The tumour suppressor and chromatin-remodelling factor BRG1 antagonizes Myc activity and promotes cell differentiation in human cancer

BRG1, a member of the SWI/SNF complex, is mutated in cancer, but it is unclear how it promotes tumourigenesis. We report that re-expression of BRG1 in lung cancer cells up-regulates lung-specific transcripts, restoring the gene expression signature of normal lung. Using cell lines from several cancer types we found that those lacking BRG1 do not respond to retinoic acid (RA) or glucocorticoids (GC), while restoration of BRG1 restores sensitivity. Conversely, in SH-SY5Y cells, a paradigm of RA-dependent differentiation, abrogation of BRG1 prevented the response to RA. Further, our data suggest an antagonistic functional connection between BRG1 and MYC, whereby, refractoriness to RA and GC by BRG1 inactivation involves deregulation of MYC activity. Mechanistically, some of these effects are mediated by BRG1 binding to MYC and MYC-target promoters. The BRG1-MYC antagonism was also evident in primary tumours. Finally, BRG1 restoration significantly dampened invasion and progression and decreased MYC in lung cancer cells orthotopically implanted in nude mice. Thus, BRG1 inactivation enables cancer cells to sustain undifferentiated gene expression programs and prevent its response to environmental stimuli.

Thank you for the submission of your manuscript to EMBO Molecular Medicine. We have now heard back from the three referees whom we asked to evaluate your manuscript. As you will see from the reports below, the referees find the overall topic of your study of potential interest. However, they raise substantial and very significant concerns on your work, which should be convincingly addressed in a major revision of the present manuscript.
As detailed in the attached reports, despite the positive comments of Referee #1, Referees #2 and #3 found significant drawbacks in your study and importantly mentioned that at this stage, the claims do not appear to be fully supported by the data. Nevertheless all three referees provide clear and very detailed comments and suggestions to improve the focus and quality of your manuscript.
I would like to point out the very important issues raised by the referees that we will insist upon: -Focus of the paper -both Referees #2 and #3 found too many data, making the storyline hard to follow -Missing controls and details both regarding experimental design and interpretation of the results. They also have highlighted some inconsistencies in some experiments that need to be clarified. -Modest in vivo effect contrasting with strong therapeutic claims in the body of the paper (Ref#3, point 5).
While it is clear that publication of the paper cannot be considered at this stage, given these overall evaluations I would be open to the submission of a revised manuscript. I must stress however, that the referees' concerns must be fully addressed and that acceptance of the manuscript would entail a second round of review. I would add that it is particularly important that all of their suggestions are taken on board as we cannot consider its publication otherwise. I realize that addressing the reviewers concerns in full will require time and additional work and experimentation. I am unsure whether you will be able or willing to address those and return a revised manuscript within the 3 months deadline.
I should remind you that it is EMBO Molecular Medicine policy to allow a single round of revision only and that, therefore, acceptance or rejection of the manuscript will depend on the completeness of your responses included in the next, final version of the manuscript. For this reason, and to save you from any frustrations in the end I would strongly advise against returning an incomplete revision and would also understand your decision if you choose to rather seek rapid publication elsewhere at this stage.
I look forward to seeing a revised form of your manuscript.
Should you find that the requested revisions are not feasible within the constraints outlined here and choose, therefore, to submit your paper elsewhere, we would welcome a message to this effect.
Yours sincerely, Editor EMBO Molecular Medicine ***** Reviewer's comments ***** Referee #1 (Comments on Novelty/Model System): Medical impact selected as "medium" because retinoic acid is not part of the lung cancer drug arsenal at present.

Referee #1 (Other Remarks):
Romero et al study the effect of the tumor suppressor BRG1 in lung cancer. Their main findings are that BRG1 antagonizes MYC function and activity, in agreement with their finding that MYC and BRG1 expression are mutually exclusive in lung cancers. A second major effect of BRG1 deletion is a defect in both responses to retinoic acid and glucocorticoids. These findings are of relevance, given the well established role of retinoic acid in lung differentiation and lung cancer.
In general, the experiments are well-performed and controlled and make an important contribution to our understanding of why BGR1 is so often mutated in lung cancers.
There are a few points the authors should consider before publication.

1.
To make a compelling case that the BRG1 mutants do not affect cell growth, their effect should be compared to mock-infected cells, which is missing in Fig. 1D.

2.
In Figure 4B, I cannot agree with the conclusion that knockdown of BRG1 prevents MYC downregulation, as it is only evident for one of the two shRNAs against BRG1 used.
3 Does BRG1 knockdown prevent inhibition of proliferation in RA-induced differentiation in BRG1 wild type lung cancer cells? This would be formal evidence for the notion that BRG1 is required for a retinoic acid response.
This manuscript examines the effects of reexpressing mutant forms of the BRG1 ATPase, the key component of the SWI/SNF chromatin remodeling complex, on gene expression, growth and differentiation in BRG1-deficient human NSCLC cell lines. Through an extensive characterization of the resulting cell lines, the authors draw several conclusions including 1) loss of BRG1 inhibits nuclear receptor responsiveness in NSCLC cells; 2) the SWI/SNF complex negatively regulates MYC family member transactivation function in gene expression and 3) reexpression of BRG1 in deficient NSCLC restores a more differentiated phenotype in the cells. Because of the potential importance of each topic, each of these results could easily merit a manuscript of its own. Therefore, one major problem with the current manuscript lies in the large breadth of data at the sacrifice of any serious depth of analysis. Each of the 6 figures has 4-10 panels of multiple figures that the manuscript does not often explain in sufficient detail for rigorous analyses. The Material and Methods and Figure Legends also contains minimal information to help decode the overwhelming amount of data presented in this manuscript. Therefore, the authors should concentrate the most straightforward observation and remove other data that does not pertain to that study. This reviewer would suggest that the authors' findings concerning the relationship among BRG1 loss, c-myc expression and GR and RAR activation would prove an exciting story for the readers of this journal with potential translational possibilities for the treatment of NSCLC. However, they need to address the issues raised below that would significantly improve the impact of the manuscript.
Major Points-1) The authors should provide a thorough explanation of the gene expression array data in Figure 1. For example, the manuscript does not include a list of the 460 differentially expressed genes. Statistical considerations for how they determined that these differentially enriched genes were enriched for myc targets or that the 15 genes highlighted of the 1355 total genes in Figure 1H showed a correlation with lung development were not presented. The authors do not discuss which genes (a subset or all 460?) were used for the GSEA analyses in Figure 1I&J. The figure legends and material and methods would also benefit from more detail. Considering the importance of these data to the central conclusion of the report, the authors should provide the maximum amount of clarity.
2) Figure 2 shows that the BRG1 lacking 90 amino acids in ATPase domain does not bind to the promoter of 4 different genes. The authors do not provide any data to explain this result. They should at least look at binding of another SWI/SNF complex, such as BAF180, to determine if the BRG1-containing complexes are completely missing from these sites. They also need to perform IP experiments to determine whether the complex forms with this mutant BRG1.
3) The data in Figure 2B&C do not support with the authors' conclusion on page 8 that "an antagonistic function between the tumor suppressor BRG1 and the MYC oncogenes that is consistent with the mutually exclusive presence of BRG1 inactivation and MYC amplification in lung cancer". In Figure 1G, both SERPINE1 and ACP5 expression increases after induction of wtBRG1 but not mtBRG1. However, Figure 2C shows strong recruitment of c-myc to the SERPINE1 promoter after wtBRG1 induction with little recruitment of c-myc to the ACP5 promoter. However, induction of the mtBRG1 causes a strong recruitment of c-myc to the ACP5 promoter with no concomitant increase in gene transcription. The interpretation of these data remains unclear to this reviewer. 4) Similar concerns about the lack of information about the genes analyzed in Figure 2D arise as mentioned above. Figure 3A-E are very clear and strong data. However, Figure 3F seems to contradict the data in Figure 1G. The heat map is Figure 3F compares the mRNA expression in H1299 cells after wtBRG1 induction to the level seen after mtBRG1 expression. According to the data in Figure 1G, genes like SERPINE1 and MCAM should increase by greater than 60X and 10X, respectively. However, in Figure 3F, they increase by 5X and 3X, respectively. The authors need to address this discrepancy.

