Low HER2 enables dedifferentiation and transformation of normal breast epithelial cells via chromatin opening

Article activity feed

1. 

   Review Commons

   Nov 17, 2022

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

   Learn more at Review Commons

   ---

   Reply to the reviewers

   We thank the reviewers for their time in evaluating of our manuscript and for the useful feedback. We are grateful that reviewers acknowledged that our study is important because it “sheds much needed light on this less documented early stage of cancer development”. The reviewers were overall positive in their assessment and, as reviewer #3 noted, our study “advances this field conceptually by highlighting the importance of targeting the cell signaling and chromatin regulation together”. The common criticism of all reviewers relates to writing style, some textual interpretation and ensuring that the number of replicates, statistical analysis, and cell …

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

   Learn more at Review Commons

   ---

   Reply to the reviewers

   We thank the reviewers for their time in evaluating of our manuscript and for the useful feedback. We are grateful that reviewers acknowledged that our study is important because it “sheds much needed light on this less documented early stage of cancer development”. The reviewers were overall positive in their assessment and, as reviewer #3 noted, our study “advances this field conceptually by highlighting the importance of targeting the cell signaling and chromatin regulation together”. The common criticism of all reviewers relates to writing style, some textual interpretation and ensuring that the number of replicates, statistical analysis, and cell culture type were appropriately mentioned. We felt these were valid points and have taken onboard all these comments. A shared concern between two of the reviewers was related to the logic behind the timepoints we chose to analyse cells in the different assays. We are confident that we have addressed this, and all other comments as detailed below.
   Please find below a point-by-point reply to the reviewer’s comments.

   Reviewer #1 (Evidence, reproducibility and clarity):

   This study aimed to identify events that happens early in malignant transformation of breast cancer (BC) cells that are driven by HER2 oncogene. Constructing a 3D inducible model to study impact of HER2 protein level on BC cell and assessment of gross morphological changes, protein phosphorylation and chromatin accessibility at different time points of HER2 activation.

   Using a controllable in vitro model is a good approach although it is not novel. Also the method used to assess HER2 protein positivity is not standardized nor clinically relevant. Positivity of HER2 in clinical practice is assessed either through immunohistochemistry (IHC 3+ or 2+ with gene amplification), however the author did not mention any control for positivity except western blot which is not used in clinical practice.

   We agree with the reviewer that we should have included our comparison of HER2 protein levels for our cells with a positive control. We have tested this, and the data will be included in the revised version of the manuscript. Briefly, both western blot (WB) and IHC are very useful methods with different benefits: WB is less cost effective but more quantitative, while IHC gives a better overview of tissue heterogeneity. Indeed, due to higher sample processing costs, WB is not used in clinical practice to assess HER2 but it has been shown that there is a high concordance (in 95% of over 300 tumours analysed) between the two methods as both techniques showed prognostic significance R. Molina et al., 1992 (PMID: 1363511). We performed comparison of HER2 protein expression levels of our subpopulations (low, medium, and high HER2 expressing cells) versus two patients’ samples that were already known to be HER2 positive using IHC 3+ or 2+. We were able to demonstrate that HER2 protein levels as measured by western blotting showed that the low HER2 expressing cells expressed less HER2 protein compared to IHC 3+ or 2+ and may be comparable to patients with IHC 1+, which are considered HER2 negative and do not qualify for anti-HER2 therapies such as Trastuzumab.

   There is difference between early HER2 positive BC and HER2 low BC. As the earlier is driven by HER2 oncogenic signalling pathway, but the latter is not.
   Identification of molecular changes that occur at HER2 low BC seems very important and clinically relevant, however HER 2 low is not fully characterized, yet. And the only definition available is either HER2 1+ or 2+ without gene amplification. The author was not very clear about threshold he followed to call the model HER2 low. Is it positive with lower limit of positivity or just small amount of protein). He also concluded that BC with sub-threshold of HER2 protein behave more aggressive than HER2 positive BC. What is the threshold and was it correlated with IHC or gene amplification level to be reliable?

   The HER2 positive population in our in vitro inducible system was determined by flow cytometry, we separated the overall (bulk) HER2 positive cells into three different subpopulations and selected the bottom 20% of HER2 expressing cells as the “low HER2” and the top 20% of HER2 expressing cells as “high HER2”. We show in figure 4C the different thresholds for low, med, high HER2 protein expression by flow cytometry. We have modified the figure and the figure legend (figure 4C) to better indicate the different subpopulations. Through western blotting we compared these population of cells with patients’ samples that had IHC 3+ or 2+ and showed that low HER2 population expressed less protein than IHC 2+, whereas the high HER2 was relatively comparable to IHC 3+ sample.

   The status of oestrogen and progesterone receptors were not highlighted. Triple negative breast cancer, for instance, is more aggressive than HER2 positive BC, this may be the reason for the worse behaviour.

