REG;Tpr53 double KO; and chemosensitivity studies with cell lines and xenografts. Overall, this is a very thorough analysis that pinpoints a specific phosphorylation switch control of proteasome regulation of p21 during cell cycle and DNA damage response. The studies are primarily cell line-based but are also complemented by xenograft, genetically modified mice and TCGA data. This work suggests a new target and approach toward effectively treating p53- dysfunctional cancers with chemotherapeutics, a major problem in the clinic. This finding is likely to be of considerable interest to the community. The major problem with the manuscript is poor preparation, improper use of English and inexact terminology. The authors should employ an English editor and science writer to resubmit a better manuscript. Specific examples of inexact terminology include “tight binding”, found in the Abstract and within the manuscript. How is “tight binding” measured and what are the values for binding constants that would indicate “tight”? Quantifying this binding is needed for such terms to be used. Showing co-localization in immunofluorescence microscopy is insufficient and non-robust (Suppl. Fig. 3A). Examples of poor manuscript preparation are the numerous grammatical errors, incorrect word usage that obscures the likely intended meaning (a couple of examples: “difference cells”? “Hiked levels”?), and some references improperly formatted. Additionally, some figure legends lack sufficient detail (Fig. 1A, what is N1 and S1, for example). References are sometimes lacking. A major example of this is the human Ser/Thr phosphatase library screen. There is no reference for this library or any data for the screen shown. These data should be included as supplemental data or referenced.	Reviewers' comments: Reviewer #1 (Remarks to the Author); expert in proteasome regulation: This is an interesting and potentially important report that describes regulation of REG gamma function via the action of two proteins: NIP30 and CDC25. The authors present data showing that NIP30 directly binds to REG gamma and decrease levels of p21, presumably by enhancing REGgamma activation of proteasome activity. They further show that NIP30 is phosphorylated and that phosphorylation attenuates NIP30/ REGgamma activated p21 degradation. They identify CDC25 as a NIP30 phosphatase and provide evidence that this activity may regulate NIP30/ REG - mediated levels of p21. In general, the data are internally consistent and offer a model for proteasome-dependent control of p21 levels and consequent effects on cell growth. A main advance of this manuscript is the identification of CDC25 as a regulator of the NIP30/ REGgamma/proteasome axis and the data that support this model. However, many aspects of the presentation are premature, descriptive and problematic. 1. The first section of the results (Figures 1 and 2) describes identification of NIP30 as a REGgamma interacting protein and show that the C-terminus of NIP has a conserved site that is phosphorylated and required for the REGgamma interaction. Similar analysis and conclusions were obtained by Jonik-Nowak et al (ref 32) but are not described in an appropriate context here. This is somewhat misleading. Describing those results more completely in the set-up to the Results section would allow the authors to streamline the paper, focus on what is new, and perhaps go deeper into the mechanisms of the effects reported here. 2. The role of NIP30 phosphorylation is a key mechanistic part of the manuscript. The authors’ data a generally consistent with this but there needs to be more direct evidence. I did not find a description of the production of phospho-specific antibodies used in the study – a complete description of the production and validation of these antibodies is essential. The use of these antibodies is introduced in experiments of Figure 3, but should be used earlier. It would be useful to have supporting evidence in the form of direct demonstration of phosphorylation with either mass-spec data, or radioactive phosphate incorporation. There must be reference to (32) where NIP30 phosphorylation of NIP30 by CKII was demonstrated directly. 3. The data in Figure 3 are not very completing. The discrimination of co-localization on the scale used here is not fine enough to demonstrate physical interactions between proteins or assessments of the strengths of these interactions, as claimed in the text. Also, there are many proteins whose highest level of expression occurs in the testis – these results are correlative and not convincing for the claims. 4. As the authors note, the molecular mechanisms by which REG gammamight mediate ubiquitin- independent protein breakdown are not very well understood. In my opinion, this is one of the major problems in understanding how REG functions in the degradation of intact proteins. Although this manuscript fills in some of this understanding, the results still lack mechanistic detail. Steady-state levels of p21 may indeed reflect degradation, but that is not shown directly. As noted above, the data are consistent with the authors’ models, but it is possible and likely that NIP30/ REGgamma could affect p21 levels indirectly by other mechanisms. In this regard, the authors should discuss the results of ref 32 that offer either conflicting or alternative possibilities and at least show the complexity of the system. Curiously, others showed that NIP30 inhibited REG proteasome activation (although phos-NIP30 increased the degree of inhibition). In any case, the authors seem to be in a position to test a direct mechanism with a reconstituted in vitro systems. They either have or should be able to prepare pure proteins involved in the proposed regulatory axis and directly test the degradation of non-ubiquitinated p21 by it. As noted early in the Introduction, but then ignored for the rest of the paper, there is controversy about the relative roles of ubiquitin-dependent and ubiquitin-independent degradation of p21. REG gammacould be part of a hybrid proteasome that degrades p21 in an ubiquitin dependent mechanism. The suggested in vitro constitution experiments would help to sort this out. 5. The entire manuscript is in need of extensive editing to correct typographical errors and errors in English usage and grammar that include the lack of necessary articles, incorrect subject-verb agreement, fractured sentences, and odd or inappropriate word choices. The bibliography contains many incomplete or incorrectly formatted citations. There are too many instances of each of these issues to cite individually here, but in total they make the manuscript distracting and annoying to read. I am sensitive to and appreciate English-as-a-second-language challenges, but it is difficult to believe that all of the authors read and approved this version of the manuscript or that other editing services are not available at Baylor. Reviewer #2 (Remarks to the Author); expert in cell cycle, DNA damage: The authors build on the work of another group (Beata Jonik-Nowak et al, Proc Natl Acad Sci, 2018, 115:E6477; PMC6048556) who showed that PIP30/FAM192A (here called NIP30) is phosphorylated by CKII, and that phosphorylated PIP30/NIP30 binds and regulates the function of REGgamma. Here Gao et al confirm those findings and further: 1) Show that overexpression of wild-type and phosphomimetic (4A) PIP30/NIP30 posttranscriptionally affects the levels of p21, which is a substrate of REGgamma. 2) Show that overexpression of nonphosphorylatable (4D) PIP30/NIP30 reverses the antiproliferative effects of REGgamma knockout 3) Claim that CDC25A dephosphorylates PIP30/NIP30, which in turn regulates p21 levels in a p53- independent manner. 4) Identify that PIP30/NIP30 is altered in a low percentage of tumors 5) Show that overexpression of wild-type and phosphophomimetic PIP30/NIP30 increases tumor sensitivity to etoposide, cisplatin, 5-fluorouracil, and doxorubicin 6) Claim that PIP30/NIP30 is a tumor suppressor 7) Present a model in which CDC25A regulates NIP30/PIP30, which in turn regulates REGgamma that then regulates p21 levels and cell cycle progression. This is an interesting study. However, there are several serious concerns. 1) Although the results showing that altering CDC25A levels affects PIP30/NIP30 phosphorylation in cells is solid, the data showing that CDC25A directly de-phosphorylates PIP30/NIP30 is very weak (Fig. 3E); this analysis used immunopurified PIP30/NIP30 and immunopurified CDC25A. It is very possible that a contaminating phosphatase is responsible for the in vitro dephosphorylation that is observed. 2) It is not clear that CDC25A’s effects on p21 levels are driven by CDC25A-mediated dephosphorylation of PIP30/NIP30. They could be through another unknown pathway. Does expression of phosphomimetic PIP30/NIP30 prevent CDC25A-induced changes in p21? 3) Most of the experiments rely on PIP30/NIP30 overexpression. This is a serious concern because overexpression studies are prone to artefacts. Additionally, given that the authors propose that PIP30/NIP30 is a tumor suppressor, a serious analysis of the effects of losing PIP30/NIP30 would be important to understanding its functions. 4) PIP30/NIP30 mutations found in tumors are very infrequent and the effects mutations that are shown in the supplemental figure are not known. Accordingly, there are no data to support the contention that PIP30/NIP30 is a tumor suppressor. 5) Although p21 levels are affected by CDC25A and PIP30/NIP30, it is not clear that changes in p21 levels are actually driving cell cycle changes induced by CDC25A and PIP30/NIP30 manipulation. Minor comments: 1) There are many grammatical errors. 2) It is not always clear how many times an experiment was done. 3) Abbreviations are not always defined. 4) Figures are too small to read, especially annotations and symbols on graphs, even when zoomed in on. 5) Some figures are not appropriately labeled. For example, in Fig. 6B, the x-axes are not labeled and the y-axes are labeled as “survival rate.” If an MTT assay was used, this is not a survival rate. Reviewer #3 (Remarks to the Author); expert in p53, mouse models: The manuscript by Gao, et al. focuses on NIP30 as an inhibitor of the REG proteasome, which is oncogenic and distinct from the 26S proteasome. The authors uncovered a tumor suppressor role for NIP30 by its interactions with REGmapping the amino acids of interaction and uncovering a phosphorylation switch for this interaction. CDC25A was identified as the major regulatory phosphatase for this inhibitory switch during cell cycle and in response to DNA damage. Cell cycle inhibitor p21 is a major target of the CDC25A-NIP30- REG axis, which induces regulation in a p53-dependent manner. Multiple approaches were employed in these studies: IP/mass spec; yeast two-hybrid; co-IP; domain and amino acid interaction mapping; xenograft tumor analyses of expressing, mutated and KD of key players; screening of a phosphatase library; cell cycle and DNA damage studies and UV-response of the skin taken from newborn REG WT, KO, Tpr53 KO and REG;Tpr53 double KO; and chemosensitivity studies with cell lines and xenografts. Overall, this is a very thorough analysis that pinpoints a specific phosphorylation switch control of proteasome regulation of p21 during cell cycle and DNA damage response. The studies are primarily cell line-based but are also complemented by xenograft, genetically modified mice and TCGA data. This work suggests a new target and approach toward effectively treating p53- dysfunctional cancers with chemotherapeutics, a major problem in the clinic. This finding is likely to be of considerable interest to the community. The major problem with the manuscript is poor preparation, improper use of English and inexact terminology. The authors should employ an English editor and science writer to resubmit a better manuscript. Specific examples of inexact terminology include “tight binding”, found in the Abstract and within the manuscript. How is “tight binding” measured and what are the values for binding constants that would indicate “tight”? Quantifying this binding is needed for such terms to be used. Showing co-localization in immunofluorescence microscopy is insufficient and non-robust (Suppl. Fig. 3A). Examples of poor manuscript preparation are the numerous grammatical errors, incorrect word usage that obscures the likely intended meaning (a couple of examples: “difference cells”? “Hiked levels”?), and some references improperly formatted. Additionally, some figure legends lack sufficient detail (Fig. 1A, what is N1 and S1, for example). References are sometimes lacking. A major example of this is the human Ser/Thr phosphatase library screen. There is no reference for this library or any data for the screen shown. These data should be included as supplemental data or referenced.	TPR	NC_BS_CAN_012_RR1.txt	{ "criticism": 31, "example": 9, "importance_and_relevance": 8, "materials_and_methods": 38, "praise": 4, "presentation_and_reporting": 25, "results_and_discussion": 9, "suggestion_and_solution": 14, "total": 93 }	{ "criticism": 0.3333333333333333, "example": 0.0967741935483871, "importance_and_relevance": 0.08602150537634409, "materials_and_methods": 0.40860215053763443, "praise": 0.043010752688172046, "presentation_and_reporting": 0.26881720430107525, "results_and_discussion": 0.0967741935483871, "suggestion_and_solution": 0.15053763440860216 }	1.483871	[ { "content": "Please review this paper.", "role": "user" }, { "content": "Reviewers' comments: \n\nReviewer #1 (Remarks to the Author); expert in proteasome regulation: \n\nThis is an interesting and potentially important report that describes regulation of REG gamma \nfunction via the action of two proteins: NIP30 and CDC25. The authors present data showing that \nNIP30 directly binds to REG gamma and decrease levels of p21, presumably by enhancing \nREGgamma activation of proteasome activity. They further show that NIP30 is phosphorylated \nand that phosphorylation attenuates NIP30/ REGgamma activated p21 degradation. They identify \nCDC25 as a NIP30 phosphatase and provide evidence that this activity may regulate NIP30/ REG - \nmediated levels of p21. In general, the data are internally consistent and offer a model for \nproteasome-dependent control of p21 levels and consequent effects on cell growth. \n\nA main advance of this manuscript is the identification of CDC25 as a regulator of the NIP30/ \nREGgamma/proteasome axis and the data that support this model. However, many aspects of the \npresentation are premature, descriptive and problematic. \n\n1. The first section of the results (Figures 1 and 2) describes identification of NIP30 as a \nREGgamma interacting protein and show that the C-terminus of NIP has a conserved site that is \nphosphorylated and required for the REGgamma interaction. Similar analysis and conclusions \nwere obtained by Jonik-Nowak et al (ref 32) but are not described in an appropriate context here. \nThis is somewhat misleading. Describing those results more completely in the set-up to the Results \nsection would allow the authors to streamline the paper, focus on what is new, and perhaps go \ndeeper into the mechanisms of the effects reported here. \n2. The role of NIP30 phosphorylation is a key mechanistic part of the manuscript. The authors’ \ndata a generally consistent with this but there needs to be more direct evidence. I did not find a \ndescription of the production of phospho-specific antibodies used in the study – a complete \ndescription of the production and validation of these antibodies is essential. The use of these \nantibodies is introduced in experiments of Figure 3, but should be used earlier. It would be useful \nto have supporting evidence in the form of direct demonstration of phosphorylation with either \nmass-spec data, or radioactive phosphate incorporation. There must be reference to (32) where \nNIP30 phosphorylation of NIP30 by CKII was demonstrated directly. \n3. The data in Figure 3 are not very completing. The discrimination of co-localization on the scale \nused here is not fine enough to demonstrate physical interactions between proteins or \nassessments of the strengths of these interactions, as claimed in the text. Also, there are many \nproteins whose highest level of expression occurs in the testis – these results are correlative and \nnot convincing for the claims. \n4. As the authors note, the molecular mechanisms by which REG gammamight mediate ubiquitin-\nindependent protein breakdown are not very well understood. In my opinion, this is one of the \nmajor problems in understanding how REG functions in the degradation of intact proteins. \nAlthough this manuscript fills in some of this understanding, the results still lack mechanistic \ndetail. Steady-state levels of p21 may indeed reflect degradation, but that is not shown directly. \nAs noted above, the data are consistent with the authors’ models, but it is possible and likely that \nNIP30/ REGgamma could affect p21 levels indirectly by other mechanisms. In this regard, the \nauthors should discuss the results of ref 32 that offer either conflicting or alternative possibilities \nand at least show the complexity of the system. Curiously, others showed that NIP30 inhibited \nREG proteasome activation (although phos-NIP30 increased the degree of inhibition). In any case, \nthe authors seem to be in a position to test a direct mechanism with a reconstituted in vitro \nsystems. They either have or should be able to prepare pure proteins involved in the proposed \nregulatory axis and directly test the degradation of non-ubiquitinated p21 by it. As noted early in \nthe Introduction, but then ignored for the rest of the paper, there is controversy about the relative \nroles of ubiquitin-dependent and ubiquitin-independent degradation of p21. REG gammacould be \npart of a hybrid proteasome that degrades p21 in an ubiquitin dependent mechanism. The \nsuggested in vitro constitution experiments would help to sort this out. \n5. The entire manuscript is in need of extensive editing to correct typographical errors and errors \n\n\fin English usage and grammar that include the lack of necessary articles, incorrect subject-verb \nagreement, fractured sentences, and odd or inappropriate word choices. The bibliography contains \nmany incomplete or incorrectly formatted citations. There are too many instances of each of these \nissues to cite individually here, but in total they make the manuscript distracting and annoying to \nread. I am sensitive to and appreciate English-as-a-second-language challenges, but it is difficult \nto believe that all of the authors read and approved this version of the manuscript or that other \nediting services are not available at Baylor. \n\nReviewer #2 (Remarks to the Author); expert in cell cycle, DNA damage: \n\nThe authors build on the work of another group (Beata Jonik-Nowak et al, Proc Natl Acad Sci, \n2018, 115:E6477; PMC6048556) who showed that PIP30/FAM192A (here called NIP30) is \nphosphorylated by CKII, and that phosphorylated PIP30/NIP30 binds and regulates the function of \nREGgamma. Here Gao et al confirm those findings and further: \n1) Show that overexpression of wild-type and phosphomimetic (4A) PIP30/NIP30 \nposttranscriptionally affects the levels of p21, which is a substrate of REGgamma. \n2) Show that overexpression of nonphosphorylatable (4D) PIP30/NIP30 reverses the \nantiproliferative effects of REGgamma knockout \n3) Claim that CDC25A dephosphorylates PIP30/NIP30, which in turn regulates p21 levels in a p53-\nindependent manner. \n4) Identify that PIP30/NIP30 is altered in a low percentage of tumors \n5) Show that overexpression of wild-type and phosphophomimetic PIP30/NIP30 increases tumor \nsensitivity to etoposide, cisplatin, 5-fluorouracil, and doxorubicin \n6) Claim that PIP30/NIP30 is a tumor suppressor \n7) Present a model in which CDC25A regulates NIP30/PIP30, which in turn regulates REGgamma \nthat then regulates p21 levels and cell cycle progression. \n\nThis is an interesting study. However, there are several serious concerns. \n\n1) Although the results showing that altering CDC25A levels affects PIP30/NIP30 phosphorylation \nin cells is solid, the data showing that CDC25A directly de-phosphorylates PIP30/NIP30 is very \nweak (Fig. 3E); this analysis used immunopurified PIP30/NIP30 and immunopurified CDC25A. It is \nvery possible that a contaminating phosphatase is responsible for the in vitro dephosphorylation \nthat is observed. \n2) It is not clear that CDC25A’s effects on p21 levels are driven by CDC25A-mediated \ndephosphorylation of PIP30/NIP30. They could be through another unknown pathway. Does \nexpression of phosphomimetic PIP30/NIP30 prevent CDC25A-induced changes in p21? \n3) Most of the experiments rely on PIP30/NIP30 overexpression. This is a serious concern because \noverexpression studies are prone to artefacts. Additionally, given that the authors propose that \nPIP30/NIP30 is a tumor suppressor, a serious analysis of the effects of losing PIP30/NIP30 would \nbe important to understanding its functions. \n4) PIP30/NIP30 mutations found in tumors are very infrequent and the effects mutations that are \nshown in the supplemental figure are not known. Accordingly, there are no data to support the \ncontention that PIP30/NIP30 is a tumor suppressor. \n5) Although p21 levels are affected by CDC25A and PIP30/NIP30, it is not clear that changes in \np21 levels are actually driving cell cycle changes induced by CDC25A and PIP30/NIP30 \nmanipulation. \n\nMinor comments: \n1) There are many grammatical errors. \n2) It is not always clear how many times an experiment was done. \n\n\f3) Abbreviations are not always defined. \n4) Figures are too small to read, especially annotations and symbols on graphs, even when \nzoomed in on. \n5) Some figures are not appropriately labeled. For example, in Fig. 6B, the x-axes are not labeled \nand the y-axes are labeled as “survival rate.” If an MTT assay was used, this is not a survival rate. \n\nReviewer #3 (Remarks to the Author); expert in p53, mouse models: \n\nThe manuscript by Gao, et al. focuses on NIP30 as an inhibitor of the REG proteasome, which is \noncogenic and distinct from the 26S proteasome. The authors uncovered a tumor suppressor role \nfor NIP30 by its interactions with REGmapping the amino acids of interaction and uncovering a \nphosphorylation switch for this interaction. CDC25A was identified as the major regulatory \nphosphatase for this inhibitory switch during cell cycle and in response to DNA damage. Cell cycle \ninhibitor p21 is a major target of the CDC25A-NIP30- REG axis, which induces regulation in a \np53-dependent manner. Multiple approaches were employed in these studies: IP/mass spec; yeast \ntwo-hybrid; co-IP; domain and amino acid interaction mapping; xenograft tumor analyses of \nexpressing, mutated and KD of key players; screening of a phosphatase library; cell cycle and DNA \ndamage studies and UV-response of the skin taken from newborn REG WT, KO, Tpr53 KO and \nREG;Tpr53 double KO; and chemosensitivity studies with cell lines and xenografts. \n\nOverall, this is a very thorough analysis that pinpoints a specific phosphorylation switch control of \nproteasome regulation of p21 during cell cycle and DNA damage response. The studies are \nprimarily cell line-based but are also complemented by xenograft, genetically modified mice and \nTCGA data. This work suggests a new target and approach toward effectively treating p53-\ndysfunctional cancers with chemotherapeutics, a major problem in the clinic. This finding is likely \nto be of considerable interest to the community. \n\nThe major problem with the manuscript is poor preparation, improper use of English and inexact \nterminology. The authors should employ an English editor and science writer to resubmit a better \nmanuscript. \n\nSpecific examples of inexact terminology include “tight binding”, found in the Abstract and within \nthe manuscript. How is “tight binding” measured and what are the values for binding constants \nthat would indicate “tight”? Quantifying this binding is needed for such terms to be used. Showing \nco-localization in immunofluorescence microscopy is insufficient and non-robust (Suppl. Fig. 3A). \n\nExamples of poor manuscript preparation are the numerous grammatical errors, incorrect word \nusage that obscures the likely intended meaning (a couple of examples: “difference cells”? “Hiked \nlevels”?), and some references improperly formatted. Additionally, some figure legends lack \nsufficient detail (Fig. 1A, what is N1 and S1, for example). References are sometimes lacking. A \nmajor example of this is the human Ser/Thr phosphatase library screen. There is no reference for \nthis library or any data for the screen shown. These data should be included as supplemental data \nor referenced. \n\n\f ", "role": "assistant" } ]	2,020
The REGγ inhibitor NIP30 increases sensitivity to chemotherapy in p53-deficient tumor cells	Reviewers' comments: Reviewer #1 (Remarks to the Author): This is a revised manuscript that examines the relative roles of CDC25, NIP30, and REGγ on proteasome-dependent degradation of p21 and consequent cell growth. As with the previous version of the manuscript, the results allow the authors to present a model whereby NIP30 regulates REGγ activated proteasome activity via CDC25 control of NIP30 phosphorylation status. This is an interesting and potentially important set of results. The authors have responded to my original concerns to improve the manuscript. However, in my opinion, their responses are not completely satisfactory and in most cases consist of minor additions to a largely unchanged text. For example, I thought that reference to previous work by others was not sufficiently acknowledged in the setup to this work and thought that a satisfactory response would include a more serious reworking of the introduction and first part of the results. Instead the authors have simply added references in several places. As noted in my original comments, I continue to think that a major unresolved issue here is the molecular mechanism by which REG gamma/NIP30 regulates proteasome action against protein substrates such as p21 and suggested an in vitro experiment that could directly examine this process and test the authors’ models. The authors have provided such an experiment (Supplemental Figure 3F), but it is incompletely described and therefore difficult to evaluate; there is no indication of relative concentrations of proteins in the assay, details of the incubation conditions, validation that the single time point is an appropriate measure of rate, etc. The use of in vitro translated (in what system?) p21 (instead of bacterially expressed and purified p21) introduces lysates into the system. Maybe it makes no difference, but who knows? The single p21- only “control” is not sufficient without a detailed description of the assay conditions and components. The model figure (6E) is somewhat confusing and perhaps misleading since there is no direct experimental evidence to show whether the effect of CDC25 action on NIP30 phosphorylation affects proteasome activity by blocking REGγ/NIP30 interaction in isolation or whether this occurs within the proteasome complex. This is an issue that could add considerable mechanistic detail to the work and could be tested directly, but is not done so here. Blanket statements in many figure legends, such as “All experiments were repeated three times.” are not entirely clear. Does that mean the complete xenograph experiment shown in Figure 6C and 6D was done independently three times? If so, why not combine all of the data? The presentation is improved but there are still many examples of grammatical errors and odd syntax throughout. Reviewer #2 (Remarks to the Author): My major concerns have been addressed. However, although the grammar and word usage have been much improved, the text is still rough. Also, the figures are still very small and very difficult to read. Reviewer #3 (Remarks to the Author): The resubmitted manuscript is much improved and addresses the majority of problems in the first submission. The major claims are substantiated and present new evidence regarding tumor suppressor and p53-independent functions. A few problems with wording remain in the Discussion: lines 495, 498, 527 and 542. Overall, a much improved submission.	Reviewers' comments: Reviewer #1 (Remarks to the Author): This is a revised manuscript that examines the relative roles of CDC25, NIP30, and REGγ on proteasome-dependent degradation of p21 and consequent cell growth. As with the previous version of the manuscript, the results allow the authors to present a model whereby NIP30 regulates REGγ activated proteasome activity via CDC25 control of NIP30 phosphorylation status. This is an interesting and potentially important set of results. The authors have responded to my original concerns to improve the manuscript. However, in my opinion, their responses are not completely satisfactory and in most cases consist of minor additions to a largely unchanged text. For example, I thought that reference to previous work by others was not sufficiently acknowledged in the setup to this work and thought that a satisfactory response would include a more serious reworking of the introduction and first part of the results. Instead the authors have simply added references in several places. As noted in my original comments, I continue to think that a major unresolved issue here is the molecular mechanism by which REG gamma/NIP30 regulates proteasome action against protein substrates such as p21 and suggested an in vitro experiment that could directly examine this process and test the authors’ models. The authors have provided such an experiment (Supplemental Figure 3F), but it is incompletely described and therefore difficult to evaluate; there is no indication of relative concentrations of proteins in the assay, details of the incubation conditions, validation that the single time point is an appropriate measure of rate, etc. The use of in vitro translated (in what system?) p21 (instead of bacterially expressed and purified p21) introduces lysates into the system. Maybe it makes no difference, but who knows? The single p21- only “control” is not sufficient without a detailed description of the assay conditions and components. The model figure (6E) is somewhat confusing and perhaps misleading since there is no direct experimental evidence to show whether the effect of CDC25 action on NIP30 phosphorylation affects proteasome activity by blocking REGγ/NIP30 interaction in isolation or whether this occurs within the proteasome complex. This is an issue that could add considerable mechanistic detail to the work and could be tested directly, but is not done so here. Blanket statements in many figure legends, such as “All experiments were repeated three times.” are not entirely clear. Does that mean the complete xenograph experiment shown in Figure 6C and 6D was done independently three times? If so, why not combine all of the data? The presentation is improved but there are still many examples of grammatical errors and odd syntax throughout. Reviewer #2 (Remarks to the Author): My major concerns have been addressed. However, although the grammar and word usage have been much improved, the text is still rough. Also, the figures are still very small and very difficult to read. Reviewer #3 (Remarks to the Author): The resubmitted manuscript is much improved and addresses the majority of problems in the first submission. The major claims are substantiated and present new evidence regarding tumor suppressor and p53-independent functions. A few problems with wording remain in the Discussion: lines 495, 498, 527 and 542. Overall, a much improved submission.	TPR	NC_BS_CAN_012_RR2.txt	{ "criticism": 11, "example": 5, "importance_and_relevance": 2, "materials_and_methods": 11, "praise": 5, "presentation_and_reporting": 10, "results_and_discussion": 4, "suggestion_and_solution": 2, "total": 26 }	{ "criticism": 0.4230769230769231, "example": 0.19230769230769232, "importance_and_relevance": 0.07692307692307693, "materials_and_methods": 0.4230769230769231, "praise": 0.19230769230769232, "presentation_and_reporting": 0.38461538461538464, "results_and_discussion": 0.15384615384615385, "suggestion_and_solution": 0.07692307692307693 }	1.923077	[ { "content": "Please review this paper.", "role": "user" }, { "content": "\n\n\fReviewers' comments: \n\nReviewer #1 (Remarks to the Author): \n\nThis is a revised manuscript that examines the relative roles of CDC25, NIP30, and REGγ on \nproteasome-dependent degradation of p21 and consequent cell growth. As with the previous \nversion of the manuscript, the results allow the authors to present a model whereby NIP30 \nregulates REGγ activated proteasome activity via CDC25 control of NIP30 phosphorylation status. \nThis is an interesting and potentially important set of results. \n\nThe authors have responded to my original concerns to improve the manuscript. However, in my \nopinion, their responses are not completely satisfactory and in most cases consist of minor \nadditions to a largely unchanged text. For example, I thought that reference to previous work by \nothers was not sufficiently acknowledged in the setup to this work and thought that a satisfactory \nresponse would include a more serious reworking of the introduction and first part of the results. \nInstead the authors have simply added references in several places. \n\nAs noted in my original comments, I continue to think that a major unresolved issue here is the \nmolecular mechanism by which REG gamma/NIP30 regulates proteasome action against protein \nsubstrates such as p21 and suggested an in vitro experiment that could directly examine this \nprocess and test the authors’ models. The authors have provided such an experiment \n(Supplemental Figure 3F), but it is incompletely described and therefore difficult to evaluate; there \nis no indication of relative concentrations of proteins in the assay, details of the incubation \nconditions, validation that the single time point is an appropriate measure of rate, etc. The use of \nin vitro translated (in what system?) p21 (instead of bacterially expressed and purified p21) \nintroduces lysates into the system. Maybe it makes no difference, but who knows? The single p21-\nonly “control” is not sufficient without a detailed description of the assay conditions and \ncomponents. \n\nThe model figure (6E) is somewhat confusing and perhaps misleading since there is no direct \nexperimental evidence to show whether the effect of CDC25 action on NIP30 phosphorylation \naffects proteasome activity by blocking REGγ/NIP30 interaction in isolation or whether this occurs \nwithin the proteasome complex. This is an issue that could add considerable mechanistic detail to \nthe work and could be tested directly, but is not done so here. \n\nBlanket statements in many figure legends, such as “All experiments were repeated three times.” \nare not entirely clear. Does that mean the complete xenograph experiment shown in Figure 6C and \n6D was done independently three times? If so, why not combine all of the data? \n\nThe presentation is improved but there are still many examples of grammatical errors and odd \nsyntax throughout. \n\nReviewer #2 (Remarks to the Author): \n\nMy major concerns have been addressed. However, although the grammar and word usage have \nbeen much improved, the text is still rough. Also, the figures are still very small and very difficult \nto read. \n\nReviewer #3 (Remarks to the Author): \n\nThe resubmitted manuscript is much improved and addresses the majority of problems in the first \nsubmission. The major claims are substantiated and present new evidence regarding tumor \n\n\fsuppressor and p53-independent functions. \n\nA few problems with wording remain in the Discussion: lines 495, 498, 527 and 542. \n\nOverall, a much improved submission.\n\n\f \n\n \n\n \n\n", "role": "assistant" } ]	2,020
