Haploinsufficiency Interactions between RALBP1 and p53 in ERBB2 and PyVT Models of Mouse Mammary Carcinogenesis

Round 1

Reviewer 1 Report

Dear Singh et al,

in your current work, you investigate the effect of genetic Rlip deficiency in tumor growth and metastasis in 2 in two transgenic breast cancer mouse models (MMTV-PyVT and MMTV-Erbb2).

The paper is nicely written however research design shows some flaws, presentation could be improved and discussion/conclusion is highly speculative.

I think the observed difference between the 2 GEMMS is very interesting, but requires further investigation.

I have indicated additional experiments and suggestions here:

Line 273: Please include tumor growth data either as a kinetics or as average TV at endpoint.

From M&M, I understand endpoint tumors  were only flash frozen, however it is best practice to half freeze and half fix tumors. It would be ideal to see both macroscopic tumor appearance, and HE staining. Histology of tumors would also allow to validate molecular findings in 3.3.

 

Line 307: Please include in vivo Rlip depletion by antisense in the MMTV-Errb2 model as well.

 

Figure 4: Missing number of mice per group, please add.

 

Line 325: Please provide the following data as well:

- average size mets

- ratio micro/macro mets

- survival analysis after primary tumor resection

 

Figure 6: Some of these blots are low quality and hard to interpret (BAX, BCl-2). Please optimize conditions and rerun lysates. Also, as mentioned above, it would be nice to validate some of this finding histologically.

 

Figure 7: Please merge Fig 6 and 7 in a panel figure showing blots and their correspondent densitometry on the side. Also please add p-values for statistical significance for protein expression in Rlip +/- vs -/- comparison for both models.

 

Figure 8: Please add p-values for statistical significance for mRNA expression in Rlip +/- vs -/- comparison for both models.

 

Line 403: Please explain why you performed computational analysis. The evidence for Rlip/P53 complex is speculative. Validation of these findings is missing.

 

Line 447: The references listed here are on preclinical work, please rectify.

 

Line 477: This is a major flaw of this work: there is no mechanistic insight, nor even an experiment aimed at investigating why this is.

 

Line 496: This hypothesis could be tested in Erbb2:Rlip+/+ mice, testing the efficacy of the combination of Rlip antisense and trastuzumab.

 

Line 512: Crystal structure or ChiP sequencing would be required to identify this interaction, in addition to computational modelling.

 

Line 554: Gene and protein expression profile are little of not interest if findings are not further pursued.

Author Response

Response to Reviewers:

We thank the Editor for giving us the opportunity to revise the manuscript and the Reviewers for their constructive comments. Based on the comments of the Reviewers, we have revised the manuscript with additional data as requested. Pointwise response to the Reviewers’ comments is offered as follows:

 

 

Reviewer #1:

in your current work, you investigate the effect of genetic Rlip deficiency in tumor growth and metastasis in 2 in two transgenic breast cancer mouse models (MMTV-PyVT and MMTV-Erbb2). The paper is nicely written however research design shows some flaws, presentation could be improved and discussion/conclusion is highly speculative. I think the observed difference between the 2 GEMMS is very interesting, but requires further investigation.

We thank the reviewer for recognizing the breadth and importance of our studies. Pointwise response is offered as follows:

Line 273: Please include tumor growth data either as a kinetics or as average TV at endpoint.

We have plotted the tumor volumes at endpoint for both models.  This data can now be found in Figure S4 in the supplemental data.

From M&M, I understand endpoint tumors were only flash frozen, however it is best practice to half freeze and half fix tumors. It would be ideal to see both macroscopic tumor appearance, and HE staining. Histology of tumors would also allow to validate molecular findings in 3.3.

H&E staining was performed on the lungs of the Rlip/PyVT genotypes in order to evaluate lung metastasis.   H&E staining was not performed on mammary tumor tissues obtained from mice. This is stated in M&M. 