   We have modified our main text in the manuscript, line 68-69, to better reflect the fact the MCF10A cells are both oestrogen (ER) and progesterone (PR) negative, this has been already characterised by Qu, Y et al., 2015 (PMID: 26147507). However, importantly, we do not think that ER and PR status is the reason these cells are relatively more aggressive, as normal MCF10A cells without HER2 expression did not display any transformative characteristics in our molecular analysis and/or in vitro functional assays, despite being ER and PR negative.

   At line 130, "The low levels of HER2 protein activation at early time point may closely mimic at least partially the signalling changes occurring in HER2 positive BC patients". This claim is not quite true, as low levels of HER2 protein activation doesn't activate HER2 oncogenic signalling pathway as HER2 positive does.

   We thank you for this insightful comment, we have modified our main text to better reflect our view (line 132-133). However, we were not sure which published data the reviewer was referring to in this case. In particular, if low HER2 levels can still form dimerisation with its family members and induce signalling via its family partners such as HER1, HER3 or HER4.

   The author aimed to study the signalling changes accompanying low levels of HER2 induction by lowering significance threshold to log2fold > 0.5. Lowering the threshold for significance will increase the total number of phosphorylated protein (both at low HER2 levels and high levels). So, studying the whole significant proteins at whole time points will not be exclusive for low HER2 levels and this was evident through activation of MAPK cascade which is one of downstream signalling pathway of HER2 positive BC.

   We agree that a log2fold change > 0.5 would increase the total number of significantly phosphorylated proteins. We first performed the analysis on a more stringent cut-off value of log2fold change > 1.5 p-value, <0.05 as shown in figure 2B. In the supplementary we also show the reduced stringency of log2fold change > 0.5, p-value <0.05, for the following reasons: when it comes to proteins, it is conceivable that a log2fold change > 0.5 is sufficient to induce molecular changes; secondly, our study investigates changes that occur just half an hour, and up to 7 hours, after HER2 protein induction. At such early time-points, proteins would be beginning to be phosphorylated and the extent of it may not be pronounced (especially in a small subset of the population); finally, we thought it is important to share this supplementary analysis with the scientific community to have access to this data so that they may further interrogate it from different perspectives.

   Combining HER2 protein level (both IHC and Western blot) to different time points will give better understanding of events associated with HER2 low, early positive or late positive.

   As above, IHC is routinely performed for clinical diagnosis because it is cost effective. Although, western blotting is laborious and expensive, it is more quantitative compared to IHC.

   Reviewer #1 (Significance):

   This work provides good evidence to changes that happen at early HER2 positive breast cancer transformation and introducing a chromatin opening and accessibility as a new target of treatment of HER2 positive breast cancer patients.

   We thank reviewer #1 for their thoughtful feedback and for their appreciation of our work.

   Reviewer #2 (Evidence, reproducibility and clarity):

   HER2 amplification is associated with poor prognosis of breast cancer. Despite it has been extensively studied, it deserves thorough study how HER2 amplification alters downstream signaling pathways, chromatin structure and gene expression, and how cells overcome the hurdles in order to transform. In this study, Hayat et al used doxycycline-induced HER2 expression in MCF10A cells to recapitulate the very early stage of HER2 expression and HER2-induced mammary epithelial cell transformation. The authors performed global phosphoproteomic, ATAC-seq and single-cell RNA-seq, and propose sub-threshold low level HER2 expression activates signaling pathways and increases chromatin accessibility required for cell transformation, while high HER2 expression level in early stages results in decreased chromatin accessibility.

   Major comments:

   1. Although it is not clearly described, it seems that phosphoproteomic and single-cell RNA-seq were performed using 2D-cultured cells, while ATAC-seq was performed using 2D (FACS sorted cells based on HER2 expression levels) or 3D (time course)-cultured cells. Cells cultured on 2D and 3D are significantly different on cell signaling, chromatin structure and gene expression, and therefore cannot be compared.

   We agree that there are differences between 2D and 3D cell cultures, which may impact on the multi-omics experiments performed in this study. In an ideal world we would have preferred to be able to conduct all experiments in 3D cell cultures, including the phosphoproteomics experiments. However, this is not feasible because the phosphoproteomics experiment requires 500ug of total protein which corresponds to approximately 10 million cells for each condition and replicates in 3D matrices. 3D structures would have also presented with accessibility issues since doxycycline might not have reached all cells equally at the 30 minutes timepoint. Since we were analysing early timepoints for phosphoproteomics, homogeneity in induction was important. We performed ATAC-seq in 3D cell culture because it was feasible as it only required 25,000-50,000 cells to be grown in small 3D cell cultures and is indeed superior for physiological relevance. We therefore had to compromise and worked with the assumption that immediate signaling events will not be fundamentally different in 2D vs 3D. We have modified the main text to better reflect this and have indicated which experiments were performed in 2D vs 3D in the figure legends and the methods section.