MGMT genomic rearrangements contribute to chemotherapy resistance in gliomas	REVIEWERS' COMMENTS: Reviewer #1 (Remarks to the Author): I have no further issues with this manuscript. I commend the authors on a job well done. Reviewer #2 (Remarks to the Author): my previous concerns have been adequately addressed	REVIEWERS' COMMENTS: Reviewer #1 (Remarks to the Author): I have no further issues with this manuscript. I commend the authors on a job well done. Reviewer #2 (Remarks to the Author): my previous concerns have been adequately addressed	TPR	NC_BS_CAN_014_RR2.txt	{ "criticism": 1, "example": 0, "importance_and_relevance": 0, "materials_and_methods": 0, "praise": 2, "presentation_and_reporting": 0, "results_and_discussion": 0, "suggestion_and_solution": 0, "total": 3 }	{ "criticism": 0.3333333333333333, "example": 0, "importance_and_relevance": 0, "materials_and_methods": 0, "praise": 0.6666666666666666, "presentation_and_reporting": 0, "results_and_discussion": 0, "suggestion_and_solution": 0 }	1	[ { "content": "Please review this paper.", "role": "user" }, { "content": "\t\n\n\fREVIEWERS' COMMENTS: \n \nReviewer #1 (Remarks to the Author): \n \nI have no further issues with this manuscript. I commend the authors on a job well done. \n \n \nReviewer #2 (Remarks to the Author): \n \nmy previous concerns have been adequately addressed \n\n\f\n", "role": "assistant" } ]	2,020
MGMT genomic rearrangements contribute to chemotherapy resistance in gliomas	Reviewers' comments: Reviewer #1 (Remarks to the Author): This manuscript describes a potential resistance mechanism in temozolomide-treated glioblastoma (GBM). In a genomic profiling of recurrent GBMs, the authors identify fusions involving the DNA repair enzyme MGMT, associated with a small percentage of tumors. They find that MGMT fusions are mutually exclusive with MGMT hypomethylation and global hypermutation, other established resistance mechanisms. They go on to perform functional analysis in U251 an U87 cells. They find that expression of MGMT fusions increases MGMT expression and decreases sensitivity to temozolomide, and that these phenomena are correlated. Moreover, MGMT fusion expression increases temozolomide resistance in a murine xenograft model. Finally, they demonstate that MGMT fusions are shed in exosomes, and can be detected in peripheral blood, pointing to a potential method for clinical screening. While the small number of MGMT-fused cases does detract somewhat from the significance of this work, it is a well-performed study describing a novel mechanisms of resistance in a deadly tumor type. My comments are below. 1) The reliance of the author’s functional analyses on U87 and U251 cells is somewhat problematic. These lines have been subjected to extensive culture over the years and may not represent the best models of glioma biology. Many patient-derived glioma stem cells (GSCs) are currently available and recapitulate the broad spectrum of molecular features characterizing gliomas in general. The authors should recapitulate their key in vitro and in vivo findings using GSCs. The authors might also consider using the RCAS/tv-a system, which they have leveraged very effectively in the past, the address some of their hypotheses in vivo. 2) Recurrent glioma specimens are notoriously heterogeneous in their composition with regard to tumor content, reactive tissue, necrotizing treatment effect, and inflammatory cells. This heterogeneity could impact the detection of genomic factors associated with resistance, including MGMT-fusions. Can the authors provide more information as to the histopathological and molecular QC measures used in their profiling studies? Was there a tumor content threshold? Who did the analysis? What were the purity/ploidy measurements of their samples? More minor 1) In the introduction, the authors should better explain the distinction between IDH-wild type and IDH-mutant glioma. General readers will not appreciate this difference. They should indicate early in the paper that malignant gliomas come in two basic varieties, IDH-mutant and IDH-wild type, and both are treated with temozolomide as the standard of care. 2) On a related note, the authors should consider presenting their genomic data more in the context of IDH-mutant vs IDH-wild type glioma, rather than HGG vs LGG. Reviewer #2 (Remarks to the Author): The manuscript “MGMT Genomic rearrangements...” submitted by Oldrini et al describes studies addressing a new form of MGMT genetic alteration in glioma that may contribute to TMZ resistance. In this study the authors show that in a small percentage of these patients, perhaps 3%, fusions of the MGMT gene with the 5’ UTR of other genes results in higher level MGMT expression that in turn appears associated with TMZ resistance. In terms of novelty this is an interesting and unique finding, and although as the authors note a similar MGMT fusion has been noted in a single medulloblastoma, the full range of such fusions in glioma has not been previously described. The work is also of interest in that there is growing interest in promoter highjacking and fusion transcripts as drivers in cancer, although this mechanism has not in the past been shown to involve MGMT. The work is also of importance as TMZ remains a mainline therapeutic for most forms of glioma, and resistance to TMZ and recurrence is common. As such the identification of a new mechanism of resistance and possibly a biomarker of this resistance, if even in a small percentage of patients is of interest to a broad range of investigators. In terms of the studies themselves, they appear carefully done, appropriately controlled, and adequately statistically evaluated. There are no instances in which the data presented do not support the conclusions reached. I have only a few questions regarding the work as presented: 1. The authors note that they have analyzed “hypermutation” in the samples and found that it was mutually exclusive with MGMT fusion. It is not clear however what type of “hypermutation” the authors are considering. Is this hypermutation the global type of hypermutation noted in TMZ- treated recurrent GBM or are the authors talking about MGMT hypermutation? If the latter, is this MGMT gene or promoter hypermutation? One really cant tell from the text or from the description provided in the methods section. This is an important point as most readers would be interested in whether MGMT fusion was mutually exclusive with TMZ-induced genomic hypermutation (as one might expect it to be). 2. Im also a little unclear about the generation of the cells carrying MGMT fusions by CRISPR. For such experiments one would typically use CRISPR-Cas to create clones, then sort through the clones to find those with the desired recombination in one or both copies of the targeted genes. Line 122 in the text does not clearly describe how the process was carried out, and the suggestion that the end result of the initial engineering was “a mixed population of cells carrying different fusion events” is puzzling. Additionally the further use of populations of cells that arose as a result of TMZ selection is also a little puzzling as one would expect all the appropriately engineered cells to be TMZ resistant. This entire creation/selection process needs to be better described.	Reviewers' comments: Reviewer #1 (Remarks to the Author): This manuscript describes a potential resistance mechanism in temozolomide-treated glioblastoma (GBM). In a genomic profiling of recurrent GBMs, the authors identify fusions involving the DNA repair enzyme MGMT, associated with a small percentage of tumors. They find that MGMT fusions are mutually exclusive with MGMT hypomethylation and global hypermutation, other established resistance mechanisms. They go on to perform functional analysis in U251 an U87 cells. They find that expression of MGMT fusions increases MGMT expression and decreases sensitivity to temozolomide, and that these phenomena are correlated. Moreover, MGMT fusion expression increases temozolomide resistance in a murine xenograft model. Finally, they demonstate that MGMT fusions are shed in exosomes, and can be detected in peripheral blood, pointing to a potential method for clinical screening. While the small number of MGMT-fused cases does detract somewhat from the significance of this work, it is a well-performed study describing a novel mechanisms of resistance in a deadly tumor type. My comments are below. 1) The reliance of the author’s functional analyses on U87 and U251 cells is somewhat problematic. These lines have been subjected to extensive culture over the years and may not represent the best models of glioma biology. Many patient-derived glioma stem cells (GSCs) are currently available and recapitulate the broad spectrum of molecular features characterizing gliomas in general. The authors should recapitulate their key in vitro and in vivo findings using GSCs. The authors might also consider using the RCAS/tv-a system, which they have leveraged very effectively in the past, the address some of their hypotheses in vivo. 2) Recurrent glioma specimens are notoriously heterogeneous in their composition with regard to tumor content, reactive tissue, necrotizing treatment effect, and inflammatory cells. This heterogeneity could impact the detection of genomic factors associated with resistance, including MGMT-fusions. Can the authors provide more information as to the histopathological and molecular QC measures used in their profiling studies? Was there a tumor content threshold? Who did the analysis? What were the purity/ploidy measurements of their samples? More minor 1) In the introduction, the authors should better explain the distinction between IDH-wild type and IDH-mutant glioma. General readers will not appreciate this difference. They should indicate early in the paper that malignant gliomas come in two basic varieties, IDH-mutant and IDH-wild type, and both are treated with temozolomide as the standard of care. 2) On a related note, the authors should consider presenting their genomic data more in the context of IDH-mutant vs IDH-wild type glioma, rather than HGG vs LGG. Reviewer #2 (Remarks to the Author): The manuscript “MGMT Genomic rearrangements...” submitted by Oldrini et al describes studies addressing a new form of MGMT genetic alteration in glioma that may contribute to TMZ resistance. In this study the authors show that in a small percentage of these patients, perhaps 3%, fusions of the MGMT gene with the 5’ UTR of other genes results in higher level MGMT expression that in turn appears associated with TMZ resistance. In terms of novelty this is an interesting and unique finding, and although as the authors note a similar MGMT fusion has been noted in a single medulloblastoma, the full range of such fusions in glioma has not been previously described. The work is also of interest in that there is growing interest in promoter highjacking and fusion transcripts as drivers in cancer, although this mechanism has not in the past been shown to involve MGMT. The work is also of importance as TMZ remains a mainline therapeutic for most forms of glioma, and resistance to TMZ and recurrence is common. As such the identification of a new mechanism of resistance and possibly a biomarker of this resistance, if even in a small percentage of patients is of interest to a broad range of investigators. In terms of the studies themselves, they appear carefully done, appropriately controlled, and adequately statistically evaluated. There are no instances in which the data presented do not support the conclusions reached. I have only a few questions regarding the work as presented: 1. The authors note that they have analyzed “hypermutation” in the samples and found that it was mutually exclusive with MGMT fusion. It is not clear however what type of “hypermutation” the authors are considering. Is this hypermutation the global type of hypermutation noted in TMZ- treated recurrent GBM or are the authors talking about MGMT hypermutation? If the latter, is this MGMT gene or promoter hypermutation? One really cant tell from the text or from the description provided in the methods section. This is an important point as most readers would be interested in whether MGMT fusion was mutually exclusive with TMZ-induced genomic hypermutation (as one might expect it to be). 2. Im also a little unclear about the generation of the cells carrying MGMT fusions by CRISPR. For such experiments one would typically use CRISPR-Cas to create clones, then sort through the clones to find those with the desired recombination in one or both copies of the targeted genes. Line 122 in the text does not clearly describe how the process was carried out, and the suggestion that the end result of the initial engineering was “a mixed population of cells carrying different fusion events” is puzzling. Additionally the further use of populations of cells that arose as a result of TMZ selection is also a little puzzling as one would expect all the appropriately engineered cells to be TMZ resistant. This entire creation/selection process needs to be better described.	TPR	NC_BS_CAN_014_RR1.txt	{ "criticism": 7, "example": 1, "importance_and_relevance": 6, "materials_and_methods": 33, "praise": 5, "presentation_and_reporting": 5, "results_and_discussion": 6, "suggestion_and_solution": 8, "total": 45 }	{ "criticism": 0.15555555555555556, "example": 0.022222222222222223, "importance_and_relevance": 0.13333333333333333, "materials_and_methods": 0.7333333333333333, "praise": 0.1111111111111111, "presentation_and_reporting": 0.1111111111111111, "results_and_discussion": 0.13333333333333333, "suggestion_and_solution": 0.17777777777777778 }	1.577778	[ { "content": "Please review this paper.", "role": "user" }, { "content": "Reviewers' comments:\nReviewer #1 (Remarks to the Author):\nThis manuscript describes a potential resistance mechanism in temozolomide-treated glioblastoma\n(GBM). In a genomic profiling of recurrent GBMs, the authors identify fusions involving the DNA\nrepair enzyme MGMT, associated with a small percentage of tumors. They find that MGMT fusions\nare mutually exclusive with MGMT hypomethylation and global hypermutation, other established\nresistance mechanisms. They go on to perform functional analysis in U251 an U87 cells. They find\nthat expression of MGMT fusions increases MGMT expression and decreases sensitivity to\ntemozolomide, and that these phenomena are correlated. Moreover, MGMT fusion expression\nincreases temozolomide resistance in a murine xenograft model. Finally, they demonstate that\nMGMT fusions are shed in exosomes, and can be detected in peripheral blood, pointing to a\npotential method for clinical screening.\nWhile the small number of MGMT-fused cases does detract somewhat from the significance of this\n work, it is a well-performed study describing a novel mechanisms of resistance in a deadly tumor\n type.\n My comments are below.\n1) The reliance of the author’s functional analyses on U87 and U251 cells is somewhat\nproblematic. These lines have been subjected to extensive culture over the years and may not\nrepresent the best models of glioma biology. Many patient-derived glioma stem cells (GSCs) are\ncurrently available and recapitulate the broad spectrum of molecular features characterizing\ngliomas in general. The authors should recapitulate their key in vitro and in vivo findings using\nGSCs. The authors might also consider using the RCAS/tv-a system, which they have leveraged\nvery effectively in the past, the address some of their hypotheses in vivo.\n2) Recurrent glioma specimens are notoriously heterogeneous in their composition with regard to\ntumor content, reactive tissue, necrotizing treatment effect, and inflammatory cells. This\nheterogeneity could impact the detection of genomic factors associated with resistance, including\nMGMT-fusions. Can the authors provide more information as to the histopathological and molecular\nQC measures used in their profiling studies? Was there a tumor content threshold? Who did the\nanalysis? What were the purity/ploidy measurements of their samples?\nMore minor\n1) In the introduction, the authors should better explain the distinction between IDH-wild type and\nIDH-mutant glioma. General readers will not appreciate this difference. They should indicate early\nin the paper that malignant gliomas come in two basic varieties, IDH-mutant and IDH-wild type,\nand both are treated with temozolomide as the standard of care.\n2) On a related note, the authors should consider presenting their genomic data more in the\n context of IDH-mutant vs IDH-wild type glioma, rather than HGG vs LGG.\n Reviewer #2 (Remarks to the Author):\nThe manuscript “MGMT Genomic rearrangements...” submitted by Oldrini et al describes studies\naddressing a new form of MGMT genetic alteration in glioma that may contribute to TMZ\nresistance. In this study the authors show that in a small percentage of these patients, perhaps\n3%, fusions of the MGMT gene with the 5’ UTR of other genes results in higher level MGMT\nexpression that in turn appears associated with TMZ resistance. In terms of novelty this is an\ninteresting and unique finding, and although as the authors note a similar MGMT fusion has been\n noted in a single medulloblastoma, the full range of such fusions in glioma has not been previously\n described. The work is also of interest in that there is growing interest in promoter highjacking and\n fusion transcripts as drivers in cancer, although this mechanism has not in the past been shown to\n involve MGMT. The work is also of importance as TMZ remains a mainline therapeutic for most\n forms of glioma, and resistance to TMZ and recurrence is common. As such the identification of a\n new mechanism of resistance and possibly a biomarker of this resistance, if even in a small\n percentage of patients is of interest to a broad range of investigators.\n In terms of the studies themselves, they appear carefully done, appropriately controlled, and\n adequately statistically evaluated. There are no instances in which the data presented do not\n support the conclusions reached. I have only a few questions regarding the work as presented:\n 1. The authors note that they have analyzed “hypermutation” in the samples and found that it was\n mutually exclusive with MGMT fusion. It is not clear however what type of “hypermutation” the\n authors are considering. Is this hypermutation the global type of hypermutation noted in TMZ-\n treated recurrent GBM or are the authors talking about MGMT hypermutation? If the latter, is this\n MGMT gene or promoter hypermutation? One really cant tell from the text or from the description\n provided in the methods section. This is an important point as most readers would be interested in\n whether MGMT fusion was mutually exclusive with TMZ-induced genomic hypermutation (as one\n might expect it to be).\n 2. Im also a little unclear about the generation of the cells carrying MGMT fusions by CRISPR. For\n such experiments one would typically use CRISPR-Cas to create clones, then sort through the\n clones to find those with the desired recombination in one or both copies of the targeted genes.\n Line 122 in the text does not clearly describe how the process was carried out, and the suggestion\n that the end result of the initial engineering was “a mixed population of cells carrying different\n fusion events” is puzzling. Additionally the further use of populations of cells that arose as a result\n of TMZ selection is also a little puzzling as one would expect all the appropriately engineered cells\n to be TMZ resistant. This entire creation/selection process needs to be better described.\n ", "role": "assistant" } ]	2,020
European polygenic risk score for prediction of breast cancer shows similar performance in	Reviewers' comments: Reviewer #1 (Remarks to the Author): This manuscript presents a very large and fairly comprehensive analysis studying the application of a polygenic risk score derived using data from European women to characterize risk in Asian women. The components of the paper are well developed but the actual presentation has many issues and the paper needs a great deal of revision for clarity. I have many minor comments and two more major concerns. Major question – it is not very clear that genetic analysis to identify 1% of women at polygenic risk would have clinical utility or be feasible. If the PRS is set to the top 5% does that provide an Odds ratio that is meaningful or a prospective risk that would lead to different management of women? Some sort of sensitivity analysis considering at least an upper 5% threshold and perhaps an upper 10% threshold should be included. What is the AUC when you include versus exclude Family History. Table 3 gives the attenuation in OR when including family history but the more interesting element to me would be to evaluate what the AUC is when you include markers and Family history. I think it would be useful to also plot the distribution of risk according to quantiles of risk and according to age. This would better illustrate the lack of information from individuals at low PRS across all ages. Minor comments: Line 143. Should read ‘mutations in these genes are rare’ not these genes are rare, since almost everyone has two copies of these genes (except rare people with deletions). Line 157 – this discussion is not very comprehensive. While it may be true that PRS has some utility how much does it improve beyond traditional and early collect information like family history? Lines 192-197 it would be helpful to have some information about what the interpretation of the PRS mean score would be. Is this score somehow standardized? If not then should it be to make the interpretation more straightforward? Lines 226-230 gives changes in percent attenuation without indicating what is being attenuated or what the relevance of this sentence is. Please clarify this section without requiring close perusal of the table since the sentence makes no sense as written. Lines 249-251. The actual AUCs should be cited in the text where it is being highlighted. These sentences do not make sense. Smaller than what? Since this sentence starts a new paragraph there is no referent for that, so the sentence is uninterpretable. ‘The OR per SD was significantly smaller (p < 0.05) than that in the Asian studies from Asia, 267 for overall breast cancer (1.36, 95% CI = 1.25 – 1.49) and ER-positive breast cancer (1.38, 268 95% CI = 1.25 – 1.53), but significant higher (p < 0.05) for ER-negative breast cancer (1.49,269 95% CI = 1.26 – 1.76, Table 2(b)).’ Around lines 284 – where do the referent risks come from for populations that are being described? It does not sound as if population risks for breast cancer are being derived from cohort studies but rather from a population database but there is no reference so it is impossible to tell how these analyses were conducted. Line 339 – effect size of PRS - the authors do not define their threshold. Do they mean top 1% of Asian PRS as a cut off? Do they mean top 1% of European? Line 347 – lower predictive accuracy in Asians- the author need to point to a table in this discussion and give some values since this information was not previously cited. Or it could be moved to the results and described more fully (which would be preferable). The discussion cites that a screening regimen based on PRS and age could reduce numbers of women undergoing screening and cites a prior paper but shouldn’t they also derive these numbers from their own analysis? Line 372 to 375 – this presents some interesting observations that have not been described previously and do not cite a figure or table so are uninterpretable. Lines 381-387- great that simulations were done. Please present the results in a figure and or table. Otherwise this is unacceptable presentation of unsupported information. Lines 406-410. Did the authors check for heterogeneity in the ORs for family history by study? Opportunistic control selection could definitely lead to enrichment of FH positive controls. That might also be yield a similar frequency of FH in case populations and variability in FH+ controls, which could be tested for. 432-435 – the sentence structure in this section is weird and confusing e.g. ‘All three studies are population-based case-control studies; and (c) 10,266 women 433 of Chinese ethnicity participating in Singapore Chinese Health Study (SCHS)’ In this section you have a period break before All so it is not very clear that you are referring to the prior sentence and then you join this unclear phrase the next unclear phrase after a semicolon. Please break unrelated phrases into sentences and join related clauses to the antecedent sentences. Line 472 ‘Post-imputation quality of all studies was based on the IMPUTE imputation quality score.’ This should not be a guessing game. Please describe what these imputation quality metrics are or explain them further in a supplementary methods that are referenced. Lines 485-490 the English is missing plurals and articles so should be revised. It is great to see a reference for how the PRS was standarised, but I think this information should also be placed in the text where the means of PRS are presented. Line 491-495 – the manuscript indicates that the PRS was stratified into quartiles and the middle quintile was used for referent but that is not very clearly relevant to the paper where top 1% of distribution was used. I think in this section the description is fairly clear but where the 1% cut off is described you need to cite that you are using the middle quintile as referent. Line 507-509 – AUC analyses were adjusted by study where relevant. This needs to be explained further. Lines 515-519 please give a reference for why this formula is appropriate. The references on line 551 should be cited in the main body of the paper as requested above. For reproducibility it would be better to specify which modules or R or Stata were used for major analyses rather than the very generic reference that is given. Table 2 needs to cite in the footnote that the cutpoint is top 1% of the PRS in controls if that is what was used for stratifying high risk individuals. Reviewer #2 (Remarks to the Author): Ho and colleges evaluated the performance of a European-derived polygenic risk score (PRS) in an Asian population of 17,262 breast cancer cases and 17,695 controls. Most of these women were living in Asian countries, with a smaller population of Asian-Americans (1,507 cases and 1,212 controls). In addition, they evaluated the PRS performance in 10,255 Chinese women from a prospective cohort (413 incident breast cancers). This is a well-conducted study with findings that have public health relevance since it provides estimates for polygenic breast cancer risk in Asian women that could be used for risk-stratified screening/prevention. The main finding is that the European-derived PRS can provide substantial risk stratification of Asian women, however the stratification is lower than for European-ancestry women. Although this is the largest study conducted in Asians to date, as the authors pointed out, even larger studies are needed to obtain precise OR estimates at the extreme of the PRS distribution, or to further improve the performance of PRS in Asian women. Below are some comments/questions for consideration: Q1: The authors used PRS weights based on estimates from European ancestry populations, rather than being re-estimated in Asian populations. This is a reasonable approach as the sample sizes to estimate weights are much larger in Europeans and the authors showed evidence of consistent effect size for genetic variants between European and Asian populations. However, given that previous publications have attempted to derived Asian-specific PRS, it would be of interest to directly compare the prediction performance between previous Asian-specific PRS (derived with lower sample size) and the European based PRS. Q2: Did the authors consider incorporating Asian-specific risk variants that are not in the 313-PRSs developed for European populations? Q3. The authors evaluated the PRS based on 313 variants reported in Mavaddat et al. 2019. However, 26 of the 313 variants were excluded due to low imputation accuracy score in Asians using the 1000 genome reference panel, resulting in a 287-variant PRS. This is a limitation since it does not allow direct comparison with the previously published PRS. Did the author consider imputing the genotypes based on larger reference panels such as HRC and TopMed ? Q4: For the Singapore Chinese Health Study (SCHS), only 229 of the 287 variants that were found to be polymorphic and imputable. Although this is a valuable study because of its prospective nature and information on ancestry, it is a bit confusing to show both the 287-SNP PRS and 229- SNP PRS as main findings. I suggest moving the 229-SNP PRS findings to supplement or make clearer what is the added value of these analyses as main findings. I do not see much value added on tables 1 or 2. Findings on Tables 4 and 5 by ancestry are of interest, are PRS means (SD) on Table 4 based on the 229 or 287 variant PRS? Q5. Are the principal components calculated separately for the 10 Asian studies in BCAC and the three Asian-American studies? Q6: The authors find a smaller RR for family history than previously reported in European-ancestry studies. Are there previous studies in Asian populations that support this difference? Q7: The 287-variant PRS was less predictive of breast cancer risk in the three Asian-American studies than in the studies conducted in Asian (which found no differences between Asians of Chinese, Malay or Indian origin). This is a potentially important finding that requires confirmation in larger studies. Could the authors formally test if these differences are statistically significant in the current population? Based on these findings, can the authors comment on what needs to be done to use PRS in Asian populations in the US? Or other countries? Minor comments: Q1: In the imputation for BCAC, the phase of the 1000 genome project should be specified. Q2: The EM algorithm needs citation Q3: The version of R and Stata needs to be listed Q4: Indicate population/sample size included in each table/figure or a reference to a table/figure showing the population/sample size. Q5: Table 1- change format to match Table 2 (with less blank cells). Q6: Consider adding the European estimates to Figure 2 Q7: Consider adding estimates from American Asians to Figure 3. Q8: Tables 3 and 4- indicate what PRS is shown – 287-SNP PRS?	