Single-cell transcriptome analysis indicates fatty acid metabolism-mediated metastasis and immunosuppression in male breast cancer

Male breast cancer (MBC) is a rare but aggressive malignancy with cellular and immunological characteristics that remain unclear. Here, we perform transcriptomic analysis for 111,038 single cells from tumor tissues of six MBC and thirteen female breast cancer (FBC) patients. We find that that MBC has significantly lower infiltration of T cells relative to FBC. Metastasis-related programs are more active in cancer cells from MBC. The activated fatty acid metabolism involved with FASN is related to cancer cell metastasis and low immune infiltration of MBC. T cells in MBC show activation of p38 MAPK and lipid oxidation pathways, indicating a dysfunctional state. In contrast, T cells in FBC exhibit higher expression of cytotoxic markers and immune activation pathways mediated by immune-modulatory cytokines. Moreover, we identify the inhibitory interactions between cancer cells and T cells in MBC. Our study provides important information for understanding the tumor immunology and metabolism of MBC.

understudied topic. The Authors performed single-cell analysis to identify notable differences in the two groups, such as ESR1 and AR activity and fatty acid metabolism mainly mediated by FASN expression. Moreover, the Authors describe higher levels of tumor purity, cell cycle genes and pathways related to tumor invasiveness, as well as lower immune infiltration in male breast cancer compared to samples from women, findings in line with bulk tumor samples in the TCGA. The Authors also describe the presence of immuneepithelial cells, and the potential relationship between fatty acid metabolism and metastasis-related programs. A prognostic role of these features is also suggested, using bulk data from the TCGA, as well as a potential therapeutic role for FASN inhibition as demonstrated by previous works. The relationship between fatty acid synthesis, immunosuppression and tumor progression is intriguing, but may require some clarification.
The following may need to be addressed: -As single-cell experiments were performed in 3 male and 2 post-menopausal female breast cancer samples, the explorative nature of the findings should be addressed throughout the manuscript.
-Clinicopathological characteristics such as tumor stage may be relevant when comparing the male and female samples, to ensure that differences observed were not related to differences in the staging (e.g. larger and more advanced tumors may be associated with immune exhaustion). As per supplementary table 1, tumor stage was not available for female breast cancer samples. The Authors should specify this in the Methods section (lines 406-411). Whether samples are from primary, untreated breast cancers should also be specified.
- Supplementary Table 1  should also be specified.
-Lines 219-226: The Authors describe a positive correlation between FASN expression and tumor purity, and a negative correlation not only with immune cells but also with CAFs and endothelial cells. The Authors then suggest that elevated expression of FASN may promote immune escape. Since higher tumor purity is also necessarily associated with lower stroma content (including immune cells), the cause-effect relationship suggested by the Authors may not be necessarily proved by these findings. This section may need to be adjusted accordingly.
-Similarly, the positive correlation with metastasis-related pathways (e.g. suggested in line 374-376) may not necessarily mean "causation", as many other factors could play a role.
-Line 270 -Paragraph "MBC-specific T cells that co-expressed epithelial and immune markers were in the apoptosis stage": here, the Authors describe the presence of T cells showing both T and epithelial cell markers, suggesting the existence of "epithelial-T cells".
The Authors tried to exclude that this finding were to be related to technical artifacts in the single-cell analysis, and showed the coexistence of CD3E and KRT8 markers with immunofluorescence experiments.
Although intriguing, further validation in other available single cell datasets (e.g. doi: 10.1038/s41588-021-00911-1) is in my opinion warranted, as it would give more robustness to this finding.
Moreover, the section included in lines 278-284 may need to be explained in a clearer way.
Indeed, as per Supplementary Figure 6, panel A (Differentially expressed genes and functional analysis of T cells between male and female patients), and as suggested by the Authors (lines 274-278), it seems that KRT genes are expressed only in T cells from sample M3 (and maybe to a lesser extent M1). Confirming these findings in other single-cell datasets would be useful to exclude that the presence of epithelial-T cells are patientspecific, as one may argue that they may not be specific of male breast cancer.