   1. Phosphoproteomic (0.5, 4 and 7 hours), ATAC-seq (1, 4, 7, 24 and 48 hours) and single-cell RNA-seq (7, 24, 48 and 72 hours) were performed on cells at different time points after doxycycline treatment. The authors need to clearly explain the rationale why such time points were chosen for each experiment in the text.

   There are indeed differences in the time-points analysed between the different multi-omics analysis. However, as mentioned above, the reason for selecting such early time points for the phosphoproteomic experiment was that signalling changes are rapid and we were focused on characterising the early signalling dynamics. With regards to the ATAC-seq and scRNA-seq, there are several shared time-points such as the 7h, 24h, and the 48h. Additionally, as the chromatin changes would be slower acting as compared to signalling changes, two later time-points were selected including the 48h (ATAC-seq) and 72h (scRNA-seq) to capture some late changes during cellular transformation.

   1. Change on chromatin accessibility does not necessarily mean change on gene expression levels. RNA-seq needs to be performed and analyzed along with ATAC-seq data.

   We agree that chromatin accessibility does not necessarily correlates with gene expression changes and the need to perform RNA-seq to make such a conclusion. This is the reason we performed single cell RNA-seq, which looks at changes in high temporal and cellular resolution. This is particularly useful for the heterogenous cell population that we worked in to better understand the differences between cell types.

   1. Analyses on multi-omics data are quite preliminary. Clustering analysis on the time course of phosphoproteomic, ATAC-seq and single-cell RNA-seq will help characterize the dynamics of cell signaling and gene expression. Integrated analyses on multi-omics data and construction of regulatory network are necessary to identify the key signaling node and key epigenetic regulators/machinery that facilitate or prevent cell transformation. Integrated analyses, of course, need to be performed on data obtained from cells cultured in the same conditions.

   We think our study is an important work and provides a strong foundation for a comprehensive, integrative multi-omics study using primary human breast cells with parallel analysis performed on the same population of cells using the latest techniques such as scATAC and RNA-seq or scNMT-seq. We are indeed in the process to apply for funding in a larger analysis that involves in vivo work and clinical samples, using this study as a foundation.

   1. The authors picked up several genes from the analyses, and discussed the potential importance in cell transformation without functional validation. It is important to show data demonstrating altered expression of certain genes and/or altered activity of certain signaling pathway/epigenetic regulators is indeed important for cell transformation in low HER2-expressing condition or preventing cell transformation in high HER2-expressing condition.

   We agree that this is important. The scope of this study is to report on the result that low HER2 was unexpectedly more aggressive compared to high HER2, which was a highly reproducible observation, and identified a molecular explanation for this behaviour (dedifferentiation and predominant chromatin opening). In terms of cross validation, we found the MUC1 protein expression to be low in low HER2 expressing cells, indicating that they are more stem-like (figure 4B). We confirmed and validated this finding in our scRNA-seq data shown in figure 4F. The pathway analysis from phosphoproteomic study shows that MAPK pathway is highly activated upon HER2 protein overexpression. To validate this claim, we performed western blotting analysis that confirm this as the ERK protein was hyperphosphorylated in HER2 expressing cells compared to controls. Thus, our resource study provides many candidates that can be tested to further explore the biology.

   1. HER2 expression in MCF10A cells is insufficient in inducing tumor formation in vivo, although HER2 expression results in disrupted acini structure and colony formation in vitro (e.g. Alajati et al. 2013 Cancer Res, 73:5320-5327 cited in the manuscript). It is interesting to investigate whether this is due to the mechanisms identified in this study.

   MFC10A cells are generally difficult to transform in vivo. It is possible that mechanisms identified in our study might be responsible for lower tumourigenicity in vivo with WT HER2 compared to HER2 variants, since our study suggests activated checkpoints in high HER2 cells. It would be interesting to compare the differential impact on chromatin for the two HER2 variants too. In our system, we think the reasons why cells form abnormal morphological changes and grow colonies in vitro is a result of HER2 overexpression, which induces aberrant signalling, and this may be leading to loss of cell-to-cell contact and disruption of adhesion molecules. However, the objective of this study was to understand the early signaling to chromatin changes in in vitro cellular transformation, and changes in cell morphology are a consequential part of the process.

   1. In Figure 2C, two replicates are completely separated and replicates of each time points are not clustered together.

   We agree that the two replicates are separated into two separate groups, this was demonstrated by the PCA analysis (Supplementary Fig 1F). We grouped these samples into “early” (0h, 1h, 4h, and 7h time-points) or “late” (24h and 48h time-points) based on them clustering well into these two groups. The subsequent analysis were performed based on these groups that clustered together. However, we still showed each replicate in figure 2C to appreciate the dynamics of chromatin accessibility between each time-point, which shows clear differences in HER2 versus Control.

   Minor comments:

   1. Essential experimental information, e.g. whether cells were cultured in 2D or 3D, needs to be clearly and accurately described in main text, figure legends and experimental procedures.

   The figure legends in the manuscript have now been modified to include information on cell culture type.

