Bcl-xL is translocated to the nucleus via CtBP2 to epigenetically promote metastasis

  1. Tiantian Zhang
  2. Sha Li
  3. Yingcai Adrian Tan
  4. Joseph HyungJoon Na
  5. Zhengming Chen
  6. Priyadarshan Damle
  7. Xiang Chen
  8. Soyoung Choi
  9. Bikash Mishra
  10. Dunrui Wang
  11. Steven R. Grossman
  12. Xuejun Jiang
  13. Yi Li
  14. Yao-Tseng Chen
  15. Jenny Z. Xiang
  16. Yi-Chieh Nancy Du

Reviewed by

Read the full article

Listed in

Log in to save this article

Abstract

Besides its mitochondria-based anti-apoptotic role, Bcl-xL also travels to the nucleus to promote cancer metastasis by upregulating global histone H3 trimethyl Lys4 (H3K4me3) and TGFβ transcription. How Bcl-xL is translocated into the nucleus and how nuclear Bcl-xL regulates H3K4me3 modification are not understood. Here, we report that C-terminal Binding Protein 2 (CtBP2) binds Bcl-xL via its N-terminus and translocates Bcl-xL into the nucleus. Knockdown of CtBP2 by shRNA decreases the nuclear portion of Bcl-xL and reverses Bcl-xL-induced cell migration and metastasis in mouse models. Furthermore, knockout of CtBP2 suppresses Bcl-xL transcription. The binding between Bcl-xL and CtBP2 is required for their interaction with MLL1, a histone H3K4 methyltransferase. Pharmacologic inhibition of MLL1 enzymatic activity reverses Bcl-xL-induced H3K4me3 and TGFβ mRNA upregulation as well as cell invasion. Moreover, cleavage under targets and release using nuclease (CUT&RUN) coupled with next generation sequencing reveals that H3K4me3 modifications are particularly enriched in the promotor region of genes encoding TGFβ and its signaling pathway in the cancer cells overexpressing Bcl-xL. Altogether, the metastatic function of Bcl-xL is mediated by its interaction with CtBP2 and MLL1.

Article activity feed

  5. Review Commons

    Note: This response was posted by the corresponding author to Review Commons. The content has not been altered except for formatting.

    Learn more at Review Commons

    Reply to the reviewers

    Response to Referees Letter

    We thank the reviewers for their constructive comments and their positive comments that this study provides insights into the non-canonical roles of Bcl-xL in cancer and may lead to therapeutic approaches to repress metastatic capacity. We have carefully read their comments and have extensively revised the manuscript accordingly. The specific points made by each reviewer are addressed below in blue color.

    Response to Reviewer #1:

    Reviewer #1 (Evidence, reproducibility and clarity (Required)):

    Summary In this study the authors build on their previous work that Bcl-xL has a role in metastasis promotion independent of it's function in the mitochondrial apoptotic pathway. They show that Bcl-xL can be found in the nucleus of some human breast cancer cells and through a mass spec approach show that CtBP2 promotes the nuclear translocation of Bcl-xL. Using various knockdown/knockout methods they show that reduced levels of CtBP2 reduces metastasis, because of loss of Bcl-xL translocation to the nucleus. The authors map this interaction and show that this interaction modulates metastasis.

    Major comments * Figure 1 - a more comprehsive analysis of nuclear Bcl-xL should be conducted. The data presented only shows 3 different samples, with no quantification. Perhaps the authors could stain a breast cancer TMA or similar?

    __Response: __We performed breast cancer TMA staining experiment as suggested. This experiment provides further support to our conclusion. We have included the following information in the revised manuscript.

    “We further evaluated human breast specimens in tissue microarrays (TMAs), consisting of 25 non-neoplastic breast tissues, 150 primary breast cancer, 55 lymph node metastases, and 99 metastatic breast cancer at various distant sites, for the expression and localization of Bcl-xL by immunohistochemistry. Compared to normal breast tissues, the intensity of Bcl-xL was significantly higher in breast cancer, including primary tumors, lymph node (LN) metastases, and distant metastases (Table 1a and 1b). The proportion of positive perinuclear/nuclear Bcl-xL cases was significantly increased in human breast cancer tissues compared to normal breast tissues (Table 1c and Figure 1d), and it showed an increasing trend towards metastases (Table 1d, p =0.004).”

    * Figure 2 - could the authors show a graph with a representation of the mass spectrometry data, so the reader can get a sense of how many proteins were found to be associated with Bcl-xL?