 Line 307: Please include in vivo Rlip depletion by antisense in the MMTV-Errb2 model as well.

We did not observe a prevention of cancer or a change in survival proportion of PyVT:Rlip^+/- and PyVT:Rlip^-/- mice compared with PyVT:Rlip^+/+ mice, therefore we have tested whether or not Rlip-LNA treatment regresses tumor or increases survival proportion. On the other hand, Rlip depletion in MMTV-Erbb2 mice it is quite evident that genetic manipulation extends survival significantly, therefore Rlip is treatment experiment was not necessary in Erbb2 mice.

Figure 4: Missing number of mice per group, please add.

This has been corrected. 

 Line 325: Please provide the following data as well:

- average size mets

- ratio micro/macro mets

- survival analysis after primary tumor resection

We have tested whether or not Rlip depletion prevents lung metastasis in the MMTV-PyVT model. For this experiment we have not performed primary tumor resection as a survival surgery in order to evaluate metastases at a later time. Rather, we have collected lung specimens at the time of euthanasia due to primary tumor endpoints and fixed the lungs for H&E staining for counting at the time of euthanasia. Thus, the survival curves corresponding to these lung metastasis results are the curves shown in Figs 2 and 3. The pathologist did not measure the size or ratio of micro/macro mets.

 Figure 6: Some of these blots are low quality and hard to interpret (BAX, BCl-2). Please optimize conditions and rerun lysates. Also, as mentioned above, it would be nice to validate some of this finding histologically.

We agree with the reviewer. We have repeated the western blots several times and also tried different antibodies to address this issue, but none produced a better image than the one reported reported. If reviewer is not convinced, we can take out this data.

 Figure 7: Please merge Fig 6 and 7 in a panel figure showing blots and their correspondent densitometry on the side. Also please add p-values for statistical significance for protein expression in Rlip +/- vs -/- comparison for both models.

We have added statistical data to the figure as requested. We agree that merging the two figures would enhance the paper’s clarity for readers.  We have merged Figs 6 and 7 and renumbered accordingly.

 Figure 8: Please add p-values for statistical significance for mRNA expression in Rlip +/- vs -/- comparison for both models.

We have updated the figure as requested.

 Line 403: Please explain why you performed computational analysis. The evidence for Rlip/P53 complex is speculative. Validation of these findings is missing.

We have previously reported that Rlip binds to p53 Singhal et al. Targeting p53 Null Neuroblastomas through RLIP76 (Cancer Prev Res; 2011). The figures use predictions based on computer modeling.  Computer modeling can provide us possible interaction model of the protein-protein complex at atomic level, especially when there is no experimental X-ray or NMR structures. We can use the computer model as a guide for our future experimental validations, such as mutagenesis. We have used the same ZDOCK procedures to support/explain several protein interactions studies.  (See Awasthi, et al. Anticancer activity of 2′-hydroxyflavanone towards lung cancer, Oncotarget 2018, and Nagaprashantha, et al.  2'-Hydroxyflavanone effectively targets RLIP76-mediated drug, Oncotarget 2018.). In the results have added additional rationale and changed the wording of the section’s figure legends to more clearly reflect the purpose and predicted nature of the models. 

Line 447: The references listed here are on preclinical work, please rectify.

  We apologize for the confusion, but thank the reviewer for paying close attention to language clarity.  Rlip-targeted therapies have never been used in human patients.  By clinical applications we meant to refer to preclinical reports studying potential clinical uses of Rlip-targeted therapies (i.e. treating melanoma).  We have restated this as “potential clinical applications”.  No changes have been made to the citations.

 Line 477: This is a major flaw of this work: there is no mechanistic insight, nor even an experiment aimed at investigating why this is.