   1. Statistic methods are not provided. In Fig. 4D, HER2-med and HER2-high need to be compared to HER2-low group.

   Statistical analyses have been added to figure 4D and HER2-med and HER2-high have been compared to HER2-low group.

   Reviewer #2 (Significance):

   The authors propose sub-threshold low level HER2 expression activates signaling pathways and increases chromatin accessibility, which facilitates mammary epithelial cell transformation, while high HER2 expression in early stages results in decreased chromatin accessibility via unknown feedback mechanisms. It is interesting to identify which signaling and epigenetic regulators are essential to cell transformation, which feedback mechanisms prevent the transformation of HER2-amplified mammary epithelial cells, whether inactivation of such feedback mechanism indeed occurs in tumorigenesis of HER2-amplified breast cancer, and whether it is a potential therapeutic target for HER2-amplified breast cancer.

   Expertise of review: breast cancer, cell signaling, tumor microenvironment.

   We thank reviewer #2 for their time and for providing such useful feedback on our work.

   Reviewer #3 (Evidence, reproducibility and clarity):

   In this paper Hayat et.al study the early transformational events that follow the activation of the oncogenic HER2 signaling pathway and its crosstalk with chromatin opening. Using an inducible in vitro model of HER2+ breast cancer they have identified that the overexpression of HER2 transforms non-tumorigenic breast epithelial cells via chromatin regulation. The study also shows that the transformative potential of the cells is inversely related their HER2 expression where the low HER2 expressing cells obtain a stem-cell like signature and increased chromatin accessibility leading to an increased transformative potential.

   Major comments:

   While the key conclusions of the paper are convincing, here are the parts of the study that need further clarification or supporting data from the authors.

   1. In Figure 1C the authors show that MCF10AHER2 cells formed complex transformed masses when grown in 3 dimensional cultures. From the figure it is evident that that the transformative potential of the HER overexpression is far more pronounced at the Day 6 and Day 9 mark. Therefore, one wonders why these time points weren’t used as the “late timepoint” in any of the sequencing studies moving forward. Can the authors comment on this choice and perform additional experiments to address the molecular changes that lead to the dramatic transformations seen at this timepoint? Since the authors have a well-established protocol in place, looking at an additional time point could be potentially feasible, provided the cells/samples have been frozen down at this stage. If unable to do so, could the authors comment on the molecular changes they would expect to see at this time point.

   In our study we primarily focused on the early events upon HER2 overexpression because the changes appear to be much more dynamic, and we hypothesised that these events are the cause of the subsequent, more pronounced featured later on. The rationale behind employing an inducible system and capturing the early changes was to identify aberrant molecular events at the earliest time possible. Indeed, numerous studies have investigated the differences between normal versus cancer cells (many of which are at later time points, that have missed the foremost aberrant molecular changes). Based on our ATAC-seq analysis at late-timepoints, 24h and 48h time-point, the number of changes in chromatin accessibility become relatively more stable as compared to early time points (supplementary figure 2A).

   1. Fig 1D the authors conclude that the overexpression of HER2 causes increased cell invasion based on the results seen in a collagen coated plate. How to the authors explain the lack of any such significant change in a Matrigel coated plate?

   To test the invasiveness of the HER2 overexpressing cells, collagen is used to increase stiffness to Matrigel. Stiffness is relevant for the type of invasion seen in these 3D cultures because it activates pathways important for invasion. We added the references to the text for clarity (PMID: 15838603 and PMID: 16472698).

   1. In Supp Fig 1D the authors use the DAVID Bioinformatics tool to identify the various signaling pathways enriched in the HER2 induced system. In addition to the MAPK pathway this analysis also shows other common cancer-related pathways (eg. The Mtor pathway) being enriched to a similar or higher extent. Can authors address why only the MAPK pathways was pursed in detail?

   HER2 is major receptor that can signal through various signalling pathways. We highlighted the MAPK pathway because it has been previously shown that MAPK cascades can modify chromatin through transcription factors and chromatin regulators Clayton and Mahadevan., 2010 (PMID:19948258). We think that when HER2 is overexpressed, it primarily signals down the MAPK pathway, resulting in the activation of transcription factors and chromatin regulators that lead to a highly accessible chromatin and ultimately contributes to transformation. To confirm this result, we did perform western blotting control analysis and found that indeed, HER2 overexpression consistently activates the MAPK pathway that shows phosphorylation of ERK but does not influence AKT phosphorylation. We can include this data in the manuscript.

   1. Figure 4B and supplementary figure 3E only show that percentage of the cells have either MUC1-ve or EpCAMlow or CD24low expression. However, Figure 4A and the corresponding text indicates that that breast stem cells are defined by a combination of MUC1-ve, EpCAMlow, and CD24low expression. If this is the case, the authors need to show the percentage of the cells within each population have an overlap of all these expression signatures, to support the claim of low HER2 expressing cells showing a more de-differentiated stem-cell like property.

