scRNA-seq and spatial transcriptomics reveal neuroendocrine-like cancer cells promote angiogenesis and EMT through neural signaling pathways in male breast cancer

doi:10.21203/rs.3.rs-4291684/v1

Download PDF

Article

scRNA-seq and spatial transcriptomics reveal neuroendocrine-like cancer cells promote angiogenesis and EMT through neural signaling pathways in male breast cancer

https://doi.org/10.21203/rs.3.rs-4291684/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Male breast cancer (MBC) is a relatively rare and inadequately researched disease, and its cellular and molecular traits remain obscure. In this study, we conducted single-cell sequencing (N=20) and spatial transcriptomics (N=14) on 34 fresh tissue samples from 27 MBC patients. We identified six major cancer cell subtypes that are associated with the development and progression of MBC. Specifically, cancer cells exhibiting neuroendocrine-like properties facilitate immune evasion, tumor angiogenesis, epithelial-to-mesenchymal transition, cell proliferation, tumor invasion, and metastasis. They do so by secreting neuro-related factors and engaging in regulating neuro-related signaling pathways, synergistically interacting with T cells, macrophages, and fibroblasts within the tumor microenvironment. Additionally, we found that mutations or copy number variations amplifications of the UTY gene on the Y chromosome and/or its high transcript expression are closely associated with adverse clinical outcomes in male cancer patients, including MBC patients. In conclusion, our study provides important data support for a deeper understanding of the molecular characteristics and tumor microenvironment of MBC, and offers important clues for developing improved therapeutic strategies to improve the prognosis of MBC patients.

Biological sciences/Cancer/Breast cancer

Biological sciences/Genetics

Biological sciences/Cancer/Cancer microenvironment

Biological sciences/Genetics/Sequencing/RNA sequencing

Most studies on breast cancer exclude male breast cancer (MBC), even though MBC accounts for approximately 1% of all breast cancers^1,2. It is a rare and inadequately researched disease. Currently, data on the cellular biology characteristics of MBC are relatively limited, and clinical understanding of MBC is primarily inferred from data on female breast cancer (FMBC), leading to MBC treatment strategies often being based on FMBC^3,4. Research have shown that the clinical and pathological features of MBC differ from those of FMBC, and even after excluding clinical factors, the overall prognosis of MBC remains significantly worse than that of FMBC⁵.

Gender differences have emerged as a critical consideration in contemporary precision medicine^6,7. Research has revealed that aberrations in hormonal signaling pathways, genetic mutations, or structural anomalies on the sex chromosomes can drive the onset, progression, and adverse tumor-related events of breast cancer and prostate cancer⁷. Specifically, investigations have suggested that abnormal expression or structural irregularities of genes on the Y chromosome are closely associated with malignant tumor metastasis in males and poorer clinical outcomes⁸. MBC is a rare malignancy whose pathogenesis is intricately linked to abnormalities in neuro-related factors and hormone regulation^9,10. Nonetheless, the cellular and molecular characteristics of MBC remain enigmatic, with unidentified features at this level potentially serving as pivotal determinants for poor prognosis in male patients. Therefore, there is an urgent need to elucidate the cellular and molecular characteristics within the tumor microenvironment (TME) of MBC to gain deeper insights into the underlying mechanisms governing its onset and progression.

The coordinated regulation of neuro-related factors and hormones plays a pivotal role in the development of diverse malignant tumors^11–13, as well as in reshaping the immune landscape within the TME^14,15. Neurogenic infiltration in breast cancer is intricately linked to disease progression, metastasis, and clinical staging¹⁶. Furthermore, neuro-related secretions such as circadian rhythm hormones and neuro-related factors have been shown to stimulate cancer cell proliferation, invasion, and metastasis^17,18. Notably, the regulatory interplay of neuron-related signal factors and hormones within the TME further fosters immune suppression¹⁹ and angiogenesis²⁰ in breast cancer. Single-cell multi-omics sequencing technology has emerged as a valuable tool for investigating intra-tumor heterogeneity and unraveling the complexities of TME across various malignant tumors^21,22. This approach holds promise for identifying potential therapeutic targets that align with precision medicine principles. However, there remains a paucity of research data pertaining to MBC, necessitating further investigation and analysis into molecular characteristics at the single-cell level.