    __Response: __As suggested, we have included the mass spectrometry data in Supplemental Table 1. Forty proteins were commonly immunoprecipitated by anti-HA magnetic beads from all three cell lines overexpressing HA-tagged wt Bcl-xL and two Bcl-xL mutants but not from the parental cells overexpressing the control vector.

    * Have the authors tried any other ways to verify the interaction between Bcl-xL and CtBP2? For instance, do they co-localise when imaged? Also, can the reverse IP be performed?

    __Response: __We have verified the interaction between Bcl-xL and CtBP2 by several methods, including IP, reverse IP, and co-immunostaining. Please find HA-Bcl-xL IP and Western for endogenous CtBP2 (Figure 2a), co-immunostaining of endogenous Bcl-xL and CtBP2-V5 (Figure 2b and 2c), co-immunostaining of endogenous Bcl-xL and endogenous CtBP2 (Figure 4e), HA-Bcl-xL IP and Western for seven different constructs of V5 tagged CtBP2 (Figure 5b and 5c), and V5-CtBP2 IP and Western for seven different constructs of Myc tagged Bcl-xL (Figure 6b).

    * Figure 2C - the authors claim that this data shows that Bcl-xL nuclear translocation is reduced in cells with reduced levels of CtBP2 - however, although they quantify this I simply do not see it from the images presented. I do not think this data supports the conclusion that knockdown of CtBP2 reduces Bcl-xL translocation to the nucleus. Furthermore, this data is only shown with overexpressed Bcl-xL - have the authors tried with endogenous staining of Bcl-xL?

    Response: To assist Reviewer #1’s visualization, below are some marked RFP+ cells that responded to Dox-inducible shRNA expression from Figure 2e. Please note that these cells were not sorted by dsRed so that they gave us a unique opportunity to determine whether the knockdown of CtBP2 affected Bcl-xL nuclear localization by comparing subcellular localization of HA-Bcl-xL in the dsRed-positive cells and the neighboring dsRed-negative cells in the same images. The nuclear-to-cytosol ratio of HA-Bcl-xL was reduced in the dsRed-positive shCtBP2 cells compared to the dsRed-negative cells in both shCtBP2 #2260 and #2403 cultures on dox, not in shRLuc #713 control cells on dox.

    In addition, we have performed endogenous staining of Bcl-xL and found that CtBP2 knockout reduced the nuclear to cytosol ratio of endogenous Bcl-xL (Figure 4f).

    * Figure 2e-f - again these data are in cells with overexpressed Bcl-xL - does the same effect on invasion happen when only CtBP2 levels are reduced, without overexpression of Bcl-xL? What happens when Bcl-xL is knocked down? Also, doxycycline has been shown to affect mitochondrial function, which might confound this data - perhaps another way to knockdown CtBP2 (e.g. CRISPR which is used later in the study) would rule this out

    Response: First, we have previously reported that CtBP2 knockdown reduced migration in cells without overexpression of Bcl-xL (Paliwal et al., 2007), and others have shown that siRNA knockdown of Bcl-xL reduces migration and invasion (Trisciuoglio et al., 2017).

    Second, to control any effect of doxycycline, we have included the doxycycline-fed control cells that express doxycycline-inducible shRNA against Renilla Luciferase (shRLuc #713) in revised Figure 2g and 2h (original Figure 2e and 2f).

    Third, the novelty of this study is that the discovery that Bcl-xL and CtBP2 interact with each other to promote metastasis. Our study showed that CtBP2 controls Bcl-xL in two ways: nuclear translocation and transcription. Because we found that knockout CtBP2 reduced transcription of endogenous Bcl-xL (Figure 4a-c), it will make the interpretation of the migration effect difficult. Using cells overexpressing HA-Bcl-xL, whose transcription is not regulated by CtBP2, we can evaluate whether the invasion effect of HA-Bcl-xL is mediated by CtBP2 when CtBP2 is knocked down. While overexpression of Bcl-xL promotes invasion (Choi et al., 2016), knockdown of CtBP2 can reverse the effect (Figure 2g).

    * Figure 3c - these blots are not labelled, but ideally this would be shown with endogenous Bcl-xL, rather that just the overexpressed HA-Bcl-xL. However these data are more convincing than the images presented in Figure 2c

    __Response: __We apologize for the missing labels in these blots of Figure 3c when we merged the graphs. We have now added them back.

    * Figure 4 - the authors use CRISPR to knockout CtBP2 - logically this data would go with the shRNA data shown before, as it seems to just repeat what has already been shown?