Mechanistic studies of Rlip inhibition have been published for other cancers including breast cancer, but this paper is novel in that it is the first such analysis pertaining to breast cancer that shows a clear difference in responsiveness to Rlip-targeted interventions between the two models used.  This is intriguing and we agree that further experiments to explain this difference in response would be valuable.  We plan to do these experiments in the future, and will also continue to explore the utility of the strategy in other models of HER2+ breast cancer.

 

 Line 496: This hypothesis could be tested in Erbb2:Rlip+/+ mice, testing the efficacy of the combination of Rlip antisense and trastuzumab.

We thank the reviewer for the suggestion, but this would be beyond scope of this manuscript.

 Line 512: Crystal structure or ChiP sequencing would be required to identify this interaction, in addition to computational modelling.

 Correct.  As stated above, the Rlip-p53 interaction has been established, and we our intent was to state a hypothesis for future study.  We have added explanation and changed the wording in the Results to make it more clear to readers that this model is based on predictive computation.

 Line 554: Gene and protein expression profile are little of not interest if findings are not further pursued.

These are exciting new findings on how Rlip differentially affects cancers from two diverse origins. We are further pursuing the findings from this study.

 

Author Response File: Author Response.docx

Reviewer 2 Report

Remarks to the Author: The authors sought to identify the mechanism of haploinsufficiency interactions between RALBP1 and p53 in ERBB2 and PyVT models of mouse mammary carcinogenesis. To prove this, they used well-designed two specific mouse models including Rlip-MMTV-Erbb2 and Rlip-MMTV-PyVT GEM mice. By mouse survival analysis, they clearly show that median survival times across three mouse conditions (Erbb2:Rlip+/+; Erbb2:Rlip-/-; Erbb2:Rlip+/-) are very different from each other and this result highly supports their hypothesis. Furthermore, data from protein expression, gene expression analysis and computational model of Rlip-TP53 complex also support their ideas. However, I have the following concerns.   1. statistical evidence From Figure 2, 3 and 4, there are no descriptions about the detailed experiments and statistical tests. For example, in the case of Figure 2, how many samples did you test in every three conditions? What are the statistical p-values between the three groups? (i: Erbb2:Rlip+/+ vs Erbb2:Rlip+/-, ii)  Erbb2:Rlip+/+ vs Erbb2:Rlip-/-, iii)  Erbb2:Rlip+/- vs Erbb2:Rlip-/-).   2. Computational modeling of Rlip-p53 complex 1: Even though models of Rlip interacting with p53 present strong evidence of Rlip-TP53 interactions and their possible interaction interfaces, there is no description or evidence of the physical interactions between Rlip-TP53 in this text. Also, computational modeling with ZDOC is a computational simulation and just conjugates them together to know the possibility of a protein-protein interaction interface. Therefore, it can't provide direct evidence of physical interactions between two proteins. What about searching the possibility of protein-protein interactions through large-scale protein-protein interactions databases (STRING, HPRD or other DBs) or computational predictions or reference searches?   2: description of the computational modeling of Rlip-p53 complex In the method section, there is no description of the computational modeling of the Rlip-p53 complex. More detailed descriptions about the simulation (how many times) and how to find the best model from simulation should be described in the text.

Author Response

Response to Reviewers:

We thank the Editor for giving us the opportunity to revise the manuscript and the Reviewers for their constructive comments. Based on the comments of the Reviewers, we have revised the manuscript with additional data as requested. Pointwise response to the Reviewers’ comments is offered as follows:

 

 

Reviewer#2:

Remarks to the Author: The authors sought to identify the mechanism of haploinsufficiency interactions between RALBP1 and p53 in ERBB2 and PyVT models of mouse mammary carcinogenesis. To prove this, they used well-designed two specific mouse models including Rlip-MMTV-Erbb2 and Rlip-MMTV-PyVT GEM mice. By mouse survival analysis, they clearly show that median survival times across three mouse conditions (Erbb2:Rlip+/+; Erbb2:Rlip-/-; Erbb2:Rlip+/-) are very different from each other and this result highly supports their hypothesis. Furthermore, data from protein expression, gene expression analysis and computational model of Rlip-TP53 complex also support their ideas.