5)
6) The Western blot in Figure 4B shows a reduction in C-MYC protein in response to RA treatment in sh#4 despite the apparent lack of BRG1 protein. Furthermore, the cell morphology pictures in Figure 4C were also unconvincing. The sh-cntl cells looked more altered by RA treatment than either the parent or the sh#4 cell line. In Figure 4E, the reduction in nucleolin in the knockdown cell lines appeared related to reduced protein loading as demonstrated by the actin controls. Therefore, this Figure should be eliminated or additional knockdown characterized. 7) Additional information about the ChIP protocol would help in the interpretation of the data in Figure 5B. The authors purportedly show recoveries of BRG1 at various promoters in the 10-60% range. Almost every other report demonstrates recovery rates in the 0.05-0.5% range for this protein.
The authors should discuss how they achieved some an amazing increase in their recovery rate.
8) The authors should omit the remainder of Figure 5. The majority of the studies involving the SWI/SNF complex have not shown specific binding to DNA, although several of the complex members have putative DNA binding motifs. If the authors are correct with their observation in the c-myc promoter, this could represent a paradigm shift in the field. Therefore, they should expand these data to provide sufficient experimental support i.e. demonstrated binding to these sequences, use of existing ChIP-seq data for BRG1 and other complex members, refinement of the binding domains, which complex members are involved, etc. 9) There is too much data presented in Figure 6 combined with too little experimental detail in the text and Material and Methods to readily evaluate the data. The H&E and IHC photographs are too small to interpret and the sources and type of c-myc and BRG1 data used to generate Figure 6G remain obscure.
Minor Points 1) Figure 1F should be moved to the supplemental data.
2) It is not clear what Supplemental Figure 2A demonstrates.
3) There are numerous typographical and grammatical errors throughout the manuscript that the authors should address.
Referee #3 (Comments on Novelty/Model System): As described in the comments to authors, key controls are missing.
Referee #3 (Other Remarks): Romero et al have submitted a manuscript "The Tumor Suppressor and Chromatin Remodeling Factor BRG1 Antagonizes Myc Activity and Promotes Cell Differentiation in Human Cancer". In this manuscript, the authors investigate the mechanisms driving BRG1-mutant lung cancers.
Growing evidence has indicated that SWI/SNF complexes serve widespread roles in tumor suppression, and the mechanisms driving tumorigenesis following mutations in BRG1, as well as mutations in other SWI/SNF subunits, have remained unclear and are of current interest. BRG1 is mutated in some non-small cell lung cancer cell lines and in some primary tumors, suggesting important tumor suppressor functions during lung tumorigenesis.
The authors report that re-expression of BRG1 in a BRG1-mutant lung cancer cell line leads to upregulation of genetic programs specific to fully differentiated lung and misregulation of Myc target genes, as well as downregulation of Myc. They also find that re-expression of BRG1 confers increased sensitivity to retinoic acid and glucocorticoid, and BRG1 knockdown decreases sensitivity in BRG1-expressing cell lines. Several papers have previously implicated SWI/SNF complexes in the regulation of both the Myc pathway and pathways mediated by nuclear hormone receptors. This paper evaluates this in the context of BRG1-mutant lung cancers and provides evidence suggesting that these pathways may serve important roles in driving tumor formation and in preventing the response to GC and RA.
While this manuscript potentially provides mechanistic insight into the function of a novel tumor suppressor, with relevance to a major lethal cancer, the manuscript is somewhat difficult to follow and contains several weaknesses that at present make it difficult to determine whether the conclusions are warranted.
First, the authors point out, based upon their prior publication and upon data provided in this manuscript, that amplification of MYC and inactivation of BRG1 are essentially mutually exclusive events. This is an interesting observation that certainly raises the prospect of overlapping mechanisms of these two cancer driving mutations.
Via the data presented in Figure 1, the authors describe finding 460 genes affected by wtBRG1 expression. According to the methods section. The authors report <15% of the genes identified and give no information on the p-value or FDR associated with the observed changes. In addition, figure  1H is confusing and appears inconsistent with Figures 1I and 1J. Figure 1H (with legend and text) seems to imply that, of the 1,355 genes differentially expressed between normal lung and primary lung tumors, only 14 of those genes have their expression affected by restoration of wtBRG1. Consequently, it is difficult to determine why particular genes/pathways became the focus of further study and whether the focus upon GC and RA is warranted. Are BRG1 mutant lung cancers differentially resistant to GC and RA compared to non-BRG1 mutant cancers? If this is not the case, the rationale for pursuing these pathways is less clear.
In Figure 2, the authors demonstrate that wtBRG1 is recruited to MYC consensus regulatory elements in four genes, and that changes in expression of wtBRG1 affect MYC occupancy at these genomic regions. However, there is no apparent pattern relating BRG1 recruitment, MYC recruitment, and transcription. For example, in an experiment designed to determine whether expression of BRG1 affects Myc binding to target, the authors report that BRG1 facilitates Myc binding at two targets but blocks it at the other two. There does not seem to be a strong correlation between these effects and gene expression. While this could be the case, such findings do not lend themselves to a simple model and raise the prospect that the relationship between BRG1 and MYC is complex or indirect.
In Figure 3, the effects of BRG1 expression upon RA and GC induced changes in morphology and gene expression is investigated. However, a critical control -for expression of BRG1 in the absence of treatment with RA and GC -is not provided. This control is essential since the authors report that expression of BRG1 alone causes growth inhibition. It is therefore unclear whether the changes in morphology are due to BRG1 mediated facilitation of RA and GC action, as claimed, or instead are simply due to re-expression of BRG1 itself, regardless of RA and/or GC treatment. The same concern exists with respect to Myc and its target signatures -might these simply be a secondary consequence of BRG1 driven arrest?
Consequently, the statement in the abstract that "refractoriness to RA and GC by BRG1 inactivation involves deregulation of MYC activity" is not clearly supported by the data. While re-expression of BRG1 restores MYC repression in response to RA or GC stimulation, whether a failure to repress MYC has anything to do with the decreased sensitivity of BRG1-mutant cell lines to RA and GC is unclear. Furthermore, it was unclear from figure 4B whether knocking down BRG1 in the SH-SY5Y cell line prevented repression of MYC. While this clearly appears to be the case with one shRNA, the other shRNA appears to have no effect. This leaves the question open as to whether sh#4 is preventing Myc repression via off-target effects.
Ultimately, the motivation for pursuing these studies is to gain insight into mechanisms of oncogenesis in lung cancers, with the potential for identifying new therapies. The authors state "BRG1 restored responsiveness to RA in lung cancer cells orthotopically implanted in nude mice." Yet, in the results (pg 14, 2nd paragraph) they state "the treatment with RA and DEX did not provide additional survival advantages or further reduction in tumor invasiveness in the wtBRG1". Given the partial in vivo response to RA, ie gene expression changes but failure to show increased survival advantages, such a strong statement in the abstract does not seem warranted.
Overall, the finding that BRG1 inactivation and amplification of c-Myc are mutually exclusive (and that this is statistically significant) carries impact as lung cancer is a major disease and the finding suggests, precisely as the authors argue, overlapping function. Consequently, there is good rationale for investigating the relationship between BRG1 and Myc. As described above, however, the absence of key controls, complex findings and a lack of in vivo effect of RA and GC treatment with respect to BRG1 mutation make it unclear that the strong conclusions are warranted.