    __Response: __In Figure 4, we examined the effect of CtBP2 knockout on the endogenous Bcl-xL. We were pleased to see that CtBP2 knockout reduced the nuclear-to-cytosol ratio of endogenous Bcl-xL. Moreover, we observed that CtBP2 knockout reduced transcription of Bcl-xL. These knockout data (Figure 4) were logically presented after the knockdown data (Figure 2 and 3).

    * Figure 4d - what does "SN" refer to? There is no loading control for this part of the fractionation - I assume this is supernatant? If so, why is there no loading control for this (same applies to figure 3c). Also, why are these not on the same blot? If CtBP2 knockdown reduces Bcl-xL mRNA level, does it also reduce Bcl-xL protein levels? We should be able to tell this from the blots in figure 4d, but since they are on different membranes this is impossible to deduce.

    __Response: __We apologize for the missing information. We have added “SN: soluble nuclear fraction” in the figure legend of Figure 4d and re-run all the samples on the same blot. No detection of cytoplasmic proteins and chromatin-bound proteins in the soluble nuclear fraction suggested good fractionation as described (Méndez and Stillman, 2000, PMID: 11046155). CtBP2 knockout indeed reduced Bcl-xL protein levels, as shown in Figure 4a.

    * Figure 5c - molecular weight markers should be included here.

    __Response: __We apologize for the missing labels of the molecular weight markers, and we have added them in the revision.

    * Figure 7a - the text says that MM102 treatment "significantly reduced" H3K4me3 levels - where is the quantification of this?

    __Response: __We appreciate the suggestion, and we have now added the quantification in Figure 7a.

    Minor comments * Some of the figures are not properly labelled * Some of the data are presented in an awkward manner - the authors should consider re-structuring either the manuscript or the figures so there is less "jumping around"

    __Response: __We apologize for the missing labels again, and we have now labeled the figures properly. We hope that the revision (with additional data and properly labelled figures) has made the structure of the manuscript sound.

    Reviewer #1 (Significance (Required)):

    General assessment * Provides new insight into non-canonical roles of Bcl-xL in cancer * Relies heavily on over-expressed proteins to draw conclusions * If the data were stronger and supported the conclusions, this study could be of interest to a broad cancer audience

    My expertise Cell biology, cell death, cancer, imaging

    __Reviewer #2 (Evidence, reproducibility and clarity (Required)): ____ __ The manuscript describes a large number of experiments each of which describes a small part of the functional cascade of Bcl-xL in nuclear function and metastatic tumor behavior. No one experiment accomplishes a lot, but taken as a total, the story is compelling and fairly complete.

    Major: Figure 1 shows Bcl-xL in one primary sample (a) but clearly not in a second one (c). The authors state 3 of 15. Can they make any comment about breast cancer subtype of these 3 or outcomes? This seems fairly thin evidence of Bcl-xL involvement in human tumorigenesis in general - a better survey might be performed with tissue microarrays of more than one cancer subtype. I'm not sure that this figure is compelling or necessary really for the rest of the manuscript. Really, the main weakness of this paper is some proof that this Bcl-xL-mediated pathway is significant in some proportion of human cancer and metastasis. Perhaps some RNASeq datasets on metastatic versus localized cancers could be mined to establish this relvance?

    __Response: __We appreciate this suggestion. We have compared the breast cancer subtypes and the outcomes of the cases used in the original immunofluorescent study. No particular cancer subtype or outcome of these cases is associated with the presence of more nuclear Bcl-xL.

    As suggested by the reviewer, we used breast cancer TMAs to investigate the involvement of Bcl-xL in human tumorigenesis in general. We have found that the cases positive of peri-nuclear and nuclear Bcl-xL showed an increasing trend of metastases (Table 1d). We have included the following information in the revised manuscript.

    “We further evaluated human breast specimens in tissue microarrays (TMAs), consisting of 25 non-neoplastic breast tissues, 150 primary breast cancer, 55 lymph node metastases, and 99 metastatic breast cancer at various distant sites, for the expression and localization of Bcl-xL by immunohistochemistry. Compared to normal breast tissues, the intensity of Bcl-xL was significantly higher in breast cancer, including primary tumors, lymph node (LN) metastases, and distant metastases (Table 1a and 1b). The proportion of positive perinuclear/nuclear Bcl-xL cases was significantly increased in human breast cancer tissues compared to normal breast tissues (Table 1c and Figure 1d), and it showed an increasing trend towards metastases (Table 1d, p =0.004).”