In this study, we explored major subtypes of cancer cells and the composition of the tumor microenvironment (TIME) in MBC. We investigated transcriptional, spatial transcriptomic, genomic, and metabolomic characteristics of major types of cells in MBC at single-cell resolution using single-cell RNA-seq (scRNA-seq) and Visium Spatial Transcriptome-seq (ST-seq) technologies. We found that neuroendocrine-like cancer cells secreting neuro-related factors play a crucial role in cancer progression, closely associated with the proliferation, infiltration, metastasis, epithelial-mesenchymal transition (EMT), and immune suppression of surrounding cancer cell subtypes. Acting as the main "agents" in the tumor stroma microenvironment, neuroendocrine-like cells participate in a "trade network" of intercellular communication signals, promoting tumor cell EMT and angiogenesis by secreting and distributing neuro-related factors and neuro-related intercellular signaling molecules produced by themselves or other non-epithelial cells. Furthermore, mutations or copy number variations amplification of the UTY gene on the Y chromosome, as well as its high transcript expression, are closely associated with adverse clinical outcomes in male cancer patients, including MBC patients. In summary, our study provides important biological insights into tumor neuro signaling and cancer development in MBC, offering clues for the development of therapeutic strategies to improve the prognosis of MBC patients.

Single-cell and spatial transcriptome characterization of male breast cancer

To delve into the cellular ecosystem and molecular characteristics of MBC, we conducted scRNA-seq and ST-seq on 34 fresh tissue samples from 27 patients. This included 20 scRNA-seq samples from 13 patients (13 in local areas, 4 near areas, and 3 lymph areas), as well as ST-seq samples from 14 patients (Supplementary Data 1). Following multiple quality control steps, we obtained single-cell transcriptomes of 204,419 cells from all scRNA-seq data for subsequent analysis. Utilizing Seurat V5²³ for integrated all scRNA-seq data, we identified 27 cell clusters (Extended Data Fig. 1a), covering 9 main cell types: epithelial cells (50.11%), NK/T cells (20.57%), fibroblasts (10.13%), B cells (6.59%), macrophages (5.21%), endothelial cells (4.77%), plasma cells (1.55%), mast cells (0.59%), and neutrophils (0.48%) (Fig. 1a). The majority of samples retained diverse cell types (Supplementary Data 2). Epithelial cells were detected in all samples (Extended Data Fig. 1b and Supplementary Fig. 1), and all types of cells showing distinct boundaries between with different cell populations (Extended Data Fig. 1c). Epithelial cells were mainly enriched in local areas and near areas (59.4% and 49.6%), while significantly fewer were found in lymph nodes (19.5%). In contrast, lymph node immune cells were enriched at 80.0%, 30.6% in local areas, and least in near areas (13.7%). Due to the small volume of MBC tissue samples used in this study, the sampling sites of local areas and near areas were very close, and the main cell types and their distributions in these two tissues were extremely similar (Extended Data Fig. 1d). All these cells were manually annotated based on their specific marker genes, which showed significant specificity in different cell populations (Fig. 1b and d, and Supplementary Data 3).

For ST-seq data, an average of 2,908 spots per slice were obtained (range: 2,097 to 4,268 spots), as detailed in Extended Data Fig. 1e and Supplementary Data 4. Using the same marker gene list as used in scRNA-seq, we calculated specific cell meta-modules for each spot and manually annotated them (Fig. 1d, e, and Supplementary Data 5). Unlike in scRNA-seq, in ST-seq data, the expression of marker genes in immune-related spots showed more overlap. Therefore, we calculated the Shannon entropy for different types of spots in each slice. The analysis results showed that the Shannon entropy of epithelial cell spots was generally lower in all slices (Extended Data Fig. 1f), indicating higher consistency in gene expression patterns among epithelial cell spots and relatively higher purity of epithelial cells. In most ST data, the median percentage of epithelial cell spots was 45.42% (range: 16.59%-62.67%), followed by fibroblast spots with a median percentage of 16.76% (range: 2.637%-36.785%). In contrast, fewer immune cell spots were obtained in ST data, especially T cell spots (5.77%-14.57%) (Fig. 1f, g).

Subtype specific energy metabolism and neuronal signaling pathways in MBC cells

Given the limited research and data on MBC cells, in this study, we focused on these critical malignant epithelial cells to deeply explore the molecular biological characteristics of MBC at the cellular level. We used the numbat²⁴ tool, a somatic haplotype-aware analysis method based on somatic copy number variation (CNV), to infer genomic aberrations from the scRNA-seq data of each patient. The analysis results showed that MBC patients exhibited extremely high genomic heterogeneity. For example, we observed minimal genomic alterations in patients P1, P2, P7, P6, and P9, while patients P8 and P12 showed mainly chromosome amplifications, and patients P4, P3, P5, and P11 exhibited mainly chromosome deletions. Patients P10 and P13 showed a diversity variety of genomic alteration types (Supplementary Fig. 2). Through numbat analysis, we identified a total of 87,218 malignant tumor cells (85.14% of total epithelial cells) in all scRNA-seq data, with the highest proportion in lymph areas samples (99.50%), followed by local areas tissues (90.70%), and the lowest in near areas tissues (65.50%) (Expanded Fig. 2a, b).