Reviewers' comments: Reviewer #1 (Remarks to the Author): This manuscript presents a very large and fairly comprehensive analysis studying the application of a polygenic risk score derived using data from European women to characterize risk in Asian women. The components of the paper are well developed but the actual presentation has many issues and the paper needs a great deal of revision for clarity. I have many minor comments and two more major concerns. Major question – it is not very clear that genetic analysis to identify 1% of women at polygenic risk would have clinical utility or be feasible. If the PRS is set to the top 5% does that provide an Odds ratio that is meaningful or a prospective risk that would lead to different management of women? Some sort of sensitivity analysis considering at least an upper 5% threshold and perhaps an upper 10% threshold should be included. What is the AUC when you include versus exclude Family History. Table 3 gives the attenuation in OR when including family history but the more interesting element to me would be to evaluate what the AUC is when you include markers and Family history. I think it would be useful to also plot the distribution of risk according to quantiles of risk and according to age. This would better illustrate the lack of information from individuals at low PRS across all ages. Minor comments: Line 143. Should read ‘mutations in these genes are rare’ not these genes are rare, since almost everyone has two copies of these genes (except rare people with deletions). Line 157 – this discussion is not very comprehensive. While it may be true that PRS has some utility how much does it improve beyond traditional and early collect information like family history? Lines 192-197 it would be helpful to have some information about what the interpretation of the PRS mean score would be. Is this score somehow standardized? If not then should it be to make the interpretation more straightforward? Lines 226-230 gives changes in percent attenuation without indicating what is being attenuated or what the relevance of this sentence is. Please clarify this section without requiring close perusal of the table since the sentence makes no sense as written. Lines 249-251. The actual AUCs should be cited in the text where it is being highlighted. These sentences do not make sense. Smaller than what? Since this sentence starts a new paragraph there is no referent for that, so the sentence is uninterpretable. ‘The OR per SD was significantly smaller (p < 0.05) than that in the Asian studies from Asia, 267 for overall breast cancer (1.36, 95% CI = 1.25 – 1.49) and ER-positive breast cancer (1.38, 268 95% CI = 1.25 – 1.53), but significant higher (p < 0.05) for ER-negative breast cancer (1.49,269 95% CI = 1.26 – 1.76, Table 2(b)).’ Around lines 284 – where do the referent risks come from for populations that are being described? It does not sound as if population risks for breast cancer are being derived from cohort studies but rather from a population database but there is no reference so it is impossible to tell how these analyses were conducted. Line 339 – effect size of PRS - the authors do not define their threshold. Do they mean top 1% of Asian PRS as a cut off? Do they mean top 1% of European? Line 347 – lower predictive accuracy in Asians- the author need to point to a table in this discussion and give some values since this information was not previously cited. Or it could be moved to the results and described more fully (which would be preferable). The discussion cites that a screening regimen based on PRS and age could reduce numbers of women undergoing screening and cites a prior paper but shouldn’t they also derive these numbers from their own analysis? Line 372 to 375 – this presents some interesting observations that have not been described previously and do not cite a figure or table so are uninterpretable. Lines 381-387- great that simulations were done. Please present the results in a figure and or table. Otherwise this is unacceptable presentation of unsupported information. Lines 406-410. Did the authors check for heterogeneity in the ORs for family history by study? Opportunistic control selection could definitely lead to enrichment of FH positive controls. That might also be yield a similar frequency of FH in case populations and variability in FH+ controls, which could be tested for. 432-435 – the sentence structure in this section is weird and confusing e.g. ‘All three studies are population-based case-control studies; and (c) 10,266 women 433 of Chinese ethnicity participating in Singapore Chinese Health Study (SCHS)’ In this section you have a period break before All so it is not very clear that you are referring to the prior sentence and then you join this unclear phrase the next unclear phrase after a semicolon. Please break unrelated phrases into sentences and join related clauses to the antecedent sentences. Line 472 ‘Post-imputation quality of all studies was based on the IMPUTE imputation quality score.’ This should not be a guessing game. Please describe what these imputation quality metrics are or explain them further in a supplementary methods that are referenced. Lines 485-490 the English is missing plurals and articles so should be revised. It is great to see a reference for how the PRS was standarised, but I think this information should also be placed in the text where the means of PRS are presented. Line 491-495 – the manuscript indicates that the PRS was stratified into quartiles and the middle quintile was used for referent but that is not very clearly relevant to the paper where top 1% of distribution was used. I think in this section the description is fairly clear but where the 1% cut off is described you need to cite that you are using the middle quintile as referent. Line 507-509 – AUC analyses were adjusted by study where relevant. This needs to be explained further. Lines 515-519 please give a reference for why this formula is appropriate. The references on line 551 should be cited in the main body of the paper as requested above. For reproducibility it would be better to specify which modules or R or Stata were used for major analyses rather than the very generic reference that is given. Table 2 needs to cite in the footnote that the cutpoint is top 1% of the PRS in controls if that is what was used for stratifying high risk individuals. Reviewer #2 (Remarks to the Author): Ho and colleges evaluated the performance of a European-derived polygenic risk score (PRS) in an Asian population of 17,262 breast cancer cases and 17,695 controls. Most of these women were living in Asian countries, with a smaller population of Asian-Americans (1,507 cases and 1,212 controls). In addition, they evaluated the PRS performance in 10,255 Chinese women from a prospective cohort (413 incident breast cancers). This is a well-conducted study with findings that have public health relevance since it provides estimates for polygenic breast cancer risk in Asian women that could be used for risk-stratified screening/prevention. The main finding is that the European-derived PRS can provide substantial risk stratification of Asian women, however the stratification is lower than for European-ancestry women. Although this is the largest study conducted in Asians to date, as the authors pointed out, even larger studies are needed to obtain precise OR estimates at the extreme of the PRS distribution, or to further improve the performance of PRS in Asian women. Below are some comments/questions for consideration: Q1: The authors used PRS weights based on estimates from European ancestry populations, rather than being re-estimated in Asian populations. This is a reasonable approach as the sample sizes to estimate weights are much larger in Europeans and the authors showed evidence of consistent effect size for genetic variants between European and Asian populations. However, given that previous publications have attempted to derived Asian-specific PRS, it would be of interest to directly compare the prediction performance between previous Asian-specific PRS (derived with lower sample size) and the European based PRS. Q2: Did the authors consider incorporating Asian-specific risk variants that are not in the 313-PRSs developed for European populations? Q3. The authors evaluated the PRS based on 313 variants reported in Mavaddat et al. 2019. However, 26 of the 313 variants were excluded due to low imputation accuracy score in Asians using the 1000 genome reference panel, resulting in a 287-variant PRS. This is a limitation since it does not allow direct comparison with the previously published PRS. Did the author consider imputing the genotypes based on larger reference panels such as HRC and TopMed ? Q4: For the Singapore Chinese Health Study (SCHS), only 229 of the 287 variants that were found to be polymorphic and imputable. Although this is a valuable study because of its prospective nature and information on ancestry, it is a bit confusing to show both the 287-SNP PRS and 229- SNP PRS as main findings. I suggest moving the 229-SNP PRS findings to supplement or make clearer what is the added value of these analyses as main findings. I do not see much value added on tables 1 or 2. Findings on Tables 4 and 5 by ancestry are of interest, are PRS means (SD) on Table 4 based on the 229 or 287 variant PRS? Q5. Are the principal components calculated separately for the 10 Asian studies in BCAC and the three Asian-American studies? Q6: The authors find a smaller RR for family history than previously reported in European-ancestry studies. Are there previous studies in Asian populations that support this difference? Q7: The 287-variant PRS was less predictive of breast cancer risk in the three Asian-American studies than in the studies conducted in Asian (which found no differences between Asians of Chinese, Malay or Indian origin). This is a potentially important finding that requires confirmation in larger studies. Could the authors formally test if these differences are statistically significant in the current population? Based on these findings, can the authors comment on what needs to be done to use PRS in Asian populations in the US? Or other countries? Minor comments: Q1: In the imputation for BCAC, the phase of the 1000 genome project should be specified. Q2: The EM algorithm needs citation Q3: The version of R and Stata needs to be listed Q4: Indicate population/sample size included in each table/figure or a reference to a table/figure showing the population/sample size. Q5: Table 1- change format to match Table 2 (with less blank cells). Q6: Consider adding the European estimates to Figure 2 Q7: Consider adding estimates from American Asians to Figure 3. Q8: Tables 3 and 4- indicate what PRS is shown – 287-SNP PRS?	TPR	NC_BS_CAN_015_RR1.txt	{ "criticism": 15, "example": 22, "importance_and_relevance": 7, "materials_and_methods": 44, "praise": 6, "presentation_and_reporting": 36, "results_and_discussion": 18, "suggestion_and_solution": 36, "total": 90 }	{ "criticism": 0.16666666666666666, "example": 0.24444444444444444, "importance_and_relevance": 0.07777777777777778, "materials_and_methods": 0.4888888888888889, "praise": 0.06666666666666667, "presentation_and_reporting": 0.4, "results_and_discussion": 0.2, "suggestion_and_solution": 0.4 }	2.044444	[ { "content": "Please review this paper.", "role": "user" }, { "content": "Reviewers' comments:\n\nReviewer #1 (Remarks to the Author):\n\nThis manuscript presents a very large and fairly comprehensive analysis studying the application of \na polygenic risk score derived using data from European women to characterize risk in Asian \nwomen. The components of the paper are well developed but the actual presentation has many \nissues and the paper needs a great deal of revision for clarity. I have many minor comments and \ntwo more major concerns.\nMajor question – it is not very clear that genetic analysis to identify 1% of women at polygenic risk \nwould have clinical utility or be feasible. If the PRS is set to the top 5% does that provide an Odds \nratio that is meaningful or a prospective risk that would lead to different management of women?\nSome sort of sensitivity analysis considering at least an upper 5% threshold and perhaps an upper \n10% threshold should be included.\n\nWhat is the AUC when you include versus exclude Family History. Table 3 gives the attenuation in \nOR when including family history but the more interesting element to me would be to evaluate \nwhat the AUC is when you include markers and Family history.\n\nI think it would be useful to also plot the distribution of risk according to quantiles of risk and \naccording to age. This would better illustrate the lack of information from individuals at low PRS \nacross all ages.\n\nMinor comments: Line 143. Should read ‘mutations in these genes are rare’ not these genes are \nrare, since almost everyone has two copies of these genes (except rare people with deletions).\nLine 157 – this discussion is not very comprehensive. While it may be true that PRS has some \nutility how much does it improve beyond traditional and early collect information like family \nhistory?\n\nLines 192-197 it would be helpful to have some information about what the interpretation of the \nPRS mean score would be. Is this score somehow standardized? If not then should it be to make \nthe interpretation more straightforward?\n\nLines 226-230 gives changes in percent attenuation without indicating what is being attenuated or \nwhat the relevance of this sentence is. Please clarify this section without requiring close perusal of \nthe table since the sentence makes no sense as written.\n\nLines 249-251. The actual AUCs should be cited in the text where it is being highlighted.\n\nThese sentences do not make sense. Smaller than what? Since this sentence starts a new \nparagraph there is no referent for that, so the sentence is uninterpretable. ‘The OR per SD was \nsignificantly smaller (p < 0.05) than that in the Asian studies from Asia, 267 for overall breast \ncancer (1.36, 95% CI = 1.25 – 1.49) and ER-positive breast cancer (1.38, 268 95% CI = 1.25 –\n1.53), but significant higher (p < 0.05) for ER-negative breast cancer (1.49,269 95% CI = 1.26 –\n1.76, Table 2(b)).’\n\nAround lines 284 – where do the referent risks come from for populations that are being \ndescribed? It does not sound as if population risks for breast cancer are being derived from cohort \nstudies but rather from a population database but there is no reference so it is impossible to tell \nhow these analyses were conducted.\n\nLine 339 – effect size of PRS - the authors do not define their threshold. Do they mean top 1% of \nAsian PRS as a cut off? Do they mean top 1% of European?\n\nLine 347 – lower predictive accuracy in Asians- the author need to point to a table in this \n\n\fdiscussion and give some values since this information was not previously cited. Or it could be \nmoved to the results and described more fully (which would be preferable).\n\nThe discussion cites that a screening regimen based on PRS and age could reduce numbers of \nwomen undergoing screening and cites a prior paper but shouldn’t they also derive these numbers \nfrom their own analysis?\n\nLine 372 to 375 – this presents some interesting observations that have not been described \npreviously and do not cite a figure or table so are uninterpretable.\n\nLines 381-387- great that simulations were done. Please present the results in a figure and or \ntable. Otherwise this is unacceptable presentation of unsupported information.\n\nLines 406-410. Did the authors check for heterogeneity in the ORs for family history by study? \nOpportunistic control selection could definitely lead to enrichment of FH positive controls. That \nmight also be yield a similar frequency of FH in case populations and variability in FH+ controls, \nwhich could be tested for.\n\n432-435 – the sentence structure in this section is weird and confusing e.g. ‘All three studies are \npopulation-based case-control studies; and (c) 10,266 women 433 of Chinese ethnicity \nparticipating in Singapore Chinese Health Study (SCHS)’ In this section you have a period break \nbefore All so it is not very clear that you are referring to the prior sentence and then you join this \nunclear phrase the next unclear phrase after a semicolon. Please break unrelated phrases into \nsentences and join related clauses to the antecedent sentences.\n\nLine 472 ‘Post-imputation quality of all studies was based on the IMPUTE imputation quality score.’ \nThis should not be a guessing game. Please describe what these imputation quality metrics are or \nexplain them further in a supplementary methods that are referenced.\n\nLines 485-490 the English is missing plurals and articles so should be revised. It is great to see a \nreference for how the PRS was standarised, but I think this information should also be placed in \nthe text where the means of PRS are presented.\n\nLine 491-495 – the manuscript indicates that the PRS was stratified into quartiles and the middle \nquintile was used for referent but that is not very clearly relevant to the paper where top 1% of \ndistribution was used. I think in this section the description is fairly clear but where the 1% cut off \nis described you need to cite that you are using the middle quintile as referent.\n\nLine 507-509 – AUC analyses were adjusted by study where relevant. This needs to be explained \nfurther.\n\nLines 515-519 please give a reference for why this formula is appropriate.\n\nThe references on line 551 should be cited in the main body of the paper as requested above.\n\nFor reproducibility it would be better to specify which modules or R or Stata were used for major \nanalyses rather than the very generic reference that is given.\n\nTable 2 needs to cite in the footnote that the cutpoint is top 1% of the PRS in controls if that is \nwhat was used for stratifying high risk individuals.\n\n\fReviewer #2 (Remarks to the Author):\n\nHo and colleges evaluated the performance of a European-derived polygenic risk score (PRS) in an \nAsian population of 17,262 breast cancer cases and 17,695 controls. Most of these women were \nliving in Asian countries, with a smaller population of Asian-Americans (1,507 cases and 1,212 \ncontrols). In addition, they evaluated the PRS performance in 10,255 Chinese women from a \nprospective cohort (413 incident breast cancers). This is a well-conducted study with findings that \nhave public health relevance since it provides estimates for polygenic breast cancer risk in Asian \nwomen that could be used for risk-stratified screening/prevention. The main finding is that the \nEuropean-derived PRS can provide substantial risk stratification of Asian women, however the \nstratification is lower than for European-ancestry women. Although this is the largest study \nconducted in Asians to date, as the authors pointed out, even larger studies are needed to obtain \nprecise OR estimates at the extreme of the PRS distribution, or to further improve the performance \nof PRS in Asian women.\n\nBelow are some comments/questions for consideration:\n\nQ1: The authors used PRS weights based on estimates from European ancestry populations, rather \nthan being re-estimated in Asian populations. This is a reasonable approach as the sample sizes to \nestimate weights are much larger in Europeans and the authors showed evidence of consistent \neffect size for genetic variants between European and Asian populations. However, given that \nprevious publications have attempted to derived Asian-specific PRS, it would be of interest to \ndirectly compare the prediction performance between previous Asian-specific PRS (derived with \nlower sample size) and the European based PRS.\n\nQ2: Did the authors consider incorporating Asian-specific risk variants that are not in the 313-PRSs \ndeveloped for European populations?\n\nQ3. The authors evaluated the PRS based on 313 variants reported in Mavaddat et al. 2019. \nHowever, 26 of the 313 variants were excluded due to low imputation accuracy score in Asians \nusing the 1000 genome reference panel, resulting in a 287-variant PRS. This is a limitation since it \ndoes not allow direct comparison with the previously published PRS. Did the author consider \nimputing the genotypes based on larger reference panels such as HRC and TopMed ?\n\nQ4: For the Singapore Chinese Health Study (SCHS), only 229 of the 287 variants that were found \nto be polymorphic and imputable. Although this is a valuable study because of its prospective \nnature and information on ancestry, it is a bit confusing to show both the 287-SNP PRS and 229-\nSNP PRS as main findings. I suggest moving the 229-SNP PRS findings to supplement or make \nclearer what is the added value of these analyses as main findings. I do not see much value added \non tables 1 or 2. Findings on Tables 4 and 5 by ancestry are of interest, are PRS means (SD) on \nTable 4 based on the 229 or 287 variant PRS?\n\nQ5. Are the principal components calculated separately for the 10 Asian studies in BCAC and the \nthree Asian-American studies?\n\nQ6: The authors find a smaller RR for family history than previously reported in European-ancestry \nstudies. Are there previous studies in Asian populations that support this difference?\n\nQ7: The 287-variant PRS was less predictive of breast cancer risk in the three Asian-American \nstudies than in the studies conducted in Asian (which found no differences between Asians of \nChinese, Malay or Indian origin). This is a potentially important finding that requires confirmation \nin larger studies. Could the authors formally test if these differences are statistically significant in \nthe current population? Based on these findings, can the authors comment on what needs to be \ndone to use PRS in Asian populations in the US? Or other countries?\n\n\fMinor comments:\n\nQ1: In the imputation for BCAC, the phase of the 1000 genome project should be specified.\n\nQ2: The EM algorithm needs citation\n\nQ3: The version of R and Stata needs to be listed\n\nQ4: Indicate population/sample size included in each table/figure or a reference to a table/figure \nshowing the population/sample size.\n\nQ5: Table 1- change format to match Table 2 (with less blank cells).\n\nQ6: Consider adding the European estimates to Figure 2\n\nQ7: Consider adding estimates from American Asians to Figure 3.\n\nQ8: Tables 3 and 4- indicate what PRS is shown – 287-SNP PRS? \n\n\f\n", "role": "assistant" } ]	2,015
European polygenic risk score for prediction of breast cancer shows similar performance in	REVIEWERS' COMMENTS: Reviewer #2 (Remarks to the Author): The authors have addressed my comments adequately Reviewer #3 (Remarks to the Author): The authors have made a comprehensive response to the questions raised by the reviewers. The addition of explanatory comments and re-editing has improved the clarity of the paper and there are no significant revisions that I would suggest. Appreciating that the paper has been well reviewed and revised there were a couple of points that might be addressed for the final text: Line 297 - the two figures here seem to be for Asians in Asia and Asians in America but it is not actually clear and could be spelled out Line 414 - it may be worth noting that this figure (20%) would then be around the same as it is in the European population - as written it leaves the impression that this is still exceptional Line 441 - "the PRS based on large European-ancestry studies may be used as the basis for Asian- specific breast cancer risk prediction models" - although it has been well covered in the text this summary statement probably still needs to have a caveat such as "that take into account the unique distribution of the PRS in the Asian population" to avoid giving the impression that the risk prediction produced by the standard European-trained model (such as generated by BOADICEA or from commercial labs) is 'good enough' for the interpretation of genotype data from an Asian woman. Reviewer #4 (Remarks to the Author): The authors have adequately responded to all queries. One point of confusion that remains is what analyses use SNP weights from Europeans vs. SNP weights from Asians and how do these results compare. This needs to be better defined in the methods, figure/table legends and in the results section.	REVIEWERS' COMMENTS: Reviewer #2 (Remarks to the Author): The authors have addressed my comments adequately Reviewer #3 (Remarks to the Author): The authors have made a comprehensive response to the questions raised by the reviewers. The addition of explanatory comments and re-editing has improved the clarity of the paper and there are no significant revisions that I would suggest. Appreciating that the paper has been well reviewed and revised there were a couple of points that might be addressed for the final text: Line 297 - the two figures here seem to be for Asians in Asia and Asians in America but it is not actually clear and could be spelled out Line 414 - it may be worth noting that this figure (20%) would then be around the same as it is in the European population - as written it leaves the impression that this is still exceptional Line 441 - "the PRS based on large European-ancestry studies may be used as the basis for Asian- specific breast cancer risk prediction models" - although it has been well covered in the text this summary statement probably still needs to have a caveat such as "that take into account the unique distribution of the PRS in the Asian population" to avoid giving the impression that the risk prediction produced by the standard European-trained model (such as generated by BOADICEA or from commercial labs) is 'good enough' for the interpretation of genotype data from an Asian woman. Reviewer #4 (Remarks to the Author): The authors have adequately responded to all queries. One point of confusion that remains is what analyses use SNP weights from Europeans vs. SNP weights from Asians and how do these results compare. This needs to be better defined in the methods, figure/table legends and in the results section.	TPR	NC_BS_CAN_015_RR2.txt	{ "criticism": 3, "example": 1, "importance_and_relevance": 0, "materials_and_methods": 3, "praise": 3, "presentation_and_reporting": 3, "results_and_discussion": 2, "suggestion_and_solution": 2, "total": 6 }	{ "criticism": 0.5, "example": 0.16666666666666666, "importance_and_relevance": 0, "materials_and_methods": 0.5, "praise": 0.5, "presentation_and_reporting": 0.5, "results_and_discussion": 0.3333333333333333, "suggestion_and_solution": 0.3333333333333333 }	2.833333	[ { "content": "Please review this paper.", "role": "user" }, { "content": "\n \n \n \n\n\fREVIEWERS' COMMENTS:\n\nReviewer #2 (Remarks to the Author):\n\nThe authors have addressed my comments adequately\n\nReviewer #3 (Remarks to the Author):\n\nThe authors have made a comprehensive response to the questions raised by the reviewers. The \naddition of explanatory comments and re-editing has improved the clarity of the paper and there \nare no significant revisions that I would suggest.\n\nAppreciating that the paper has been well reviewed and revised there were a couple of points that \nmight be addressed for the final text:\n\nLine 297 - the two figures here seem to be for Asians in Asia and Asians in America but it is not \nactually clear and could be spelled out\n\nLine 414 - it may be worth noting that this figure (20%) would then be around the same as it is in \nthe European population - as written it leaves the impression that this is still exceptional\n\nLine 441 - \"the PRS based on large European-ancestry studies may be used as the basis for Asian-\nspecific breast cancer risk prediction models\" - although it has been well covered in the text this \nsummary statement probably still needs to have a caveat such as \"that take into account the \nunique distribution of the PRS in the Asian population\" to avoid giving the impression that the risk \nprediction produced by the standard European-trained model (such as generated by BOADICEA or \nfrom commercial labs) is 'good enough' for the interpretation of genotype data from an Asian \nwoman.\n\nReviewer #4 (Remarks to the Author):\n\nThe authors have adequately responded to all queries. One point of confusion that remains is what \nanalyses use SNP weights from Europeans vs. SNP weights from Asians and how do these results \ncompare. This needs to be better defined in the methods, figure/table legends and in the results \nsection. \n\n\f", "role": "assistant" } ]	2,015