    Most other experiments and figures are well explained. The only one I have some trouble with is Figure 8 CUT and RUN data where we are only presented with peaks around six genes. Is there a way to summarize data for the rest of the genome? Or to display a composite of CUT and RUN data on promoters that are not predicted to be targets of Bcl-xL and MLL1 activity (compared to those that are)?

    __Response: __We have deposited the entire CUT&RUN-Seq datasets in Gene Expression Omnibus (accession #GSE221629, https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc= GSE221629), which will become publicly available when the manuscript is published.

    It is very challenging to present 1,190 unique H3K4me3 histone modification regions, and we tried our best to present the CUT&RUN-Seq data in the revised manuscript. In addition to the differential H3K4me3 peaks around promoters of six genes, we have included genome browser view, including the whole gene body by zooming out in Supplementary Figure S7 and peaks for 9 regions that are not targets of Bcl-xL and MLL1 activity in Supplementary Figure S8. Furthermore, we used Hypergeometric Optimization of Motif EnRichment (HOMER) to perform motif analysis for the differential H3K4me3 peaks. Enrichment p-values of the motifs were between 1e-12 and 1e-2 (Supplementary Table S5). It is of note that motifs with a p-value of more than 1e-10 or even 1e-12 are likely to be false positives (http://homer.ucsd.edu/homer/introduction/basics.html). The result revealed the limitation to identify motifs around the H3K4me3 CUT&RUN peaks recognized by the nuclear Bcl-xL complex.

    Minor: While the main future direction pointed out by the manuscript was made in the last sentence of the Discussion, it could be spelled out in more detail to enforce the manuscript's impact.

    __Response: __We appreciate this suggestion and expanded the discussion in the revised manuscript to enforce the impact of this work.

    Reviewer #2 (Significance (Required)):

    The authors describe nuclear targets and functions of the anti-apoptotic protein TF Bcl-xL, which has long been of research interest to this group. Specifically, this manuscript follows up on Choi 2016 which established that nuclear localization seemed to be critical for promotion of metastatic/invasion properties of Bcl-xL independent of its anti-apoptotic function. Due to the membrane localization in cells, it was unclear how Bcl-xL entered the nucleus, simulating the current paper. Here the authors (i) demonstrate this nuclear localization happens without mutation to the protein, (ii) localization is promoted by binding to CtBP2 in co-precipitations, (iii) enforced loss of CtBP2 expression correlated with lower metastasis, (iii) specific domains within the two proteins are necessary for physical interaction and function (iii) the histone methyltransferase MLL is critical for downstream transcriptomic impacts which include upregulation of the TGFbeta pathway. Description of this pathway and the specific protein domains necessary may lead to therapeutic targets to repress metastatic capacity. This reviewer is an expert as a cancer biologist and epidemiologist.

    __Reviewer #3 (Evidence, reproducibility and clarity (Required)): __ Summary Zhang et al. investigated new roles of Bcl-xL and CtBP2 in cancer progression. They previously reported that Bcl-xL is nuclear localized and promotes cancer metastasis by inducing global histone H3 trimethyl Lys4 (H3K4me3) independent of its anti-apoptotic activity. In this study, they found that CtBP2 is a key factor for promoting the nuclear translocation of Bcl-xL. Furthermore, they showed that the binding between Bcl-xL and CtBP2 is required for MLL1 activation. MLL1 mediates the Bcl-xL-induced H3K4me3 activation and upregulation of TGFβ mRNA level. By global analysis of histone H3K4me3, the authors demonstrated that H3K4me3 modifications are enriched in the promoter regions of genes encoding TGFβ and related signaling pathways in cancer cells overexpressing Bcl-xL. Therefore, they concluded that Bcl-xL exerts its metastatic function by interacting with CtBP2 and MLL1. The mechanism for histone modification by Bcl-xL is interesting and this study expanded our current understanding of epigenetic regulation in cancer. However, the mechanism for MLL1 activation induced by Bcl-xL is not fully demonstrated.

    Major points

    1. Figure 1) The number of primary breast cancer and lymph node specimens is too small. The authors analyzed only two cases of primer breast cancer and one case of lymph node metastasis. They should also present the result of normal breast tissues to show increased nuclear enrichment during disease progression. In addition, quantification of nuclear signals and statistical analysis are necessary. More importantly, the expression of CtBP2 and MLL1 should be evaluated in these clinical samples because they claimed that the interaction of Bcl-xL/CtBP2/MLL1 is important for tumor metastasis in this study.