   Our results confirm that upon HER2 overexpression, cells become MUC1-ve, EpCAMlow and CD24-ve, acquiring the breast stem cell signature. We did not show the CD24 expression because all the cells that were MUC1-ve and EpCAMlow were also 100% CD24-ve. We have now modified figure 4B and the figure legend to reflect this change, additionally, we added another figure (supplementary figure 4) that shows how the analysis was performed systematically.

   1. The authors also state 'other biological effects being responsible for the lower capacity in anchorage-independent growth of high HER2 expressing cells' that is shown in fig 4d. While an experimental investigation of these effects may be out of the scope of this study, the authors may consider commenting (and referencing additional literature) on the other biological effects they think may result in this phenomenon.

   We have modified the manuscript (lines 294-296) and added further explanation as to what other biological effects may be responsible for lack of colony growth in high HER2 expressing cells in lines.

   1. The authors do a great job providing details about all statistical analyses performed, however the details regarding the experimental replicates are only provided for some experiments making it difficult to infer if the experiments have been adequately replicated before concluding results. Can the authors please add the n - value for all applicable experiments in the figure legend or the methods section?

   The number of replicates has now been added to the respective figure legends.

   1. What is the scope for validation of these findings in vivo and in human samples? Could the authors please comment on this in the discussion section of the manuscript.

   The primary goal of this study was to understand the early transformational events in a simple in vitro, yet a robust model that is highly accessible. We have analysed some human samples to compare the HER2 protein expression levels. However, the findings from this manuscript could be validated in more precious models such as primary human cells, human tumours samples and in vivo in animals. We have modified the end of discussion to address these points (lines 394-399).

   Minor comments:

   1. In figure 1B the authors show a western blot analysis for HER2 expression over time while using GAPDH as a loading control. However, GADPH control seems to be unequal, especially in the 1ug/ml Dox lane. This needs to be addressed.

   We agree that there is a slight difference in the GAPDH levels in this western blot. We have carried out densitometry analysis which could be added to the supplementary data if required, to show that even though the GAPDH appears to be slightly less in the 1ug/ml of dox (last lane), it shows that HER2 levels are even greater than what appears on the blot, thereby confirming the trend we have observed in the current western blot.

   1. In figure 1C, it is unclear if the images shown are representative of the exact same spot over a 9-day period or of different spots.

   In figure 1C, the morphological regions are representative of the whole well in which the cells were growing but not the exact same spot. This is because nearly all the cells (>90%) transformed from round, organised acini to the fibroblastic, invasive morphology by day 9. We have captured multiple images of different areas in the well using confocal microscopy, and this can be added in the supplementary data.

   1. In Supplementary figure 3E, labeling the y-axis on the figure as opposed to just in the legends would make it easy for the reader.

   The figure has now been appropriately labelled.

   1. With respect to presentation: In figures involving single cell RNA sequencing and phosphoproteome analyses, highlighting the specific genes that are focused in detail on the manuscript would aid the reading process. The current format makes it difficult for the reader to spot the specific genes that are the points of focus within each heat map.

   We modified the figures concerning the phosphoproteomic analysis and scRNA-seq and have highlighted important genes for readers’ ease.

   Reviewer #3 (Significance):

   I have close to a decade's experience in working on breast cancer. In the past I focused on studying intratumor genetic heterogeneity and cell signaling pathway interactions. I am currently working on identifying novel therapeutic targets for the treatment of ER+ breast cancer. My expertise lies in understanding molecular biology of the disease. While I have worked with and understand most techniques used in this study, I would like to indicate that I do not have sufficient expertise in ATAC seq and am unable to evaluate the intricacies of this technique.

   While molecular changes that occur in HER2+ breast cancer have been highly investigated, the changes that occur at an early pre-cancerous stage of the disease aren't as well documented. The work by Hayat et al., sheds much needed light on this less documented early stage of cancer development. The past decade has shown an increased focus on epigenetic therapy with more chromatin targeting drugs entering clinic (Siklos et al., 2022). There has also been increased clinical evidence underlining the efficiency of combining epigenetic therapy and with hormonal and other anticancer therapies in solid tumors (Jin et al., 2021). Phase II clinical trials combining HDAC inhibitors with aromatase inhibitor have shown to improve clinical outcomes in patients (Yardley et al., 2013). Similarly, pre-clinical studies have shown that combination therapy with BET inhibitors improved treatment efficacy and circumvented drug resistance in fulvestrant (Feng et al., 2014) and everolimus (Bihani et al., 2015) treatments. Conclusions from the work by Hayat et.al, although based on in vitro analyses, advances this field conceptually by highlighting the importance of targeting the cell signaling and chromatin regulation together. If validated in in vivo models and clinical samples, this may open up potential possibilities of combining anti-HER2 therapies with epigenetic therapies. Additionally, the study also makes an interesting observation that low HER2 expression could result in increased tumorigenicity of cells which is in contrary to current clinical norm of looking at increased HER2 expression as a sign of aggressive disease. These findings are of interest to the scientific and clinical community working on discovering novel therapeutic targets and biomarkers for treatment of HER2+ breast cancer.