AXL confers cell migration and invasion by hijacking a PEAK1-regulated focal adhesion	Reviewers' comments: Reviewer #1 (Remarks to the Author): This manuscript is improved over the previous version, but a few issues remain to be resolved before the work is complete 1) The method of quantifying PLA signal seems odd (total average intensity versus number of puncta). The method of analysis does not tell you how much more PLA signal is there when an interaction is detected? 2)Figure 7 still seems a mess to me. Many of the blots are blown out/have poor exposures, and the graphs of FA lifetime have no primary data, such as the examples presented in Fig 3. Reviewer #2 (Remarks to the Author): The authors miss the point of the Gas6 issue. Yes, gamma-carboxylation is required for Gas6- mediated AXL activation but so is association of Gas6 with PS lipids -- indeed, the role of gamma- carboxylation is to facilitate Gas6/PS association. This has been established by multiple literature reports from a variety of laboratories. The extent to which Gas6 obtained from one source or another is gamma-carboxylated is relevant only in terms of its interaction with PS. Thus, the presence of PS lipids, such as from apoptotic cellular debris or other conditions, in the experimental assay cultures is critical in governing AXL activation, regardless of the extent of Gas6 gamma-carboxylation. Now, if one has a source of Gas6 that is more extensively gamma- carboxylated than a different source, the amount of PS lipids needed to be present in the assay environment to obtain a given level of AXL activation may be lesser, but Gas6 by itself cannot activate AXL to an appreciable degree even if extensively gamma-carboxylated. Reviewer #3 (Remarks to the Author): The authors have clarified my previous concerns and have addressed previous points/concerns raised by carrying out further experiments. Reviewer #4 (Remarks to the Author): The manuscript by Afnan Abu-Thuraia, et al investigated the signaling mechanisms mediated by AXL, one of the receptor tyrosine kinases (RTKs) that are activated during metastasis of breast cancer. Using quantitative phosphoproteomics, the authors identified PEAK1 as a critical downstream factor of the AXL activation and PEAK1 promoted the turnover of focal adhesion (FA) by recruiting the FA turnover complex to paxillin. Knockout of PEAK1 inhibited breast cancer cell metastasis, which can be rescued by wild-type PEAK1 but not mutant PEAK1 that lost interaction with the pathway. Overall, the study reported an interesting finding on the important role of PEAK1 in mediating the activity of FA turnover upon AXL activation. However, a number of key mechanisms are not clear despite the observation of AXL-mediated phosphorylation events. In addition, the logic and study design of the overall study described in the manuscript was quite complicated and not all experiments were necessary to be included in the manuscript. Some major and minor concerns are listed below. Due to these concerns, the study is not yet ready for publication in its current form. 1. Identification of PEAK1 was achieved by overlapping the phosphoproteomics data of AXL activation with the interactome data of NEDD9. However, it is quite strange that the interaction between NEDD9 and PEAK1 could not be validated (line 265). This cast doubts on the validity of the NEDD9 interactome analysis with Bio-ID method. 2. The mechanism of how AXL activates PEAK1 was still not clear. Since the interaction between PEAK1 and AXL was validated and then PEAK1 phosphorylation increased upon AXL activation, the authors hypothesized that PEAK1 was a direct AXL target. However, mutation of the identified tyrosine residues did not inhibit AXL-mediated PEAK1 phosphorylation. So the key question is that whether PEAK1 phosphorylation was required for its activation. It is also possible that AXL activates PEAK1 indirectly or through protein-protein interactions. 3. For validation of AXL control of PEAK1 localization at FA, the authors applied siRNA and chemical treatment to knock down or inhibit AXL activity. However, these experiments are not sufficient to demonstrate the importance of AXL kinase activity in this process. Activation of AXL and kinase dead mutant by overexpression and GAS6 treatment would be more informative to determine if PEAK1 phosphorylation by AXL plays an important role during this process. 4. The overall role of NEDD9 in PEAK1-mediated phosphorylation and activation of FA turnover is not very clear. The manuscript included Figure 4 and a large section of texts describing the validation and interaction of NEDD9 with FA. However, such information only led to the identification of PEAK1 and not quite relevant with the key findings of the current manuscript. Therefore, the inclusion of these data on NEDD9 made it difficult to follow the logic flow of the study. 5. The mechanisms on how AXL-mediated CRKII phosphorylation regulates its interaction with PXN was not clear. Since the AXL-mediated phosphorylation sites on CRKII have been identified, it would be straightforward to test if these phosphorylation events are important for CRKII interaction with PXN. 6. The importance of PEAK1-mediated phosphorylation of PXN was not clear. The authors showed that PEAK1 may be able to directly phosphorylate PXN and it was also possible that CSK through PRAG phosphorylates PXN. These possibilities were not rigorously studied and should not be included if the findings are not conclusive. It is also unclear what may be the site-specific effect of PEAK1-mediated PXN phosphorylation in FA turnover. 7. The interaction between PEAK1 and PAK1 complex was identified and the authors hypothesized that these interactions are important for FA turnover. However, other than the knockdown study, there is no further mechanistic study to prove the hypothesis. 8. The authors demonstrated the important role of PEAK1 and CRKII complex in FA turnover using 3PA PEAK1 mutant. However, the key question is whether AXL mediates the interaction between PEAK1 and CRKII. 9. Since 3PA PEAK1 mutant is the key to demonstrate the important role of activation of PEAK1 by AXL in vivo, it is important to determine if the mutation affect the interaction between PEAK1 and b-PIX/GIT1/PAK1 complex. 10. The proteomics data analysis figure (Figure 1c) is confusing. For quantitative proteomics data, please perform three biological replicates for each time point and use volcano plots to identify significantly changed hits. Minor concern: 1. Figure 1d, how the relative abundance of phosphorylation was defined? Why was the 0 min abundance for each site not shown? The figure may be better presented as line graphs. 2. Figure 2 can be put in Supplementary material as the analysis does not contribute significantly to identification of PEAK1 and other targets. Reviewer #5 (Remarks to the Author): The revised manuscript by Abu-Thuraia et al., addresses both the scientific concerns and the issues raised by reviewers with the presentation of the proposal during the prior cycle of review. The manuscript addresses the potential role of an AXL/PEAK regulated signaling axis consortium in promoting cell adhesion and cell migration. The work provides a new rationale for the regulation of cell motility downstream of AXL via CrkII and NEDD9, which may be important not only in pathologies such as cancer, but also during development. The authors have done a good job of addressing most of the prior issues. There are still a couple issues related to prior points which should be addressed: The authors conclude that PEAK1/CrkII interaction is required for “localization” at FAs. It remains possible that PEAK1 is not responsible for localization/targeting, but acts to stabilize CrkII within the FAs. This should be addressed within the text, at a minimum. The studies are highly dependent upon cell lines and upon Axl over-expression, and this seems partly due to a lab-constructed system. Axl is expressed in only a fraction of invasive breast cancers and it is a stretch to claim it is “highly expressed.” In fact, Axl is a significant negative prognostic only in renal cancer (where PEAK1 is abundant). In other tissues, which express high levels of both PEAK1 and Axl, (such as testicles, where co-expression is the highest) Axl expression is typically lost during oncogenesis. The authors should address this dichotomy in order to provide their results in a correct context. Minor: The presentation of the bioinformatics includes gratuitous graphics (which are generally intended for the presentation of much bigger data matrixes) (eg., Figure 2a, d). In these cases, a simple table would provide more information to the reader while occupying less space.	Reviewers' comments: Reviewer #1 (Remarks to the Author): This manuscript is improved over the previous version, but a few issues remain to be resolved before the work is complete 1) The method of quantifying PLA signal seems odd (total average intensity versus number of puncta). The method of analysis does not tell you how much more PLA signal is there when an interaction is detected? 2)Figure 7 still seems a mess to me. Many of the blots are blown out/have poor exposures, and the graphs of FA lifetime have no primary data, such as the examples presented in Fig 3. Reviewer #2 (Remarks to the Author): The authors miss the point of the Gas6 issue. Yes, gamma-carboxylation is required for Gas6- mediated AXL activation but so is association of Gas6 with PS lipids -- indeed, the role of gamma- carboxylation is to facilitate Gas6/PS association. This has been established by multiple literature reports from a variety of laboratories. The extent to which Gas6 obtained from one source or another is gamma-carboxylated is relevant only in terms of its interaction with PS. Thus, the presence of PS lipids, such as from apoptotic cellular debris or other conditions, in the experimental assay cultures is critical in governing AXL activation, regardless of the extent of Gas6 gamma-carboxylation. Now, if one has a source of Gas6 that is more extensively gamma- carboxylated than a different source, the amount of PS lipids needed to be present in the assay environment to obtain a given level of AXL activation may be lesser, but Gas6 by itself cannot activate AXL to an appreciable degree even if extensively gamma-carboxylated. Reviewer #3 (Remarks to the Author): The authors have clarified my previous concerns and have addressed previous points/concerns raised by carrying out further experiments. Reviewer #4 (Remarks to the Author): The manuscript by Afnan Abu-Thuraia, et al investigated the signaling mechanisms mediated by AXL, one of the receptor tyrosine kinases (RTKs) that are activated during metastasis of breast cancer. Using quantitative phosphoproteomics, the authors identified PEAK1 as a critical downstream factor of the AXL activation and PEAK1 promoted the turnover of focal adhesion (FA) by recruiting the FA turnover complex to paxillin. Knockout of PEAK1 inhibited breast cancer cell metastasis, which can be rescued by wild-type PEAK1 but not mutant PEAK1 that lost interaction with the pathway. Overall, the study reported an interesting finding on the important role of PEAK1 in mediating the activity of FA turnover upon AXL activation. However, a number of key mechanisms are not clear despite the observation of AXL-mediated phosphorylation events. In addition, the logic and study design of the overall study described in the manuscript was quite complicated and not all experiments were necessary to be included in the manuscript. Some major and minor concerns are listed below. Due to these concerns, the study is not yet ready for publication in its current form. 1. Identification of PEAK1 was achieved by overlapping the phosphoproteomics data of AXL activation with the interactome data of NEDD9. However, it is quite strange that the interaction between NEDD9 and PEAK1 could not be validated (line 265). This cast doubts on the validity of the NEDD9 interactome analysis with Bio-ID method. 2. The mechanism of how AXL activates PEAK1 was still not clear. Since the interaction between PEAK1 and AXL was validated and then PEAK1 phosphorylation increased upon AXL activation, the authors hypothesized that PEAK1 was a direct AXL target. However, mutation of the identified tyrosine residues did not inhibit AXL-mediated PEAK1 phosphorylation. So the key question is that whether PEAK1 phosphorylation was required for its activation. It is also possible that AXL activates PEAK1 indirectly or through protein-protein interactions. 3. For validation of AXL control of PEAK1 localization at FA, the authors applied siRNA and chemical treatment to knock down or inhibit AXL activity. However, these experiments are not sufficient to demonstrate the importance of AXL kinase activity in this process. Activation of AXL and kinase dead mutant by overexpression and GAS6 treatment would be more informative to determine if PEAK1 phosphorylation by AXL plays an important role during this process. 4. The overall role of NEDD9 in PEAK1-mediated phosphorylation and activation of FA turnover is not very clear. The manuscript included Figure 4 and a large section of texts describing the validation and interaction of NEDD9 with FA. However, such information only led to the identification of PEAK1 and not quite relevant with the key findings of the current manuscript. Therefore, the inclusion of these data on NEDD9 made it difficult to follow the logic flow of the study. 5. The mechanisms on how AXL-mediated CRKII phosphorylation regulates its interaction with PXN was not clear. Since the AXL-mediated phosphorylation sites on CRKII have been identified, it would be straightforward to test if these phosphorylation events are important for CRKII interaction with PXN. 6. The importance of PEAK1-mediated phosphorylation of PXN was not clear. The authors showed that PEAK1 may be able to directly phosphorylate PXN and it was also possible that CSK through PRAG phosphorylates PXN. These possibilities were not rigorously studied and should not be included if the findings are not conclusive. It is also unclear what may be the site-specific effect of PEAK1-mediated PXN phosphorylation in FA turnover. 7. The interaction between PEAK1 and PAK1 complex was identified and the authors hypothesized that these interactions are important for FA turnover. However, other than the knockdown study, there is no further mechanistic study to prove the hypothesis. 8. The authors demonstrated the important role of PEAK1 and CRKII complex in FA turnover using 3PA PEAK1 mutant. However, the key question is whether AXL mediates the interaction between PEAK1 and CRKII. 9. Since 3PA PEAK1 mutant is the key to demonstrate the important role of activation of PEAK1 by AXL in vivo, it is important to determine if the mutation affect the interaction between PEAK1 and b-PIX/GIT1/PAK1 complex. 10. The proteomics data analysis figure (Figure 1c) is confusing. For quantitative proteomics data, please perform three biological replicates for each time point and use volcano plots to identify significantly changed hits. Minor concern: 1. Figure 1d, how the relative abundance of phosphorylation was defined? Why was the 0 min abundance for each site not shown? The figure may be better presented as line graphs. 2. Figure 2 can be put in Supplementary material as the analysis does not contribute significantly to identification of PEAK1 and other targets. Reviewer #5 (Remarks to the Author): The revised manuscript by Abu-Thuraia et al., addresses both the scientific concerns and the issues raised by reviewers with the presentation of the proposal during the prior cycle of review. The manuscript addresses the potential role of an AXL/PEAK regulated signaling axis consortium in promoting cell adhesion and cell migration. The work provides a new rationale for the regulation of cell motility downstream of AXL via CrkII and NEDD9, which may be important not only in pathologies such as cancer, but also during development. The authors have done a good job of addressing most of the prior issues. There are still a couple issues related to prior points which should be addressed: The authors conclude that PEAK1/CrkII interaction is required for “localization” at FAs. It remains possible that PEAK1 is not responsible for localization/targeting, but acts to stabilize CrkII within the FAs. This should be addressed within the text, at a minimum. The studies are highly dependent upon cell lines and upon Axl over-expression, and this seems partly due to a lab-constructed system. Axl is expressed in only a fraction of invasive breast cancers and it is a stretch to claim it is “highly expressed.” In fact, Axl is a significant negative prognostic only in renal cancer (where PEAK1 is abundant). In other tissues, which express high levels of both PEAK1 and Axl, (such as testicles, where co-expression is the highest) Axl expression is typically lost during oncogenesis. The authors should address this dichotomy in order to provide their results in a correct context. Minor: The presentation of the bioinformatics includes gratuitous graphics (which are generally intended for the presentation of much bigger data matrixes) (eg., Figure 2a, d). In these cases, a simple table would provide more information to the reader while occupying less space.	TPR	NC_BS_CBIO_003_RR2.txt	{ "criticism": 20, "example": 7, "importance_and_relevance": 4, "materials_and_methods": 44, "praise": 2, "presentation_and_reporting": 8, "results_and_discussion": 4, "suggestion_and_solution": 12, "total": 76 }	{ "criticism": 0.2631578947368421, "example": 0.09210526315789473, "importance_and_relevance": 0.05263157894736842, "materials_and_methods": 0.5789473684210527, "praise": 0.02631578947368421, "presentation_and_reporting": 0.10526315789473684, "results_and_discussion": 0.05263157894736842, "suggestion_and_solution": 0.15789473684210525 }	1.328947	[ { "content": "Please review this paper.", "role": "user" }, { "content": "\n\n\fReviewers' comments: \n\nReviewer #1 (Remarks to the Author): \n\nThis manuscript is improved over the previous version, but a few issues remain to be resolved \nbefore the work is complete \n\n1) The method of quantifying PLA signal seems odd (total average intensity versus number of \npuncta). The method of analysis does not tell you how much more PLA signal is there when an \ninteraction is detected? \n2)Figure 7 still seems a mess to me. Many of the blots are blown out/have poor exposures, and \nthe graphs of FA lifetime have no primary data, such as the examples presented in Fig 3. \n\nReviewer #2 (Remarks to the Author): \n\nThe authors miss the point of the Gas6 issue. Yes, gamma-carboxylation is required for Gas6-\nmediated AXL activation but so is association of Gas6 with PS lipids -- indeed, the role of gamma-\ncarboxylation is to facilitate Gas6/PS association. This has been established by multiple literature \nreports from a variety of laboratories. The extent to which Gas6 obtained from one source or \nanother is gamma-carboxylated is relevant only in terms of its interaction with PS. Thus, the \npresence of PS lipids, such as from apoptotic cellular debris or other conditions, in the \nexperimental assay cultures is critical in governing AXL activation, regardless of the extent of Gas6 \ngamma-carboxylation. Now, if one has a source of Gas6 that is more extensively gamma-\ncarboxylated than a different source, the amount of PS lipids needed to be present in the assay \nenvironment to obtain a given level of AXL activation may be lesser, but Gas6 by itself cannot \nactivate AXL to an appreciable degree even if extensively gamma-carboxylated. \n\nReviewer #3 (Remarks to the Author): \n\nThe authors have clarified my previous concerns and have addressed previous points/concerns \nraised by carrying out further experiments. \n\nReviewer #4 (Remarks to the Author): \n\nThe manuscript by Afnan Abu-Thuraia, et al investigated the signaling mechanisms mediated by \nAXL, one of the receptor tyrosine kinases (RTKs) that are activated during metastasis of breast \ncancer. Using quantitative phosphoproteomics, the authors identified PEAK1 as a critical \ndownstream factor of the AXL activation and PEAK1 promoted the turnover of focal adhesion (FA) \nby recruiting the FA turnover complex to paxillin. Knockout of PEAK1 inhibited breast cancer cell \nmetastasis, which can be rescued by wild-type PEAK1 but not mutant PEAK1 that lost interaction \nwith the pathway. \nOverall, the study reported an interesting finding on the important role of PEAK1 in mediating the \nactivity of FA turnover upon AXL activation. However, a number of key mechanisms are not clear \ndespite the observation of AXL-mediated phosphorylation events. In addition, the logic and study \ndesign of the overall study described in the manuscript was quite complicated and not all \nexperiments were necessary to be included in the manuscript. Some major and minor concerns are \nlisted below. Due to these concerns, the study is not yet ready for publication in its current form. \n1. Identification of PEAK1 was achieved by overlapping the phosphoproteomics data of AXL \nactivation with the interactome data of NEDD9. However, it is quite strange that the interaction \nbetween NEDD9 and PEAK1 could not be validated (line 265). This cast doubts on the validity of \n\n\fthe NEDD9 interactome analysis with Bio-ID method. \n2. The mechanism of how AXL activates PEAK1 was still not clear. Since the interaction between \nPEAK1 and AXL was validated and then PEAK1 phosphorylation increased upon AXL activation, the \nauthors hypothesized that PEAK1 was a direct AXL target. However, mutation of the identified \ntyrosine residues did not inhibit AXL-mediated PEAK1 phosphorylation. So the key question is that \nwhether PEAK1 phosphorylation was required for its activation. It is also possible that AXL \nactivates PEAK1 indirectly or through protein-protein interactions. \n3. For validation of AXL control of PEAK1 localization at FA, the authors applied siRNA and chemical \ntreatment to knock down or inhibit AXL activity. However, these experiments are not sufficient to \ndemonstrate the importance of AXL kinase activity in this process. Activation of AXL and kinase \ndead mutant by overexpression and GAS6 treatment would be more informative to determine if \nPEAK1 phosphorylation by AXL plays an important role during this process. \n4. The overall role of NEDD9 in PEAK1-mediated phosphorylation and activation of FA turnover is \nnot very clear. The manuscript included Figure 4 and a large section of texts describing the \nvalidation and interaction of NEDD9 with FA. However, such information only led to the \nidentification of PEAK1 and not quite relevant with the key findings of the current manuscript. \nTherefore, the inclusion of these data on NEDD9 made it difficult to follow the logic flow of the \nstudy. \n5. The mechanisms on how AXL-mediated CRKII phosphorylation regulates its interaction with PXN \nwas not clear. Since the AXL-mediated phosphorylation sites on CRKII have been identified, it \nwould be straightforward to test if these phosphorylation events are important for CRKII \ninteraction with PXN. \n6. The importance of PEAK1-mediated phosphorylation of PXN was not clear. The authors showed \nthat PEAK1 may be able to directly phosphorylate PXN and it was also possible that CSK through \nPRAG phosphorylates PXN. These possibilities were not rigorously studied and should not be \nincluded if the findings are not conclusive. It is also unclear what may be the site-specific effect of \nPEAK1-mediated PXN phosphorylation in FA turnover. \n7. The interaction between PEAK1 and PAK1 complex was identified and the authors hypothesized \nthat these interactions are important for FA turnover. However, other than the knockdown study, \nthere is no further mechanistic study to prove the hypothesis. \n8. The authors demonstrated the important role of PEAK1 and CRKII complex in FA turnover using \n3PA PEAK1 mutant. However, the key question is whether AXL mediates the interaction between \nPEAK1 and CRKII. \n9. Since 3PA PEAK1 mutant is the key to demonstrate the important role of activation of PEAK1 by \nAXL in vivo, it is important to determine if the mutation affect the interaction between PEAK1 and \nb-PIX/GIT1/PAK1 complex. \n10. The proteomics data analysis figure (Figure 1c) is confusing. For quantitative proteomics data, \nplease perform three biological replicates for each time point and use volcano plots to identify \nsignificantly changed hits. \nMinor concern: \n1. Figure 1d, how the relative abundance of phosphorylation was defined? Why was the 0 min \nabundance for each site not shown? The figure may be better presented as line graphs. \n2. Figure 2 can be put in Supplementary material as the analysis does not contribute significantly \nto identification of PEAK1 and other targets. \n\nReviewer #5 (Remarks to the Author): \n\nThe revised manuscript by Abu-Thuraia et al., addresses both the scientific concerns and the \nissues raised by reviewers with the presentation of the proposal during the prior cycle of review. \nThe manuscript addresses the potential role of an AXL/PEAK regulated signaling axis consortium in \npromoting cell adhesion and cell migration. The work provides a new rationale for the regulation of \ncell motility downstream of AXL via CrkII and NEDD9, which may be important not only in \n\n\fpathologies such as cancer, but also during development. The authors have done a good job of \naddressing most of the prior issues. There are still a couple issues related to prior points which \nshould be addressed: \n\nThe authors conclude that PEAK1/CrkII interaction is required for “localization” at FAs. It remains \npossible that PEAK1 is not responsible for localization/targeting, but acts to stabilize CrkII within \nthe FAs. This should be addressed within the text, at a minimum. \n\nThe studies are highly dependent upon cell lines and upon Axl over-expression, and this seems \npartly due to a lab-constructed system. Axl is expressed in only a fraction of invasive breast \ncancers and it is a stretch to claim it is “highly expressed.” In fact, Axl is a significant negative \nprognostic only in renal cancer (where PEAK1 is abundant). In other tissues, which express high \nlevels of both PEAK1 and Axl, (such as testicles, where co-expression is the highest) Axl \nexpression is typically lost during oncogenesis. The authors should address this dichotomy in order \nto provide their results in a correct context. \n\nMinor: The presentation of the bioinformatics includes gratuitous graphics (which are generally \nintended for the presentation of much bigger data matrixes) (eg., Figure 2a, d). In these cases, a \nsimple table would provide more information to the reader while occupying less space.\n\n\f \n\n", "role": "assistant" } ]	2,020
AXL confers cell migration and invasion by hijacking a PEAK1-regulated focal adhesion	Reviewers' comments: Reviewer #1 (Remarks to the Author): This is an involved study that implicates receptor tyrosine kinase organized by the enzyme AXL as a potential mechanism to enhance metastasis and acquisition of resistance to cancer drugs. This is a very complicated study with many parts that is compounded by the frequent reference to supplemental data. Some of which should be omitted and some of which should be included in the body of the text. Using a variety of indirect methods, the authors propose that AXL signaling contributes to FA dynamics. This work could benefit from more mechanistic experiments that clearly define the role of AXL and the pseudokinase PEAK1 in tumor growth and metastasis. More specific criticisms are listed below. On page 4 it states “We identified an overrepresentation of MAPK, CDK, PKA and PKC kinase motifs in most of the clusters, suggesting that AXL may signal 5via regulation of these kinases (Supplementary Information, Fig. S2d). This data is waek and extremely speculative. Since it covers several of the predominant kinases it is hard to know what this means. I recommend removal of this data. The proximity ligation data presented in fig 2a is poor and hard to evaluated. Likewise, The data in figures 2b and c is not very convincing. First, quantitation of the PLA data should and Quantitation of the number of focal adhesions as dot scatter plots is a more compelling way to show the relative effect of changing AXL expression or action. Evidence for changes in the number of focal adhesions presented in figure 2c and the supplementary information is far from convincing. The Bio ID proteomic data in figures 3f and g is poorly connected to the rest of the paper. This is a rather speculative element of the work that needs some more experimental evidence to warrant its inclusion in the study. Reviewer #2 (Remarks to the Author): This manuscript offers a fascinating and important study of AXL signaling, which is timely because of the growing appreciation of the key role AXL plays in tumor resistance to targeted and chemotherapeutic drugs. The kind of understanding being pursued here, of nodes downstream of AXL activation could be very useful in suggesting effective combination treatments as well as in gaining insights into novel receptor tyrosine kinase signaling biochemistry. The authors bring together multiple approaches, experimental and computational, to ascertain relevant pathway elements including NEDD9, PEAK1, CSK, and Paxillin, toward enhancing invasive tumor cell migration via focal adhesion turnover. The resulting story is appealing and appears to be generally sound. My main point of concern relates to the technical details of Gas6 treatment for AXL stimulation. The authors seem to be not familiar with certain complications inherent in the Gas6/AXL system, such that mere treatment with soluble Gas6 is well-known to have variable effects depending on availability and characteristics of lipid vesicle-mediated interactions (for instance, as studied by Meyer et al in Cell Systems [2015]), such that AXL activation effects will vary with factors that are not explicitly considered in the experimental protocols outlined here. I believe that the authors need to provide more detailed characterization of their Gas6 protocols, including whether the cells being treated express Gas6 themselves in autocrine manner and to what degree apoptotic cell debris might be present due to low levels of nonviability in the culture conditions. These details will be vital for other laboratories in any attempts to reproduce and build upon the findings here. Reviewer #3 (Remarks to the Author): In this paper the authors try to elucidate the major signaling pathways activated by the receptor tyrosine kinase, AXL. The authors have used systems biology and proteomic approaches and find that AXL modulates the phosphorylation of a network of focal adhesion (FA) proteins that culminates in faster FA disassembly. They go on to deliniate a rather convoluted pathway whereby AXL phosphorylates the FA protein NEDD9, which then promotes NEDD9/CRKII-mediated phosphorylation of PEAK1. This pseudokinase then complexes wilt CSK kinase to phosphorylate paxillin, resulting in FA turnover. The strengths of this work lie in the biochemistry and systems biology, which are carried out very well. The finding that a major network of AXL-mediated phosphorylation events occur in FA proteins is novel, and this work does suggest that AXL can modulate FA turnover. The weaknesses lie in the biology and physiological relevance of the findings to tumor progression, especially the relevance of focal adhesion turnover-per se to AXL-mediated metastasis. Thus while the final data in Figs 5 and S7 using PEAK KO MDA-MB-231 cells suggest that PEAK may contribute to metastasis, The authors have not connected the pathway back to AXL. Thus, PEAK may have completely AXL-independent role in regulating metastasis. To make the case that it is the FA turnover pathway mediated by AXL that promotes metastasis, the authors need to demonstrate that some of the key phosphorylation and protein-protein interactions leading PEAK-mediated to paxillin phosphorylation and FA turnover also promote metastasis. Otherwise the paper does not hang together very well as it represents some excellent biochemistry and a separate finding that PEAK KO modulates metastasis. Specific Points that should be addressed: 1. Introduction and Discussion sections are missing. These need to be added to provide a context to this work and to demonstrate originality of the work 2. All the phospho-blots lack blots of the total proteins. This is essential to show that it is the phosporylation of the protein and not the total expression of the proteins that is changing. 