    __Response: __We appreciate this suggestion to increase the number of the clinical samples. We have stained breast cancer TMAs and included normal breast tissues to show increased nuclear enrichment during disease progression (Table 1). We have included the following information in the revised manuscript. Although we would also like to co-stain these breast cancer TMAs with CtBP2 and MLL1, there are no suitable antibodies for co-staining these two proteins with Bcl-xL in these FFPE sections.

    “We further evaluated breast cancer specimens in tissue microarrays (TMAs) for the expression and localization of Bcl-xL by immunohistochemistry. Compared to normal breast tissues, the intensity of Bcl-xL was significantly higher in breast cancer, including primary tumors, lymph node (LN) metastases, and distant metastases (Table 1a and 1b). Perinuclear/nuclear Bcl-xL is significantly increased in human breast cancer tissues compared to normal breast tissues (Table 1c and Figure 1d). The proportion of peri-nuclear and nuclear Bcl-xL positive cases showed an increasing trend towards metastasis (Table 1d).”

    1. (Figure 2c) In this experiment, the expression of Bcl-xL is mainly observed in the cytoplasm even in the condition of shControl. Therefore, I think that the nuclear localization of Bcl-xL is not convincingly regulated by CtBP2 expression change. Overexpression of CtBP2 is also necessary to show CtBP2-dependent nuclear localization of Bcl-xL.

    __Response: __We appreciate this suggestion to overexpress CtBP2. We have performed this experiment by transiently transfecting cells with CtBP2 and found that overexpression of CtBP2 increased the nuclear to cytosol ratio of Bcl-xL (new Figure 2b and 2c) and included the following information in the revised manuscript.

    “To determine the role of CtBP2 in mediating Bcl-xL’s nuclear translocation, we employed overexpression and knockdown of CtBP2 approaches. To overexpress CtBP2, we transfected a V5-tagged CtBP2 construct (Paliwal et al., 2006) into 293T cells and performed immunofluorescent staining using anti-V5 and anti-Bcl-xL antibodies. We observed an increased nuclear-to-cytosol ratio of endogenous Bcl-xL in cells overexpressing CtBP2-V5 (Figure 2b and 2c).”

    1. (Figure 6d-e) These results are important because the anti-apoptotic activity is not inhibited even if the interaction between CtBP2 and Bcl-xL is lost. I wonder whether the authors analyzed the cellular localization of each mutant protein (particularly, wt, construct #5 and #6) in the presence of CtBP2. In addition, the authors should examine how the histone K4me3 and MLL1 activity is affected by overexpressing construct #5 and #6 to elucidate the metastatic ability by these constructs (Figure 6e). The authors should describe whether wt Bcl-xL is constract #2 or not in the legends.

    __Response: __We appreciate that the reviewer pointed out the importance of our finding that even if the interaction between CtBP2 and Bcl-xL is lost, the anti-apoptotic activity of Bcl-xL is not inhibited. As suggested by the reviewer, we described wt Bcl-xL as construct #2 in the manuscript, and we analyzed the subcellular localization of wt HA-Bcl-xL (construct #2, which binds to CtBP2), construct #5 (which binds to CtBP2), and construct #6 (which does not bind to CtBP2), in the presence of endogenous CtBP2 in N134 mouse PNET cells. We found that the nuclear to cytosol ratio of wt HA-Bcl-xL (construct #2) and construct #5 was similar to each other, and we observed a reduction in the nuclear-to-cytosol ratio of construct #6 (Figure 6f and 6g). This is in consistent of the reduction of the metastatic ability of construct #6.

    Further, we examined H3K4me3 and MLL1 in these cells and found that H3K4me3 was reduced in construct #6 compared to wt HA-Bcl-xL (construct #2) and construct #5 (Figure 6c). We also found that H3K4me3 levels were reduced in the CtBP2 knockout cells (Supplementary Figure S5b).

    Minor points

    1. (Figure 2d) Labels for these graphs are lacking.

    __Response: __We apologize for the missing labels when we merged the graphs. We have added them back (new Figure 2f).

    1. (Figure 2e, f) The authors should label in these graphs whether these results are statistically significant or not.

    __Response: __Thanks for the suggestion. We have labeled * for statistically significant (P 6) (Figure 3c) No labels for these blots.

    __Response: __We apologize for the missing labels when we merged the graphs. We have added them back.

    1. (Figure 3b) They should describe the full spell of n/a in the legends.

    __Response: __Thanks for the suggestion. We have described “n/a: non-sorted parental cells” in the legends in the revision.