   We thank reviewer #3 for his/her overall assessment and for appreciating this work. There is a significant focus regarding low HER2 positive breast cancers in the field. Approximately 50-60% of breast cancers have "low" HER2 expression and in many cases, this low HER2 is seen together with metastatic cancer. The FDA has very recently approved fam-trastuzumab deruxtecan-nxki aka Enhertu, which appears to target these cancers with low HER2 well and is shown to be relatively effective in a phase 3 clinical trial known as Destiny Breast-04. However, it is not yet clear how low HER2 expressing cells drive the metastatic spread of breast cancers or why they are so aggressive. Our work sheds a light that increased chromatin accessibility could be a route of transformation in low HER2 cancers. Therefore, providing an alternative platform to target these cancers and why it is crucial that this work reaches the clinical and scientific community as soon as possible.

   Read the original source
2. 

   Review Commons

   Nov 17, 2022

   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

   In this paper Hayat et.al study the early transformational events that follow the activation of the oncogenic HER2 signaling pathway and its crosstalk with chromatin opening. Using an inducible in vitro model of HER2+ breast cancer they have identified that the overexpression of HER2 transforms non-tumorigenic breast epithelial cells via chromatin regulation. The study also shows that the transformative potential of the cells is inversely related their HER2 expression where the low HER2 expressing cells obtain a stem-cell like signature and increased chromatin accessibility leading to an increased transformative potential.

   Major …

   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

   In this paper Hayat et.al study the early transformational events that follow the activation of the oncogenic HER2 signaling pathway and its crosstalk with chromatin opening. Using an inducible in vitro model of HER2+ breast cancer they have identified that the overexpression of HER2 transforms non-tumorigenic breast epithelial cells via chromatin regulation. The study also shows that the transformative potential of the cells is inversely related their HER2 expression where the low HER2 expressing cells obtain a stem-cell like signature and increased chromatin accessibility leading to an increased transformative potential.

   Major comments:

   While the key conclusions of the paper are convincing, here are the parts of the study that need further clarification or supporting data from the authors.

   1. In Figure 1C the authors show that MCF10AHER2 cells formed complex transformed masses when grown in 3 dimensional cultures. From the figure it is evident that that the transformative potential of the HER overexpression is far more pronounced at the Day 6 and Day 9 mark. Therefore, one wonders why these time points weren't used as the "late timepoint" in any of the sequencing studies moving forward. Can the authors comment on this choice and perform additional experiments to address the molecular changes that lead to the dramatic transformations seen at this timepoint? Since the authors have a well-established protocol in place, looking at an additional time point could be potentially feasible, provided the cells/samples have been frozen down at this stage. If unable to do so, could the authors comment on the molecular changes they would expect to see at this time point.
   2. Fig 1D the authors conclude that the overexpression of HER2 causes increased cell invasion based on the results seen in a collagen coated plate. How to the authors explain the lack of any such significant change in a Matrigel coated plate?
   3. In Supp Fig 1D the authors use the DAVID Bioinformatics tool to identify the various signaling pathways enriched in the HER2 induced system. In addition to the MAPK pathway this analysis also shows other common cancer-related pathways (eg. the mTOR pathway) being enriched to a similar or higher extent. Can authors address why only the MAPK pathways was pursed in detail?
   4. Figure 4B and supplementary figure 3E only show that percentage of the cells have either MUC1-ve or EpCAMlow or CD24low expression. However, Figure 4A and the corresponding text indicates that that breast stem cells are defined by a combination of MUC1-ve, EpCAMlow, and CD24low expression. If this is the case, the authors need to show the percentage of the cells within each population have an overlap of all these expression signatures, to support the claim of low HER2 expressing cells showing a more de-differentiated stem-cell like property.
   5. The authors also state 'other biological effects being responsible for the lower capacity in anchorage-independent growth of high HER2 expressing cells' that is shown in fig 4d. While an experimental investigation of these effects may be out of the scope of this study, the authors may consider commenting (and referencing additional literature) on the other biological effects they think may result in this phenomenon.
   6. The authors do a great job providing details about all statistical analyses performed, however the details regarding the experimental replicates are only provided for some experiments making it difficult to infer if the experiments have been adequately replicated before concluding results. Can the authors please add the n - value for all applicable experiments in the figure legend or the methods section?
   7. What is the scope for validation of these findings in vivo and in human samples? Could the authors please comment on this in the discussion section of the manuscript.