3.In Fig 2 the analyses of the FAs should also include effects of GAS6 induced effects in addition to the constitutive situation, i.e. effects of GAS6 + R428. 4. p12. Does inhibition of paxillin phosphorylation ( directly or by using phospho-mutants) have the same effect on metastasis as PEAK KO? 5. The effect of AXL KO or the effect of AXL inhibitor on metastasis has to be carried out. Is this effect rescued by PEAK? 6. The metastasis data ( PEAK KO) need to be carried out in another TNBC cell line in addition to MDA-MB-231 cells 7 Fig s7j, why were the lungs probed ex vivo. Could in vivo signal be detected? 8. Fig s7d/e: Tumor volumes should be measured , not only IVIS signal. 9. The pathway model should be depicted in the main figures.	Reviewers' comments: Reviewer #1 (Remarks to the Author): This is an involved study that implicates receptor tyrosine kinase organized by the enzyme AXL as a potential mechanism to enhance metastasis and acquisition of resistance to cancer drugs. This is a very complicated study with many parts that is compounded by the frequent reference to supplemental data. Some of which should be omitted and some of which should be included in the body of the text. Using a variety of indirect methods, the authors propose that AXL signaling contributes to FA dynamics. This work could benefit from more mechanistic experiments that clearly define the role of AXL and the pseudokinase PEAK1 in tumor growth and metastasis. More specific criticisms are listed below. On page 4 it states “We identified an overrepresentation of MAPK, CDK, PKA and PKC kinase motifs in most of the clusters, suggesting that AXL may signal 5via regulation of these kinases (Supplementary Information, Fig. S2d). This data is waek and extremely speculative. Since it covers several of the predominant kinases it is hard to know what this means. I recommend removal of this data. The proximity ligation data presented in fig 2a is poor and hard to evaluated. Likewise, The data in figures 2b and c is not very convincing. First, quantitation of the PLA data should and Quantitation of the number of focal adhesions as dot scatter plots is a more compelling way to show the relative effect of changing AXL expression or action. Evidence for changes in the number of focal adhesions presented in figure 2c and the supplementary information is far from convincing. The Bio ID proteomic data in figures 3f and g is poorly connected to the rest of the paper. This is a rather speculative element of the work that needs some more experimental evidence to warrant its inclusion in the study. Reviewer #2 (Remarks to the Author): This manuscript offers a fascinating and important study of AXL signaling, which is timely because of the growing appreciation of the key role AXL plays in tumor resistance to targeted and chemotherapeutic drugs. The kind of understanding being pursued here, of nodes downstream of AXL activation could be very useful in suggesting effective combination treatments as well as in gaining insights into novel receptor tyrosine kinase signaling biochemistry. The authors bring together multiple approaches, experimental and computational, to ascertain relevant pathway elements including NEDD9, PEAK1, CSK, and Paxillin, toward enhancing invasive tumor cell migration via focal adhesion turnover. The resulting story is appealing and appears to be generally sound. My main point of concern relates to the technical details of Gas6 treatment for AXL stimulation. The authors seem to be not familiar with certain complications inherent in the Gas6/AXL system, such that mere treatment with soluble Gas6 is well-known to have variable effects depending on availability and characteristics of lipid vesicle-mediated interactions (for instance, as studied by Meyer et al in Cell Systems [2015]), such that AXL activation effects will vary with factors that are not explicitly considered in the experimental protocols outlined here. I believe that the authors need to provide more detailed characterization of their Gas6 protocols, including whether the cells being treated express Gas6 themselves in autocrine manner and to what degree apoptotic cell debris might be present due to low levels of nonviability in the culture conditions. These details will be vital for other laboratories in any attempts to reproduce and build upon the findings here. Reviewer #3 (Remarks to the Author): In this paper the authors try to elucidate the major signaling pathways activated by the receptor tyrosine kinase, AXL. The authors have used systems biology and proteomic approaches and find that AXL modulates the phosphorylation of a network of focal adhesion (FA) proteins that culminates in faster FA disassembly. They go on to deliniate a rather convoluted pathway whereby AXL phosphorylates the FA protein NEDD9, which then promotes NEDD9/CRKII-mediated phosphorylation of PEAK1. This pseudokinase then complexes wilt CSK kinase to phosphorylate paxillin, resulting in FA turnover. The strengths of this work lie in the biochemistry and systems biology, which are carried out very well. The finding that a major network of AXL-mediated phosphorylation events occur in FA proteins is novel, and this work does suggest that AXL can modulate FA turnover. The weaknesses lie in the biology and physiological relevance of the findings to tumor progression, especially the relevance of focal adhesion turnover-per se to AXL-mediated metastasis. Thus while the final data in Figs 5 and S7 using PEAK KO MDA-MB-231 cells suggest that PEAK may contribute to metastasis, The authors have not connected the pathway back to AXL. Thus, PEAK may have completely AXL-independent role in regulating metastasis. To make the case that it is the FA turnover pathway mediated by AXL that promotes metastasis, the authors need to demonstrate that some of the key phosphorylation and protein-protein interactions leading PEAK-mediated to paxillin phosphorylation and FA turnover also promote metastasis. Otherwise the paper does not hang together very well as it represents some excellent biochemistry and a separate finding that PEAK KO modulates metastasis. Specific Points that should be addressed: 1. Introduction and Discussion sections are missing. These need to be added to provide a context to this work and to demonstrate originality of the work 2. All the phospho-blots lack blots of the total proteins. This is essential to show that it is the phosporylation of the protein and not the total expression of the proteins that is changing. 3.In Fig 2 the analyses of the FAs should also include effects of GAS6 induced effects in addition to the constitutive situation, i.e. effects of GAS6 + R428. 4. p12. Does inhibition of paxillin phosphorylation ( directly or by using phospho-mutants) have the same effect on metastasis as PEAK KO? 5. The effect of AXL KO or the effect of AXL inhibitor on metastasis has to be carried out. Is this effect rescued by PEAK? 6. The metastasis data ( PEAK KO) need to be carried out in another TNBC cell line in addition to MDA-MB-231 cells 7 Fig s7j, why were the lungs probed ex vivo. Could in vivo signal be detected? 8. Fig s7d/e: Tumor volumes should be measured , not only IVIS signal. 9. The pathway model should be depicted in the main figures.	TPR	NC_BS_CBIO_003_RR1.txt	{ "criticism": 12, "example": 10, "importance_and_relevance": 9, "materials_and_methods": 25, "praise": 4, "presentation_and_reporting": 7, "results_and_discussion": 7, "suggestion_and_solution": 15, "total": 55 }	{ "criticism": 0.21818181818181817, "example": 0.18181818181818182, "importance_and_relevance": 0.16363636363636364, "materials_and_methods": 0.45454545454545453, "praise": 0.07272727272727272, "presentation_and_reporting": 0.12727272727272726, "results_and_discussion": 0.12727272727272726, "suggestion_and_solution": 0.2727272727272727 }	1.618182	[ { "content": "Please review this paper.", "role": "user" }, { "content": "Reviewers' comments: \n\nReviewer #1 (Remarks to the Author): \n\nThis is an involved study that implicates receptor tyrosine kinase organized by the enzyme AXL as \na potential mechanism to enhance metastasis and acquisition of resistance to cancer drugs. This is \na very complicated study with many parts that is compounded by the frequent reference to \nsupplemental data. Some of which should be omitted and some of which should be included in the \nbody of the text. Using a variety of indirect methods, the authors propose that AXL signaling \ncontributes to FA dynamics. This work could benefit from more mechanistic experiments that \nclearly define the role of AXL and the pseudokinase PEAK1 in tumor growth and metastasis. More \nspecific criticisms are listed below. \n\nOn page 4 it states “We identified an overrepresentation of MAPK, CDK, PKA and PKC kinase motifs \nin most of the clusters, suggesting that AXL may signal 5via regulation of these kinases \n(Supplementary Information, Fig. S2d). This data is waek and extremely speculative. Since it \ncovers several of the predominant kinases it is hard to know what this means. I recommend \nremoval of this data. \n\nThe proximity ligation data presented in fig 2a is poor and hard to evaluated. Likewise, The data in \nfigures 2b and c is not very convincing. First, quantitation of the PLA data should and Quantitation \nof the number of focal adhesions as dot scatter plots is a more compelling way to show the relative \neffect of changing AXL expression or action. \n\nEvidence for changes in the number of focal adhesions presented in figure 2c and the \nsupplementary information is far from convincing. \n\nThe Bio ID proteomic data in figures 3f and g is poorly connected to the rest of the paper. This is a \nrather speculative element of the work that needs some more experimental evidence to warrant its \ninclusion in the study. \n\nReviewer #2 (Remarks to the Author): \n\nThis manuscript offers a fascinating and important study of AXL signaling, which is timely because \nof the growing appreciation of the key role AXL plays in tumor resistance to targeted and \nchemotherapeutic drugs. The kind of understanding being pursued here, of nodes downstream of \nAXL activation could be very useful in suggesting effective combination treatments as well as in \ngaining insights into novel receptor tyrosine kinase signaling biochemistry. The authors bring \ntogether multiple approaches, experimental and computational, to ascertain relevant pathway \nelements including NEDD9, PEAK1, CSK, and Paxillin, toward enhancing invasive tumor cell \nmigration via focal adhesion turnover. The resulting story is appealing and appears to be generally \nsound. \n\nMy main point of concern relates to the technical details of Gas6 treatment for AXL stimulation. \nThe authors seem to be not familiar with certain complications inherent in the Gas6/AXL system, \nsuch that mere treatment with soluble Gas6 is well-known to have variable effects depending on \navailability and characteristics of lipid vesicle-mediated interactions (for instance, as studied by \nMeyer et al in Cell Systems [2015]), such that AXL activation effects will vary with factors that are \nnot explicitly considered in the experimental protocols outlined here. I believe that the authors \nneed to provide more detailed characterization of their Gas6 protocols, including whether the cells \nbeing treated express Gas6 themselves in autocrine manner and to what degree apoptotic cell \ndebris might be present due to low levels of nonviability in the culture conditions. These details will \n\n\fbe vital for other laboratories in any attempts to reproduce and build upon the findings here. \n\nReviewer #3 (Remarks to the Author): \n\nIn this paper the authors try to elucidate the major signaling pathways activated by the receptor \ntyrosine kinase, AXL. \nThe authors have used systems biology and proteomic approaches and find that AXL modulates \nthe phosphorylation of a network of focal adhesion (FA) proteins that culminates in faster FA \ndisassembly. They go on to deliniate a rather convoluted pathway whereby AXL phosphorylates the \nFA protein NEDD9, which then promotes NEDD9/CRKII-mediated phosphorylation of PEAK1. This \npseudokinase then complexes wilt CSK kinase to phosphorylate paxillin, resulting in FA turnover. \n\nThe strengths of this work lie in the biochemistry and systems biology, which are carried out very \nwell. The finding that a major network of AXL-mediated phosphorylation events occur in FA \nproteins is novel, and this work does suggest that AXL can modulate FA turnover. \n\nThe weaknesses lie in the biology and physiological relevance of the findings to tumor progression, \nespecially the relevance of focal adhesion turnover-per se to AXL-mediated metastasis. \nThus while the final data in Figs 5 and S7 using PEAK KO MDA-MB-231 cells suggest that PEAK \nmay contribute to metastasis, The authors have not connected the pathway back to AXL. Thus, \nPEAK may have completely AXL-independent role in regulating metastasis. To make the case that \nit is the FA turnover pathway \nmediated by AXL that promotes metastasis, the authors need to demonstrate that some of the key \nphosphorylation and protein-protein interactions leading PEAK-mediated to paxillin phosphorylation \nand FA turnover also promote metastasis. \nOtherwise the paper does not hang together very well as it represents some excellent biochemistry \nand a separate finding that PEAK KO modulates metastasis. \n\nSpecific Points that should be addressed: \n\n1. Introduction and Discussion sections are missing. These need to be added to provide a context \nto this work and to demonstrate originality of the work \n\n2. All the phospho-blots lack blots of the total proteins. This is essential to show that it is the \nphosporylation of the protein and not the total expression of the proteins that is changing. \n\n3.In Fig 2 the analyses of the FAs should also include effects of GAS6 induced effects in addition to \nthe constitutive \nsituation, i.e. effects of GAS6 + R428. \n\n4. p12. Does inhibition of paxillin phosphorylation ( directly or by using phospho-mutants) have \nthe same effect on metastasis as PEAK KO? \n\n5. The effect of AXL KO or the effect of AXL inhibitor on metastasis has to be carried out. Is this \neffect rescued by PEAK? \n\n6. The metastasis data ( PEAK KO) need to be carried out in another TNBC cell line in addition to \nMDA-MB-231 cells \n\n7 Fig s7j, why were the lungs probed ex vivo. Could in vivo signal be detected? \n\n8. Fig s7d/e: Tumor volumes should be measured , not only IVIS signal. \n\n9. The pathway model should be depicted in the main figures.\n\n\f", "role": "assistant" } ]	2,020
AXL confers cell migration and invasion by hijacking a PEAK1-regulated focal adhesion	REVIEWERS' COMMENTS: Reviewer #1 (Remarks to the Author): The authors have satisfactorily addressed all of the points necessary for publication. Reviewer #2 (Remarks to the Author): The authors state that now on page 6 that there is minimal phosphatidylserine due to apoptotic cells in their assay. Nonetheless, it remains imperative that their Gas6-His be complexed with phosphatidylserine in order to activate AXL. Thus, the authors need to explain where the phosphatidylserine is coming from in order to account for their Gas6-His mediated AXL phosphorylation. Reviewer #4 (Remarks to the Author): The reviewer appreciates the efforts of the authors. The revised manuscripts with new experiments have addressed most of the reviewer’s concerns. Although some caveats remain, including the functional relevance of PEAK1 phosphorylation and duplicate analysis of SILAC quantification, this reviewer agrees that they are not essential to this study and the major conclusions are supported by a large body of evidence. Overall, the work identified an important signaling pathway with functional significance in cancer metastasis and it is recommended for publication.	REVIEWERS' COMMENTS: Reviewer #1 (Remarks to the Author): The authors have satisfactorily addressed all of the points necessary for publication. Reviewer #2 (Remarks to the Author): The authors state that now on page 6 that there is minimal phosphatidylserine due to apoptotic cells in their assay. Nonetheless, it remains imperative that their Gas6-His be complexed with phosphatidylserine in order to activate AXL. Thus, the authors need to explain where the phosphatidylserine is coming from in order to account for their Gas6-His mediated AXL phosphorylation. Reviewer #4 (Remarks to the Author): The reviewer appreciates the efforts of the authors. The revised manuscripts with new experiments have addressed most of the reviewer’s concerns. Although some caveats remain, including the functional relevance of PEAK1 phosphorylation and duplicate analysis of SILAC quantification, this reviewer agrees that they are not essential to this study and the major conclusions are supported by a large body of evidence. Overall, the work identified an important signaling pathway with functional significance in cancer metastasis and it is recommended for publication.	TPR	NC_BS_CBIO_003_RR3.txt	{ "criticism": 1, "example": 1, "importance_and_relevance": 2, "materials_and_methods": 5, "praise": 4, "presentation_and_reporting": 1, "results_and_discussion": 1, "suggestion_and_solution": 2, "total": 8 }	{ "criticism": 0.125, "example": 0.125, "importance_and_relevance": 0.25, "materials_and_methods": 0.625, "praise": 0.5, "presentation_and_reporting": 0.125, "results_and_discussion": 0.125, "suggestion_and_solution": 0.25 }	2.125	[ { "content": "Please review this paper.", "role": "user" }, { "content": "\n\n\fREVIEWERS' COMMENTS: \n\nReviewer #1 (Remarks to the Author): \n\nThe authors have satisfactorily addressed all of the points necessary for publication. \n\nReviewer #2 (Remarks to the Author): \n\nThe authors state that now on page 6 that there is minimal phosphatidylserine due to apoptotic \ncells in their assay. Nonetheless, it remains imperative that their Gas6-His be complexed with \nphosphatidylserine in order to activate AXL. Thus, the authors need to explain where the \nphosphatidylserine is coming from in order to account for their Gas6-His mediated AXL \nphosphorylation. \n\nReviewer #4 (Remarks to the Author): \n\nThe reviewer appreciates the efforts of the authors. The revised manuscripts with new experiments \nhave addressed most of the reviewer’s concerns. Although some caveats remain, including the \nfunctional relevance of PEAK1 phosphorylation and duplicate analysis of SILAC quantification, this \nreviewer agrees that they are not essential to this study and the major conclusions are supported \nby a large body of evidence. Overall, the work identified an important signaling pathway with \nfunctional significance in cancer metastasis and it is recommended for publication.\n\n\f", "role": "assistant" } ]	2,020
Micro-environmental sensing by bone marrow stroma identifies IL-6 and TGFβ1 as	REVIEWERS' COMMENTS: Reviewer #1 (Remarks to the Author): The authors have adequately addressed the comments of all reviewers, and have considerabley improved the manuscript. Reviewer #2 (Remarks to the Author): The authors have satisfactorily addressed my concerns. Reviewer #3 (Remarks to the Author): In this study, the authors interrogate the basis for the altered hematopoietic stem cell function with age. Using a combinatorial approach of probing stromal cells as molecular sensors for cytokine signalling, they reveal that both Tgfb and IL6 signalling is altered with age. By a series of functional experiments (predominantly by inhibiting the activity of these mitogens) they are able to separate the functional effects of these pathways on overall lineage potential and erythropoiesis, respectively. The paper has many original lines of experimentation and is likely to be influential in the field. The results are clearly presented, the methods are well described, and they have adequately adressed all the concerns I raised in connection to my review of their original submitted manuscript.	REVIEWERS' COMMENTS: Reviewer #1 (Remarks to the Author): The authors have adequately addressed the comments of all reviewers, and have considerabley improved the manuscript. Reviewer #2 (Remarks to the Author): The authors have satisfactorily addressed my concerns. Reviewer #3 (Remarks to the Author): In this study, the authors interrogate the basis for the altered hematopoietic stem cell function with age. Using a combinatorial approach of probing stromal cells as molecular sensors for cytokine signalling, they reveal that both Tgfb and IL6 signalling is altered with age. By a series of functional experiments (predominantly by inhibiting the activity of these mitogens) they are able to separate the functional effects of these pathways on overall lineage potential and erythropoiesis, respectively. The paper has many original lines of experimentation and is likely to be influential in the field. The results are clearly presented, the methods are well described, and they have adequately adressed all the concerns I raised in connection to my review of their original submitted manuscript.	TPR	NC_BS_CBIO_008_RR2.txt	{ "criticism": 0, "example": 0, "importance_and_relevance": 1, "materials_and_methods": 4, "praise": 4, "presentation_and_reporting": 0, "results_and_discussion": 1, "suggestion_and_solution": 0, "total": 7 }	{ "criticism": 0, "example": 0, "importance_and_relevance": 0.14285714285714285, "materials_and_methods": 0.5714285714285714, "praise": 0.5714285714285714, "presentation_and_reporting": 0, "results_and_discussion": 0.14285714285714285, "suggestion_and_solution": 0 }	1.428571	[ { "content": "Please review this paper.", "role": "user" }, { "content": "\fREVIEWERS' COMMENTS: \n\nReviewer #1 (Remarks to the Author): \n\nThe authors have adequately addressed the comments of all reviewers, and have considerabley \nimproved the manuscript. \n\nReviewer #2 (Remarks to the Author): \n\nThe authors have satisfactorily addressed my concerns. \n\nReviewer #3 (Remarks to the Author): \n\nIn this study, the authors interrogate the basis for the altered hematopoietic stem cell function with \nage. Using a combinatorial approach of probing stromal cells as molecular sensors for cytokine \nsignalling, they reveal that both Tgfb and IL6 signalling is altered with age. By a series of functional \nexperiments (predominantly by inhibiting the activity of these mitogens) they are able to separate the \nfunctional effects of these pathways on overall lineage potential and erythropoiesis, respectively. \n\nThe paper has many original lines of experimentation and is likely to be influential in the field. The \nresults are clearly presented, the methods are well described, and they have adequately adressed all \nthe concerns I raised in connection to my review of their original submitted manuscript.\n\n\f\t\n\n\f\n", "role": "assistant" } ]	2,020
Micro-environmental sensing by bone marrow stroma identifies IL-6 and TGFβ1 as	Reviewers' comments: Reviewer #1 (Remarks to the Author): In this manuscript, Valletta et al. analyzed the stromal cells of the bone marrow to address the decline of erythropoiesis and lymphopoiesis upon ageing. They found upregulation of IL-6 and TGFb signaling induced gene expression in BM stromal cells of aged mice, and could subsequently demonstrate that both cytokines play an important, but distinct role in this process. The authors demonstrate HSC- intrinsic and extrinsic mechanisms that drive hematopoietic decline during ageing, and associate the BM stroma in this process. The amount of work presented is substantial, and the fact that the involvement of not one, but two cytokines is investigated, provides extra value to this manuscript. This study is important for the field, as it helps us to understand how vital physiological processes deteriorate during ageing, and what could be done to counteract this. The manuscript is very well written, and the experiments are of a high technical level and well- performed. Data is generally well displayed and conclusions are in line with the findings. Although most of the findings and claims are novel, it should be noted that the contribution of IL-6 to erythroid decline during ageing has been shown before in several studies (some of them also referenced by the authors; Refs 21-24). Eventhough not all of the observations are experimentally addressed and still remain unclear (eg. what is the contribution of the other inflammatory networks, beyond IL-6 and TGFb, that are also observed in BM stromal cells of aged mice?), it is obvious that gene expression analysis evokes new questions for future studies. Hence, I restrict my feedback to the questions that the authors experimentally addressed and the conclusions they drew from their experiments. Major comments: A. Based on the stromal cell analysis, the authors hypothesized that increased IL-6 signaling could underly (part of) the ageing phenotype. Indeed, based on blocking experiments with anti-IL-6 antibodies (Figure 3), the authors conclude that IL-6 is an important driver of the age-mediated decline in erythropoiesis, which agrees with several other studies. However, it is not clear whether the authors now conclude that IL-6 affects erythropoiesis through the BM stroma or that it acts directly on the erythroid progenitor cells? It could well be that the observed IL-6 mediated changes in BM stroma are an epiphenomenon and not causally related to the erythroid decline, but this should be made clear, given the important focus in the manuscript on the stromal cells in the BM. B. The authors argue that TGFb acts independently of IL-6, and plays a different role in the ageing of HSCs, LSK and myelo-erythroid progenitor compartments; here, it is made likely that the stroma itself is actively contributing, as the authors found more TGF-b1 in ageing compared to young stroma. However, it has been well established that TGFb is a major determinant of HSC quiescence, which is neither referred to nor experimentally addressed here. It is conceivable that the effects observed with the TGFbR-inhibitor in Figures 4 and 5 are simply due to a loss of HSC quiescence, which subsequently resulted in enhanced/increased differentiation of the HSCs and downstream progenitors. It is important that the authors adequately address this possibility and, if necessary, adjust their conclusions accordingly. Minor comments: C. The authors show a lot of GSEA graphs, but it is unclear for the reader which genes are part of the indicated enrichment. For instance, in Figure 2h there is an increase in so-called “preGM-associated genes” in CFU-Es from old mice, but the identity of these genes is not revealed and it is also not clear which genes within these gene-sets are indeed higher or lower. It would be highly valuable if the GSEA analyses that are shown throughout this manuscript would be expanded by including the related genes in a supplemental figure or table. D. In figure 2, the authors describe that purified MPP2, preCFU-Es and CD71+ CFU-Es from aged mice all have decreased erythroid differentiation (Figure 2f). The prediction is that they all show a gain in myeloid differentiation, but this is only shown for the preCFU-E population in Figure 2i. It would be worth to also show the myeloid potential of the MPP2 population, as this would reveal how far up in the hematopoietic hierarchy the switch from erythroid to myeloid occurs. E. In Figure 3g&h, the authors show in a GSEA that the loss of preCFU-E genes upon ageing is prevented by depletion of IL-6. It is important to also show whether the enrichment of preGM genes upon ageing (Fig 2h) is also lost by the IL-6 depletion. This would denote whether IL-6 truly affects a shift in the balance between erythroid and myeloid commitment, or only affects erythroid differentiation. F. At the bottom of page 5, the authors refer to Suppl Fig 3e, but this does not exist and should be Suppl Fig 4a. Reviewer #2 (Remarks to the Author): In the manuscript, “Using bone marrow stromal cells as micro-environmental sensors identifies IL-6 and TGFβ signaling as regulators of declining erythropoiesis and lymphopoiesis during hematopoietic ageing,” Valletta et al. utilized bone marrow (BM) stromal cells to determine consistent differences in cellular signaling across multiple cell types between young and aged BM and find enhanced IL-6 and TGF beta gene programs in aged BM. Next, hematopoietic stem cells (HSC) were assessed functionally by transplantation to determine if age related defects in erythropoiesis and lymphopoiesis are cell- autonomous, and the author observed small non-significant changes in the intrinsic ability for aged HSC to initiate erythropoiesis while declining lymphopoiesis was a cell-autonomous effect. Taken together these data indicated both cell-autonomous and non-cell-autonomous effects on the ability of aged HSC to produce normal lineage output. Overall, the work shows some important findings relating to cell-extrinsic signaling networks in aging impacting platelet and erythroid output, which are often ignored in favor of myeloid vs. lymphoid lineage alterations only. Major concerns: Inflammatory gene signatures in aging HSC are well described in prior work (Chambers, Plos Biology 2007) Here, RNA-seq analysis of stromal cells identified a similarly inflammatory signature that impacts hematopoiesis. Thus, the value of analyzing gene expression in the stroma or calling them 'sensors' is not clear, given gene expression analysis of HSC/HSPC would yield similar results. Value would be added to this analysis if the impact of the inflammatory signature on HSC maintenance function of the stromal cells themselves is addressed. Indeed a larger issue with the work as a whole is that a preliminary level of mechanistic insight with respect to the impact of TGF-beta and/or IL-6 on aged hematopoietic function is provided. Increased levels of IL-6/TGF (as opposed to increased signaling activity in these pathways due to cell- intrinsic changes) in the BM are not shown. Indeed other previous analysis of aged BM (Ergen et al, Blood 2012) show decreased IL-6 levels in BM fluid by direct interrogation, whereas the study cited by the authors (Mei et al, 2018) only shows protein levels in serum and not BM (indirect measurement of IL-6 was performed by qRT-PCR in this study). Echoing the point above relating to mechanism, it will be important for the authors to distinguish between the alternative hypotheses of increased levels of these factors in situ in the BM vs. increased signaling activity due to intrinsic deregulations. Reviewer #3 (Remarks to the Author): In this paper from Valetta et al., the authors attempt to find an answer to the molecular features underlying ageing hematopoiesis/aging HSC behavior. The studies try to argue for a role for IL6 and Tgfb signaling in controlling hematopoiesis at two levels (erythroid progenitor cells and HSCs, respectively). The interpretations appear for the most part to be supported by the data provided and statistics (where applied) seem adequate. The authors first describe the general features of ageing hematopoiesis, focusing on decreased erythropoiesis and enhanced platelet (biogenesis?). They validate previous findings and apply a transgenic mouse model that permit tracing of these two lineages (demonstrating that also this transgenic strain exhibits these well-known age-associated features) (Figure 1). In essence, the authors conclude that while there is a general reduction of the contribution of aged HSCs to multilineage hematopoiesis, this is more pronounced for the lymphoid lineages. Not sure if there is anything wrong with the resolution in my figures, but I cannot really see any cells in the gates connected to Figure 1b (plots to far right) and Figure 1d (CD11b vs Gr1 plot; no myeloid cells?). Having verified/established these basic features, the authors next investigate myeloerythroid progenitor cells in young and aged mice. They demonstrate an increase in megakaryocyte progenitors and a decrease of early erythroid progenitors (including MPP2s – it could benefit the general reader to explain a bit more why these cells were investigated). Finally, authors conduct some gene expression studies and culture experiments to demonstrate that the myeloid activity of aged preCFU-Es is higher from aged cells. These latter experiments seem to indicate that the myeloid (as opposed to erythroid) activity/cell goes up with age. This raises some concerns as to the way the prospective isolation was made – if f.i. CD55 had been included (positivity) as a marker for candidate preCFU-Es, would the authors obtain the same results? The authors next go on to use the rather clever approach of using stromal/microenvironmental cells as probes for what might underlie the hematopoietic phenotypes of aging. This lead to the identification of two candidate signaling pathways; IL6 and Tgfb. Further experimentation is focusing on the neutralisation of the two pathways, followed by re-evaluation of hematopoiesis. From these lines of studies, the conclusions are: 1) that IL6 exerts negative effects of erythropoiesis (rather than having direct effects on HSCs) while 2) Tgfb-inhibition appears to restore the lineage (or rather progenitor) output from aged HSCs (including the overproduction of megakaryocyte progenitors; what about platelets?). They provide some molecular evidence (several types of genome-wide RNA expression experiments) to support their conclusions. Finally, the investigators conduct transplantation experiment using purified HSCs from vehicle or Tgfb- inhibited treated mice. They try to demonstrate that this treatment ”normalizes” the output from HSCs with enhanced relative output of non-platelet lineages. It appears that the overall reconstitution of Tgfbi-inhibited cells is dramatically lower than that from vehicle-treated. What is going on here? It appears to this reviewer that Tgfb-inhibition is causing an overall decline in HSC function, that perhaps has some selectivity for HSCs with a ”myeloid- bias”/”compromised lymphopoiesis”. But this is very difficult to conclude if the variation in overall reconstitution is very compromised – f.i. low-level chimeras tend to be skewed towards “lymphoid- biased” cells (which is most likely reflecting more on the longevity of mature lymphoid cells). Some emphasis is on the “rescue” of B lymphopoiesis (Figure 5e) as a consequence of Tgfb-inhibition, but details on whether there is restoration of B lymphopoiesis in the bone marrow is lacking. It would have been very interesting for the authors to provide some information specifically on IL7-signaling, as compromised IL7 signaling in the past has been proposed to be a major mechanism for the reduced B (and/or T) lymphopoiesis with age.	