    1. (Figure 4f) The label of Y-axis should be corrected.

    __Response: __Thanks for the suggestion. We have corrected the label of Y-axis.

    1. (Figure 8c) The location of gene transcriptional start site and ChIP signal level should be shown. In addition, the genome browser view including whole gene body by zooming out should be shown.

    __Response: __In addition to the differential peaks around promoters of six genes in Fig. 8, we have included the whole gene body with the location of the gene transcriptional start site in Supplementary Figure S7.

    Reviewer #3 (Significance (Required)):

    It is interesting that Bcl-xL can be transported to the nucleus and modulate the entire epigenetic condition for promoting metastatic ability. In the previous study, this group highlighted the nuclear function of Bcl-xL in cancer cells. This concept, Bcl-xL functions independent of its anti-apoptotic activity (Choi et al. Nat Commun 2016;7:10384.), is highly original and will bring some impacts on cancer research. In this study, the authors revealed molecular mechanisms to elucidate this nuclear translocation of Bcl-xL and how Bcl-xL regulate the epigenetic condition. However, the authors should present more evidences to demonstrate the mechanism that CtBP2/Bcl-xL interaction with MLL1 regulate global K4me3 levels in the nucleus to promote metastasis.

    1. First of all, there are insufficient data to demonstrate how the interaction with Bcl-xL is involved in MLL1 activation. In Figure 7e, the authors analyzed H3K4me3 level by only inhibiting MLL1 expression and activity. However, the authors should investigate whether Bcl-xL and CtBP2 knockdown or overexpression modulate MLL1-mediated histone H3K4me3 regulation.

    Response: We appreciate that Reviewer #3 considered our work to be highly original. As suggested, we investigated whether CtBP2 knockout affected H3K4me3 levels and found that H3K4me3 levels were reduced in the CtBP2 knockout cells (Supplementary Figure S5b). Conversely, we have reported that Bcl-xL overexpression increases H3K4me3 levels (Choi et al., 2016). The main take-home message of this study is the discovery of the nuclear translocation mechanism of Bcl-xL through a novel interaction with CtBP2. We have shown that Bcl-xL or CtBP2 binds to MLL1 only when Bcl-xL and CtB2 bind to each other (Figure 5b, 5c, and 6b).

    1. (Figure 8) The authors should explain why MLL1 activation specifically affect the K4me3 levels of TGFβ signal-associated genes. I wonder whether Bcl-xL/MLL1/CtBP2 functions as cofactors by binding to certain transcription factors. In addition, Bcl-xL, CtBP2 and MLL1 ChIP-seq/CUT & RUN analysis would be preferable.

    __Response: __We have tried but have not been able to successfully establish the CUT&RUN conditions using Bcl-xL, CtBP2, and MLL1 antibodies. Whether Bcl-xL/MLL1/CtBP2 functions as cofactors by binding to certain transcription factors is a very interesting question. Additional studies are required to identify the other components of this Bcl-xL/CtBP2/MLL1 protein complex, which is beyond the scope of this work. This is added in the Discussion of the revised manuscript.

  6. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #3

    Evidence, reproducibility and clarity

    Summary

    Zhang et al. investigated new roles of Bcl-xL and CtBP2 in cancer progression. They previously reported that Bcl-xL is nuclear localized and promotes cancer metastasis by inducing global histone H3 trimethyl Lys4 (H3K4me3) independent of its anti-apoptotic activity. In this study, they found that CtBP2 is a key factor for promoting the nuclear translocation of Bcl-xL. Furthermore, they showed that the binding between Bcl-xL and CtBP2 is required for MLL1 activation. MLL1 mediates the Bcl-xL-induced H3K4me3 activation and upregulation of TGFβ mRNA level. By global analysis of histone H3K4me3, the authors demonstrated that H3K4me3 modifications are enriched in the promoter regions of genes encoding TGFβ and related signaling pathways in cancer cells overexpressing Bcl-xL. Therefore, they concluded that Bcl-xL exerts its metastatic function by interacting with CtBP2 and MLL1. The mechanism for histone modification by Bcl-xL is interesting and this study expanded our current understanding of epigenetic regulation in cancer. However, the mechanism for MLL1 activation induced by Bcl-xL is not fully demonstrated.