   Minor comments:

   1. In figure 1B the authors show a western blot analysis for HER2 expression over time while using GAPDH as a loading control. However, GADPH control seems to be unequal, especially in the 1ug/ml Dox lane. This needs to be addressed.
   2. In figure 1C, it is unclear if the images shown are representative of the exact same spot over a 9-day period or of different spots.
   3. In Supplementary figure 3E, labeling the y-axis on the figure as opposed to just in the legends would make it easy for the reader.
   4. With respect to presentation: In figures involving single cell RNA sequencing and phosphoproteome analyses, highlighting the specific genes that are focused in detail on the manuscript would aid the reading process. The current format makes it difficult for the reader to spot the specific genes that are the points of focus within each heat map.

   Significance

   I have close to a decade's experience in working on breast cancer. In the past I focused on studying intratumor genetic heterogeneity and cell signaling pathway interactions. I am currently working on identifying novel therapeutic targets for the treatment of ER+ breast cancer. My expertise lies in understanding molecular biology of the disease. While I have worked with and understand most techniques used in this study, I would like to indicate that I do not have sufficient expertise in ATAC seq and am unable to evaluate the intricacies of this technique.

   While molecular changes that occur in HER2+ breast cancer have been highly investigated, the changes that occur at an early pre-cancerous stage of the disease aren't as well documented. The work by Hayat et al., sheds much needed light on this less documented early stage of cancer development. The past decade has shown an increased focus on epigenetic therapy with more chromatin targeting drugs entering clinic (Siklos et al., 2022). There has also been increased clinical evidence underlining the efficiency of combining epigenetic therapy and with hormonal and other anticancer therapies in solid tumors (Jin et al., 2021). Phase II clinical trials combining HDAC inhibitors with aromatase inhibitor have shown to improve clinical outcomes in patients (Yardley et al., 2013). Similarly, pre-clinical studies have shown that combination therapy with BET inhibitors improved treatment efficacy and circumvented drug resistance in fulvestrant (Feng et al., 2014) and everolimus (Bihani et al., 2015) treatments. Conclusions from the work by Hayat et.al, although based on in vitro analyses, advances this field conceptually by highlighting the importance of targeting the cell signaling and chromatin regulation together. If validated in in vivo models and clinical samples, this may open up potential possibilities of combining anti-HER2 therapies with epigenetic therapies. Additionally, the study also makes an interesting observation that low HER2 expression could result in increased tumorigenicity of cells which is in contrary to current clinical norm of looking at increased HER2 expression as a sign of aggressive disease. These findings are of interest to the scientific and clinical community working on discovering novel therapeutic targets and biomarkers for treatment of HER2+ breast cancer.

   Read the original source
3. 

   Review Commons

   Nov 17, 2022

   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

   HER2 amplification is associated with poor prognosis of breast cancer. Despite it has been extensively studied, it deserves thorough study how HER2 amplification alters downstream signaling pathways, chromatin structure and gene expression, and how cells overcome the hurdles in order to transform. In this study, Hayat et al used doxycycline-induced HER2 expression in MCF10A cells to recapitulate the very early stage of HER2 expression and HER2-induced mammary epithelial cell transformation. The authors performed global phosphoproteomic, ATAC-seq and single-cell RNA-seq, and propose sub-threshold low level HER2 expression activates …

   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

   HER2 amplification is associated with poor prognosis of breast cancer. Despite it has been extensively studied, it deserves thorough study how HER2 amplification alters downstream signaling pathways, chromatin structure and gene expression, and how cells overcome the hurdles in order to transform. In this study, Hayat et al used doxycycline-induced HER2 expression in MCF10A cells to recapitulate the very early stage of HER2 expression and HER2-induced mammary epithelial cell transformation. The authors performed global phosphoproteomic, ATAC-seq and single-cell RNA-seq, and propose sub-threshold low level HER2 expression activates signaling pathways and increases chromatin accessibility required for cell transformation, while high HER2 expression level in early stages results in decreased chromatin accessibility.

   Major comments:

   1. Although it is not clearly described, it seems that phosphoproteomic and single-cell RNA-seq were performed using 2D-cultured cells, while ATAC-seq was performed using 2D (FACS sorted cells based on HER2 expression levels) or 3D (time course)-cultured cells. Cells cultured on 2D and 3D are significantly different on cell signaling, chromatin structure and gene expression, and therefore cannot be compared.
   2. Phosphoproteomic (0.5, 4 and 7 hours), ATAC-seq (1, 4, 7, 24 and 48 hours) and single-cell RNA-seq (7, 24, 48 and 72 hours) were performed on cells at different time points after doxycycline treatment. The authors need to clearly explain the rationale why such time points were chosen for each experiment in the text.
   3. Change on chromatin accessibility does not necessarily mean change on gene expression levels. RNA-seq needs to be performed and analyzed along with ATAC-seq data.
   4. Analyses on multi-omics data are quite preliminary. Clustering analysis on the time course of phosphoproteomic, ATAC-seq and single-cell RNA-seq will help characterize the dynamics of cell signaling and gene expression. Integrated analyses on multi-omics data and construction of regulatory network are necessary to identify the key signaling node and key epigenetic regulators/machinery that facilitate or prevent cell transformation. Integrated analyses, of course, need to be performed on data obtained from cells cultured in the same conditions.
   5. The authors picked up several genes from the analyses, and discussed the potential importance in cell transformation without functional validation. It is important to show data demonstrating altered expression of certain genes and/or altered activity of certain signaling pathway/epigenetic regulators is indeed important for cell transformation in low HER2-expressing condition or preventing cell transformation in high HER2-expressing condition.
   6. HER2 expression in MCF10A cells is insufficient in inducing tumor formation in vivo, although HER2 expression results in disrupted acini structure and colony formation in vitro (e.g. Alajati et al. 2013 Cancer Res, 73:5320-5327 cited in the manuscript). It is interesting to investigate whether this is due to the mechanisms identified in this study.
   7. In Figure 2C, two replicates are completely separated and replicates of each time points are not clustered together.