Reviewers' comments: Reviewer #1 (Remarks to the Author): In this manuscript, Valletta et al. analyzed the stromal cells of the bone marrow to address the decline of erythropoiesis and lymphopoiesis upon ageing. They found upregulation of IL-6 and TGFb signaling induced gene expression in BM stromal cells of aged mice, and could subsequently demonstrate that both cytokines play an important, but distinct role in this process. The authors demonstrate HSC- intrinsic and extrinsic mechanisms that drive hematopoietic decline during ageing, and associate the BM stroma in this process. The amount of work presented is substantial, and the fact that the involvement of not one, but two cytokines is investigated, provides extra value to this manuscript. This study is important for the field, as it helps us to understand how vital physiological processes deteriorate during ageing, and what could be done to counteract this. The manuscript is very well written, and the experiments are of a high technical level and well- performed. Data is generally well displayed and conclusions are in line with the findings. Although most of the findings and claims are novel, it should be noted that the contribution of IL-6 to erythroid decline during ageing has been shown before in several studies (some of them also referenced by the authors; Refs 21-24). Eventhough not all of the observations are experimentally addressed and still remain unclear (eg. what is the contribution of the other inflammatory networks, beyond IL-6 and TGFb, that are also observed in BM stromal cells of aged mice?), it is obvious that gene expression analysis evokes new questions for future studies. Hence, I restrict my feedback to the questions that the authors experimentally addressed and the conclusions they drew from their experiments. Major comments: A. Based on the stromal cell analysis, the authors hypothesized that increased IL-6 signaling could underly (part of) the ageing phenotype. Indeed, based on blocking experiments with anti-IL-6 antibodies (Figure 3), the authors conclude that IL-6 is an important driver of the age-mediated decline in erythropoiesis, which agrees with several other studies. However, it is not clear whether the authors now conclude that IL-6 affects erythropoiesis through the BM stroma or that it acts directly on the erythroid progenitor cells? It could well be that the observed IL-6 mediated changes in BM stroma are an epiphenomenon and not causally related to the erythroid decline, but this should be made clear, given the important focus in the manuscript on the stromal cells in the BM. B. The authors argue that TGFb acts independently of IL-6, and plays a different role in the ageing of HSCs, LSK and myelo-erythroid progenitor compartments; here, it is made likely that the stroma itself is actively contributing, as the authors found more TGF-b1 in ageing compared to young stroma. However, it has been well established that TGFb is a major determinant of HSC quiescence, which is neither referred to nor experimentally addressed here. It is conceivable that the effects observed with the TGFbR-inhibitor in Figures 4 and 5 are simply due to a loss of HSC quiescence, which subsequently resulted in enhanced/increased differentiation of the HSCs and downstream progenitors. It is important that the authors adequately address this possibility and, if necessary, adjust their conclusions accordingly. Minor comments: C. The authors show a lot of GSEA graphs, but it is unclear for the reader which genes are part of the indicated enrichment. For instance, in Figure 2h there is an increase in so-called “preGM-associated genes” in CFU-Es from old mice, but the identity of these genes is not revealed and it is also not clear which genes within these gene-sets are indeed higher or lower. It would be highly valuable if the GSEA analyses that are shown throughout this manuscript would be expanded by including the related genes in a supplemental figure or table. D. In figure 2, the authors describe that purified MPP2, preCFU-Es and CD71+ CFU-Es from aged mice all have decreased erythroid differentiation (Figure 2f). The prediction is that they all show a gain in myeloid differentiation, but this is only shown for the preCFU-E population in Figure 2i. It would be worth to also show the myeloid potential of the MPP2 population, as this would reveal how far up in the hematopoietic hierarchy the switch from erythroid to myeloid occurs. E. In Figure 3g&h, the authors show in a GSEA that the loss of preCFU-E genes upon ageing is prevented by depletion of IL-6. It is important to also show whether the enrichment of preGM genes upon ageing (Fig 2h) is also lost by the IL-6 depletion. This would denote whether IL-6 truly affects a shift in the balance between erythroid and myeloid commitment, or only affects erythroid differentiation. F. At the bottom of page 5, the authors refer to Suppl Fig 3e, but this does not exist and should be Suppl Fig 4a. Reviewer #2 (Remarks to the Author): In the manuscript, “Using bone marrow stromal cells as micro-environmental sensors identifies IL-6 and TGFβ signaling as regulators of declining erythropoiesis and lymphopoiesis during hematopoietic ageing,” Valletta et al. utilized bone marrow (BM) stromal cells to determine consistent differences in cellular signaling across multiple cell types between young and aged BM and find enhanced IL-6 and TGF beta gene programs in aged BM. Next, hematopoietic stem cells (HSC) were assessed functionally by transplantation to determine if age related defects in erythropoiesis and lymphopoiesis are cell- autonomous, and the author observed small non-significant changes in the intrinsic ability for aged HSC to initiate erythropoiesis while declining lymphopoiesis was a cell-autonomous effect. Taken together these data indicated both cell-autonomous and non-cell-autonomous effects on the ability of aged HSC to produce normal lineage output. Overall, the work shows some important findings relating to cell-extrinsic signaling networks in aging impacting platelet and erythroid output, which are often ignored in favor of myeloid vs. lymphoid lineage alterations only. Major concerns: Inflammatory gene signatures in aging HSC are well described in prior work (Chambers, Plos Biology 2007) Here, RNA-seq analysis of stromal cells identified a similarly inflammatory signature that impacts hematopoiesis. Thus, the value of analyzing gene expression in the stroma or calling them 'sensors' is not clear, given gene expression analysis of HSC/HSPC would yield similar results. Value would be added to this analysis if the impact of the inflammatory signature on HSC maintenance function of the stromal cells themselves is addressed. Indeed a larger issue with the work as a whole is that a preliminary level of mechanistic insight with respect to the impact of TGF-beta and/or IL-6 on aged hematopoietic function is provided. Increased levels of IL-6/TGF (as opposed to increased signaling activity in these pathways due to cell- intrinsic changes) in the BM are not shown. Indeed other previous analysis of aged BM (Ergen et al, Blood 2012) show decreased IL-6 levels in BM fluid by direct interrogation, whereas the study cited by the authors (Mei et al, 2018) only shows protein levels in serum and not BM (indirect measurement of IL-6 was performed by qRT-PCR in this study). Echoing the point above relating to mechanism, it will be important for the authors to distinguish between the alternative hypotheses of increased levels of these factors in situ in the BM vs. increased signaling activity due to intrinsic deregulations. Reviewer #3 (Remarks to the Author): In this paper from Valetta et al., the authors attempt to find an answer to the molecular features underlying ageing hematopoiesis/aging HSC behavior. The studies try to argue for a role for IL6 and Tgfb signaling in controlling hematopoiesis at two levels (erythroid progenitor cells and HSCs, respectively). The interpretations appear for the most part to be supported by the data provided and statistics (where applied) seem adequate. The authors first describe the general features of ageing hematopoiesis, focusing on decreased erythropoiesis and enhanced platelet (biogenesis?). They validate previous findings and apply a transgenic mouse model that permit tracing of these two lineages (demonstrating that also this transgenic strain exhibits these well-known age-associated features) (Figure 1). In essence, the authors conclude that while there is a general reduction of the contribution of aged HSCs to multilineage hematopoiesis, this is more pronounced for the lymphoid lineages. Not sure if there is anything wrong with the resolution in my figures, but I cannot really see any cells in the gates connected to Figure 1b (plots to far right) and Figure 1d (CD11b vs Gr1 plot; no myeloid cells?). Having verified/established these basic features, the authors next investigate myeloerythroid progenitor cells in young and aged mice. They demonstrate an increase in megakaryocyte progenitors and a decrease of early erythroid progenitors (including MPP2s – it could benefit the general reader to explain a bit more why these cells were investigated). Finally, authors conduct some gene expression studies and culture experiments to demonstrate that the myeloid activity of aged preCFU-Es is higher from aged cells. These latter experiments seem to indicate that the myeloid (as opposed to erythroid) activity/cell goes up with age. This raises some concerns as to the way the prospective isolation was made – if f.i. CD55 had been included (positivity) as a marker for candidate preCFU-Es, would the authors obtain the same results? The authors next go on to use the rather clever approach of using stromal/microenvironmental cells as probes for what might underlie the hematopoietic phenotypes of aging. This lead to the identification of two candidate signaling pathways; IL6 and Tgfb. Further experimentation is focusing on the neutralisation of the two pathways, followed by re-evaluation of hematopoiesis. From these lines of studies, the conclusions are: 1) that IL6 exerts negative effects of erythropoiesis (rather than having direct effects on HSCs) while 2) Tgfb-inhibition appears to restore the lineage (or rather progenitor) output from aged HSCs (including the overproduction of megakaryocyte progenitors; what about platelets?). They provide some molecular evidence (several types of genome-wide RNA expression experiments) to support their conclusions. Finally, the investigators conduct transplantation experiment using purified HSCs from vehicle or Tgfb- inhibited treated mice. They try to demonstrate that this treatment ”normalizes” the output from HSCs with enhanced relative output of non-platelet lineages. It appears that the overall reconstitution of Tgfbi-inhibited cells is dramatically lower than that from vehicle-treated. What is going on here? It appears to this reviewer that Tgfb-inhibition is causing an overall decline in HSC function, that perhaps has some selectivity for HSCs with a ”myeloid- bias”/”compromised lymphopoiesis”. But this is very difficult to conclude if the variation in overall reconstitution is very compromised – f.i. low-level chimeras tend to be skewed towards “lymphoid- biased” cells (which is most likely reflecting more on the longevity of mature lymphoid cells). Some emphasis is on the “rescue” of B lymphopoiesis (Figure 5e) as a consequence of Tgfb-inhibition, but details on whether there is restoration of B lymphopoiesis in the bone marrow is lacking. It would have been very interesting for the authors to provide some information specifically on IL7-signaling, as compromised IL7 signaling in the past has been proposed to be a major mechanism for the reduced B (and/or T) lymphopoiesis with age.	TPR	NC_BS_CBIO_008_RR1.txt	{ "criticism": 9, "example": 10, "importance_and_relevance": 8, "materials_and_methods": 42, "praise": 7, "presentation_and_reporting": 5, "results_and_discussion": 16, "suggestion_and_solution": 11, "total": 72 }	{ "criticism": 0.125, "example": 0.1388888888888889, "importance_and_relevance": 0.1111111111111111, "materials_and_methods": 0.5833333333333334, "praise": 0.09722222222222222, "presentation_and_reporting": 0.06944444444444445, "results_and_discussion": 0.2222222222222222, "suggestion_and_solution": 0.1527777777777778 }	1.5	[ { "content": "Please review this paper.", "role": "user" }, { "content": "Reviewers' comments: \n\nReviewer #1 (Remarks to the Author): \n\nIn this manuscript, Valletta et al. analyzed the stromal cells of the bone marrow to address the decline \nof erythropoiesis and lymphopoiesis upon ageing. They found upregulation of IL-6 and TGFb signaling \ninduced gene expression in BM stromal cells of aged mice, and could subsequently demonstrate that \nboth cytokines play an important, but distinct role in this process. The authors demonstrate HSC-\nintrinsic and extrinsic mechanisms that drive hematopoietic decline during ageing, and associate the \nBM stroma in this process. The amount of work presented is substantial, and the fact that the \ninvolvement of not one, but two cytokines is investigated, provides extra value to this manuscript. \nThis study is important for the field, as it helps us to understand how vital physiological processes \ndeteriorate during ageing, and what could be done to counteract this. \nThe manuscript is very well written, and the experiments are of a high technical level and well-\nperformed. Data is generally well displayed and conclusions are in line with the findings. Although \nmost of the findings and claims are novel, it should be noted that the contribution of IL-6 to erythroid \ndecline during ageing has been shown before in several studies (some of them also referenced by the \nauthors; Refs 21-24). \nEventhough not all of the observations are experimentally addressed and still remain unclear (eg. \nwhat is the contribution of the other inflammatory networks, beyond IL-6 and TGFb, that are also \nobserved in BM stromal cells of aged mice?), it is obvious that gene expression analysis evokes new \nquestions for future studies. Hence, I restrict my feedback to the questions that the authors \nexperimentally addressed and the conclusions they drew from their experiments. \n\nMajor comments: \nA. Based on the stromal cell analysis, the authors hypothesized that increased IL-6 signaling could \nunderly (part of) the ageing phenotype. Indeed, based on blocking experiments with anti-IL-6 \nantibodies (Figure 3), the authors conclude that IL-6 is an important driver of the age-mediated \ndecline in erythropoiesis, which agrees with several other studies. However, it is not clear whether the \nauthors now conclude that IL-6 affects erythropoiesis through the BM stroma or that it acts directly on \nthe erythroid progenitor cells? It could well be that the observed IL-6 mediated changes in BM stroma \nare an epiphenomenon and not causally related to the erythroid decline, but this should be made clear, \ngiven the important focus in the manuscript on the stromal cells in the BM. \nB. The authors argue that TGFb acts independently of IL-6, and plays a different role in the ageing of \nHSCs, LSK and myelo-erythroid progenitor compartments; here, it is made likely that the stroma itself \nis actively contributing, as the authors found more TGF-b1 in ageing compared to young stroma. \nHowever, it has been well established that TGFb is a major determinant of HSC quiescence, which is \nneither referred to nor experimentally addressed here. It is conceivable that the effects observed with \nthe TGFbR-inhibitor in Figures 4 and 5 are simply due to a loss of HSC quiescence, which subsequently \nresulted in enhanced/increased differentiation of the HSCs and downstream progenitors. It is \nimportant that the authors adequately address this possibility and, if necessary, adjust their \nconclusions accordingly. \n\nMinor comments: \nC. The authors show a lot of GSEA graphs, but it is unclear for the reader which genes are part of the \nindicated enrichment. For instance, in Figure 2h there is an increase in so-called “preGM-associated \ngenes” in CFU-Es from old mice, but the identity of these genes is not revealed and it is also not clear \nwhich genes within these gene-sets are indeed higher or lower. It would be highly valuable if the \nGSEA analyses that are shown throughout this manuscript would be expanded by including the related \ngenes in a supplemental figure or table. \nD. In figure 2, the authors describe that purified MPP2, preCFU-Es and CD71+ CFU-Es from aged mice \n\n\fall have decreased erythroid differentiation (Figure 2f). The prediction is that they all show a gain in \nmyeloid differentiation, but this is only shown for the preCFU-E population in Figure 2i. It would be \nworth to also show the myeloid potential of the MPP2 population, as this would reveal how far up in \nthe hematopoietic hierarchy the switch from erythroid to myeloid occurs. \nE. In Figure 3g&h, the authors show in a GSEA that the loss of preCFU-E genes upon ageing is \nprevented by depletion of IL-6. It is important to also show whether the enrichment of preGM genes \nupon ageing (Fig 2h) is also lost by the IL-6 depletion. This would denote whether IL-6 truly affects a \nshift in the balance between erythroid and myeloid commitment, or only affects erythroid \ndifferentiation. \nF. At the bottom of page 5, the authors refer to Suppl Fig 3e, but this does not exist and should be \nSuppl Fig 4a. \n\nReviewer #2 (Remarks to the Author): \n\nIn the manuscript, “Using bone marrow stromal cells as micro-environmental sensors identifies IL-6 \nand TGFβ signaling as regulators of declining erythropoiesis and lymphopoiesis during hematopoietic \nageing,” Valletta et al. utilized bone marrow (BM) stromal cells to determine consistent differences in \ncellular signaling across multiple cell types between young and aged BM and find enhanced IL-6 and \nTGF beta gene programs in aged BM. Next, hematopoietic stem cells (HSC) were assessed functionally \nby transplantation to determine if age related defects in erythropoiesis and lymphopoiesis are cell-\nautonomous, and the author observed small non-significant changes in the intrinsic ability for aged \nHSC to initiate erythropoiesis while declining lymphopoiesis was a cell-autonomous effect. Taken \ntogether these data indicated both cell-autonomous and non-cell-autonomous effects on the ability of \naged HSC to produce normal lineage output. \n\nOverall, the work shows some important findings relating to cell-extrinsic signaling networks in aging \nimpacting platelet and erythroid output, which are often ignored in favor of myeloid vs. lymphoid \nlineage alterations only. \n\nMajor concerns: \n\nInflammatory gene signatures in aging HSC are well described in prior work (Chambers, Plos Biology \n2007) Here, RNA-seq analysis of stromal cells identified a similarly inflammatory signature that \nimpacts hematopoiesis. Thus, the value of analyzing gene expression in the stroma or calling them \n'sensors' is not clear, given gene expression analysis of HSC/HSPC would yield similar results. Value \nwould be added to this analysis if the impact of the inflammatory signature on HSC maintenance \nfunction of the stromal cells themselves is addressed. Indeed a larger issue with the work as a whole \nis that a preliminary level of mechanistic insight with respect to the impact of TGF-beta and/or IL-6 on \naged hematopoietic function is provided. \n\nIncreased levels of IL-6/TGF (as opposed to increased signaling activity in these pathways due to cell-\nintrinsic changes) in the BM are not shown. Indeed other previous analysis of aged BM (Ergen et al, \nBlood 2012) show decreased IL-6 levels in BM fluid by direct interrogation, whereas the study cited by \nthe authors (Mei et al, 2018) only shows protein levels in serum and not BM (indirect measurement of \nIL-6 was performed by qRT-PCR in this study). Echoing the point above relating to mechanism, it will \nbe important for the authors to distinguish between the alternative hypotheses of increased levels of \nthese factors in situ in the BM vs. increased signaling activity due to intrinsic deregulations. \n\n\fReviewer #3 (Remarks to the Author): \n\nIn this paper from Valetta et al., the authors attempt to find an answer to the molecular features \nunderlying ageing hematopoiesis/aging HSC behavior. The studies try to argue for a role for IL6 and \nTgfb signaling in controlling hematopoiesis at two levels (erythroid progenitor cells and HSCs, \nrespectively). The interpretations appear for the most part to be supported by the data provided and \nstatistics (where applied) seem adequate. \n\nThe authors first describe the general features of ageing hematopoiesis, focusing on decreased \nerythropoiesis and enhanced platelet (biogenesis?). They validate previous findings and apply a \ntransgenic mouse model that permit tracing of these two lineages (demonstrating that also this \ntransgenic strain exhibits these well-known age-associated features) (Figure 1). In essence, the \nauthors conclude that while there is a general reduction of the contribution of aged HSCs to \nmultilineage hematopoiesis, this is more pronounced for the lymphoid lineages. \n\nNot sure if there is anything wrong with the resolution in my figures, but I cannot really see any cells \nin the gates connected to Figure 1b (plots to far right) and Figure 1d (CD11b vs Gr1 plot; no myeloid \ncells?). \n\nHaving verified/established these basic features, the authors next investigate myeloerythroid \nprogenitor cells in young and aged mice. They demonstrate an increase in megakaryocyte progenitors \nand a decrease of early erythroid progenitors (including MPP2s – it could benefit the general reader to \nexplain a bit more why these cells were investigated). Finally, authors conduct some gene expression \nstudies and culture experiments to demonstrate that the myeloid activity of aged preCFU-Es is higher \nfrom aged cells. These latter experiments seem to indicate that the myeloid (as opposed to erythroid) \nactivity/cell goes up with age. This raises some concerns as to the way the prospective isolation was \nmade – if f.i. CD55 had been included (positivity) as a marker for candidate preCFU-Es, would the \nauthors obtain the same results? \n\nThe authors next go on to use the rather clever approach of using stromal/microenvironmental cells as \nprobes for what might underlie the hematopoietic phenotypes of aging. This lead to the identification \nof two candidate signaling pathways; IL6 and Tgfb. Further experimentation is focusing on the \nneutralisation of the two pathways, followed by re-evaluation of hematopoiesis. From these lines of \nstudies, the conclusions are: 1) that IL6 exerts negative effects of erythropoiesis (rather than having \ndirect effects on HSCs) while 2) Tgfb-inhibition appears to restore the lineage (or rather progenitor) \noutput from aged HSCs (including the overproduction of megakaryocyte progenitors; what about \nplatelets?). They provide some molecular evidence (several types of genome-wide RNA expression \nexperiments) to support their conclusions. \n\nFinally, the investigators conduct transplantation experiment using purified HSCs from vehicle or Tgfb-\ninhibited treated mice. They try to demonstrate that this treatment ”normalizes” the output from HSCs \nwith enhanced relative output of non-platelet lineages. \n\nIt appears that the overall reconstitution of Tgfbi-inhibited cells is dramatically lower than that from \nvehicle-treated. What is going on here? It appears to this reviewer that Tgfb-inhibition is causing an \noverall decline in HSC function, that perhaps has some selectivity for HSCs with a ”myeloid-\nbias”/”compromised lymphopoiesis”. But this is very difficult to conclude if the variation in overall \nreconstitution is very compromised – f.i. low-level chimeras tend to be skewed towards “lymphoid-\nbiased” cells (which is most likely reflecting more on the longevity of mature lymphoid cells). \n\n\fSome emphasis is on the “rescue” of B lymphopoiesis (Figure 5e) as a consequence of Tgfb-inhibition, \nbut details on whether there is restoration of B lymphopoiesis in the bone marrow is lacking. It would \nhave been very interesting for the authors to provide some information specifically on IL7-signaling, \nas compromised IL7 signaling in the past has been proposed to be a major mechanism for the reduced \nB (and/or T) lymphopoiesis with age.\n\n\f", "role": "assistant" } ]	2,020