Moreover, the presence of several genes related to dissociation or cellular stress (e.g. mitochondrial gene, FOSB, JUNB, heat-shock protein genes) and ribosomal genes, may raise a concern regarding contamination at the droplet level by dying cells (that would go in the same direction of what is stated in lines 308-313, when the Authors mention the high expression of apoptosis-related genes in this cell type). In this regard, can the Authors exclude that the finding of epithelial-T cells is not related to a potential issue of contamination? Indeed, in case of contamination, filtering by the number of genes may not be enough to identify technical artifacts. -Lines 454-455: Please rephrase specifying in a clearer way if UMI count and MT genes were used as regression terms in the ScaleData functon(also, Authors may replace "ScaleDate" with "ScaleData" in the text). The is a well written and comprehensive manuscript describing the immune and metabolic landscape of male breast cancer.
The premise of this paper that male and female breast cancers are immunological and metabolically different is very compelling and may potentially provide new insights into therapeutic strategies. The investigators have carefully evaluated a broad range of proliferation, angiogenesis, and metabolic pathways as well as detailed immune characterization. The study includes a limited number (3 and 2) reference cases. The study is expanded by data from the TCGA.
Strength of the study include the clearly distinctive patterns that the evaluated male and female breast cancers. The single cell sequencing is elegantly done, and the figures are beautifully outlined and clearly delineated.
A major concern of the study is that the female breast cancers neither have ER expression (ESR1) nor ER activity. Male breast cancer is mostly ER+, whereas female breast cancer has a broad diversity ranging from triple negative disease to ER+ and HER2 positive disease. The immune landscape, EMT, angiogenesis is vastly different in these subtypes. Particularly, TNBC stand out in their immune profile. The data would be very much strengthened if the authors provided data on ER+ female breast cancer, to show how this is similar or different from an ER+ male breast cancer. Male breast cancer (MBC) is associated with worse prognosis compared to female breast cancer and the cellular and molecular differences between the two remain unclear. The researchers used single-cell RNA (scRNA) sequencing and T cell receptor (scTCR) sequencing characterize the tumor microenvironment of MBC. They sequenced three MBC and two post-menopausal ER+ female breast cancers (FBC) and show evidence that MBC have lower immune infiltration, activated ER and AR regulons, higher fatty acid synthase (FASN) expression, and exhausted CD8 T cells. The authors identify a subset of T-cells that express epithelial cytokeratins. However, the manuscript is lacking good quality evidence for the existence of these epithelial-T cells. The authors should consider removing that entire section or provide additional experiments to validate their findings. Androgens have long been known to drive fatty acid synthase PMID: 9067276, and the authors show good evidence of AR regulon activation in MBC, perhaps more focus on the androgen receptor would tie this story together. Overall, the study is of interest, but more experiments and analysis are needed for this study.
Specific comments 1. While two of the three MBC samples have low immune infiltrate, one actually has similar levels to the two other FBC samples (Figure 1e). Therefore, on cannot conclude that there are less immune cells in MBC, as this may just be a sampling artefact.
2. Please supply raw p-value and statistical test used in Fig.1g. There are only 12 male samples compared to 1085 female samples in the TCGA, therefore one likely cannot assume the MBC will represent a normal distribution unless proven. Figure 1i  7. Supplementary Fig. 2 legend description inadequate. What fold change and significance and testing performed? 8. Supplementary Fig. 4 legend needs more detail. How were FASN high and low cutoffs determined? 9. The fact that FASN and the ER-and AR-response genesets were significantly enriched by the up-regulated genes of "epithelial-T" co-expression cells, suggests that there may be mixing of epithelial and T cell RNA in these dual positive cells. Therefore, additional experiments are needed for the existence of "epithelial-T cells". The authors provide dual immunoflouresence (IF), however the staining in Figure 5B is unconvincing. The legend states the scale bar is 50uM, but there is no scale bar and thus hard to interpret. It is not clear whether the staining is from a single mitotic cell or many cells at a distance. The DAPI does not even show uniform nuclear localization. The staining appears to be an artifact. The researchers need to show additional validation of the for IF using positive and negative control tissues. In addition, the investigators need to quantify the CD3 only and epithelial T cells for the IF. The authors should also provide another independent method to support their findings such as flow cytometry (KRT and CD3) of dissociated T cells from fresh tumor tissue if possible. 13. Data availability section is weak, and data are not publicly deposited (this can be blinded until publication but available for reviewers).