 

We thank the reviewer for characterizing the results presented as “highly supports their hypothesis”.  We believe that the extensive studies performed and the multiple lines of evidence presented are strong evidence of the rigorous validation of our findings.   

 

1. statistical evidence From Figure 2, 3 and 4, there are no descriptions about the detailed experiments and statistical tests. For example, in the case of Figure 2, how many samples did you test in every three conditions? What are the statistical p-values between the three groups? (i: Erbb2:Rlip+/+ vs Erbb2:Rlip+/-, ii)  Erbb2:Rlip+/+ vs Erbb2:Rlip-/-, iii)  Erbb2:Rlip+/- vs Erbb2:Rlip-/-).  

We have updated the figures as suggested.

 

2. Computational modeling of Rlip-p53 complex 1: Even though models of Rlip interacting with p53 present strong evidence of Rlip-TP53 interactions and their possible interaction interfaces, there is no description or evidence of the physical interactions between Rlip-TP53 in this text. Also, computational modeling with ZDOC is a computational simulation and just conjugates them together to know the possibility of a protein-protein interaction interface. Therefore, it can't provide direct evidence of physical interactions between two proteins. What about searching the possibility of protein-protein interactions through large-scale protein-protein interactions databases (STRING, HPRD or other DBs) or computational predictions or reference searches?   2: description of the computational modeling of Rlip-p53 complex In the method section, there is no description of the computational modeling of the Rlip-p53 complex. More detailed descriptions about the simulation (how many times) and how to find the best model from simulation should be described in the text.

We have previously reported that Rlip binds to p53 (Singhal et al. Targeting p53 Null Neuroblastomas through RLIP76. Cancer Prev Res (2011)).  Per the reviewers’ helpful comments, we have further clarified several experimental details and have modified the results to clearly point out potential limitations of the conclusions.

Author Response File: Author Response.docx

Reviewer 3 Report

The authors sought to address whether Rlip serves as a haploinsufficient  oncogenesis factor in mouse models of breast cancer. The main spontaneous mouse models were Erbb2 overexpression or PyVT expression, both driven by the MMTV promoter. The authors claim their data support a selective modulation by Rlip of metastatic ability in the PyVT model while the Erbb2 models the effect was pronounced for primary tumors as well as metastatic rates. Correlative studies were performed with apoptotic and cell cycle markers as well as angiogenic markers. The authors speculate that Ras or p53 may be reasons for the difference in the models used.

 

For significance, this study is timely and performed in a way that allows for critical evaluation of the questions presented. Haploinsufficiency is far more common in genuine human tumor genetics than complete gene knockouts, yet studies on the subject remain rare. The spontaneous mouse models used are well-validated in the literature and the knockout transgenics used to cross in Rlip complete knockout or haploinsufficiency are clearly among the best models to address the authors’ hypotheses. The authors have previously performed MCF7 and MDA-MB-231 xenograft studies of Rlip depletion in Cancers, leaving this spontaneous mouse model as a clear next step.

 

For scientific rigor, the authors chose to utilize a Cre-LoxP model to remove exon4. The authors backcrossed the mice for 7 generations prior to the presented experiments. To assess genetic states after tumor formation, the authors used previously validated shRNA targeting Rlip. Experiments were adequately powered for moderate effect size.

 

The authors stick to the data with their language in the paper and offer only a brief discussion, as perhaps appropriate for this straight-forward study addressing a very specific question. All conclusions were adequately supported by their presented data, with the singular exception that it should be made clear whether the studies indicate solely genetic-phenotype interaction (as per only measuring DNA content of Rlip) or if the conclusions can be related to RNA or protein levels of Rlip.