    Major points

    1. Figure 1) The number of primary breast cancer and lymph node specimens is too small. The authors analyzed only two cases of primer breast cancer and one case of lymph node metastasis. They should also present the result of normal breast tissues to show increased nuclear enrichment during disease progression. In addition, quantification of nuclear signals and statistical analysis are necessary. More importantly, the expression of CtBP2 and MLL1 should be evaluated in these clinical samples because they claimed that the interaction of Bcl-xL/CtBP2/MLL1 is important for tumor metastasis in this study.
    2. (Figure 2c) In this experiment, the expression of Bcl-xL is mainly observed in the cytoplasm even in the condition of shControl. Therefore, I think that the nuclear localization of Bcl-xL is not convincingly regulated by CtBP2 expression change. Overexpression of CtBP2 is also necessary to show CtBP2-dependent nuclear localization of Bcl-xL.
    3. (Figure 6d-e) These results are important because the anti-apoptotic activity is not inhibited even if the interaction between CtBP2 and Bcl-xL is lost. I wonder whether the authors analyzed the cellular localization of each mutant protein (particularly, wt, construct #5 and #6) in the presence of CtBP2. In addition, the authors should examine how the histone K4me3 and MLL1 activity is affected by overexpressing construct #5 and #6 to elucidate the metastatic ability by these constructs (Figure 6e). The authors should describe whether wt Bcl-xL is constract #2 or not in the legends.

    Minor points

    1. (Figure 2d) Labels for these graphs are lacking.
    2. (Figure 2e, f) The authors should label in these graphs whether these results are statistically significant or not.
    3. (Figure 3c) No labels for these blots.
    4. (Figure 3b) They should describe the full spell of n/a in the legends.
    5. (Figure 4f) The label of Y-axis should be corrected.
    6. (Figure 8c) The location of gene transcriptional start site and ChIP signal level should be shown. In addition, the genome browser view including whole gene body by zooming out should be shown.

    Significance

    It is interesting that Bcl-xL can be transported to the nucleus and modulate the entire epigenetic condition for promoting metastatic ability. In the previous study, this group highlighted the nuclear function of Bcl-xL in cancer cells. This concept, Bcl-xL functions independent of its anti-apoptotic activity (Choi et al. Nat Commun 2016;7:10384.), is highly original and will bring some impacts on cancer research. In this study, the authors revealed molecular mechanisms to elucidate this nuclear translocation of Bcl-xL and how Bcl-xL regulate the epigenetic condition. However, the authors should present more evidences to demonstrate the mechanism that CtBP2/Bcl-xL interaction with MLL1 regulate global K4me3 levels in the nucleus to promote metastasis.

    1. First of all, there are insufficient data to demonstrate how the interaction with Bcl-xL is involved in MLL1 activation. In Figure 7e, the authors analyzed H3K4me3 level by only inhibiting MLL1 expression and activity. However, the authors should investigate whether Bcl-xL and CtBP2 knockdown or overexpression modulate MLL1-mediated histone H3K4me3 regulation.
    2. (Figure 8) The authors should explain why MLL1 activation specifically affect the K4me3 levels of TGFβ signal-associated genes. I wonder whether Bcl-xL/MLL1/CtBP2 functions as cofactors by binding to certain transcription factors. In addition, Bcl-xL, CtBP2 and MLL1 ChIP-seq/CUT & RUN analysis would be preferable.
  7. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #2

    Evidence, reproducibility and clarity

    The manuscript describes a large number of experiments each of which describes a small part of the functional cascade of Bcl-xL in nuclear function and metastatic tumor behavior. No one experiment accomplishes a lot, but taken as a total, the story is compelling and fairly complete.

    Major:

    Figure 1 shows Bcl-xL in one primary sample (a) but clearly not in a second one (c). The authors state 3 of 15. Can they make any comment about breast cancer subtype of these 3 or outcomes? This seems fairly thin evidence of Bcl-xL involvement in human tumorigenesis in general - a better survey might be performed with tissue microarrays of more than one cancer subtype. I'm not sure that this figure is compelling or necessary really for the rest of the manuscript. Really, the main weakness of this paper is some proof that this Bcl-xL-mediated pathway is significant in some proportion of human cancer and metastasis. Perhaps some RNASeq datasets on metastatic versus localized cancers could be mined to establish this relvance?

    Most other experiments and figures are well explained. The only one I have some trouble with is Figure 8 CUT and RUN data where we are only presented with peaks around six genes. Is there a way to summarize data for the rest of the genome? Or to display a composite of CUT and RUN data on promoters that are not predicted to be targets of Bcl-xL and MLL1 activity (compared to those that are)?

    Minor:

    While the main future direction pointed out by the manuscript was made in the last sentence of the Discussion, it could be spelled out in more detail to enforce the manuscript's impact.