Statistical test for
14. The authors should consider evaluating the role of AR in MBC in more detail. Such as performing IHC on specimens, evaluating the RNA-seq for existence to alternative splicing in the androgen receptor. The manuscript "Single-cell transcriptome analysis reveals fatty acid metabolism 4 mediated metastasis and immunosuppression in male breast cancer" is an interesting 5 effort to characterize the differences between breast cancer in men and women, which 6 is un understudied topic. The Authors performed single-cell analysis to identify notable 7 differences in the two groups, such as ESR1 and AR activity and fatty acid metabolism  The following may need to be addressed:   performing the same analysis procedure using this updated dataset, we found that the 31 main results were consistent with the previous version, and demonstrated the followings:

32
(1) scRNA-seq, bulk transcriptome, and immunohistochemistry consistently 33 demonstrated that MBC had a significantly lower degree of T cell infiltration than FBC;

34
(2) metastasis-related programs such as cell migration, epithelial-mesenchymal     showed that there were no significant differences in age, HER2 status, KI67 level, and 75 extent of the tumor (T) between the FBC and MBC groups (Response In the latter case, "IHC classification" may be more appropriate. Furthermore, while it 104 is true that all samples were ER-positive, Authors should also mention the presence of  We defined the ER, PR, HER2, and KI67 status using IHC, and further evaluated the     We added the corresponding description in the revised Method section as follows  Response: Thank you for pointing this out. We apologize for not making this clear. The   This section may need to be adjusted accordingly.

222
Response: Thank you for your comments. We agree with the reviewer's concern.
223 Accordingly, we revised the corresponding part as follows (Lines 273-279): "Thus, we 224 performed a pan-cancer analysis to evaluate the association between FASN expression  Response: Thank you for your comments. We apologize for the inappropriate statement. 235 We revised the corresponding part as follows (Lines 502-505): "Notably, the fatty acid   11. Moreover, the presence of several genes related to dissociation or cellular stress (e.g. 355 mitochondrial gene, FOSB, JUNB, heat-shock protein genes) and ribosomal genes, may 356 raise a concern regarding contamination at the droplet level by dying cells (that would  Response: Thank you for your insightful comments. We agree with the reviewer that 363 some low-quality cells would be possibly included during the tissue dissociation, 364 including the stressed, broken, or dying cells, and doublets or multiplets. Firstly, by 365 performing the standard cell-filtering procedures that are commonly used in many 366 scRNA-seq studies, we had tried to limit the dissociation-related artifacts of multiplets 367 and broken/dying cells. Specifically, cells with expressed genes less than 200 or greater 368 than 6000 were excluded to remove the empty droplets and multiplets. Considering that 369 dying cells often exhibit extensive mitochondrial contamination, we calculated the 370 percentage of reads that mapped to the mitochondrial genome and filtered cells that 371 had >25% mitochondrial reads. Secondly, gradient cell-filtering criteria were 372 performed to limit the number of expressed genes and mitochondrial reads percentage.

373
Results showed that the percentage of CD3E + KRT8 + T cells did not decline with the                  Response: Thank you for pointing this out. In order to address this concern, we used 853 the violin-boxplots to better visualize the distribution of data in the revised Figure 4 854 and 5 (Response Figure 19-21). Specifically, the shape of violins represents the data's   These discrepancies need to be addressed to conclusively state that KRT8 positive T cell exist.