To further explore the heterogeneity of tumor cells in MBC, we classified these cancer cells into six different subtypes based on functional genes associated with cancer development, including neuroendocrine-like cancer cells (GFRA1, NF1, GRIA2)^25–27, proliferation-like cancer cells (CD24, TUBB3, TCIM), metastasis-like cancer cells (RNF181, MDK, BST2), invasion-like cancer cells (MKI67, TOP2A, NDC80), immune-like cancer cells (CD74, PTPRC, PARP14), and EMT-like cancer cells (VIM, FN1, MMP2) (Fig. 2a, b, and Supplementary Data 6). Overall, proliferation-like cells exhibited higher Shannon diversity than other subtypes in MBC (Expanded Fig. 2c), and the Shannon diversity of different subtypes varied among different patients (Expanded Fig. 2d).

In current clinical practice, the treatment strategies for MBC often reference those for FMBC, primarily because the majority of MBC patients are hormone receptor-positive, similar to postmenopausal ER-positive and/or PR-positive FMBC^28,29. Our research data shows, at the single-cell level, genes associated with FMBC, such as ESR1, PGR, and PRLR, are highly enriched in neuroendocrine-like cancer cells (Fig. 2b). Additionally, we found significant differences in the number of specific marker genes among different subtypes of cancer cells, with neuroendocrine-like cancer cells having the most abundant unique marker genes (expressed in over 75% of the population), while proliferation-like cancer cells did not exhibit population-specific marker genes (Fig. 2c and Supplementary Data 5). This phenomenon may be due to the high Shannon diversity within the proliferation-like cell population, leading to less specific expression of marker genes.

Furthermore, we calculated cell cycle scoring to assess the proliferative capacity of different subtype cancer cells, and the results showed that invasion-like cancer cells exhibited the strongest proliferative ability (Fig. 2d). To explore the representative biological functions of different subtype cells, we conducted Gene Ontology (GO) enrichment analysis. The analysis results indicate that neuroendocrine-like cancer cells are closely associated with regulating circadian rhythm events, which may suggest their involvement in inducing and/or promoting malignant events such as distant metastasis of tumor cells^30–32 (Fig. 2e).

We defined four tumor-associated meta-module signal scores, including neurotrophic, angiogenesis, glycolysis, and fatty acid metabolism. To validate the biological and clinical significance of these four signal scores, we analyzed their correlation with male cancers in the TCGA pan-cancer dataset. The results showed that these four signal scores were associated with poor prognosis in male malignant tumors (Supplementary Fig. 3).

Next, we analyzed the expression levels of these signal scores in six different subtypes of cancer cells in MBC. We found that invasion-like and metastasis-like cancer cells mainly rely on glycolysis for energy metabolism, while neuroendocrine-like cancer cells primarily depend on fatty acid metabolism. Furthermore, EMT-like, proliferation-like, and neuroendocrine-like cancer cells, which are associated with tumor proliferation and metastasis, exhibited highly active neurotrophic and angiogenesis signals (Fig. 2f, Supplementary Data 8).

We further observed the spatial distribution of these signals within solid tumors and found that the neurotrophic, glycolysis, and fatty acid metabolism signals were more intense in areas with dense cancer cells, while the angiogenesis signal was widely distributed throughout the tumor tissue (Fig. 2g-m, Supplementary Fig. 4, Supplementary Fig. 5). At the single-cell level, neurotrophic and angiogenesis signals showed positive correlations among different subtypes of tumor cells (Fig. 2o, Extended Data Fig. 2e), and at the sample level, the intensity of neurotrophic signals was positively correlated with the number of cancer cells (Extended Data Fig. 2f). Previous studies^33,34 and current evidence suggest that neurotrophic signals may be involved in regulating tumor angiogenesis and the proliferation of cancer cells in MBC.

To further validate these observations, we analyzed the correlation between neurotrophic and angiogenesis signals in 29 types of male cancers in the TCGA database, particularly finding significant correlations in breast cancer (BRCA), prostate cancer (PRAD), and pancreatic adenocarcinoma (PAAD) (Fig. 2p, q and Supplementary Fig. 6). High-intensity neurotrophic and angiogenesis signals were closely associated with poor prognosis in 14 types of male malignant tumors (Supplementary Fig. 7).