Minor points:
Figure panel 1D is missing a negative control construct. While it is clear that the mutant BRG1 constructs have less effect than the BRG1 WT construct, it is entirely unclear whether they have some effect themselves in the absence of such a control.
The model figure was too small and not legible -it's unclear whether this was true of the original or whether this was an effect of PDF conversion.
It is unclear how the GSEA analyses were performed in figure 1 and 2. Are the BRG1wt (+Dox) being compared to BRG1wt (-Dox) or the BRG1wt (-Dox) or to the BRG1wt (+Dox)? This should be made clear in the figure legends or by labeling the graphs themselves. It was difficult to determine whether the gene sets tested were being positively or negatively enriched following reintroduction of BRG1.
Responses to the questions/concerns raised by the reviewers.

REVIEWER No.1.
We are particularly grateful to this reviewer for the general positive comments about our work.
Response to each specific comment: Q1.-To make a compelling case that the BRG1 mutants do not affect cell growth, their effect should be compared to mock-infected cells, which is missing in Fig. 1D. A: We agree with the reviewer and have added the mock control (empty vector). This control was already included in the first assay, together with the wt and mutants, but due to space constraints was not part of the original figure. The figure has now been modified and simplified to accommodate this new information (previous Figure 1E has been moved to supplementary information-now Figure S1A). Figure 4B, I cannot agree with the conclusion that knockdown of BRG1 prevents MYC down-regulation, as it is only evident for one of the two shRNAs against BRG1 used. A: This same concern has been raised by all three reviewers. We understand and agree with this point but we would like to point out that this is a technical issue. SH-SY5Y cells infected with sh#1 did not deplete BRG1 protein levels as efficiently as sh#4 did. This was not evident in the western blot of previous figure 4B because of the total protein loading, which was not equal in all lanes (it is reduced in the lane corresponding to sh#1 compared with sh#4, sh-scramble and parental cells). In previous figure S4A, the differences in the efficiency between SH-SY5Y cells infected with shRNA#1 and #4 were clearer. As shown in the old figure, these differences affected the neuroblastoma (SH-SY5Y) cells but not the lung cancer cells (H446), indicating that the ability of both short hairpins to deplete BRG1 protein is similar but that in the SH-SY5Y the percentage of cells infected with the sh#1 is less than 70%. In Annex 1 (for reviewer purposes only) we provide a quantification of the western blot to show this. We agree that, as it stands, the western blot in figure 4B could lead to confusion among readers. To solve this problem, we selected for SH-SY5Y cells infected with sh#1 by culturing the cells in a medium with puromycin for one week. The selection was significantly enriched in cells infected with the shRNA#1. After that, the entire experiment was replicated and the new western blot is shown in figure 4B. The right panels of figure S4A have also been replaced and the text modified accordingly. The quantification of this new western blot is also shown in Annex 1.

Q3.-Does BRG1 knockdown prevent inhibition of proliferation in RA-induced differentiation in BRG1 wild type lung cancer cells? This would be formal evidence for the notion that BRG1 is required for a retinoic acid response.
A: We agree with the reviewer on this point. However, most lung cancer cell lines are quite refractory to RA treatment, including some of the BRG1 wild type (see figure 3F), possibly due to alterations of other members of the SWI/SNF chromatin remodelling complex. In fact, several recent papers (Wilson BG, Roberts CW. Nat Rev Cancer. 2011;Lai AY, Wade PA. Nat Rev Cancer. 2011,Varela I, et al. Nature. 2011 report inactivating mutations at different SWI/SNF chromatin remodelling complex components (ARID1A, ARID1B, ARID2, PBRM1, among others) in different types of cancer, including lung. Therefore, following the reviewer's recommendation, we have used the SH-SY5Y neuroblastoma cells and performed MTT analysis after abrogation of BRG1 expression and treatment with RA. The new results (now included in Supplementary Figure 4C) provide clear evidence that BRG1 knockdown prevents inhibition of proliferation in RA-induced differentiation in the BRG1 wild type SH-SY5Y cancer cells.

REVIEWER No.2.
As a general comment, the reviewer highlights the large amount of data provided in the manuscript:

.) the authors draw several conclusions including 1) loss of BRG1 inhibits nuclear receptor responsiveness in NSCLC cells; 2) the SWI/SNF complex negatively regulates MYC family member transactivation function in gene expression and 3) reexpression of BRG1 in deficient NSCLC restores a more differentiated phenotype in the cells. Because of the potential importance of each topic, each of these results could easily merit a manuscript of its own."
The reviewer suggests that this amount of data prevents an in-depth analysis: "Therefore, one major problem with the current manuscript lies in the large breadth of data at the sacrifice of any serious depth of analysis. Each of the 6 figures has 4-10 panels of multiple figures that the manuscript does not often explain in sufficient detail for rigorous analyses. (...) Therefore, the authors should concentrate the most straightforward observation and remove other data that does not pertain to that study." Of course, we agree with the reviewer's comments about the large amount of data and, consequently, in this revised version of the manuscript we have made efforts to present the figures better. However, in our opinion, the large number of experiments and data provided here do not prevent a serious in-depth analysis. We have used different experimental approaches and a variety of cancer models to demonstrate each of the conclusions reached. Finally, we thank the reviewer for his/her very positive comments about the exciting story that our findings provide for the readers of this journal and about the potential translational possibilities for the treatment of NSCLC.
Response to each specific comment: Q1.-The authors should provide a thorough explanation of the gene expression array data in Figure 1. For example, the manuscript does not include a list of the 460 differentially expressed genes. Statistical considerations for how they determined that these differentially enriched genes were enriched for myc targets or that the 15 genes highlighted of the 1355 total genes in Figure 1H showed a correlation with lung development were not presented. The authors do not discuss which genes (a subset or all 460?) were used for the GSEA analyses in Figure 1I&J. The figure legends and material and methods would also benefit from more detail. Considering the importance of these data to the central conclusion of the report, the authors should provide the maximum amount of clarity. A: We apologize for including little information regarding the bioinformatics and gene expression data. Very similar comments were raised by reviewer No.3. In this revised version of the manuscript we provide this information.
-The list of 456 genes was generated by selecting those transcripts from known genes that fulfilled the following criteria (now described in the supplementary material and methods section as follows): i) induced or repressed by at least 1.2 times in the wild type-induced H1299tr-BRG1wt (dox+) compared with wild type-uninduced H1299tr-BRG1wt (dox-); ii) increased or repressed in the wild type H1299tr-BRG1wt (dox+) in at least two different time-frames and iii) unchanged expression levels in the mutant-induced clones H1299tr-BRG1mut (dox+) versus mutant-uninduced H1299tr-BRG1mut (dox-). We have added a new supplementary table containing this entire list of transcripts (Table S1) which contains additional information such as the P-values. The previous Table S1 is  now Table S2. This contains some selected transcripts, among those from Table S1, with strong down-regulation or up-regulation.
-We used this list to perform the three GSEA analyses contained in figure 1. However, after restricting ourselves to genes that were common in the datasets the list of genes decreased by up to 152 in the GSEA in figure 1H, to 233 in the GSEA of Dataset GDS1650 and 185 of Dataset GSE8569, in figure 1G. We have followed the reviewer's recommendations and included this information in the material and methods section.
-We determined that the BRG1 gene expression signature contained MYC and several MYC target genes by visual examination of individual targets. To clarify this we have modified the previous sentence "This signature involved retinoic acid or glucocorticoid receptor-targets and MYC-targets (Fernandez et al, 2003) " by this other one: "Manual examination of individual targets of this signature revealed the presence of retinoic acid, glucocorticoid receptor and MYC targets (Fernandez et al, 2003)". The statistical correlation with lung development is included in the GSEA with dataset GDS 3447/3448. -Finally, we also agree that we did not provide clear information about what exactly is represented in the previous figure 1H. In fact, it was not our intention to provide statistical information and the 14 genes indicated by an arrow were chosen from the previous Supplementary Table S1 (i.e., lungspecific transcripts and/or MYC targets). The statistics of this dataset (GSE8569) are included in the GSEA analysis from figure 1G (comparative analysis of the six normal lungs and matched tumors). To avoid confusing the readers and, since the bioinformatics are already included, we have now removed the previous figure 1H. Figure 2 shows that the BRG1 lacking 90 amino acids in ATPase domain does not bind to the promoter of 4 different genes. The authors do not provide any data to explain this result. They should at least look at binding of another SWI/SNF complex, such as BAF180, to determine if the BRG1-containing complexes are completely missing from these sites. They also need to perform IP experiments to determine whether the complex forms with this mutant BRG1. A: Reviewer No 2 suggests that the manuscript is too dense and contains too much data. However, she/he asks us to perform additional experiments, all of which are aimed at explaining how the mutant form of BRG1 fails to bind the promoters of several genes. The additional experiments requested include ChIP of other members of the SWI/SNF complex, preferentially BAF180, and immunoprecipitation of the mutant with other SWI/SNF complex members. Of course, we agree that understanding how this mutation affects the function of the SWI/SNF complex is a fundamental question. However, the mutant used in our work, coming from a cancer cell line, is only used here for control purposes. In our opinion, performing these additional experiments is beyond the scope of our current manuscript. We feel that if we were to provide these extra data, the manuscript would be to wide-ranging, thereby detracting from the main focus of the work. Therefore, we have chosen to direct our efforts towards the reviewer's recommendation to concentrate on the most straightforward observations that, according to the reviewer, are BRG1 loss, cmyc expression and GR and RAR activation. We hope that the reviewer will understand and agree with our decision. Figure 2B&C do not support with the authors' conclusion on page 8 that "an antagonistic function between the tumor suppressor BRG1 and the MYC oncogenes that is consistent with the mutually exclusive presence of BRG1 inactivation and MYC amplification in lung cancer". In Figure 1G, both SERPINE1 and ACP5 expression increases after induction of wtBRG1 but not mtBRG1. However, Figure 2C shows strong recruitment of c-myc to the SERPINE1 promoter after wtBRG1 induction with little recruitment of c-myc to the ACP5 promoter. However, induction of the mtBRG1 causes a strong recruitment of c-myc to the ACP5 promoter with no concomitant increase in gene transcription. The interpretation of these data remains unclear to this reviewer. A: The reviewer raises a very valid point, and one that has also puzzled us. We spent a lot of time in the laboratory discussing these experimental observations and trying to decide whether these agree well with the MYC and BRG1 mutually exclusive gene alterations in human cancer. After doing a thorough literature review, we arrived at the following possible explanation: the recruitment of BRG1 in gene promoters facilitates the expression of "good genes", i.e., those associated with cell differentiation (e.g., ACP5, SERPINE1, AQP1, etc.) while repressing the expression of "bad genes", i.e., those associated with cell un-differentiation or stemness (e.g., HES1). In contrast, the pattern of MYC recruitment to gene promoters and its association with gene expression is more complex and depends on the target gene. Therefore, in cooperation with wtBRG1 (SWI/SNF complex), MYC will facilitate the expression of "good genes" (AQP1, SERPINE1), while in the absence of wtBRG1 it will either repress the expression of "good genes" (ACP5) or activate the expression of "bad genes" (HES1). Although MYC is generally considered a transcriptional activator it can also act as a transcriptional repressor (Reviewed in Herkert B, Eilers M (2010) Transcriptional repression: the dark side of myc. Genes Cancer 1:580-586.) Therefore, the recruitment of BRG1 to the E-boxes of AQP1 and SERPINE promoters will have the purpose of opening up the chromatin to allow transcriptional activation mediated by MYC. In contrast, the recruitment of BRG1 to the E-boxes of ACP5 and HES1 promoters will serve the purpose of closing the chromatin to prevent MYC transcriptional repression (ACP5) or MYC transcriptional activation (HES1). In Annex 2, we provide a possible model (for review purposes only), but we would like to make it clear that other models and scenarios are possible.

Q3.-The data in
We understand that these observations need clarification and thus, we have now added the following paragraph to the discussion section (page 16): "However, the relationship of BRG1 and MYC in the control of the expression of MYC target genes is complex and depends on the target gene. In some cases (i.e., AQP1 and SERPINE) the recruitment of BRG1 to the E-boxes will enable MYC-mediated transcriptional activation. In contrast, the recruitment of BRG1 to the E-boxes of ACP5 and HES1 promoters will prevent MYC transcriptional repression (ACP5) or MYC transcriptional activation (HES1)." However, this model still does not explain all the observations, and we would expect that in tumor cells with MYC amplified and BRG1wt, the expression of some "good genes" will be high (e.g., AQP1 and SERPINE1). The GSEA analysis (Fig 2D) shows an inverse correlation between the BRG1 gene expression signature and the expression profile of lungs from mice overexpressing MYC and NMYC. Interestingly, these genes include AQP1, which is down-regulated in NMYCoverexpressing lungs. While this is evidence of the inverse MYC and BRG1 functional relationship, it also suggests that some pieces are missing in the model and therefore that we do not understand the whole mechanism. Therefore, to tone down our previous statement we have modified the sentence from the abstract: "We also demonstrated an antagonistic functional connection between BRG1 and MYC, whereby refractoriness to RA and GC by BRG1 inactivation involves deregulation of MYC activity." to "Further, our data suggest an antagonistic functional connection between BRG1 and MYC, whereby refractoriness to RA and GC by BRG1 inactivation involves deregulation of MYC activity." and from page 8 in the results section: "(...) This suggests an antagonistic function between the tumor suppressor BRG1 and the MYC oncogenes that is consistent with the mutually exclusive presence of BRG1 inactivation and MYC amplification in lung cancer (...)" to this: "(...) Although more experimental evidence is required to draw definitive conclusions and to understand the mechanisms involved, these data suggest an antagonistic function between the tumor suppressor BRG1 and the MYC oncogenes that is consistent with the mutually exclusive presence of BRG1 inactivation and MYC amplification in lung cancer (...) Figure 2D arise as mentioned above. A: Here, the analysis was initially performed with the shorter list of genes that previously constituted the Table S1 (now Table S2). In the new version, to homogenize all the GSEAs we have repeated the analysis with the set of 456 genes. The similarities are maintained while the FDR values are more significant. The gene set sizes were 310 for GSE6077 and 369 for dataset GSE10954, as indicated in the Material and Methods section.

Q4.-Similar concerns about the lack of information about the genes analyzed in
Q5.- Figure 3A-E are very clear and strong data. However, Figure 3F seems to contradict the data in Figure 1G. The heat map is Figure 3F compares the mRNA expression in H1299 cells after wtBRG1 induction to the level seen after mtBRG1 expression. According to the data in Figure 1G, genes like SERPINE1 and MCAM should increase by greater than 60X and 10X, respectively. However, in Figure 3F, they increase by 5X and 3X, respectively. The authors need to address this discrepancy.
A: While in current Figure 1F (previous Fig1G) the changes in gene expression are relative to the indicated internal control, in the heat map in Figure 3E (previous Fig 3F) these changes are also relative to the mutant. This is the reason for the differences. We apologize for the misunderstanding and have now clarified this in the figure legend:

"Heat map of the expression levels, assessed by real time Q-RT-PCR, of the indicated genes and conditions (RA, DEX or FBS, fetal bovine serum) in H1299tr-BRG1wt cells, relative to the levels of expression in H1299tr-BRG1mut, matched for each condition. Asterisks indicate changes of gene expression independent of BRG1."
Q6.-The western blot in Figure 4B shows a reduction in C-MYC protein in response to RA treatment in sh#4 despite the apparent lack of BRG1 protein. Furthermore, the cell morphology pictures in Figure 4C were also unconvincing. The sh-cntl cells looked more altered by RA treatment than either the parent or the sh#4 cell line. In Figure 4E, the reduction in nucleolin in the knockdown cell lines appeared related to reduced protein loading as demonstrated by the actin controls. Therefore, this Figure should be eliminated or additional knockdown characterized. A: First of all, we have modified Figure 4C to show better images of the changes in cell morphology. Second, we understand and agree with this point but we would like to point out that this is a technical issue. SH-SY5Y cells infected with sh#1 did not deplete BRG1 protein levels as efficiently as sh#4 did. This was not evident in the western blot of previous figure 4B because of the total protein loading, which was not equal in all lanes (it is reduced in the lane corresponding to sh#1 compared with sh#4, sh-scramble and parental cells). In previous figure S4A, the differences in the efficiency between SH-SY5Y cells infected with shRNA#1 and #4 were clearer. As shown in the old figure, these differences affected the neuroblastoma (SH-SY5Y) cells but not the lung cancer cells (H446), indicating that the ability of both short hairpins to deplete BRG1 protein is similar but that in the SH-SY5Y the percentage of cells infected with the sh#1 is less than 70%. In Annex 1 (for reviewer purposes only) we provide a quantification of the western blot to show this. We agree that, as it stands, the western blot in figure 4B could lead to confusion among readers. To solve this problem, we selected for SH-SY5Y cells infected with sh#1 by culturing the cells in a medium with puromycin for one week. The selection was significantly enriched in cells infected with the shRNA#1. After that, the entire experiment was replicated and the new western blot is shown in figure 4B. The right panels of figure S4A have also been replaced and the text modified accordingly. The quantification of this new WB is also shown in Annex 1, for review purposes only. Third, we have removed the nucleolin from the WB. Figure 5B. The authors purportedly show recoveries of BRG1 at various promoters in the 10-60% range. Almost every other report demonstrates recovery rates in the 0.05-0.5% range for this protein. The authors should discuss how they achieved some an amazing increase in their recovery rate. A: Following the reviewer's comment, we have gone back to the raw data and revised everything carefully. This has highlighted a mistake in the calculations of the recoveries, which was overestimated by a factor of 10. We have used the protocol described in the Diagenode Manual LowCell#ChIP Kit (http://www.diagenode.com/) and have calculated the enrichment using: % input= AE^ (Ctinput -CtChIP) x Fd x 100% AE is the amplification efficiency of each primer set using the formula: AE= 10^(-1 /slope). CtChIP and Ctinput are threshold values obtained from the exponential phase of qPCR; Fd is a dilution factor of the input DNA used to balance the difference in amounts of ChIP and input DNA taken for qPCR. We had incorrectly used 100 as the Fd instead of the true value of 10.

Q7.-Additional information about the ChIP protocol would help in the interpretation of the data in
The mistake affected all the Q-ChIPs (both for the ChIP with MYC and BRG1 antibodies). We are grateful to the reviewer to have identified this error, which we have now corrected in the corresponding graphs ( Fig. 2 and Fig. 5).
Q8.-The authors should omit the remainder of Figure 5. The majority of the studies involving the SWI/SNF complex have not shown specific binding to DNA, although several of the complex members have putative DNA binding motifs. If the authors are correct with their observation in the c-myc promoter, this could represent a paradigm shift in the field. Therefore, they should expand these data to provide sufficient experimental support i.e. demonstrated binding to these sequences, use of existing ChIP-seq data for BRG1 and other complex members, refinement of the binding domains, which complex members are involved, etc. A: First, we have followed the reviewer's recommendations and removed the remainder of Figure 5. Second, we do not completely understand the reviewer's concerns here. From our observations it cannot be concluded that BRG1 or the SWI/SNF complex directly binds DNA in the MYC locus. Of course, the recruitment of BRG1 to the MYC locus and to the E-boxes of other target genes can be indirect and, could have the purpose of remodelling chromatin (opening/closing chromatin to control gene expression). To our knowledge, we are not the first to show Brg1 recruitment to the Myc promoter, as stated in the discussion section, on page 19: " It is also known that the SWI/SNF complex is recruited to the MYC promoter to activate or repress its transcription in proliferating cells and during cell differentiation, respectively (Chi et al, 2003;Nagl et al, 2006)." The paper by Chi et al. reports that Brg1 selectively binds to the Myc promoter in immature T cells, but not in peripheral resting T cells. In their work, Brg1 is selectively enriched in the Tcr sites of the Myc promoter to enhance Myc gene expression. In this case, since the interest of the authors was the relationship with the Wnt pathway, Brg1 enrichment was only tested at two Tcr binding sites located at about -4kb of the Myc transcriptional start site (TSS) and in a control region at about 12kb downstream the TSS, outside the regions tested in our analysis. The manuscript by Nagl et al., report that the SWI/SNF complex is required, directly, for repression of c-myc during cell differentiation. This involves direct promoter targeting of c-myc gene by these complexes. Furthermore, we have followed the recommendations of the reviewer and compared our current data with that recently reported in the ChIP-seq of BRG1 (Euskirchen, G. et al. 2011). We are pleased to report a strong overlap among the regions found to recruit BRG1 (and other members to the SWI/SNF complex) in the MYC locus and the regions we report in our current study. Specifically, the coincidence is only found with the regions of the MYC locus, recruiting BRG1 in the cells cultured in regular (FBS) conditions. In Annex 3 (for review purposes only) we provide information on the comparison of the BRG1 binding to Myc promoters in the different reports and how these compare with our current data.

Moreover, we have included this information in the discussion section (page 18): "Furthermore, in growing cells, the BRG1-enriched regions within the MYC locus strongly overlap with those recently reported in HeLa cells, using ChIP-Seq to map the binding regions for components of the SWI/SNF complex (Euskirchen et al, 2011)."
Overall, we believe that our data do not represent a paradigm shift in the field but agree closely with all the previous information. We think that our results provide another piece of the puzzle about how the SWI/SNF complex binds to the Myc promoter to permit or to prevent Myc gene expression. Our data also serve to link this with the previous knowledge about how treatment with RA or GC reduced MYC levels. We agree that we could provide additional experimental support and test other complex members, refine the binding domains, analyze which complex members are involved, etc,. However, all this work and additional data, although very interesting and necessary, is beyond the scope of our current manuscript. We hope that the reviewer will understand and agree with our decision. Figure 6 combined with too little experimental detail in the text and Material and Methods to readily evaluate the data. The H&E and IHC photographs are too small to interpret and the sources and type of c-myc and BRG1 data used to generate Figure 6G remain obscure. A: We agree with this comment and to make the figure more readable we have changed/removed various panels in the figure, as indicated below: -To make the MYC immunostaining picture more clear, in the tumors grown in mice, we have removed the 50x and have left only the 200x magnification.

Q9.-There is too much data presented in
-We have moved the H&E staining corresponding to the tumors without induction of BRG1-wt to supplementary figures ( Figure S6A). -We have moved the quantitative RT-PCRs corresponding to the tumors without induction of BRG1-wt to supplementary figures ( Figure S6B). -We have made other changes in the figure to make it more readable and have clarified the sources of material (in the material and methods section) and all the data used to generate figure 6G. We have also included additional experimental details in the material and methods section, regarding the source of the lung primary tumors used. A: In this picture we wanted to show that the chromatin immunoprecipitates contained the corresponding immunoprecipitated proteins (either BRG1 in the BRG1-ChIP or MYC in the MYC-ChIP). We think this is an important technical control that demonstrates that the ChIP worked properly. To clarify this we have slightly changed the figure legend so that it now reads: " Western blot of BRG1 and MYC in the corresponding chromatin immunoprecipitates using antibodies against BRG1 andMYC, respectively, in H1299tr-BRG1wt andH1299tr-BRG1mut cells, with (+)

or without (-) induction of BRG1 expression with 2 ng/ul doxycycline induction for four days."
Q3.-There are numerous typographical and grammatical errors throughout the manuscript that the authors should address. A: We have checked and corrected the errors throughout the manuscript.