Reviewer #1 (Remarks to the Author):
The authors addressed adequately my comments and improved considerably the manuscript. I still have a few minor questions and remarks: 1. Regarding the existence of CD3D+/KRT8+ cells, did the authors try to use a software such as CellBender to decontaminate the raw UMI matrices and further investigate if this specific population of cells still remain in the resulting clustering?
Response: Thank you for your valuable suggestion. Accordingly, we used CellBender [1] to decontaminate the in-house scRNA-seq data, of which the raw UMI matrices were available. After removing the empty droplets and retrieving background-free gene expression profiles by CellBender, we found that CD3E + KRT8 + cells still existed in all samples (Response 2. It seems that the authors are using CD79A to identify B-cells. However, CD79A is expressed by both B and plasma cells, and two different clusters are visible within the umap. Did the authors try to use MS4A1 and JCHAIN to identify B-cells and plasma cells respectively? Response: We didn't notice this problem before. Many thanks for the reviewer's suggestion. Accordingly, we evaluated the expression levels of MS4A1 and JCHAIN in each single-cell cluster. It turned out that the smaller B/plasma cluster (1356 cells) had a specific JCHAIN expression, and was annotated as plasma cluster; the larger B/plasma cluster (2292 cells) had specific MS4A1 expression and was annotated as B cluster (Response Figure 2). The corresponding figures and text were modified in the revised manuscript (please also see the response to the relevant Comment #3).    Figure 5). But we did not identify the lymphatic endothelial cells in our data (Response Figure 6). We updated the feature-plots and cell type annotations in Figure 1 of the revised manuscript (Response    Response: We agree with the reviewer that providing solid evidence for the existence of CD3 + KRT8 + cells is essential. According to your suggestion, we found some instances of CD3 + KRT8 + cells that were away from tumor or T cells to avoid the exposure artifact, although finding these cells is challenging due to the relatively low proportion (please also see the response to the relevant Comment #3). As shown in Response Figure 12 and 13, the CD3 + KRT8 + cells were not necessarily to be located 4. Clearly all the T cells (CD3D) in figure 1B are negative for KRT8 and I suspect KRT18 as well. In Figure 5 F, all the male T cells are either KRT18 or KRT8 positive, which is different than figure 1B. These discrepancies need to be addressed to conclusively state that KRT8 positive T cell exist.

Response
Response: We apologize for the confusing visualization of Figure 5F in the previous submission and appreciate the reviewer's comment. Because as many as 100 differentially expressed genes (including 50 up-regulated genes and 50 down-regulated genes) were included in the heatmap, it is impossible to show the names of all genes in Figure 5F due to the limited space. Thus, only some representative gene names were shown beside the heatmap. Maybe the inexact pointing of KRT8/KRT18/KRT19 in the previous submission caused the misunderstanding of the proportion of MBC T cells with positive expression. We showed the original heatmaps for CD8 + , CD4 + , and NKT cells, with all gene names being displayed in Response Figures 16, 17, and 18. These figures showed that nearly half of MBC T cells were positive for KRT8/KRT18/KRT19 expression. Accordingly, we corrected the pointing of gene names and tried our best to make sure that the representative genes were exactly pointed beside the corresponding rows in the revised Figure 5F (Response Figure 19). Besides, the whole list of differentially expressed genes between MBC and FBC T cells, including CD4 + , CD8 + , and NKT cells.
The reason for why T cell clusters seem to be KRT8-negative in Figure 1B Figure   5F (Response Figure 16-19). In contrast, only 2.1% of T cells were KRT8 + in FBC (Response Figure 20c). Thus, we showed the KRT8 expression intensity on the t-SNE plot based on sex and whether T cells were KRT8 + (Response Figure 21). The results clearly showed that KRT8 was expressed on some T cells, especially the T cells from MBC samples (Response Figure 21). These results were added in the revised