Pseudo-evolution tree of cancer cell Subtypes

To explore the evolutionary trajectories of cancer cells in MBC, we conducted pseudo-time analysis on the six cancer cell subtypes and constructed a pseudo-evolution tree. In the pseudo-evolution tree, we identified one main branch and four minor branches. The pseudo-evolution tree depicted that invasion-like, EMT-like, and metastasis-like cancer cells predominantly located in the main trunk of the pseudo-evolution tree, while neuroendocrine-like cancer cells highly enriched at the terminal branches (Fig. 3a), and proliferation-like cancer cells did not enrich at the main nodes of the pseudo-evolution tree. This finding may suggest that in the early stage of MBC, tumor development mainly focuses on local tissue invasion and begins to undergo EMT and metastasis. In the mid-term, it shifts towards to suppress immune cells for evading immune attacks, while in the late stage, it mainly transforms into Neuroendocrine-like cancer cells to regulate the cells in the TME. Additionally, we used the SComatic³⁵ tool, a somatic mutations detection method from sequencing reads, to calculate the somatic mutation burden in different subtype cancer cells (Supplementary Data 9). The results showed that at the population level, the mutation burden of neuroendocrine-like cells was significantly higher than that of other subtypes (Fig. 3b), but there were differences in different tissue spaces (Extended Data Fig. 3a).

Further analysis revealed that among the six subtypes of cells, the top ten genes with the highest mutation frequencies include ACACA, which plays a crucial role in fatty acid biosynthesis, ESR1 closely related to FMBC, and UTY gene located on the Y chromosome (Fig. 3c and Extended Data Fig. 3b). Previous studies have confirmed the importance of cancer gender dimorphism and the stability of the Y chromosome in the prognosis of male cancer patients^36,37. To explore the relationship between gene mutation burden on different chromosomes and different subtypes of cells in MBC, we analyzed the distribution of gene mutations on all chromosomes in different subgroups (Fig. 3d), and the gene mutation frequency in different chromosomes (Extended Data Fig. 3c). The results showed that neuroendocrine-like cells had the poorest chromosomal stability, while immune-like cells had relatively stable chromosomes. We particularly noticed that among the top ten genes with the highest mutation frequencies on the Y chromosome in MBC, UTY had the highest mutation burden, and the NLGN4Y gene related to neurotrophic factors also showed a high mutation burden in MBC (Fig. 3e). Further analysis on male-related malignant tumors (such as BRCA and PRAD) in TCGA revealed that the UTY gene may affect the occurrence and development of MBC by influencing the intensity of tumor angiogenesis signaling (Fig. 3f). This result suggests that the UTY gene may be a key risk factor in MBC patients, which is an important factor independent of previously discovered in FMBC.

Regulation of different subtypes of cancer cells by neuroendocrine-like cancer cells in MBC

We further analyzed the area distribution differences of different subtypes of tumor cells in MBC. In local areas, neuroendocrine-like, metastasis-like, and proliferation-like cells dominate, while in near areas, proliferation-like cells and EMT-like cells predominate. In lymph areas, immune-like and invasion-like cells are predominant (Fig. 4a). Additionally, the heterogeneity of cancer cells exhibits significant differences in different sample spaces. Invasion-like and immune-like cells show higher Shannon diversity in near areas, while neuroendocrine-like, metastasis-like, and proliferation-like cells exhibit higher diversity in lymph areas (Fig. 4b). EMT-like cells show the highest diversity in local areas.

To explore how the concentration of neurotrophic factors in tumor tissue alters the metabolic capabilities of tumor cells and promotes angiogenesis, we utilized the Compass³⁸ tool, a single-cell metabolism profiling algorithm, to conduct metabolic functional analysis of cancer cell subpopulations in different tissue compartments. The results indicate that, in MBC, cancer cells from different subgroups exhibit higher activities in amino acid metabolism, steroid metabolism, glycolysis, and fatty acid metabolism in local areas (Fig. 4C, Extended Data Fig. 4a, Supplementary Data 10), consistent with previous research findings suggesting a high enrichment of fatty acid metabolism signals in tumor-dense regions. We speculate that this phenomenon may be associated with the highly enriched neurotrophic signals and angiogenesis signals in the tumor regions of MBC. Therefore, we analyzed the cell-cell communications among these six subtypes of cancer cells to identify the pathways involved in their interactions. The result suggested that in MBC samples, neuroendocrine-like cancer cells extensively produce and receives neurotrophic signals (such as NRG and NPY signals) and growth factor-related signals (such as FGF signals), promoting the development of the population. Additionally, neuroendocrine-like cancer cells also synergistically promote the process of EMT in cancer cells, through the SEMA3 neurotrophic pathway and the angiogenesis-related pathway ANGPTL (Fig. 4d, Extended Data Fig. 4b-f).