 

Specific comments:

-Please provide details on how Cre was used to generate your founder mice. It is unclear how penetrant this deletion is throughout the tissues of the mouse. You state “we found that exon 4 (ENSMUSE00000304776) is conditionally removed”, how precisely do you define this - what is the exact evidence of conditional removal? To be clear, supplemental S1 does not explain the methods used for Cre.

-Without the understanding of the above knockout status, the authors need to show what evidence they have that Rlip was depleted at the mRNA or protein levels. If it was assayed in parallel, Rlip should be displayed in Fig 6, 7, and 8. While the tumors which formed could potentially re-express Rlip, these data are essential to understanding the claims presented. If the result is null, it can be added to the supplement.

-For Figure 7, the authors state “The bar 371 diagrams represent the fold change in protein levels in Rlip knockout genotypes”. Given this stated normalization, can the authors kindly explain or clarify why the -/- genotypes are not all = 1 on the y-axis?

-For any claims made regarding Fig6-8, please provide statistics in the text (eg, P < 0.05 by student’s t-test, N=3 tumors) or as marked asterisks within the figure panels

-The authors state that the shRNA targeting murine Rlip perfectly matches human Rlip. Since Fig 4 does not show differences, it is important to know if Rlip was knocked down or not. Please show direct evidence that the shRNA targeting murine Rlip knocks down Rlip, and to what extent. In vitro would be adequate, but if RNA is available from the tumors tested, that would be more convincing.

-Can the authors please explain the expected cleavage size of PARP1 (and uncleaved PARP1) and if that matches the Erbb2 panel in S13.

-Please specify number of mice per group in each figure panel or panel legend (not ≥10, for example, but specific N = 9 for group (a), N= 9 for group (b). This is currently unclear.

-Please explain if any blinding was performed for counting the number of pulmonary metastases. If no blinding was used, was there a size cutoff?

Author Response

Response to Reviewers:

We thank the Editor for giving us the opportunity to revise the manuscript and the Reviewers for their constructive comments. Based on the comments of the Reviewers, we have revised the manuscript with additional data as requested. Pointwise response to the Reviewers’ comments is offered as follows:

 

Reviewer#3:
The authors sought to address whether Rlip serves as a haploinsufficient oncogenesis factor in mouse models of breast cancer. The main spontaneous mouse models were Erbb2 overexpression or PyVT expression, both driven by the MMTV promoter. The authors claim their data support a selective modulation by Rlip of metastatic ability in the PyVT model while the Erbb2 models the effect was pronounced for primary tumors as well as metastatic rates. Correlative studies were performed with apoptotic and cell cycle markers as well as angiogenic markers. The authors speculate that Ras or p53 may be reasons for the difference in the models used. For significance, this study is timely and performed in a way that allows for critical evaluation of the questions presented. Haploinsufficiency is far more common in genuine human tumor genetics than complete gene knockouts, yet studies on the subject remain rare. The spontaneous mouse models used are well-validated in the literature and the knockout transgenics used to cross in Rlip complete knockout or haploinsufficiency are clearly among the best models to address the authors’ hypotheses. The authors have previously performed MCF7 and MDA-MB-231 xenograft studies of Rlip depletion in Cancers, leaving this spontaneous mouse model as a clear next step.For scientific rigor, the authors chose to utilize a Cre-LoxP model to remove exon4. The authors backcrossed the mice for 7 generations prior to the presented experiments. To assess genetic states after tumor formation, the authors used previously validated shRNA targeting Rlip. Experiments were adequately powered for moderate effect size. The authors stick to the data with their language in the paper and offer only a brief discussion, as perhaps appropriate for this straight-forward study addressing a very specific question. All conclusions were adequately supported by their presented data, with the singular exception that it should be made clear whether the studies indicate solely genetic-phenotype interaction (as per only measuring DNA content of Rlip) or if the conclusions can be related to RNA or protein levels of Rlip.