E2F-dependent transcription determines replication capacity and S phase length	Reviewers' comments: Reviewer #1 (Remarks to the Author): In this manuscript, Pennycook et al. intend to propose a new concept, called ‘replication capacity’, which the authors define as the maximum amount of DNA a cell is able to synthesize per unit of time throughout S phase. The authors argue that E2F-dependent transcription determines the replication capacity of a cell. The manuscript is well written, however, the conclusions are overstated and the experimental evidence is rather preliminary. Some of the results are inconsistent between experiments and some statistical differences are arguable. In general, the authors test the effect of knocking-down E2F6 during S phase. With this, the authors claim in the title that E2F activity determines replication capacity, which is an overstatement. The comments below indicate the multiple problems with this manuscript, in terms of experimental design, controls and interpretation. Overall, the present results do not support the author’s model and conclusions. I suggest addressing at least some of the specific comments listed below before submitting the manuscript elsewhere. Major comments: 1. The authors modulate E2F activity only via E2F6, but claim that general E2F activity is the driver for replication capacity (the maximal amount of DNA a cell is able to synthesise per unit time throughout S phase). Proof for their hypothesis falls short as no general E2F activity readout is provided and a positive E2F transcritption factor (e.g. E2F1) is not modulated whatsoever. The authors should therefore focus on claims for E2F6-dependent mechanisms on the title and their conclusions or include more comprehensive E2F readouts. 2. Related to 1 above, the authors should also evaluate the effect of knocking-down E2F4 and E2F3 (Giangrande et al., Genes and Dev, 2004) in S phase using RPE1 as a cellular model. (currently only E2F6 has been looked at). 3. Many controls are missing in the experimental settings of the cell cycle profile analysis. Cells were synchronized by serum starvation and then treated with nocodazole. The authors should therefore include siControl and siE2F6 cell cycle profiles without synchronization, only synchronized without nocodazole treatment and follow them up to 48 h. 4. The assessment of DNA damage (Fig. 4) is not sufficient. No proper readouts of replication stress, such as RPA foci detection or 53BP1 body formation in the following G1 were used. For RPA2 foci detection, cells have to be pre-extracted before fixation and analysis, as free nucleoplasmic RPA would largely abolish foci detection. Alternatively, WB analysis for phosphorylated ATR or Chk1 could have been used. The mere fact that higher levels of p21 exist in E2F6-depleted cells in the second cell cycle after G1 release shows that replication stress must be present in the first cell cycle. The provided analysis is not sufficient and comprehensive enough to draw any conclusions. 5. Figure 1b. The authors should show western blots of some representative examples of cyclins, licensing factors, E2F6a, DNA synthesis and repair proteins and repressors of G1/S, by synchronizing cells as in C and blotting different proteins after the release. 6. Figure 1c. These cell cycle profiles of T98G cells should be replaced by an equivalent experiment using RPE1 cells. T98G cells are p53 mutant and clearly progress differently through the cell cycle after arrest than RPE1 cells (see S3b). Especially that this is the only one experiment where the authors used T98G. Perhaps it is due to some historical reasons that this cell line is included, as the authors have published other papers using T98G (Bertoli et al., Curr Biol, 2013). 7. Figure 2a. A color code should be presented in the scattering plots for siControl and siE2F6, including gating strategy. The differences in EdU intensity are not convincing and the results from triplicates should be presented as a bar plot with S.D. 8. Figure 2b. The authors must measure the distance between origins using DNA fibre/combing techniques. PCNA intensity is not a direct measure of origin activity. PCNA is also needed to repair DNA and it can be engaged in replication factories for longer time in the presence of arrested forks. In any case, the authors should present the scattering plots from flow cytometry from siControl and siE2F6 cells. In the current figure, it is not clear, which cells were used as the example. 9. Figure 2c. Approximately 200 forks were counted for each condition. Perhaps the number of analyzed forks is not sufficient, as there is a discrepancy in the fork length with other similar experiments. There is plenty of room in the supplementary figures to show individual fibre experiments to judge the difference between experiments. 10. ‘E2F-dependant transcription in S phase results in an increase in replication fork speed, thus increasing the overall DNA synthesis rate without any detectable change in origin licensing or firing’ – again this sentence is overstated. The authors should confirm their findings using classical biochemistry experiments, like tritium-labelled thymidine incorporation and measurement of the ori-to-ori distance by immunofluorescence. 11. Why do siE2F6a-treated cells show no faster S phase after release from contact inhibition (Suppl. Fig 3), but faster replication forks? This contradicts the main message of the paper. It rather seems that contact inhibition and serum starvation have different impact on S phase length after E2F6-depletion despite increased fork speed, which is not addressed by the authors (Fig. 1 and Suppl Fig. 1 in contrast to Suppl. Fig. 3). This study is indicative: Yang HW, Chung M, Kudo T, Meyer T. Competing memories of mitogen and p53 signalling control cell-cycle entry. Nature. 2017 Sep 21;549(7672):404-408. doi: 10.1038/nature23880. Epub 2017 Sep 6. PubMed PMID: 28869970. 12. Also, the cell cycle profiles in the Figure S3b at 24 h look similar between siControl and siE2F6. The authors should quantify the amount of cells in early-mid-late S by EdU incorporation after the release 13. Figure 3a. The authors should confirm their findings by investigating the level of PCNA by immunoblotting in the same experimental settings. 14. Figure 3b. Interestingly, the authors found that the fork length was similar in siControl cells in early and mid/late S phase, (the average length in these cells was ~16 um). In the Figure 2c, for siControl the average length in non-synchronized cells was 17.8. Is this difference due to technical problems of the DNA fibres or due to synchronization? Fibre results from equivalent experiments should be presented and the number of scored fibres for each experiment should be increased. 15. Figure 3c. Is the fork length significantly different between early vs mid/late S, 17.2 vs 18.6? Any explanation why are these numbers different from the ones presented in the Figure 3b for siControl untreated cells? 16. Figure 4. gH2AX and RPA intensities can be correlated with DNA content. The cell cycle profiles from generations 0, 1, 2 should be shown. I speculate that the cells start to arrest in G1 due to DNA damage. Minor comments: • It probably is just semantic but the conclusion from results presented in the Figure 1 are overstated (Page 4, the end of paragraph 1). A more precise conclusion should be – kd-E2F6 cells spend less time in S phase. • In line 106, incomplete sentence: “This allowed us to monitor.” Reviewer #2 (Remarks to the Author): The study by Pennycook et al. analyzes the role of E2F in controlling S phase length by determining what the authors term “replication capacity”. Replication capacity is a novel and useful parameter of a cell’s DNA synthesis kinetics. The authors essentially demonstrate that E2F regulates the expression of S phase factors and actively adjusts those expression levels to ensure completion of genome duplication with a given number of active replication origins. Importantly, the study modulates E2F6, which represses E2F-driven transcription of S phase proteins in late S phase. The authors show convincingly that E2F6 knockdown accelerates S phase progression in 2 different (a transformed and nontransformed) cell lines by increasing fork speed in late S phase. The data is well-controlled, statistics are solid and the conclusions are well justified. Previous studies have shown that an increase in fork speed is associated with DNA damage. The authors show here that this is not necessarily a direct effect. This is an important observation. The finding that low level damage can be observed in the second cell-cycle after E2F6 knockdown is theoretically a consequence of diminished origin licensing, which would explain why p21 expression is increased and G1 is extended in the KD cells. This has not been explored rigorously but could be easily done by analyzing Mcm2 and Mcm7 chromatin association at the end of G1 of the second cycle. Examining this point would make this particular section of the paper clearer. Overall, this is a rigorous and important study. Reviewer #3 (Remarks to the Author): In this manuscript Pennycook et al. explore the impact of E2F transcriptional activity de-repression in S-phase dynamics of human cells. Authors observe that increasing E2F transcription by interference of the repressor E2F6 results in a faster S-phase completion. A slight increase in EdU incorporation is observed in E2F6 interfered cells, which correlates with longer replication tracks in DNA fibers, suggesting that enhanced E2F transcription increases fork speed. Conversely, overexpression of E2F6 in S-phase reduces replicated fiber length and EdU incorporation. Lastly, a slight increase is observed in gH2AX phosphorylation and RPA2 intensity, proxies of DNA damage signaling and replication stress, respectively, in E2F6 interfered cells, which correlate with an increase in p21 foci and slight lengthening in G1 duration. Based on these results authors propose that E2F controls the replication capacity of the cell (defined by fork speed and number of active forks), as part of a mechanism that would respond to fluctuations in replication initiation events to maintain absolute synthesis rates and S-phase duration. The data presented are in general of good quality and the conclusions are original and of potential interest for researchers in the replication and cell cycle fields. However, there are some issues about experimental design and data interpretation raising concerns that should be addressed to support the main conclusions and for the model to be convincing. Major points: - Authors define the concept of a replication capacity, which should be kept somewhat constant to promote optimal S-phase length and replication speed, and propose that E2F transcription works within a mechanism controlling replication capacity. Replication rates and fork speed seem increased upon E2F6 interference. However, to conclude that replication capacity is enhanced by E2F transcription in this context authors should prove that the frequency of initiation events is unchanged. This should be straight forward, as origin firing frequencies could be measured in the double label fiber experiments shown. MCM7 data on figure 2b seem to be an estimate of nuclear MCM levels rather than a measure of origin licensing/firing. - Authors should provide evidence that E2F transcription responds to changes in replication initiation event numbers to support the notion that it acts within a mechanism regulating replication capacity. - A key point is which E2F targets promote an enhancement of replication fork speed. A straightforward mechanism would be upregulating dNTP pools via induction of ribonucleotide reductase expression. Authors should measure dNTP levels in control and E2F6 interfered cells to rule out this possibility. Minor point: - DNA damage signaling and replication stress are addressed 21 hours after cell cycle block release when a small proportion of cells (5-7%) is in S-phase. These experiments would be more informative if performed at later time points when more cells have ongoing replication.	Reviewers' comments: Reviewer #1 (Remarks to the Author): In this manuscript, Pennycook et al. intend to propose a new concept, called ‘replication capacity’, which the authors define as the maximum amount of DNA a cell is able to synthesize per unit of time throughout S phase. The authors argue that E2F-dependent transcription determines the replication capacity of a cell. The manuscript is well written, however, the conclusions are overstated and the experimental evidence is rather preliminary. Some of the results are inconsistent between experiments and some statistical differences are arguable. In general, the authors test the effect of knocking-down E2F6 during S phase. With this, the authors claim in the title that E2F activity determines replication capacity, which is an overstatement. The comments below indicate the multiple problems with this manuscript, in terms of experimental design, controls and interpretation. Overall, the present results do not support the author’s model and conclusions. I suggest addressing at least some of the specific comments listed below before submitting the manuscript elsewhere. Major comments: 1. The authors modulate E2F activity only via E2F6, but claim that general E2F activity is the driver for replication capacity (the maximal amount of DNA a cell is able to synthesise per unit time throughout S phase). Proof for their hypothesis falls short as no general E2F activity readout is provided and a positive E2F transcritption factor (e.g. E2F1) is not modulated whatsoever. The authors should therefore focus on claims for E2F6-dependent mechanisms on the title and their conclusions or include more comprehensive E2F readouts. 2. Related to 1 above, the authors should also evaluate the effect of knocking-down E2F4 and E2F3 (Giangrande et al., Genes and Dev, 2004) in S phase using RPE1 as a cellular model. (currently only E2F6 has been looked at). 3. Many controls are missing in the experimental settings of the cell cycle profile analysis. Cells were synchronized by serum starvation and then treated with nocodazole. The authors should therefore include siControl and siE2F6 cell cycle profiles without synchronization, only synchronized without nocodazole treatment and follow them up to 48 h. 4. The assessment of DNA damage (Fig. 4) is not sufficient. No proper readouts of replication stress, such as RPA foci detection or 53BP1 body formation in the following G1 were used. For RPA2 foci detection, cells have to be pre-extracted before fixation and analysis, as free nucleoplasmic RPA would largely abolish foci detection. Alternatively, WB analysis for phosphorylated ATR or Chk1 could have been used. The mere fact that higher levels of p21 exist in E2F6-depleted cells in the second cell cycle after G1 release shows that replication stress must be present in the first cell cycle. The provided analysis is not sufficient and comprehensive enough to draw any conclusions. 5. Figure 1b. The authors should show western blots of some representative examples of cyclins, licensing factors, E2F6a, DNA synthesis and repair proteins and repressors of G1/S, by synchronizing cells as in C and blotting different proteins after the release. 6. Figure 1c. These cell cycle profiles of T98G cells should be replaced by an equivalent experiment using RPE1 cells. T98G cells are p53 mutant and clearly progress differently through the cell cycle after arrest than RPE1 cells (see S3b). Especially that this is the only one experiment where the authors used T98G. Perhaps it is due to some historical reasons that this cell line is included, as the authors have published other papers using T98G (Bertoli et al., Curr Biol, 2013). 7. Figure 2a. A color code should be presented in the scattering plots for siControl and siE2F6, including gating strategy. The differences in EdU intensity are not convincing and the results from triplicates should be presented as a bar plot with S.D. 8. Figure 2b. The authors must measure the distance between origins using DNA fibre/combing techniques. PCNA intensity is not a direct measure of origin activity. PCNA is also needed to repair DNA and it can be engaged in replication factories for longer time in the presence of arrested forks. In any case, the authors should present the scattering plots from flow cytometry from siControl and siE2F6 cells. In the current figure, it is not clear, which cells were used as the example. 9. Figure 2c. Approximately 200 forks were counted for each condition. Perhaps the number of analyzed forks is not sufficient, as there is a discrepancy in the fork length with other similar experiments. There is plenty of room in the supplementary figures to show individual fibre experiments to judge the difference between experiments. 10. ‘E2F-dependant transcription in S phase results in an increase in replication fork speed, thus increasing the overall DNA synthesis rate without any detectable change in origin licensing or firing’ – again this sentence is overstated. The authors should confirm their findings using classical biochemistry experiments, like tritium-labelled thymidine incorporation and measurement of the ori-to-ori distance by immunofluorescence. 11. Why do siE2F6a-treated cells show no faster S phase after release from contact inhibition (Suppl. Fig 3), but faster replication forks? This contradicts the main message of the paper. It rather seems that contact inhibition and serum starvation have different impact on S phase length after E2F6-depletion despite increased fork speed, which is not addressed by the authors (Fig. 1 and Suppl Fig. 1 in contrast to Suppl. Fig. 3). This study is indicative: Yang HW, Chung M, Kudo T, Meyer T. Competing memories of mitogen and p53 signalling control cell-cycle entry. Nature. 2017 Sep 21;549(7672):404-408. doi: 10.1038/nature23880. Epub 2017 Sep 6. PubMed PMID: 28869970. 12. Also, the cell cycle profiles in the Figure S3b at 24 h look similar between siControl and siE2F6. The authors should quantify the amount of cells in early-mid-late S by EdU incorporation after the release 13. Figure 3a. The authors should confirm their findings by investigating the level of PCNA by immunoblotting in the same experimental settings. 14. Figure 3b. Interestingly, the authors found that the fork length was similar in siControl cells in early and mid/late S phase, (the average length in these cells was ~16 um). In the Figure 2c, for siControl the average length in non-synchronized cells was 17.8. Is this difference due to technical problems of the DNA fibres or due to synchronization? Fibre results from equivalent experiments should be presented and the number of scored fibres for each experiment should be increased. 15. Figure 3c. Is the fork length significantly different between early vs mid/late S, 17.2 vs 18.6? Any explanation why are these numbers different from the ones presented in the Figure 3b for siControl untreated cells? 16. Figure 4. gH2AX and RPA intensities can be correlated with DNA content. The cell cycle profiles from generations 0, 1, 2 should be shown. I speculate that the cells start to arrest in G1 due to DNA damage. Minor comments: • It probably is just semantic but the conclusion from results presented in the Figure 1 are overstated (Page 4, the end of paragraph 1). A more precise conclusion should be – kd-E2F6 cells spend less time in S phase. • In line 106, incomplete sentence: “This allowed us to monitor.” Reviewer #2 (Remarks to the Author): The study by Pennycook et al. analyzes the role of E2F in controlling S phase length by determining what the authors term “replication capacity”. Replication capacity is a novel and useful parameter of a cell’s DNA synthesis kinetics. The authors essentially demonstrate that E2F regulates the expression of S phase factors and actively adjusts those expression levels to ensure completion of genome duplication with a given number of active replication origins. Importantly, the study modulates E2F6, which represses E2F-driven transcription of S phase proteins in late S phase. The authors show convincingly that E2F6 knockdown accelerates S phase progression in 2 different (a transformed and nontransformed) cell lines by increasing fork speed in late S phase. The data is well-controlled, statistics are solid and the conclusions are well justified. Previous studies have shown that an increase in fork speed is associated with DNA damage. The authors show here that this is not necessarily a direct effect. This is an important observation. The finding that low level damage can be observed in the second cell-cycle after E2F6 knockdown is theoretically a consequence of diminished origin licensing, which would explain why p21 expression is increased and G1 is extended in the KD cells. This has not been explored rigorously but could be easily done by analyzing Mcm2 and Mcm7 chromatin association at the end of G1 of the second cycle. Examining this point would make this particular section of the paper clearer. Overall, this is a rigorous and important study. Reviewer #3 (Remarks to the Author): In this manuscript Pennycook et al. explore the impact of E2F transcriptional activity de-repression in S-phase dynamics of human cells. Authors observe that increasing E2F transcription by interference of the repressor E2F6 results in a faster S-phase completion. A slight increase in EdU incorporation is observed in E2F6 interfered cells, which correlates with longer replication tracks in DNA fibers, suggesting that enhanced E2F transcription increases fork speed. Conversely, overexpression of E2F6 in S-phase reduces replicated fiber length and EdU incorporation. Lastly, a slight increase is observed in gH2AX phosphorylation and RPA2 intensity, proxies of DNA damage signaling and replication stress, respectively, in E2F6 interfered cells, which correlate with an increase in p21 foci and slight lengthening in G1 duration. Based on these results authors propose that E2F controls the replication capacity of the cell (defined by fork speed and number of active forks), as part of a mechanism that would respond to fluctuations in replication initiation events to maintain absolute synthesis rates and S-phase duration. The data presented are in general of good quality and the conclusions are original and of potential interest for researchers in the replication and cell cycle fields. However, there are some issues about experimental design and data interpretation raising concerns that should be addressed to support the main conclusions and for the model to be convincing. Major points: - Authors define the concept of a replication capacity, which should be kept somewhat constant to promote optimal S-phase length and replication speed, and propose that E2F transcription works within a mechanism controlling replication capacity. Replication rates and fork speed seem increased upon E2F6 interference. However, to conclude that replication capacity is enhanced by E2F transcription in this context authors should prove that the frequency of initiation events is unchanged. This should be straight forward, as origin firing frequencies could be measured in the double label fiber experiments shown. MCM7 data on figure 2b seem to be an estimate of nuclear MCM levels rather than a measure of origin licensing/firing. - Authors should provide evidence that E2F transcription responds to changes in replication initiation event numbers to support the notion that it acts within a mechanism regulating replication capacity. - A key point is which E2F targets promote an enhancement of replication fork speed. A straightforward mechanism would be upregulating dNTP pools via induction of ribonucleotide reductase expression. Authors should measure dNTP levels in control and E2F6 interfered cells to rule out this possibility. Minor point: - DNA damage signaling and replication stress are addressed 21 hours after cell cycle block release when a small proportion of cells (5-7%) is in S-phase. These experiments would be more informative if performed at later time points when more cells have ongoing replication.	TPR	NC_BS_CBIO_010_RR1.txt	{ "criticism": 15, "example": 20, "importance_and_relevance": 5, "materials_and_methods": 65, "praise": 6, "presentation_and_reporting": 20, "results_and_discussion": 18, "suggestion_and_solution": 27, "total": 126 }	{ "criticism": 0.11904761904761904, "example": 0.15873015873015872, "importance_and_relevance": 0.03968253968253968, "materials_and_methods": 0.5158730158730159, "praise": 0.047619047619047616, "presentation_and_reporting": 0.15873015873015872, "results_and_discussion": 0.14285714285714285, "suggestion_and_solution": 0.21428571428571427 }	1.396825	[ { "content": "Please review this paper.", "role": "user" }, { "content": "Reviewers' comments: \n \nReviewer #1 (Remarks to the Author): \n \nIn this manuscript, Pennycook et al. intend to propose a new concept, called ‘replication capacity’, \nwhich the authors define as the maximum amount of DNA a cell is able to synthesize per unit of \ntime throughout S phase. The authors argue that E2F-dependent transcription determines the \nreplication capacity of a cell. The manuscript is well written, however, the conclusions are \noverstated and the experimental evidence is rather preliminary. Some of the results are \ninconsistent between experiments and some statistical differences are arguable. In general, the \nauthors test the effect of knocking-down E2F6 during S phase. With this, the authors claim in the \ntitle that E2F activity determines replication capacity, which is an overstatement. The comments \nbelow indicate the multiple problems with this manuscript, in terms of experimental design, \ncontrols and interpretation. Overall, the present results do not support the author’s model and \nconclusions. I suggest addressing at least some of the specific comments listed below before \nsubmitting the manuscript elsewhere. \n \nMajor comments: \n \n1. The authors modulate E2F activity only via E2F6, but claim that general E2F activity is the driver \nfor replication capacity (the maximal amount of DNA a cell is able to synthesise per unit time \nthroughout S phase). Proof for their hypothesis falls short as no general E2F activity readout is \nprovided and a positive E2F transcritption factor (e.g. E2F1) is not modulated whatsoever. The \nauthors should therefore focus on claims for E2F6-dependent mechanisms on the title and their \nconclusions or include more comprehensive E2F readouts. \n2. Related to 1 above, the authors should also evaluate the effect of knocking-down E2F4 and \nE2F3 (Giangrande et al., Genes and Dev, 2004) in S phase using RPE1 as a cellular model. \n(currently only E2F6 has been looked at). \n3. Many controls are missing in the experimental settings of the cell cycle profile analysis. Cells \nwere synchronized by serum starvation and then treated with nocodazole. The authors should \ntherefore include siControl and siE2F6 cell cycle profiles without synchronization, only synchronized \nwithout nocodazole treatment and follow them up to 48 h. \n \n4. The assessment of DNA damage (Fig. 4) is not sufficient. No proper readouts of replication \nstress, such as RPA foci detection or 53BP1 body formation in the following G1 were used. For \nRPA2 foci detection, cells have to be pre-extracted before fixation and analysis, as free \nnucleoplasmic RPA would largely abolish foci detection. Alternatively, WB analysis for \nphosphorylated ATR or Chk1 could have been used. The mere fact that higher levels of p21 exist in \nE2F6-depleted cells in the second cell cycle after G1 release shows that replication stress must be \npresent in the first cell cycle. The provided analysis is not sufficient and comprehensive enough to \ndraw any conclusions. \n \n \n \n5. Figure 1b. The authors should show western blots of some representative examples of cyclins, \nlicensing factors, E2F6a, DNA synthesis and repair proteins and repressors of G1/S, by \nsynchronizing cells as in C and blotting different proteins after the release. \n \n6. Figure 1c. These cell cycle profiles of T98G cells should be replaced by an equivalent experiment \nusing RPE1 cells. T98G cells are p53 mutant and clearly progress differently through the cell cycle \nafter arrest than RPE1 cells (see S3b). Especially that this is the only one experiment where the \nauthors used T98G. Perhaps it is due to some historical reasons that this cell line is included, as \nthe authors have published other papers using T98G (Bertoli et al., Curr Biol, 2013). \n \n \n\n\f \n7. Figure 2a. A color code should be presented in the scattering plots for siControl and siE2F6, \nincluding gating strategy. The differences in EdU intensity are not convincing and the results from \ntriplicates should be presented as a bar plot with S.D. \n \n8. Figure 2b. The authors must measure the distance between origins using DNA fibre/combing \ntechniques. PCNA intensity is not a direct measure of origin activity. PCNA is also needed to repair \nDNA and it can be engaged in replication factories for longer time in the presence of arrested \nforks. In any case, the authors should present the scattering plots from flow cytometry from \nsiControl and siE2F6 cells. In the current figure, it is not clear, which cells were used as the \nexample. \n \n9. Figure 2c. Approximately 200 forks were counted for each condition. Perhaps the number of \nanalyzed forks is not sufficient, as there is a discrepancy in the fork length with other similar \nexperiments. There is plenty of room in the supplementary figures to show individual fibre \nexperiments to judge the difference between experiments. \n \n10. ‘E2F-dependant transcription in S phase results in an increase in replication fork speed, thus \nincreasing the overall DNA synthesis rate without any detectable change in origin licensing or \nfiring’ – again this sentence is overstated. The authors should confirm their findings using classical \nbiochemistry experiments, like tritium-labelled thymidine incorporation and measurement of the \nori-to-ori distance by immunofluorescence. \n \n11. Why do siE2F6a-treated cells show no faster S phase after release from contact inhibition \n(Suppl. Fig 3), but faster replication forks? This contradicts the main message of the paper. It \nrather seems that contact inhibition and serum starvation have different impact on S phase length \nafter E2F6-depletion despite increased fork speed, which is not addressed by the authors (Fig. 1 \nand Suppl Fig. 1 in contrast to Suppl. Fig. 3). \n \nThis study is indicative: \nYang HW, Chung M, Kudo T, Meyer T. Competing memories of mitogen and p53 \nsignalling control cell-cycle entry. Nature. 2017 Sep 21;549(7672):404-408. doi: \n10.1038/nature23880. Epub 2017 Sep 6. PubMed PMID: 28869970. \n \n \n12. Also, the cell cycle profiles in the Figure S3b at 24 h look similar between siControl and siE2F6. \nThe authors should quantify the amount of cells in early-mid-late S by EdU incorporation after the \nrelease \n \n13. Figure 3a. The authors should confirm their findings by investigating the level of PCNA by \nimmunoblotting in the same experimental settings. \n \n14. Figure 3b. Interestingly, the authors found that the fork length was similar in siControl cells in \nearly and mid/late S phase, (the average length in these cells was ~16 um). In the Figure 2c, for \nsiControl the average length in non-synchronized cells was 17.8. Is this difference due to technical \nproblems of the DNA fibres or due to synchronization? Fibre results from equivalent experiments \nshould be presented and the number of scored fibres for each experiment should be increased. \n \n15. Figure 3c. Is the fork length significantly different between early vs mid/late S, 17.2 vs 18.6? \nAny explanation why are these numbers different from the ones presented in the Figure 3b for \nsiControl untreated cells? \n \n16. Figure 4. gH2AX and RPA intensities can be correlated with DNA content. The cell cycle profiles \nfrom generations 0, 1, 2 should be shown. I speculate that the cells start to arrest in G1 due to \nDNA damage. \n\n\f \n \nMinor comments: \n \n• It probably is just semantic but the conclusion from results presented in the Figure 1 are \noverstated (Page 4, the end of paragraph 1). A more precise conclusion should be – kd-E2F6 cells \nspend less time in S phase. \n \n \n• In line 106, incomplete sentence: “This allowed us to monitor.” \n \n \n \nReviewer #2 (Remarks to the Author): \n \nThe study by Pennycook et al. analyzes the role of E2F in controlling S phase length by \ndetermining what the authors term “replication capacity”. Replication capacity is a novel and useful \nparameter of a cell’s DNA synthesis kinetics. The authors essentially demonstrate that E2F \nregulates the expression of S phase factors and actively adjusts those expression levels to ensure \ncompletion of genome duplication with a given number of active replication origins. Importantly, \nthe study modulates E2F6, which represses E2F-driven transcription of S phase proteins in late S \nphase. The authors show convincingly that E2F6 knockdown accelerates S phase progression in 2 \ndifferent (a transformed and nontransformed) cell lines by increasing fork speed in late S phase. \nThe data is well-controlled, statistics are solid and the conclusions are well justified. Previous \nstudies have shown that an increase in fork speed is associated with DNA damage. The authors \nshow here that this is not necessarily a direct effect. This is an important observation. The finding \nthat low level damage can be observed in the second cell-cycle after E2F6 knockdown is \ntheoretically a consequence of diminished origin licensing, which would explain why p21 \nexpression is increased and G1 is extended in the KD cells. This has not been explored rigorously \nbut could be easily done by analyzing Mcm2 and Mcm7 chromatin association at the end of G1 of \nthe second cycle. Examining this point would make this particular section of the paper clearer. \n \nOverall, this is a rigorous and important study. \n \n \nReviewer #3 (Remarks to the Author): \n \nIn this manuscript Pennycook et al. explore the impact of E2F transcriptional activity de-repression \nin S-phase dynamics of human cells. Authors observe that increasing E2F transcription by \ninterference of the repressor E2F6 results in a faster S-phase completion. A slight increase in EdU \nincorporation is observed in E2F6 interfered cells, which correlates with longer replication tracks in \nDNA fibers, suggesting that enhanced E2F transcription increases fork speed. Conversely, \noverexpression of E2F6 in S-phase reduces replicated fiber length and EdU incorporation. Lastly, a \nslight increase is observed in gH2AX phosphorylation and RPA2 intensity, proxies of DNA damage \nsignaling and replication stress, respectively, in E2F6 interfered cells, which correlate with an \nincrease in p21 foci and slight lengthening in G1 duration. Based on these results authors propose \nthat E2F controls the replication capacity of the cell (defined by fork speed and number of active \nforks), as part of a mechanism that would respond to fluctuations in replication initiation events to \nmaintain absolute synthesis rates and S-phase duration. \n \nThe data presented are in general of good quality and the conclusions are original and of potential \ninterest for researchers in the replication and cell cycle fields. However, there are some issues \nabout experimental design and data interpretation raising concerns that should be addressed to \nsupport the main conclusions and for the model to be convincing. \n \n\n\f \nMajor points: \n \n- Authors define the concept of a replication capacity, which should be kept somewhat constant to \npromote optimal S-phase length and replication speed, and propose that E2F transcription works \nwithin a mechanism controlling replication capacity. Replication rates and fork speed seem \nincreased upon E2F6 interference. However, to conclude that replication capacity is enhanced by \nE2F transcription in this context authors should prove that the frequency of initiation events is \nunchanged. This should be straight forward, as origin firing frequencies could be measured in the \ndouble label fiber experiments shown. MCM7 data on figure 2b seem to be an estimate of nuclear \nMCM levels rather than a measure of origin licensing/firing. \n \n- Authors should provide evidence that E2F transcription responds to changes in replication \ninitiation event numbers to support the notion that it acts within a mechanism regulating \nreplication capacity. \n \n- A key point is which E2F targets promote an enhancement of replication fork speed. A \nstraightforward mechanism would be upregulating dNTP pools via induction of ribonucleotide \nreductase expression. Authors should measure dNTP levels in control and E2F6 interfered cells to \nrule out this possibility. \n \nMinor point: \n \n- DNA damage signaling and replication stress are addressed 21 hours after cell cycle block release \nwhen a small proportion of cells (5-7%) is in S-phase. These experiments would be more \ninformative if performed at later time points when more cells have ongoing replication. \n\n\f", "role": "assistant" } ]	2,020