We found that these signaling pathways of cell-cell communications exhibit spatial distribution differences, with the SEMA3 signaling pathway highly enriched in the core region with high cancer cell density, while the ANGPTL signaling pathway is highly enriched in the peripheral region with low cancer cell density (Fig. 4i, Extended Data Fig. 4g, Supplementary Fig. 8). Regardless of the spatial distribution of these signals, their intensity consistently increases as they approach tumor cells (Figure j, k, Extended Data Fig. 4h-j). Further analysis revealed that the average intensities of SEMA3 and ANGPTL signaling pathways among MBC samples are highly positively correlated with the proportion of EMT-like cells (Fig. 4l), further confirming the potential critical role of SEMA3 and ANGPTL signaling pathways in promoting the process of EMT in MBC. Validation results from male pan-cancer tissue samples from TCGA indicate that the intensities of SEMA3 signaling and receptor signaling are higher in metastatic male cancer patients than in non-metastatic patients (Fig. 4m). In the male pan-cancer metastatic population, the intensity of the SEMA3 signaling pathway is significantly associated with poor prognosis, especially in pancreatic adenocarcinoma (PAAD) and esophageal carcinoma (ESCA) (Fig. 4n-p).

Characterization of T cell subtypes in MBC and their cell-cell communications with cancer cells

Tumor-infiltrating T cells are core players in the TIME and are closely associated with clinical outcomes. Immune checkpoint blockade (ICB) has achieved tremendous success clinically, but its efficacy varies significantly across different types of cancers³⁹. In order to gain a deeper understanding of the interactions between different subtypes of cancer cells and T cell in MBC, we conducted clustering analysis of T cells. We identified a total of 29 distinct cell clusters, categorized into 3 major T cell subtypes and 9 minor T cell subtypes, defined by their specific expression of marker genes (Fig. 5a, b, Supplementary Data 11). Among the minor subtypes are naïve T cells, cytotoxic T cells, Treg cells, NK/NKT cells, inhibitory T cells, exhausted T cells, and proliferating T cells. Notably, proliferating T cells account for the smallest proportion (0.43%), while naïve T cells are the most abundant (43.13%). We found significant differences in gene expression and location distribution of different T cell subtypes in MBC, for instance, lymph areas are predominantly enriched with naïve T cells; in local areas, minor T cell subtypes are more abundant, with proliferating T cells mainly concentrated (Fig. 5b, c). Within the major T cell subtypes, CD4, CD8, and NK/NKT cells exhibit distinct biological functions in different sampling areas, primarily determined by differential expression of their marker genes. In local areas, the three major T cell subtypes show similar expression patterns of NEAT1 and COX6C genes (Fig. 5d). Additionally, we conducted metabolic analysis of T cells in local areas and others (Supplementary Data 12), assessing differences in metabolic capabilities of different T cell subtypes in different areas. The results show that the metabolic capacity of the three major T cell subtypes in local areas is generally lower than in other areas, especially in terms of fatty acid metabolism and anaerobic glycolysis (Fig. 5e, Extended Data Fig. 5a). To explore the interactions between different cancers and T cells in MBC, we conducted cell-cell communications analysis of all T cells and six subtypes of cancer cells in different areas (Fig. 5f-q, Extended Data Fig. 5b-j). We found that in local and near areas, all six major cancer cell subtypes and all minor T cell subtypes exhibit strong intercellular interactions, while in lymph areas, cancer cells primarily interact with immune cells through the MIF signaling pathway, which shows strong signal strength in all three areas (Fig. 5f, j, n, Extended Data Fig. 5b, e, h). Furthermore, we observed that in local areas, immune-like cells play the role of "Sender" in cell-cell communications signaling pathways (Fig. 5i), while in near and lymph areas, they act as “Mediator” regulating the communication between other subtypes of cancer cells and T cells (Fig. 5m, q).

Neuroendocrine-like cells work with TAMs to drive EMT and angiogenesis in MBC

Tumor-associated macrophages (TAMs) play crucial roles in TIME^40,41, participating in tumor progression, promoting angiogenesis, and immune suppression, closely associated with adverse prognosis and/or clinical events in solid tumor patients. To gain insight into the role of TAMs in MBC, we clustered TAMs into 21 distinct cell clusters and defined these clusters into six major subtypes of TAMs based on specific marker genes (Supplementary Data 13), including Mph SPP1+, Mph CCL4+, Mph MERTK+, Mph S100A9+, Mph STNM1+, and Mph CXCL9+. The results showed that in MBC, the proportion of Mph SPP1 + cells, associated with promoting tumor progression⁴², was the highest in TIME (total proportion 45.45%, 51.4% in local areas), while the proportion of Mph CXCL9 + cells, associated with inhibiting tumor progression⁴³, was the lowest (total proportion 2.02%, 1.48% in local areas) (Fig. 6a, b). We further investigated the functional differences of subtype of TMAs and found that TMAs enriched different biological functions in different areas (Fig. 6c). Similarly, we found that the metabolic capacity of TAMs in local areas was generally lower than that in others (Supplementary Data 14); due to the small number of Mph CXCL9 + and Mph STNM1 + cells, they were not included in the inferred metabolic functions (Fig. 6d, Extended Data Fig. 6a). The interaction between TAMs and cancer cells greatly affects the occurrence of tumor-related malignant events. Therefore, we conducted cell-cell communications analysis of TAMs and cancer cells in local areas of MBC.