    Significance

    The authors describe nuclear targets and functions of the anti-apoptotic protein TF Bcl-xL, which has long been of research interest to this group. Specifically, this manuscript follows up on Choi 2016 which established that nuclear localization seemed to be critical for promotion of metastatic/invasion properties of Bcl-xL independent of its anti-apoptotic function. Due to the membrane localization in cells, it was unclear how Bcl-xL entered the nucleus, simulating the current paper. Here the authors (i) demonstrate this nuclear localization happens without mutation to the protein, (ii) localization is promoted by binding to CtBP2 in co-precipitations, (iii) enforced loss of CtBP2 expression correlated with lower metastasis, (iii) specific domains within the two proteins are necessary for physical interaction and function (iii) the histone methyltransferase MLL is critical for downstream transcriptomic impacts which include upregulation of the TGFbeta pathway. Description of this pathway and the specific protein domains necessary may lead to therapeutic targets to repress metastatic capacity. This reviewer is an expert as a cancer biologist and epidemiologist.

  8. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #1

    Evidence, reproducibility and clarity

    Summary

    In this study the authors build on their previous work that Bcl-xL has a role in metastasis promotion independent of it's function in the mitochondrial apoptotic pathway. They show that Bcl-xL can be found in the nucleus of some human breast cancer cells and through a mass spec approach show that CtBP2 promotes the nuclear translocation of Bcl-xL. Using various knockdown/knockout methods they show that reduced levels of CtBP2 reduces metastasis, because of loss of Bcl-xL translocation to the nucleus. The authors map this interaction and show that this interaction modulates metastasis.

    Major comments

    • Figure 1 - a more comprehsive analysis of nuclear Bcl-xL should be conducted. The data presented only shows 3 different samples, with no quantification. Perhaps the authors could stain a breast cancer TMA or simiilar?
    • Figure 2 - could the authors show the a graph with a representation of the mass spectrometry data, so the reader can get a sense of how many proteins were found to be associated with Bcl-xL?
    • Have the authors tried any other ways to verify the interaction between Bcl-xL and CtBP2? For instance, do they co-localise when imaged? Also, can the reverse IP be performed?
    • Figure 2C - the authors claim that this data shows that Bcl-xL nuclear translocation is reduced in cells with reduced levels of CtBP2 - however, although they quantify this I simply do not see it from the images presented. I do not think this data supports the conclusion that knockdown of CtBP2 reduces Bcl-xL translocation to the nucleus.Furthermore, this data is only shown with overexpressed Bcl-xL - have the authors tried with endogenous staining of Bcl-xL?
    • Figure 2e-f - again these data are in cells with overexpressed Bcl-xL - does the same effect on invasion happen when only CtBP2 levels are reduced, without overexpression of Bcl-xL? What happens when Bcl-xL is knocked down? Also, doxycycline has been shown to affect mitochondrial function, which might confound this data - perhaps another way to knockdown CtBP2 (e.g. CRISPR which is used later in the study) would rule this out
    • Figure 3c - these blots are not labelled, but ideally this would be shown with endogenous Bcl-xL, rather that just the overexpressed HA-Bcl-xL. However these data are more convincing than the images presented in Figure 2c
    • Figure 4 - the authors use CRISPR to knockout CtBP2 - logically this data would go with the shRNA data shown before, as it seems to just repeat what has already been shown?
    • Figure 4d - what does "SN" refer to? There is no loading control for this part of the fractionation - I assume this is supernatant? If so, why is there no loading control for this (same applies to figure 3c). Also, why are these not on the same blot? If CtBP2 knockdown reduces Bcl-xL mRNA level, does it also reduce Bcl-xL protein levels? We should be able to tell this from the blots in figure 4d, but since they are on different membranes this is impossible to deduce
    • Figure 5c - molecular weight markers should be included here
    • Figure 7a - the text says that MM102 treatment "significantly reduced" H3K4me3 levels - where is the quantification of this?

    Minor comments

    • Some of the figures are not properly labelled
    • Some of the data are presented in an awkward manner - the authors should consider re-structuring either the manuscript or the figures so there is less "jumping around"

    Significance

    General assessment

    • Provides new insight into non-canonical roles of Bcl-xL in cancer
    • Relies heavily on over-expressed proteins to draw conclusions
    • If the data were stronger and supported the conclusions, this study could be of interest to a broad cancer audience

    My expertise

    Cell biology, cell death, cancer, imaging

  9. Version published to 10.1101/2023.04.26.538373v1 on bioRxiv