We are very grateful for the Reviewer’s appreciation of our findings. Based on the comments of the Reviewer, we have now extensively revised and streamlined the manuscript with additional supplemental materials as requested below. Pointwise responses to comments are offered as follows:

 Specific comments:

-Please provide details on how Cre was used to generate your founder mice. It is unclear how penetrant this deletion is throughout the tissues of the mouse. You state “we found that exon 4 (ENSMUSE00000304776) is conditionally removed”, how precisely do you define this - what is the exact evidence of conditional removal? To be clear, supplemental S1 does not explain the methods used for Cre.

We understand the confusion and apologize for our misstatement.  The mice we used in these experiments were standard Rlip knockout mice (exon 4), not conditional knockouts.  We have corrected this in the manuscript.

-Without the understanding of the above knockout status, the authors need to show what evidence they have that Rlip was depleted at the mRNA or protein levels. If it was assayed in parallel, Rlip should be displayed in Fig 6, 7, and 8. While the tumors which formed could potentially re-express Rlip, these data are essential to understanding the claims presented. If the result is null, it can be added to the supplement.

As clarified above, these mice are not conditional knockouts.  Exon 4 of RALBP1 is permanently deleted in these chromosomes.

-For Figure 7, the authors state “The bar 371 diagrams represent the fold change in protein levels in Rlip knockout genotypes”. Given this stated normalization, can the authors kindly explain or clarify why the -/- genotypes are not all = 1 on the y-axis?

Figure legend states that the bar diagrams in figure 7 A&B (now figures 6B and 6C after merging) represent the fold change in protein levels in Rlip knockout genotypes relative to the corresponding Rlip wild-type genotypes. i.e. Rlip^+/- and Rlip^-/- as compared to Rlip^+/+ genotypes which were defined as = 1, as indicated by the solid black line at y=1). We have updated legend for more clarity.

-For any claims made regarding Fig6-8, please provide statistics in the text (eg, P < 0.05 by student’s t-test, N=3 tumors) or as marked asterisks within the figure panels

This information has been added in the figures. 

-The authors state that the shRNA targeting murine Rlip perfectly matches human Rlip. Since Fig 4 does not show differences, it is important to know if Rlip was knocked down or not. Please show direct evidence that the shRNA targeting murine Rlip knocks down Rlip, and to what extent. In vitro would be adequate, but if RNA is available from the tumors tested, that would be more convincing.We extensively studied and validated an siRNA and the phosphorothioated Rlip-specific antisense R508, targeted to residues 508-528 of Rlip mRNA, in several previous studies for its effects on Rlip depletion, tumor regression, apoptosis, cell cycling, and cell survival in human as well as murine cells (Bose, C., et al. Topical 2'-Hydroxyflavanone for Cutaneous Melanoma. Cancers (Basel), 2019, 11, and Singhal, et al. Regression of melanoma in a murine model by RLIP76 depletion. Cancer research 2006, 66, 2354-2360.) Previously, we also demonstrated that R508-mediated Rlip depletion caused a concentration-dependent decrease in proliferation of the breast cancer cell lines in culture. Locked nucleic-acid-modified antisense oligonucleotides (LNAs) are third generation antisense molecules with improved pharmacological properties. Therefore, we used Rlip antisense locked nucleic acid (Rlip-LNA; sequence identical to R508), with a control scrambled antisense (CAS), and have previously tested its effects on breast cancer using in vitro and in vivo models (Bose C, et.al. (2020) Rlip Depletion Suppresses Growth of Breast Cancer. Cancers (Basel) 12 (6). doi:10.3390/cancers12061446). Additionally, therefore, Rlip knockdown data were not included in this manuscript, but we already had references for this information in this paper.

-Can the authors please explain the expected cleavage size of PARP1 (and uncleaved PARP1) and if that matches the Erbb2 panel in S13.