The results showed that six different subtypes of cancer cell promote the growth, proliferation, and differentiation of different subtypes of TAMs through CSF and GDF signaling pathways, especially Mph MERTK + cells (Fig. 6e, f, Extended Data Fig. 6b, c). In addition, tumor cells suppress TAMs through the MIF intercellular signaling pathway to achieve immune escape (Fig. 6g, Extended Data Fig. 6d). Interestingly, although TAMs send high-intensity TNF-related signals, these signals are not received by cancer cells in MBC (Fig. 6h, Extended Data Fig. 6e). In MBC, cancer cells promote the development of the TAMs community by secreting cytokines. Similarly, Mph MERTK + cells promote the proliferation and development of neuroendocrine-like cells by secreting the signal of IGF, and promote EMT of cancer cells through the HGF signaling pathway (Fig. 6i, j, Extended Data Fig. 6f, g). To validate this conclusion, we calculated the combined signal strength of HGF and IGF in the TCGA male malignant tumor patient dataset and found a strong correlation between the combined signal strength of HGF and IGF and the signal strength of neurotrophic signal and angiogenesis signal in metastatic and non-metastatic male malignant tumor patients, and the combined signal strength of HGF and IGF was closely related to the adverse prognosis of male malignant tumors (Fig. 6k-m, n).

Neuroendocrine-like cells act as signal distributors in the regulation of CAFs to promote cancer cell EMT and invasion

Cancer stroma plays a significant role in promoting tumor proliferation, recurrence, and treatment resistance⁴⁴. Among all the stromal cells, in TIME, cancer-associated fibroblasts (CAFs) are abundant and closely associated with cancer progression⁴⁵. To understand how CAFs regulate tumor cells in MBC, we clustered fibroblasts into 23 clusters and annotated four major subtypes of CAFs: inflammatory CAFs, vascular CAFs, EMT-like CAFs, and antigen CAFs (Fig. 7a, b, Supplementary Data 15). By analyzing the marker genes of these major CAFs subtypes, we calculated meta-module scores and found that, except for vascular CAFs, which displayed a more diverse meta-module, the meta-module expression of the other three CAFs subtypes was highly specific (Fig. 7c). Using ssGSEA, we compared the enrichment differences of marker genes of CAFs subtypes in biological signaling pathways in different areas and found that the impact of different environment on the functional expression of CAFs was more significant than the differences between subtypes (Fig. 7d). Additionally, we found that CAFs in near areas had stronger comprehensive metabolic capacity compared to those in local areas. This suggests that the density of cancer cells may affect the biological functions and metabolic capacity of CAFs (Fig. 7e, Extended Data Fig. 7a, Supplementary Data 14).

Further analysis of the cell-cell communications between CAFs and six different subtypes of cancer cells in MBC revealed that proliferation-like cancer cells affected the development of vascular CAFs through the EDN signaling pathway (Fig. 7f, Extended Data Fig. 7b), while vascular CAFs promoted the formation of cancer cell EMT process through the VEGF signaling pathway (Fig. 7g, Extended Data Fig. 7c). Simultaneously, immune-like cancer cells and EMT-like cells affected the development of CAFs community through the SPP1 signaling pathway (Fig. 7h, Extended Data Fig. 7d). Furthermore, CAFs community affected the proliferation and development of proliferation-like cancer cells, neuroendocrine-like cells, and other cancer cells through the TWEAK signaling pathway and NPY signaling pathway (Fig. 7i, j, Extended Data Fig. 7e). In the cancer stroma microenvironment of MBC, the NPY signal produced by CAFs and other tumor cells is mainly received by neuroendocrine-like cells and transmitted to other subtypes of cancers cells (Fig. 7k).

In this study, we delved into the functionality and distribution of major tumor cell subtypes in MBC, as well as their interactions with immune cells, TAMs, and CAFs in the TME. This work represents the most extensive and comprehensive single-cell multi-omics analysis of MBC to date, providing valuable data resources for understanding the cellular and molecular characteristics of MBC.