While this manuscript potentially provides mechanistic insight into the function of a novel tumor suppressor, with relevance to a major lethal cancer, the manuscript is somewhat difficult to follow and contains several weaknesses that at present make it difficult to determine whether the conclusions are warranted.
The main concern of Reviewer No.3 is the lack of controls. This differs from the comments of reviewer No.1 who says that our experiments were very well performed and controlled. We think that this is because of a misunderstanding. In her/his third comment, the reviewer refers to a lack of a particular control. The mentioned control was, in fact, included in a supplementary figure, as indicated below. Here, we provided additional data to demonstrate that all the controls have been included in each single experiment.
Specific comments: Q1.-Via the data presented in Figure 1, the authors describe finding 460 genes affected by wtBRG1 expression. According to the methods section. The authors report <15% of the genes identified and give no information on the p-value or FDR associated with the observed changes. In addition, figure 1H is confusing and appears inconsistent with Figures 1I and 1J. Figure 1H (with legend and text) seems to imply that, of the 1,355 genes differentially expressed between normal lung and primary lung tumors, only 14 of those genes have their expression affected by restoration of wtBRG1. Consequently, it is difficult to determine why particular genes/pathways became the focus of further study and whether the focus upon GC and RA is warranted.
-The list of 456 genes was generated selecting those transcripts, from known genes, that fulfilled the following criteria (now described in the material and methods section as follows: i) induced or repressed by at least 1.2 times in the wild type-induced H1299tr-BRG1wt (dox+) compared with the wild type uninduced H1299tr-BRG1wt (dox-); ii) were increased or repressed in the wild type H1299tr-BRG1wt (dox+) in at least two different time-frames, and iii) unchanged expression levels in the mutant induced clones H1299tr-BRG1mut (dox+) versus mutant uninduced H1299tr-BRG1mut (dox-) We have now added a new supplementary table that contains the entire list of transcripts (Table S1) and additional information such as the P-values. The previous Table S1 is  now Table S2. Table S2 contains some selected transcripts, among those from Table S1, with strong down-regulation or up-regulation.
-We also agree that we did not provide clear information about what exactly is represented in the previous figure 1H. In fact, it was not our intention to provide statistical information and the 14 genes indicated by an arrow were chosen from the previous Supplementary Table S1 (i.e., lungspecific transcripts and/or MYC targets). The statistics of this dataset (GSE8569) are included in the GSEA analysis from figure 1G (comparative analysis of the six normal lungs and matched tumors). To avoid confusing the readers and, since the bioinformatics are already included, we have now removed the previous figure 1H.
-The genes selected for analysis (current Table S2) were those among the 456 genes that were in the top range within the list of up-regulated/down-regulated genes. Among these, we also took into account that the genes had functional relevance: known targets of MYC (e.g. ACP5, FUT1, HES1, SERPINE1) or targets of RAR/GR (e.g. IL11, SPARC, TIE1) or were very highly expressed in normal lung as compared to lung tumors (e.g. ACP5, AQP1, TIE1).

Q2.-Are BRG1 mutant lung cancers differentially resistant to GC and RA compared to non-BRG1 mutant cancers? If this is not the case, the rationale for pursuing these pathways is less clear.
A: We agree with the reviewer that, ideally, this is a good rationale. However, we have to be very cautious, because BRG1wt cancer cells may still be defective in SWI/SNF activity, through gene alterations at other components of the pathway. In agreement with that, several recent papers (Wilson BG, Roberts CW. Nat Rev Cancer. 2011;Lai AY, Wade PA. Nat Rev Cancer. 2011,Varela I, et al. Nature. 2011 have shown inactivating mutations at other members of the SWI/SNF chromatin remodelling complex (ARID1A, ARID1B, ARID2, PBRM1, among others) in different types of cancer, including lung cancer. These alterations should also lead to refractoriness to RA and GC. In this regard, one of the BRG1-wt lung cancer cells (H1437) tested in our study is refractory to RA ( Figure 3F). This cell line also happens to be wild type for MYC, which suggests that other members of the SWI/SNF complex may be inactivated in these cells. We have realized that we have failed to provide an explanation for this, and have now included the following sentences: On page 11 of the results section: Intriguingly, the H1437 cells, which are wild type for BRG1 and MYC, were completely resistant to RA. and On page 17 of the discussion section: Intriguingly, the H1437 cells, which are wild type for BRG1 and MYC, showed resistance to RA, indicating a defect in SWI/SNF activity probably caused by alterations of genes responsible for other components of the pathway (Wilson et al, 2011).
In conclusion, we believe that since cancer cells that are wild type for BRG1 may still have impaired SWI/SNF activity due to gene alterations in other members of the complex, the premise proposed by the reviewer is not so straightforward. Here we have restored the response to RA after restoring BRG1 in BRG1-deficient cells and have reverted the RA response after abrogating BRG1 wild type expression in a wide variety of cancer cells. In our opinion, these constitute a good proof-ofprinciple to link BRG1 with the response to these agents. We hope that future work will yield other experimental approaches that will add to these. We hope the reviewer will understand and agree with our explanation. Figure 2, the authors demonstrate that wtBRG1 is recruited to MYC consensus regulatory elements in four genes, and that changes in expression of wtBRG1 affect MYC occupancy at these genomic regions. However, there is no apparent pattern relating BRG1 recruitment, MYC recruitment, and transcription. For example, in an experiment designed to determine whether expression of BRG1 affects Myc binding to target, the authors report that BRG1 facilitates Myc binding at two targets but blocks it at the other two. There does not seem to be a strong correlation between these effects and gene expression. While this could be the case, such findings do not lend themselves to a simple model and raise the prospect that the relationship between BRG1 and MYC is complex or indirect. A: The reviewer raises a very valid point, and one that has also puzzled us. We spent a lot of time in the laboratory discussing these experimental observations and trying to decide whether these agrees well with the MYC and BRG1 mutually exclusive gene alterations in human cancer. After doing a thorough literature review, we arrived at the following explanation: the recruitment of BRG1 in gene promoters facilitates the expression of "good genes", i.e., those associated with cell differentiation (e.g., ACP5, SERPINE1, AQP1, etc.) while repressing the expression of "bad genes", i.e., those associated with cell un-differentiation (e.g., HES1). In contrast, the pattern of MYC recruitment to gene promoters and its association with gene expression is more complex and depends on the target gene. Therefore, in cooperation with wtBRG1 (SWI/SNF complex), MYC will facilitate the expression of "good genes" (AQP1, SERPINE1), while in the absence of wtBRG1 it will either repress the expression of "good genes" (ACP5) or activate the expression of "bad genes" (HES1). Although MYC is generally considered a transcriptional activator it can also act as a transcriptional repressor (Reviewed in Herkert B, Eilers M (2010) Transcriptional repression: the dark side of myc. Genes Cancer 1:580-586.) Therefore, the recruitment of BRG1 to the E-boxes of AQP1 and SERPINE promoters will have the purpose of opening up the chromatin to allow transcriptional activation mediated by MYC. In contrast, the recruitment of BRG1 to the E-boxes of ACP5 and HES1 promoters will serve the purpose of closing the chromatin to prevent MYC transcriptional repression (ACP5) or MYC transcriptional activation (HES1). In Annex 2, we provide a possible model (for review purposes only), but we would like to make it clear that other models and scenarios are possible.