E2F-dependent transcription determines replication capacity and S phase length	Reviewers' comments: Reviewer #1 (Remarks to the Author): In the revised manuscript, the authors have made a good effort to answer my previous comments On the other hand, I still believe that the differences in the length of S phase (around 10% faster in siE2F6) are too small and such modest differences can be the result of experimental variation. Also, the difference of fork speed after siE2F6, around 10%, is too small to make any solid conclusions along the lines the authors put forward. Below, I specifically comment on the issues that still remain to be convincingly addressed, in the interest of authors' own reputation and providing solid and well supported evidence. Hence, before suggesting the acceptance of their manuscript, the authors should show convincingly that E2F6- depletion indeed does accelerate fork speed during mid/late S phase. This is an important part of the mechanism to explain how low-level E2F6 reduces the length of S phase without altering origin activity. At this stage, I am not satisfied with their answers to the following related questions raised in the original submission. Specifically, 11. (original concern): Why do siE2F6a-treated cells show no faster S phase after release from contact inhibition (Suppl. Fig 3), but faster replication forks? This contradicts the main message of the paper. It rather seems that contact inhibition and serum starvation have different impact on S phase length after E2F6-depletion despite increased fork speed, which is not addressed by the authors (Fig. 1 and Suppl Fig. 1 in contrast to Suppl. Fig. 3). This study is indicative: Yang HW, Chung M, Kudo T, Meyer T. Competing memories of mitogen and p53 signalling control cell-cycle entry. Nature. 2017 Sep 21;549(7672):404-408. doi: 10.1038/nature23880. Epub 2017 Sep 6. PubMed PMID: 28869970. Authors response: We agree with the reviewer that the effect on S phase length, seen in the single cell time lapse analysis, is not obvious in a synchronized RPE1 cell population. We believe that this is due to the lack of a robust synchronous release with RPE1 cells, which probably masks the decrease in S phase length in this setup. RPE1 cells are hard to arrest and synchronously release. Whilst the enrichment in the S phase population does allow us to observe an increase in DNA fibre length in middle/late S phase the extent of the synchrony is insufficient to evaluate S phase length. It is therefore, unfortunately, not possible to draw any conclusions about S phase length from this data. My new comment on the rebuttal in this point: Therefore it is not possible to draw any conclusion about different stages of S phase and correlate those with increased fork speed in mid/late S phase due to the lack of robust synchrony that the authors are experiencing with this model. Authors could evaluate fork speed using synchronised T98G cells +/-siE2F6 to test fork speed at mid/late S phase. Alternatively, following the logic of the results presented in Fig 3, authors could test the length of S phase in RPE1 cells that have been manipulated to overexpress E2F6. 12. Original concern: Also, the cell cycle profiles in the Figure S3b at 24 h look similar between siControl and siE2F6. The authors should quantify the amount of cells in early-mid-late S by EdU incorporation after the release Authors' response: Unfortunately, as mentioned above the lack of synchrony, from a release form a contact inhibition arrest, masks synchronous progression through S phase of a RPE1 cell population. We have therefore established EdU incorporation in asynchronous cells, Fig 2a. My new commenton the rebuttal in this point: This just further supports my previous serious concern and the fact that the data is still inconclusive. 13. Original concern: Figure 3a. The authors should confirm their findings by investigating the level of PCNA by immunoblotting in the same experimental settings. Authors' response: We have added this analysis in Supplementary Figs 2d (asynchronous cells) and 3a (synchronised cells). My new related comment on the rebuttal in this point: Other comments were addressed adequately. Reviewer #2 (Remarks to the Author): Authors could analyse and plot the PCNA mean intensity vs DNA content using high-content image analysis. This will show directly that cells in mid/late S phase contain higher level of PCNA. In Cimprich et al, (Science 2018) there are examples of this type of analysis using EdU vs DNA content. In Pennycook’s manuscript, authors should pre-extract cells to quantify PCNA associated to chromatin and correlate the level with the DNA content +/-siE2F6. I am utterly confused by the response to my request to analyze the status of Mcm2-7 binding in the second G1 phase (Fig. 4). The authors basically argue that the cells are no longer synchronized. However, they don't say this in the results section of their manuscript, and claim that the second G1 phase is longer (Fig. 4c). They can't have it both ways. I don't understand why they didn't attempt to assess Mcm2-7 loading dynamics in the second G1. If cells are labeled in parallel by Edu and DAPI, then the portion of cells in G1 can be clearly quantified and Mcm2-7 loading can be determined. The authors claim in their response that "Therefore, if there is a diminished origin licensing in the second cycle we should expect a decrease in Mcm binding, which we don’t observe." Where is the data that shows that Mcm2-7 binding is not decreased? That's the very piece of data I was asking for. The fact that p21 expression is increased and G1 is longer in the granddaughters clearly points to a potential licensing defect in the second G1 phase. As I said in my original review, this part of the study needs to be further corroborated. Understanding why DNA damage occurs in the 2nd cell cycle is critical to understanding the consequences of changes in S phase dynamics. Reviewer #3 (Remarks to the Author): While authors show that replication dynamics change upon alteration of E2F transcription, in order to claim that E2F determines “replication capacity” to regulate S-phase length authors should demonstrate that E2F transcription actually responds to changes in replication dynamics (e.g. origin firing frequency). In the reviewed manuscript authors have not addressed this major concern, nor they have investigated whether regulation of dNTP levels is a key driver for the observed changes of replication dynamics upon manipulation of E2F transcription (as opposed to a general increase in replication factors postulated to be limiting for replication speed). Hence, the data presented fail to provide strong support to the manuscript’s central conclusion. In order for the study to be suitable for publication in Nature Communications authors should provide strong evidence that E2F transcription actually dynamically responds to replication changes (origin firing or fork speed) to maintain “replication capacity” constant or strictly limit their conclusions to the impact of E2F transcription on replication fork speed throughout the manuscript.	Reviewers' comments: Reviewer #1 (Remarks to the Author): In the revised manuscript, the authors have made a good effort to answer my previous comments On the other hand, I still believe that the differences in the length of S phase (around 10% faster in siE2F6) are too small and such modest differences can be the result of experimental variation. Also, the difference of fork speed after siE2F6, around 10%, is too small to make any solid conclusions along the lines the authors put forward. Below, I specifically comment on the issues that still remain to be convincingly addressed, in the interest of authors' own reputation and providing solid and well supported evidence. Hence, before suggesting the acceptance of their manuscript, the authors should show convincingly that E2F6- depletion indeed does accelerate fork speed during mid/late S phase. This is an important part of the mechanism to explain how low-level E2F6 reduces the length of S phase without altering origin activity. At this stage, I am not satisfied with their answers to the following related questions raised in the original submission. Specifically, 11. (original concern): Why do siE2F6a-treated cells show no faster S phase after release from contact inhibition (Suppl. Fig 3), but faster replication forks? This contradicts the main message of the paper. It rather seems that contact inhibition and serum starvation have different impact on S phase length after E2F6-depletion despite increased fork speed, which is not addressed by the authors (Fig. 1 and Suppl Fig. 1 in contrast to Suppl. Fig. 3). This study is indicative: Yang HW, Chung M, Kudo T, Meyer T. Competing memories of mitogen and p53 signalling control cell-cycle entry. Nature. 2017 Sep 21;549(7672):404-408. doi: 10.1038/nature23880. Epub 2017 Sep 6. PubMed PMID: 28869970. Authors response: We agree with the reviewer that the effect on S phase length, seen in the single cell time lapse analysis, is not obvious in a synchronized RPE1 cell population. We believe that this is due to the lack of a robust synchronous release with RPE1 cells, which probably masks the decrease in S phase length in this setup. RPE1 cells are hard to arrest and synchronously release. Whilst the enrichment in the S phase population does allow us to observe an increase in DNA fibre length in middle/late S phase the extent of the synchrony is insufficient to evaluate S phase length. It is therefore, unfortunately, not possible to draw any conclusions about S phase length from this data. My new comment on the rebuttal in this point: Therefore it is not possible to draw any conclusion about different stages of S phase and correlate those with increased fork speed in mid/late S phase due to the lack of robust synchrony that the authors are experiencing with this model. Authors could evaluate fork speed using synchronised T98G cells +/-siE2F6 to test fork speed at mid/late S phase. Alternatively, following the logic of the results presented in Fig 3, authors could test the length of S phase in RPE1 cells that have been manipulated to overexpress E2F6. 12. Original concern: Also, the cell cycle profiles in the Figure S3b at 24 h look similar between siControl and siE2F6. The authors should quantify the amount of cells in early-mid-late S by EdU incorporation after the release Authors' response: Unfortunately, as mentioned above the lack of synchrony, from a release form a contact inhibition arrest, masks synchronous progression through S phase of a RPE1 cell population. We have therefore established EdU incorporation in asynchronous cells, Fig 2a. My new commenton the rebuttal in this point: This just further supports my previous serious concern and the fact that the data is still inconclusive. 13. Original concern: Figure 3a. The authors should confirm their findings by investigating the level of PCNA by immunoblotting in the same experimental settings. Authors' response: We have added this analysis in Supplementary Figs 2d (asynchronous cells) and 3a (synchronised cells). My new related comment on the rebuttal in this point: Other comments were addressed adequately. Reviewer #2 (Remarks to the Author): Authors could analyse and plot the PCNA mean intensity vs DNA content using high-content image analysis. This will show directly that cells in mid/late S phase contain higher level of PCNA. In Cimprich et al, (Science 2018) there are examples of this type of analysis using EdU vs DNA content. In Pennycook’s manuscript, authors should pre-extract cells to quantify PCNA associated to chromatin and correlate the level with the DNA content +/-siE2F6. I am utterly confused by the response to my request to analyze the status of Mcm2-7 binding in the second G1 phase (Fig. 4). The authors basically argue that the cells are no longer synchronized. However, they don't say this in the results section of their manuscript, and claim that the second G1 phase is longer (Fig. 4c). They can't have it both ways. I don't understand why they didn't attempt to assess Mcm2-7 loading dynamics in the second G1. If cells are labeled in parallel by Edu and DAPI, then the portion of cells in G1 can be clearly quantified and Mcm2-7 loading can be determined. The authors claim in their response that "Therefore, if there is a diminished origin licensing in the second cycle we should expect a decrease in Mcm binding, which we don’t observe." Where is the data that shows that Mcm2-7 binding is not decreased? That's the very piece of data I was asking for. The fact that p21 expression is increased and G1 is longer in the granddaughters clearly points to a potential licensing defect in the second G1 phase. As I said in my original review, this part of the study needs to be further corroborated. Understanding why DNA damage occurs in the 2nd cell cycle is critical to understanding the consequences of changes in S phase dynamics. Reviewer #3 (Remarks to the Author): While authors show that replication dynamics change upon alteration of E2F transcription, in order to claim that E2F determines “replication capacity” to regulate S-phase length authors should demonstrate that E2F transcription actually responds to changes in replication dynamics (e.g. origin firing frequency). In the reviewed manuscript authors have not addressed this major concern, nor they have investigated whether regulation of dNTP levels is a key driver for the observed changes of replication dynamics upon manipulation of E2F transcription (as opposed to a general increase in replication factors postulated to be limiting for replication speed). Hence, the data presented fail to provide strong support to the manuscript’s central conclusion. In order for the study to be suitable for publication in Nature Communications authors should provide strong evidence that E2F transcription actually dynamically responds to replication changes (origin firing or fork speed) to maintain “replication capacity” constant or strictly limit their conclusions to the impact of E2F transcription on replication fork speed throughout the manuscript.	TPR	NC_BS_CBIO_010_RR2.txt	{ "criticism": 15, "example": 11, "importance_and_relevance": 3, "materials_and_methods": 28, "praise": 4, "presentation_and_reporting": 1, "results_and_discussion": 12, "suggestion_and_solution": 11, "total": 61 }	{ "criticism": 0.2459016393442623, "example": 0.18032786885245902, "importance_and_relevance": 0.04918032786885246, "materials_and_methods": 0.45901639344262296, "praise": 0.06557377049180328, "presentation_and_reporting": 0.01639344262295082, "results_and_discussion": 0.19672131147540983, "suggestion_and_solution": 0.18032786885245902 }	1.393443	[ { "content": "Please review this paper.", "role": "user" }, { "content": "\nReviewers' comments:\nReviewer #1 (Remarks to the Author):\nIn the revised manuscript, the authors have made a good effort to answer my previous comments\nOn the other hand, I still believe that the differences in the length of S phase (around 10% faster\nin siE2F6) are too small and such modest differences can be the result of experimental variation.\nAlso, the difference of fork speed after siE2F6, around 10%, is too small to make any solid\nconclusions along the lines the authors put forward.\nBelow, I specifically comment on the issues that still remain to be convincingly addressed, in the\ninterest of authors' own reputation and providing solid and well supported evidence. Hence, before\nsuggesting the acceptance of their manuscript, the authors should show convincingly that E2F6-\ndepletion indeed does accelerate fork speed during mid/late S phase. This is an important part of\nthe mechanism to explain how low-level E2F6 reduces the length of S phase without altering origin\nactivity. At this stage, I am not satisfied with their answers to the following related questions\nraised in the original submission.\nSpecifically,\n11. (original concern): Why do siE2F6a-treated cells show no faster S phase after release from\ncontact inhibition (Suppl. Fig 3), but faster replication forks? This contradicts the main message of\nthe paper. It rather seems that contact inhibition and serum starvation have different impact on S\nphase length after E2F6-depletion despite increased fork speed, which is not addressed by the\nauthors (Fig. 1 and Suppl Fig. 1 in contrast to Suppl. Fig. 3).\nThis study is indicative:\nYang HW, Chung M, Kudo T, Meyer T. Competing memories of mitogen and p53\nsignalling control cell-cycle entry. Nature. 2017 Sep 21;549(7672):404-408. doi:\n10.1038/nature23880. Epub 2017 Sep 6. PubMed PMID: 28869970.\nAuthors response: We agree with the reviewer that the effect on S phase length, seen in the single\ncell time lapse analysis, is not obvious in a synchronized RPE1 cell population. We believe that this\nis due to the lack of a robust synchronous release with RPE1 cells, which probably masks the\ndecrease in S phase length in this setup. RPE1 cells are hard to arrest and synchronously release.\nWhilst the enrichment in the S phase population does allow us to observe an increase in DNA fibre\nlength in middle/late S phase the extent of the synchrony is insufficient to evaluate S phase\nlength. It is therefore, unfortunately, not possible to draw any conclusions about S phase length\nfrom this data.\nMy new comment on the rebuttal in this point:\nTherefore it is not possible to draw any conclusion about different stages of S phase and correlate\nthose with increased fork speed in mid/late S phase due to the lack of robust synchrony that the\nauthors are experiencing with this model.\nAuthors could evaluate fork speed using synchronised T98G cells +/-siE2F6 to test fork speed at\nmid/late S phase.\nAlternatively, following the logic of the results presented in Fig 3, authors could test the length of\nS phase in RPE1 cells that have been manipulated to overexpress E2F6.\n12. Original concern: Also, the cell cycle profiles in the Figure S3b at 24 h look similar between\n siControl and siE2F6. The authors should quantify the amount of cells in early-mid-late S by EdU\n incorporation after the release\n Authors' response: Unfortunately, as mentioned above the lack of synchrony, from a release form\na contact inhibition arrest, masks synchronous progression through S phase of a RPE1 cell\n population. We have therefore established EdU incorporation in asynchronous cells, Fig 2a.\n My new commenton the rebuttal in this point:\n This just further supports my previous serious concern and the fact that the data is still\n inconclusive.\n 13. Original concern: Figure 3a. The authors should confirm their findings by investigating the\n level of PCNA by immunoblotting in the same experimental settings.\n Authors' response: We have added this analysis in Supplementary Figs 2d (asynchronous cells)\n and 3a (synchronised cells).\n My new related comment on the rebuttal in this point:\nOther comments were addressed adequately.\nReviewer #2 (Remarks to the Author):\n Authors could analyse and plot the PCNA mean intensity vs DNA content using high-content image\n analysis. This will show directly that cells in mid/late S phase contain higher level of PCNA. In\n Cimprich et al, (Science 2018) there are examples of this type of analysis using EdU vs DNA\n content. In Pennycook’s manuscript, authors should pre-extract cells to quantify PCNA associated\n to chromatin and correlate the level with the DNA content +/-siE2F6.\n I am utterly confused by the response to my request to analyze the status of Mcm2-7 binding in\n the second G1 phase (Fig. 4). The authors basically argue that the cells are no longer\n synchronized. However, they don't say this in the results section of their manuscript, and claim\n that the second G1 phase is longer (Fig. 4c). They can't have it both ways. I don't understand why\n they didn't attempt to assess Mcm2-7 loading dynamics in the second G1. If cells are labeled in\n parallel by Edu and DAPI, then the portion of cells in G1 can be clearly quantified and Mcm2-7\n loading can be determined. The authors claim in their response that \"Therefore, if there is a\n diminished origin\n licensing in the second cycle we should expect a decrease in Mcm binding,\n which we don’t observe.\" Where is the data that shows that Mcm2-7 binding is not decreased?\n That's the very piece of data I was asking for. The fact that p21 expression is increased and G1 is\n longer in the granddaughters clearly points to a potential licensing defect in the second G1 phase.\n As I said in my original review, this part of the study needs to be further corroborated.\n Understanding why DNA damage occurs in the 2nd cell cycle is critical to understanding the\n consequences of changes in S phase dynamics.\n Reviewer #3 (Remarks to the Author):\n While authors show that replication dynamics change upon alteration of E2F transcription, in order\n to claim that E2F determines “replication capacity” to regulate S-phase length authors should\n demonstrate that E2F transcription actually responds to changes in replication dynamics (e.g.\n origin firing frequency). In the reviewed manuscript authors have not addressed this major\n concern, nor they have investigated whether regulation of dNTP levels is a key driver for the\n observed changes of replication dynamics upon manipulation of E2F transcription (as opposed to a\n general increase in replication factors postulated to be limiting for replication speed). Hence, the\n \ndata presented fail to provide strong support to the manuscript’s central conclusion. In order for\n the study to be suitable for publication in Nature Communications authors should provide strong\n evidence that E2F transcription actually dynamically responds to replication changes (origin firing\n or fork speed) to maintain “replication capacity” constant or strictly limit their conclusions to the\n impact of E2F transcription on replication fork speed throughout the manuscript.\n ", "role": "assistant" } ]	2,020
E2F-dependent transcription determines replication capacity and S phase length	REVIEWERS' COMMENTS: Reviewer #1 (Remarks to the Author): Reviewer #2 (Remarks to the Author): I am now fully satisfied with the revisions. The authors have answered all of my questions. Anja Katrin Bielinsky Reviewer #3 (Remarks to the Author): I am now satisfied by the newly added data and authors' responses, and recommend this manuscript for acceptance Throughout the text authors convey the idea that E2F-depedent transcription controls the replication capacity of cells as part of a regulatory mechanism to ensure timely replication completion and to optimise replication fork speed: P1. “Thus, at its most fundamental level, E2F-dependent transcription determines the replication capacity of a cell during S phase, controlling the time it takes to duplicate the genome. More generally, our work shows that controlling replication capacity during S phase is a robust mechanism to ensure timely completion of DNA replication, which could be at the basis of keeping replication fork speed within an optimal range independently of fluctuations in replication initiation events. P2. “Limiting the replication capacity of a cell would provide an elegant mechanism to regulate the global rate of replication during S phase, largely independent of the number of active replication forks.” P6. “This suggests that E2F-dependent transcription controls the amount of DNA a cell can synthesis per unit time in S phase, which we define as replication capacity, likely through regulating the expression of limiting factors for DNA replication.” P9. “Overall our work suggests that the tight regulation of E2F transcription in S phase is an important determinant in DNA replication control and genomic stability, which has important implications for basic biology and the understanding of cancer.” However, evidence is not provided on what cues would this mechanism respond to in order to be effective, how E2F-dependent transcription would be modulated to achieve this regulation, or through what factors would replication speed be controlled. Since the expression of many factors important for replication depends on E2F-dependent transcription, it is a logical expectation that increasing E2F-dependent transcription might augment replication rates and the overall replication capacity. The authors nicely show this in this manuscript and demonstrate this preferentially occurs through fork speed and not through increased origin firing, but the notion that E2F- transcription is part of a mechanism controlling replication rates to preserve genome stability remains premature at this point. Regardless the significant amount of work reported in the manuscript, in my opinion evidence supporting these claims is required to firmly support its publication in Nature Communications.	REVIEWERS' COMMENTS: Reviewer #1 (Remarks to the Author): Reviewer #2 (Remarks to the Author): I am now fully satisfied with the revisions. The authors have answered all of my questions. Anja Katrin Bielinsky Reviewer #3 (Remarks to the Author): I am now satisfied by the newly added data and authors' responses, and recommend this manuscript for acceptance Throughout the text authors convey the idea that E2F-depedent transcription controls the replication capacity of cells as part of a regulatory mechanism to ensure timely replication completion and to optimise replication fork speed: P1. “Thus, at its most fundamental level, E2F-dependent transcription determines the replication capacity of a cell during S phase, controlling the time it takes to duplicate the genome. More generally, our work shows that controlling replication capacity during S phase is a robust mechanism to ensure timely completion of DNA replication, which could be at the basis of keeping replication fork speed within an optimal range independently of fluctuations in replication initiation events. P2. “Limiting the replication capacity of a cell would provide an elegant mechanism to regulate the global rate of replication during S phase, largely independent of the number of active replication forks.” P6. “This suggests that E2F-dependent transcription controls the amount of DNA a cell can synthesis per unit time in S phase, which we define as replication capacity, likely through regulating the expression of limiting factors for DNA replication.” P9. “Overall our work suggests that the tight regulation of E2F transcription in S phase is an important determinant in DNA replication control and genomic stability, which has important implications for basic biology and the understanding of cancer.” However, evidence is not provided on what cues would this mechanism respond to in order to be effective, how E2F-dependent transcription would be modulated to achieve this regulation, or through what factors would replication speed be controlled. Since the expression of many factors important for replication depends on E2F-dependent transcription, it is a logical expectation that increasing E2F-dependent transcription might augment replication rates and the overall replication capacity. The authors nicely show this in this manuscript and demonstrate this preferentially occurs through fork speed and not through increased origin firing, but the notion that E2F- transcription is part of a mechanism controlling replication rates to preserve genome stability remains premature at this point. Regardless the significant amount of work reported in the manuscript, in my opinion evidence supporting these claims is required to firmly support its publication in Nature Communications.	TPR	NC_BS_CBIO_010_RR3.txt	{ "criticism": 1, "example": 2, "importance_and_relevance": 1, "materials_and_methods": 7, "praise": 2, "presentation_and_reporting": 0, "results_and_discussion": 0, "suggestion_and_solution": 1, "total": 12 }	{ "criticism": 0.08333333333333333, "example": 0.16666666666666666, "importance_and_relevance": 0.08333333333333333, "materials_and_methods": 0.5833333333333334, "praise": 0.16666666666666666, "presentation_and_reporting": 0, "results_and_discussion": 0, "suggestion_and_solution": 0.08333333333333333 }	1.166667	[ { "content": "Please review this paper.", "role": "user" }, { "content": "REVIEWERS' COMMENTS:\nReviewer #1 (Remarks to the Author):\nReviewer #2 (Remarks to the Author):\nI am now fully satisfied with the revisions. The authors have answered all of my questions. Anja Katrin Bielinsky\nReviewer #3 (Remarks to the Author):\nI am now satisfied by the newly added data and authors' responses, and recommend this\n manuscript for acceptance\n \nThroughout the text authors convey the idea that E2F-depedent transcription controls the\n replication capacity of cells as part of a regulatory mechanism to ensure timely replication\n completion and to optimise replication fork speed:\n P1. “Thus, at its most fundamental level, E2F-dependent transcription determines the replication\ncapacity of a cell during S phase, controlling the time it takes to duplicate the genome. More\ngenerally, our work shows that controlling replication capacity during S phase is a robust\nmechanism to ensure timely completion of DNA replication, which could be at the basis of keeping\nreplication fork speed within an optimal range independently of fluctuations in replication initiation\nevents.\nP2. “Limiting the replication capacity of a cell would provide an elegant mechanism to regulate the\n global rate of replication during S phase, largely independent of the number of active replication\n forks.”\n P6. “This suggests that E2F-dependent transcription controls the amount of DNA a cell can\n synthesis per unit time in S phase, which we define as replication capacity, likely through\n regulating the expression of limiting factors for DNA replication.”\n P9. “Overall our work suggests that the tight regulation of E2F transcription in S phase is an\n important determinant in DNA replication control and genomic stability, which has important\n implications for basic biology and the understanding of cancer.”\n However, evidence is not provided on what cues would this mechanism respond to in order to be\neffective, how E2F-dependent transcription would be modulated to achieve this regulation, or\nthrough what factors would replication speed be controlled. Since the expression of many factors\nimportant for replication depends on E2F-dependent transcription, it is a logical expectation that\nincreasing E2F-dependent transcription might augment replication rates and the overall replication\ncapacity. The authors nicely show this in this manuscript and demonstrate this preferentially\noccurs through fork speed and not through increased origin firing, but the notion that E2F-\ntranscription is part of a mechanism controlling replication rates to preserve genome stability\nremains premature at this point. Regardless the significant amount of work reported in the\nmanuscript, in my opinion evidence supporting these claims is required to firmly support its\npublication in Nature Communications.", "role": "assistant" } ]	2,020

Subsets and Splits

No community queries yet

The top public SQL queries from the community will appear here once available.