By analyzing data from scRNA-seq and ST-seq, we observed a lower degree of immune cell infiltration⁴⁶ in MBC, and across different slides, we found that the meta-module scores associated with epithelial cells exhibited the strongest specificity. Moreover, the Shannon entropy of epithelial-related spots was lower than that of immune-related spots. These may be attributed to the tendency of cancer cells in MBC to form high-density tumor masses, resulting in high purity of cancer cells within each spot, thereby preventing immune cells from penetrating into the interior of the tumor and instead distributing around the periphery of the tumor. In our study, we classified cancer cells in MBC into six subtypes, among which neuroendocrine-like cancer cells were the most abundant and closely associated with the occurrence and development of MBC. Additionally, the expression of hormone receptor genes PGR and ESR1, which are related to MBC treatment, was highly enriched in neuroendocrine-like cancer cells.

In the metabolic analysis at the single-cell level, we found differences in the metabolic capabilities of cells within TME across different environment. The metabolic capacity of cancer cells in local areas higher than in others, while other cell types in TME exhibited the opposite trend. This phenomenon may be attributed to the high density of cancer cells and abundant vasculature in the core region of the tumor, allowing cancer cells to gain a "competitive advantage" in this area and monopolize nutrients, thereby further impeding the infiltration of immune cells.

In the cell-cell communications analysis of cancer cell subtypes, we found that neuroendocrine-like cells engage in communications with different subtypes of cancer cells through neuro-related signaling pathways, thereby promoting tumor proliferation, invasion, metastasis, immune suppression, EMT, and nourish themselves. Additionally, other cancer cell subtypes also promote the activity of neuroendocrine-like cells through neuro-related signaling pathways, particularly NRG. In cancer cells pseudo-evolution tree, neuroendocrine-like cells highly enriched at the terminal branches, suggesting that in the process of MBC development, cancer cells may transform into neuroendocrine-like cells through autocrine neuro-related signaling factors, thereby secreting more neuro-related signaling factors and promoting the dissemination and proliferation of the tumor cell population. Therefore, we believe that in MBC, Neuroendocrine-like cells and their neuro-related signaling pathways may drive a malignant cycle of tumor occurrence and development.

In our study, we collected data from different areas of MBC patients and observed an interesting biological phenomenon: in MBC, areas with a high abundance of cells of a certain subtype have a lower Shannon diversity of cells of that subtype. Therefore, we speculate that in MBC, tumor cells of different subtypes play primary roles in their "exclusive" domains (for example, the "exclusive" domain of immune-like cells is lymph), and due to their specialized functions in these "exclusive" domains, the transcriptional heterogeneity of these subtype cancer cells is relatively low. For example, in local areas, immune-like cells act as "attackers," actively sending immune-related signals to interact with immune cells; while in their "exclusive" domains, they play more of a "coordinator" role, reducing immune cell attacks on other subtype cancer cells by modulating immune-related signaling pathways.

TAMs are one of the most important components of TIME, and the distribution of subtypes is closely associated with tumor progression, immune evasion, and other tumor-related events. Our research data indicate that in TIME of MBC, MERTK + macrophages act as the primary "partners" of neuroendocrine-like cancer cells, jointly promoting the malignant development and EMT of tumor cells. Cancer cells induce and promote the formation of MERTK + macrophages through CSF and GDF signals, which subsequently facilitate the development of neuroendocrine-like cancer cells via the IGF signaling pathway, thereby driving the proliferation and dissemination of the entire cancer cell population.

The cancer stromal microenvironment is a critical component of TME, with CAFs playing a vital role as one of the key drivers of tumor progression and angiogenesis. Our research data reveal a complex network of cell-cell communications, akin to a "trade network", between CAFs and different subtypes of cancer cells in MBC. Through this "trade network", CAFs and cancer cells mutually promote each other's development. Within this network, neuroendocrine-like cancer cells act as the primary "agents" of the NPY signaling pathway, participating in the distribution of NPY signals generated by CAFs to other subtypes of tumor cells. Simultaneously, different subtypes of tumor cells promote the development of different CAFs subtypes through their respective cell-cell communications.

Due to the extreme rarity of MBC, the samples available for experimental analysis are extremely scarce. Therefore, this study did not conduct any genomic sequencing research related to MBC. Nevertheless, we still attempted to extract potentially useful information from the raw fastq data of scRNA-seq using the de novo mutation calling method. In this exploratory study, we found that gene mutations on the Y chromosome may be associated with tumor-related adverse events in MBC. Among the 5 genes with frequent mutations on the Y chromosome, KDM5D has been proven to be a key factor leading to poor prognosis in male colorectal cancer patients^47,48. Our statistical analysis showed that UTY had the highest mutation frequency in MBC, and the high expression of the UTY gene was closely related to the angiogenesis-related signaling pathways in MBC patients and prostate cancer patients in the TCGA database. These signaling pathways are closely related to the prognosis of male malignant tumors, suggesting that the UTY gene may be associated with the clinical prognosis of MBC patients.