Q3.-In
We understand that these observations need clarification and thus, we have now added the following paragraph to the discussion section (page 16): "However, the relationship of BRG1 and MYC in the control of the expression of MYC target genes is complex and depends on the target gene. In some cases (i.e., AQP1 and SERPINE) the recruitment of BRG1 to the E-boxes will enable MYC-mediated transcriptional activation. In contrast, the recruitment of BRG1 to the E-boxes of ACP5 and HES1 promoters will prevent MYC transcriptional repression (ACP5) or MYC transcriptional activation (HES1)." However, this model still does not explain all the observations, and we would expect that in tumor cells with MYC amplified and BRG1wt, the expression of some "good genes" will be high (e.g., AQP1 and SERPINE1). The GSEAs analysis (Fig 2D) shows an inverse correlation between the BRG1 gene expression signature and the expression profile of lungs from mice overexpressing MYC and NMYC. Interestingly, these genes include AQP1, which is down-regulated in NMYCoverexpressing lungs. While this is evidence of the inverse MYC and BRG1 functional relationship, it also suggests that some pieces are missing in the model and therefore that we do not understand the whole mechanism. Therefore, to tone down our previous statement we have modified the sentence from the abstract: "We also demonstrated an antagonistic functional connection between BRG1 and MYC, whereby refractoriness to RA and GC by BRG1 inactivation involves deregulation of MYC activity." to "Further, our data suggest an antagonistic functional connection between BRG1 and MYC, whereby refractoriness to RA and GC by BRG1 inactivation involves deregulation of MYC activity." and from page 8 in the results section: "(...) This suggests an antagonistic function between the tumor suppressor BRG1 and the MYC oncogenes that is consistent with the mutually exclusive presence of BRG1 inactivation and MYC amplification in lung cancer (...)" to this: "(...) Although more experimental evidence is required to draw definitive conclusions and to understand the mechanisms involved, these data suggest an antagonistic function between the tumor suppressor BRG1 and the MYC oncogenes that is consistent with the mutually exclusive presence of BRG1 inactivation and MYC amplification in lung cancer (...) Q4.-In Figure 3, the effects of BRG1 expression upon RA and GC induced changes in morphology and gene expression is investigated. However, a critical control -for expression of BRG1 in the absence of treatment with RA and GC -is not provided. This control is essential since the authors report that expression of BRG1 alone causes growth inhibition. It is therefore unclear whether the changes in morphology are due to BRG1 mediated facilitation of RA and GC action, as claimed, or instead are simply due to re-expression of BRG1 itself, regardless of RA and/or GC treatment. A: We apologize for the mistake in the legend of figure S3, which we have now corrected. Also, we had not made it clear enough that the critical control mentioned by the reviewer is included. Some of them pertain to the supplementary material, for example, in former Figure S3. This picture shows that the morphology of the green cells is completely different after treatment with RA or DEX or both, compared with cells cultured in regular FBS. To clarify this, we have added more pictures of these controls in current figure 3A and in the supplementary figure 3 ( Figure S3C-E). We have included them as supplementary material because the main figures have already too many panels and data and the other reviewers have drawn attention to this.
In addition, in previous figure 3F (now Figure 3E), the data under FBS refer to fetal bovine serum, that is the gene expression for the indicated cells growth in regular culture conditions. Further supporting the role of BRG1 in mediating gene expression of RA and GC targets is the fact that the values in this figure are relative to the BRG1mut, as indicated in the figure legend. We have slightly modified the figure legend that now reads: Heat map of the expression levels, assessed by real time Q-RT-PCR, of the indicated genes and conditions (RA, DEX or FBS, fetal bovine serum) in H1299tr-BRG1wt cells. Gene expression relative to an internal control (GUSB) and to the levels of expression in H1299tr-BRG1mut, matched for each condition. Asterisks indicate changes of gene expression independent of BRG1. Finally, the response to RA, measured as changes in the levels of the RA-dependent transcripts CYP26A1 or RARB, is also evident only in the BRG1wt cells treated with RA (Fig. 3G, Fig. 4A and Fig. S3B). In all cases, the BRG1wt cells without RA treatment are included, as control.
We thank the reviewer for the opportunity to correct some errors and to present our results in a better way. We hope that the explanations, changes and new data in the new version of the manuscript will serve to clarify the doubts raised by the reviewer.
Q5.-The same concern exists with respect to Myc and its target signatures -might these simply be a secondary consequence of BRG1 driven arrest? Consequently, the statement in the abstract that "refractoriness to RA and GC by BRG1 inactivation involves deregulation of MYC activity" is not clearly supported by the data. A: The response to this comment is answered below in response to Q6.
Q6: While re-expression of BRG1 restores MYC repression in response to RA or GC stimulation, whether a failure to repress MYC has anything to do with the decreased sensitivity of BRG1-mutant cell lines to RA and GC is unclear. Furthermore, it was unclear from figure 4B whether knocking down BRG1 in the SH-SY5Y cell line prevented repression of MYC. While this clearly appears to be the case with one shRNA, the other shRNA appears to have no effect. This leaves the question open as to whether sh#4 is preventing Myc repression via off-target effects.
A: We understand the reviewer's point of view. However, we disagree to some extent and would like to add some comments by way of a reply. In addition to changes in morphology, compatible with cell differentiation, one of the features of the response to RA is reduction in cell growth (Doyle et al 1989;Påhlman S, et al. 1984, among many others). Therefore, it seems appropriate to propose that a failure to repress MYC is, at least, one of the reasons for the decreased sensitivity to RA in BRG1mutant cells. However, it is clear that MYC repression is not the only reason for the resistance to RA-treatment in BRG1-mutant cells. To clearly state that the reduction of MYC levels is important, but is not the only factor responsible for the RA-dependent cell differentiation effects after restituting BRG1 in cancer cells, we have modified some of the sentences in the manuscript, as follows: Regarding the second comment of the reviewer, we apologize for the problem with this figure. The same concern was raised by the other two reviewers. We understand and agree with this point but we would like to point out that this is a technical issue. SH-SY5Y cells infected with sh#1 did not deplete BRG1 protein levels as efficiently as sh#4 did. This was not evident in the western blot of former Figure 4B, because of total protein loading, which was not equal in all lanes (reduced in the lane corresponding to sh#1 as compared to sh#4, shscramble and parental cells). In former figure S4A, the differences in the efficiency between SH-SY5Y cells infected with shRNA#1 and #4 are clearer. As shown in the figure, these differences affect the neuroblastoma (SH-SY5Y) cells but not the lung cancer cells (H446), indicating that the capability of both short hairpins to deplete BRG1 protein is similar but that in the SH-SY5Y the percentage of cells infected with the sh#1 is less than 70%. In Annex 1 (for review purposes only) we provide a quantification of the western blot to show this. We agree that, as it was originally presented, the previous western blot in figure 4B could be confusing for readers. To solve this problem, we have selected for SH-SY5Y cells infected with sh#1 by culturing the cells in a medium with puromycin for one week. The selection significantly enriched the cells infected with the shRNA#1. After that, the entire experiment was replicated and the new western blot is shown in Figure 4B. The right panels of figure S4A have also been replaced and the text modified accordingly. The quantification of this new western blot is also shown in Annex 1.