   Minor comments:

   1. Essential experimental information, e.g. whether cells were cultured in 2D or 3D, needs to be clearly and accurately described in main text, figure legends and experimental procedures.
   2. Statistic methods are not provided. In Fig. 4D, HER2-med and HER2-high need to be compared to HER2-low group.

   Significance

   The authors propose sub-threshold low level HER2 expression activates signaling pathways and increases chromatin accessibility, which facilitates mammary epithelial cell transformation, while high HER2 expression in early stages results in decreased chromatin accessibility via unknown feedback mechanisms. It is interesting to identify which signaling and epigenetic regulators are essential to cell transformation, which feedback mechanisms prevent the transformation of HER2-amplified mammary epithelial cells, whether inactivation of such feedback mechanism indeed occurs in tumorigenesis of HER2-amplified breast cancer, and whether it is a potential therapeutic target for HER2-amplified breast cancer.

   Expertise of review: breast cancer, cell signaling, tumor microenvironment.

   Read the original source
4. 

   Review Commons

   Nov 17, 2022

   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

   This study aimed to identify events that happens early in malignant transformation of breast cancer (BC) cells that are driven by HER2 oncogene. Constructing a 3D inducible model to study impact of HER2 protein level on BC cell and assessment of gross morphological changes, protein phosphorylation and chromatin accessibility at different time points of HER2 activation.

   Using a controllable in vitro model is a good approach although it is not novel. Also the method used to assess HER2 protein positivity is not standardized nor clinically relevant. Positivity of HER2 in clinical practice is assessed either through immunohistochemistry …

   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

   This study aimed to identify events that happens early in malignant transformation of breast cancer (BC) cells that are driven by HER2 oncogene. Constructing a 3D inducible model to study impact of HER2 protein level on BC cell and assessment of gross morphological changes, protein phosphorylation and chromatin accessibility at different time points of HER2 activation.

   Using a controllable in vitro model is a good approach although it is not novel. Also the method used to assess HER2 protein positivity is not standardized nor clinically relevant. Positivity of HER2 in clinical practice is assessed either through immunohistochemistry (IHC 3+ or 2+ with gene amplification), however the author did not mention any control for positivity except western blot which is not used in clinical practice.
   There is difference between early HER2 positive BC and HER2 low BC. As the earlier is driven by HER2 oncogenic signalling pathway, but the latter is not.

   Identification of molecular changes that occur at HER2 low BC seems very important and clinically relevant, however HER 2 low is not fully characterized, yet. And the only definition available is either HER2 1+ or 2+ without gene amplification. The author was not very clear about threshold he followed to call the model HER2 low. Is it positive with lower limit of positivity or just small amount of protein). He also concluded that BC with sub-threshold of HER2 protein behave more aggressive than HER2 positive BC. What is the threshold and was it correlated with IHC or gene amplification level to be reliable?

   The status of oestrogen and progesterone receptors were not highlighted. Triple negative breast cancer, for instance, is more aggressive than HER2 positive BC, this may be the reason for the worse behaviour.
   At line 130, "The low levels of HER2 protein activation at early time point may closely mimic at least partially the signalling changes occurring in HER2 positive BC patients". This claim is not quite true, as low levels of HER2 protein activation doesn't activate HER2 oncogenic signalling pathway as HER2 positive does.
   The author aimed to study the signalling changes accompanying low levels of HER2 induction by lowering significance threshold to log2fold > 0.5. Lowering the threshold for significance will increase the total number of phosphorylated protein (both at low HER2 levels and high levels). So, studying the whole significant proteins at whole time points will not be exclusive for low HER2 levels and this was evident through activation of MAPK cascade which is one of downstream signalling pathway of HER2 positive BC.

   Combining HER2 protein level (both IHC and Western blot) to different time points will give better understanding of events associated with HER2 low, early positive or late positive.

   Significance

   This work provides good evidence to changes that happen at early HER2 positive breast cancer transformation and introducing a chromatin opening and accessibility as a new target of treatment of HER2 positive breast cancer patients.

   Read the original source
5. 

   Version published to 10.1101/2022.09.06.506760v1 on bioRxiv

   Sep 6, 2022