Uncleaved PARP1 is 116 kDa and PARP cleavage has been reported to yield fragments of many sizes, including 24, 42, 62, 72, and 89 kDa. (Gobeil S, et al. Characterization of the necrotic cleavage of poly(ADP-ribose) polymerase (PARP-1): implication of lysosomal proteases. Cell Death & Differentiation volume 8, pages588–594).  The band shown in S13 (now S14 after addition of another supplemental figure) corresponds most closely to a fragment size of 62 kDa. This information has been added to the figure legend. 

-Please specify number of mice per group in each figure panel or panel legend (not ≥10, for example, but specific N = 9 for group (a), N= 9 for group (b). This is currently unclear.

This information has been added in figure legend. 

-Please explain if any blinding was performed for counting the number of pulmonary metastases. If no blinding was used, was there a size cutoff?

Number of metastatic foci in lung were counted in a blinded fashion by a pathologist. This has been added to the figure legends. 

 

 

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

I thank the  authors for addressing my comments constructively.  Please find my comments as below for your responses:

Response to Reviewers:

Reviewer #1:

in your current work, you investigate the effect of genetic Rlip deficiency in tumor growth and metastasis in 2 in two transgenic breast cancer mouse models (MMTV-PyVT and MMTV-Erbb2). The paper is nicely written however research design shows some flaws, presentation could be improved and discussion/conclusion is highly speculative. I think the observed difference between the 2 GEMMS is very interesting, but requires further investigation.

We thank the reviewer for recognizing the breadth and importance of our studies. Pointwise response is offered as follows:

Line 273: Please include tumor growth data either as a kinetics or as average TV at endpoint.

We have plotted the tumor volumes at endpoint for both models.  This data can now be found in Figure S4 in the supplemental data. R1 Thanks.

From M&M, I understand endpoint tumors were only flash frozen, however it is best practice to half freeze and half fix tumors. It would be ideal to see both macroscopic tumor appearance, and HE staining. Histology of tumors would also allow to validate molecular findings in 3.3.

H&E staining was performed on the lungs of the Rlip/PyVT genotypes in order to evaluate lung metastasis.   H&E staining was not performed on mammary tumor tissues obtained from mice. This is stated in M&M.  R1 It is stated, I am providing a suggestion for future in vivo studies.

 Line 307: Please include in vivo Rlip depletion by antisense in the MMTV-Errb2 model as well.

We did not observe a prevention of cancer or a change in survival proportion of PyVT:Rlip^+/- and PyVT:Rlip^-/- mice compared with PyVT:Rlip^+/+ mice, therefore we have tested whether or not Rlip-LNA treatment regresses tumor or increases survival proportion. On the other hand, Rlip depletion in MMTV-Erbb2 mice it is quite evident that genetic manipulation extends survival significantly, therefore Rlip is treatment experiment was not necessary in Erbb2 mice. R1 I disagree on this, precisely because you observe a quite dramatic effect, it would be worth looking into the effect of Rlip treatment in the Erbb2 model as well. Not always pharmacological inhibition mimics the genetic one, and this would be relevant in the contest of translational studies.

Figure 4: Missing number of mice per group, please add.

This has been corrected.  R1 Thanks.

 Line 325: Please provide the following data as well:

- average size mets

- ratio micro/macro mets

- survival analysis after primary tumor resection

We have tested whether or not Rlip depletion prevents lung metastasis in the MMTV-PyVT model. For this experiment we have not performed primary tumor resection as a survival surgery in order to evaluate metastases at a later time. Rather, we have collected lung specimens at the time of euthanasia due to primary tumor endpoints and fixed the lungs for H&E staining for counting at the time of euthanasia. R1 This is clear, I was genuinely curious about the results in the resection model.  Thus, the survival curves corresponding to these lung metastasis results are the curves shown in Figs 2 and 3. The pathologist did not measure the size or ratio of micro/macro mets. R1 Thanks, FYI this could be easily done via opensource image analysis software, i.e ImageJ.