Q7.-Ultimately, the motivation for pursuing these studies is to gain insight into mechanisms of oncogenesis in lung cancers, with the potential for identifying new therapies. The authors state "BRG1 restored responsiveness to RA in lung cancer cells orthotopically implanted in nude mice." Yet, in the results (pg 14, 2nd paragraph) they state "the treatment with RA and DEX did not provide additional survival advantages or further reduction in tumor invasiveness in the wtBRG1". Given the partial in vivo response to RA, i.e. gene expression changes but failure to show increased survival advantages, such a strong statement in the abstract does not seem warranted. A: The sentence "BRG1 restored responsiveness to RA in lung cancer cells orthotopically implanted in nude mice" in the abstract refers to the observations in the mice models, regarding increased in gene expression of two targets of RA: CYP26A1 and RARB and of MYC. These differences are highly significant when comparing BRG1wt (H1299tr-BRG1wt, dox+) versus BRG1mut (H1299tr-BRG1mut, dox+) tumors. If we only consider the group of mice carrying tumors with restored BRG1wt (H1299tr-BRG1wt, dox+), the treatment with RA increased the tumor levels of RA-target genes only moderately (the differences were not statistically significant). If we look at the data ( Fig  S6B), this is because the levels of the RA targets were already high in the BRG1wt (not treated with RA), compared with the BRG1 mutant (with or without RA treatment). In contrast to cancer cell lines grown in culture, the tumors derived from the same cell lines grown in mice cannot be controlled for environmental influences. The blood of the animals contain hormones, including RA or GC. These could be even higher in our population of mice, compared with normal individuals, as a consequence of the physiological stress associated with malignant tumor development. In the results section we have proposed the following explanation for that: " It is interesting to note that the tumor levels of CYP26A1, MYC and RARB were higher in the wtBRG1-expressing mice, which received RA treatment, compared with the wtBRG1-expressing mice that did not receive RA-treatment (groups 2 and 5) (Fig S6B), although the differences were not statistically significant. This was probably due to the presence of hormones in the blood of these animals. The endogenous levels of RA and GC were probably high in these mice in response to the injury caused by the experimental procedures and the tumor development. This also helps explain the lack of additional survival benefits in the RA-treated mice." We also acknowledge that, in spite of the difficulties in controlling the levels of the naturally released hormones in the blood, the failure of RA treatment to increase overall survival in BRG1wt-expressing mice, requires us to make less bold conclusions. Therefore, we have changed the statement in the abstract from: "Finally, BRG1 restoration significantly dampened invasion and progression and decreased MYC and restored responsiveness to RA in lung cancer cells orthotopically implanted in nude mice." to "Finally, BRG1 restoration significantly dampened invasion and progression and decreased MYC in lung cancer cells orthotopically implanted in nude mice." Minor points: Q1: Figure panel 1D is missing a negative control construct. While it is clear that the mutant BRG1 constructs have less effect than the BRG1 WT construct, it is entirely unclear whether they have some effect themselves in the absence of such a control. A: We agree with this reviewer and have added the mock control. The mock controls had already been done, at the same time as the wt and mutants, but we had not included them because of space constraints. The new pictures, corresponding to the empty vector (mock), have now been included in Figure 1D. To allocate the new information and following the reviewer's recommendations, the panels of figure 1 have been simplified. Q2: The model figure was too small and not legible -it's unclear whether this was true of the original or whether this was an effect of PDF conversion. A: We have provided a higher quality figure.
Q3: It is unclear how the GSEA analyses were performed in figure 1 and 2. Are the BRG1wt (+Dox) being compared to BRG1wt (-Dox) or the BRG1wt (-Dox) or to the BRG1wt (+Dox)? This should be made clear in the figure legends or by labeling the graphs themselves. It was difficult to determine whether the gene sets tested were being positively or negatively enriched following re-introduction of BRG1. A: We apologize for the lack of clarity in this regard. The BRG1 signature to perform the GSEAs comparisons has been included in the current Table S1. We have now explained this in the material and methods section. We have also clarified in the figure legends (as shown below) whether the different gene sets correlate positively or negatively with the BRG1 gene expression signature.  Table S1). The genes up-regulated in the lung cancer cell line (H1299tr-BRG1wt) upon  are significantly similar to those up-regulated in normal lungs and the lungs from late embryos. P-values and false discovery rates (FDRs) are indicated. Error bars, SD of three replicates." Figure 2D. Ranked  Thank you for the submission of your revised manuscript to EMBO Molecular Medicine. We have now received the enclosed reports from the referees that were asked to re-assess it. As you will see the reviewers are now supportive and I am pleased to inform you that we will be able to accept your manuscript pending the following final amendments: -As you will see Ref.
3 is suggesting few text amendements in the main document and in The Paper Explained.
Please submit your revised manuscript within two weeks. I look forward to seeing a revised form of your manuscript as soon as possible.
I look forward to reading a new revised version of your manuscript as soon as possible.
Yours sincerely, Editor EMBO Molecular Medicine ***** Reviewer's comments ***** Referee #1 (Comments on Novelty/Model System): RA is not particularly relevant to the treatment of lung cancer, hence the medical impact "medium".
Referee #1 (Other Remarks): The revised manuscript by Romero et al has addressed the three major points that I raised in a satisfactory fashion. I support publication in the present form.
Referee #3 (Comments on Novelty/Model System): I now favor publication for the reasons outlined in the response to the author. Overall, the data in this manuscript provide important mechanistic insight into the role of BRG1 and the SWI/SNF complex, which are frequently mutated in a common type of lung cancer, as well as in other cancers.
Referee #3 (Other Remarks): The authors have added appropriate controls, clarified text and performed additional experiments.
They have now substantially addressed my concerns.
Q1. With respect to their response to my "Q1", they have now provided methodological information on how the gene expression data in Figure 1 was obtained and analyzed. Now understanding how/why they were performed, the GSEA analyses showing enrichment of the BRG1 upregulated signature in normal lung compared to lung cancer provides substantial support to the authors' contention that re-expression of BRG1 leads to upregulation of normal lung developmental programs.
Q2. With respect to Q2, the authors make a reasonable counter-argument for why the experiment I proposed would be difficult to interpret given that mutations of other SWI/SNF subunits are turning out to be so frequent. With respect to their new text on p 17 explaining this result however, I think that it should be changed from "indicating a defect in SWI/SNF activity probably caused by alterations of genes responsible for other components of the pathway" to "...conceivably caused...". The former language is too strong and not warranted.
Q3. I agree that the data are complex but there is enough data to support the relationship, particularly given the mutually exclusive nature of Brg1 and myc mutations. Therefore, the proposed text revisions now provide an appropriate context for the data.
Q4. The new appropriately resolve this concern.
Q5-7. The clarifications and revised text have provide additional context and more appropriately frame the data as well as acknowledge its limitations.
New point: In "The Paper Explained" section, the authors state "One third of the lung tumors endure inactivation of BRG1, a component of a chromatin remodeling complex, which is the fourth most commonly altered gene in this type of cancer." It's not clear to me that this statement is warranted. From the references, I was under the impression that BRG1 is mutated in non-small cell lung cancer only, not lung cancer in general. Further, while it's inactivation is frequent in cell lines (roughly 1/3), it's not clear from the data from primary cancers that the rate is this high -it seems lower based upon the authors paper. Therefore, I suggest that this text be revised.
2nd Revision -authors' response 27 February 2012 Changes to the manuscript to address the comments of Reviewer No 3 Q2. With respect to Q2, the authors make a reasonable counter-argument for why the experiment I proposed would be difficult to interpret given that mutations of other SWI/SNF subunits are turning out to be so frequent. With respect to their new text on p 17 explaining this result however, I think that it should be changed from "indicating a defect in SWI/SNF activity probably caused by alterations of genes responsible for other components of the pathway" to "...conceivably caused...". The former language is too strong and not warranted.
We agree with the reviewer and have toned down the statement from: "Intriguingly, the H1437 cells, which are wild type for BRG1 and MYC, showed resistance to RA, indicating a defect in SWI/SNF activity probably caused by alterations of genes responsible for other components of the pathway (Wilson et al, 2011)." to the new one: "Intriguingly, the H1437 cells, which are wild type for BRG1 and MYC, showed resistance to RA, suggesting a defect in SWI/SNF activity that could be caused by gene alterations at other components of the complex (Wilson et al, 2011)." New point: In "The Paper Explained" section, the authors state "One third of the lung tumors endure inactivation of BRG1, a component of a chromatin remodeling complex, which is the fourth most commonly altered gene in this type of cancer." It's not clear to me that this statement is warranted. From the references, I was under the impression that BRG1 is mutated in non-small cell lung cancer only, not lung cancer in general. Further, while it's inactivation is frequent in cell lines (roughly 1/3), it's not clear from the data from primary cancers that the rate is this high -it seems lower based upon the authors paper. Therefore, I suggest that this text be revised.