 Figure 6: Some of these blots are low quality and hard to interpret (BAX, BCl-2). Please optimize conditions and rerun lysates. Also, as mentioned above, it would be nice to validate some of this finding histologically.

We agree with the reviewer. We have repeated the western blots several times and also tried different antibodies to address this issue, but none produced a better image than the one reported reported. If reviewer is not convinced, we can take out this data. R1 Thanks for disclosing this, no need to remove data.

 Figure 7: Please merge Fig 6 and 7 in a panel figure showing blots and their correspondent densitometry on the side. Also please add p-values for statistical significance for protein expression in Rlip +/- vs -/- comparison for both models.

We have added statistical data to the figure as requested. We agree that merging the two figures would enhance the paper’s clarity for readers.  We have merged Figs 6 and 7 and renumbered accordingly. R1 Thanks

 Figure 8: Please add p-values for statistical significance for mRNA expression in Rlip +/- vs -/- comparison for both models.

We have updated the figure as requested. R1 Thanks.

 Line 403: Please explain why you performed computational analysis. The evidence for Rlip/P53 complex is speculative. Validation of these findings is missing.

We have previously reported that Rlip binds to p53 Singhal et al. Targeting p53 Null Neuroblastomas through RLIP76 (Cancer Prev Res; 2011). The figures use predictions based on computer modeling.  Computer modeling can provide us possible interaction model of the protein-protein complex at atomic level, especially when there is no experimental X-ray or NMR structures. We can use the computer model as a guide for our future experimental validations, such as mutagenesis. We have used the same ZDOCK procedures to support/explain several protein interactions studies.  (See Awasthi, et al. Anticancer activity of 2′-hydroxyflavanone towards lung cancer, Oncotarget 2018, and Nagaprashantha, et al.  2'-Hydroxyflavanone effectively targets RLIP76-mediated drug, Oncotarget 2018.). In the results have added additional rationale and changed the wording of the section’s figure legends to more clearly reflect the purpose and predicted nature of the models. R1 Thanks for clarifying.

Line 447: The references listed here are on preclinical work, please rectify.

  We apologize for the confusion, but thank the reviewer for paying close attention to language clarity.  Rlip-targeted therapies have never been used in human patients.  By clinical applications we meant to refer to preclinical reports studying potential clinical uses of Rlip-targeted therapies (i.e. treating melanoma).  We have restated this as “potential clinical applications”.  No changes have been made to the citations. R1 Thanks for rectifying.

 Line 477: This is a major flaw of this work: there is no mechanistic insight, nor even an experiment aimed at investigating why this is.

Mechanistic studies of Rlip inhibition have been published for other cancers including breast cancer, but this paper is novel in that it is the first such analysis pertaining to breast cancer that shows a clear difference in responsiveness to Rlip-targeted interventions between the two models used.  This is intriguing and we agree that further experiments to explain this difference in response would be valuable.  We plan to do these experiments in the future, and will also continue to explore the utility of the strategy in other models of HER2+ breast cancer. R1 Thanks and I apologize for not being updated with whole Rlip literature.

 

 Line 496: This hypothesis could be tested in Erbb2:Rlip+/+ mice, testing the efficacy of the combination of Rlip antisense and trastuzumab.

We thank the reviewer for the suggestion, but this would be beyond scope of this manuscript. R1 Thanks.

 Line 512: Crystal structure or ChiP sequencing would be required to identify this interaction, in addition to computational modelling.

 Correct.  As stated above, the Rlip-p53 interaction has been established, and we our intent was to state a hypothesis for future study.  We have added explanation and changed the wording in the Results to make it more clear to readers that this model is based on predictive computation.  R1 Thanks.

 Line 554: Gene and protein expression profile are little of not interest if findings are not further pursued.

These are exciting new findings on how Rlip differentially affects cancers from two diverse origins. We are further pursuing the findings from this study. R1 I agree and look forward to more insights on these findings.