In summary, our study utilized scRNA-seq and ST-seq technologies to comprehensively analyze the cellular and molecular characteristics of MBC, spanning transcriptomic, metabolomic, and genomic levels. Our findings revealed there are lower immune infiltration in MBC patients, as well as cancer cells exhibiting higher metabolic activity than others in situ TME. Particularly, neuroendocrine-like cancer cells, as the predominant cancer cell subtype in MBC, demonstrated in cancer cells pseudo-evolution tree and cell-cell communications analyses that they may mature into tumor cells induced by neuro-related signaling pathways either through autocrine or paracrine mechanisms from other subtypes of cancer cells. Neuroendocrine-like cells act as core "organizers" in TME, collaborating with other cells in the microenvironment to collectively drive tumor development and progression.

Patient sample collection

This study was reviewed and approved by the Ethics Committee of the Medical Ethics Committee of Affiliated Hospital of Qingdao University (Ethics code: QYFY-WZLL-28676). The study was done by the principles of the Declaration of Helsinki and the International Conference on Harmonization Guidelines for Good Clinical Practice. All patients whose human specimens were collected provided written informed consent. All the authors approved the manuscript for submission for publication and vouch for the accuracy and completeness of the data reported. From January 2021 to October 2023, we collected surgically resected primary breast cancer tissues and lymph node metastases from MBC patients treated in hospitals across China. These samples were analyzed using scRNA-seq and ST-seq.

scRNA-seq library preparations

Tissues were transported in sterile culture dishes containing 10 ml of 1x Dulbecco's Phosphate-Buffered Saline (DPBS; Thermo Fisher, Cat. No. 14190144) on ice to remove any residual storage solution. The tissues were then minced on ice. We used a dissociation solution containing 0.25% Trypsin (Thermo Fisher, Cat. No. 25200-072) and 10 µg/mL DNase I (Sigma, Cat. No. 11284932001) dissolved in PBS with 5% Fetal Bovine Serum (FBS; Thermo Fisher, Cat. No. SV30087.02) to digest the tissues. The tissues were dissociated at 37°C with a shaking speed of 50 rpm for approximately 40 minutes. We periodically collected the dissociated cells every 20 minutes to enhance cell yield and viability. Cell suspensions were filtered through a 40 µm nylon cell strainer, and red blood cells were lysed using 1x Red Blood Cell Lysis Solution (Thermo Fisher, Cat. No. 00-4333-57). The dissociated cells were then washed with 1x DPBS containing 2% FBS. Cell viability was assessed by staining with 0.4% Trypan blue (Thermo Fisher, Cat. No. 14190144) using the Countess® II Automated Cell Counter (Thermo Fisher).

10X library construction and sequencing

Saturation into Gel Beads-in-emulsion (GEMs), ensuring each cell was paired with a bead. Upon exposure to cell lysis buffer, polyadenylated RNA molecules hybridized to the beads. These beads were then collected into a single tube for reverse transcription, where each cDNA molecule was tagged at the 5’ end—corresponding to the 3’ end of a messenger RNA transcript—with a UMI and a cell label denoting its cell of origin. Subsequently, the beads underwent second-strand cDNA synthesis, adaptor ligation, and universal amplification. Sequencing libraries were prepared from randomly interrupted whole-transcriptome amplification products to enrich for the 3' ends of transcripts associated with the cell barcode and UMI. All remaining procedures, including library construction, were conducted according to the standard protocol provided by the manufacturer (CG000206 Rev D). The sequencing libraries were quantified using a High Sensitivity DNA Chip (Agilent) on a Bioanalyzer 2100 and the Qubit High Sensitivity DNA Assay (Thermo Fisher Scientific). The libraries were then sequenced on a NovaSeq 6000 (Illumina) using 2x150 chemistry.

Spatial sequencing library preparation

Formalin-fixed, paraffin-embedded (FFPE) samples that passed RNA quality control (DV200 > 30%) were used for spatial transcriptomic analysis and sequencing. Sections 5 microns thick were mounted onto a Visium Gene Expression slide (10X Genomics), baked at 42°C for 3 hours, and dried overnight at room temperature in a desiccator. For deparaffinization, the slide was incubated at 60°C for 2 hours, immersed in xylene, and rehydrated through an ethanol gradient. Hematoxylin and eosin (H&E) staining was performed using Mayer’s hematoxylin (Millipore Sigma), bluing reagent (Dako, Agilent), and alcoholic eosin (Millipore Sigma). After staining, the slides were scanned under a microscope. Deparaffinized sections were decrosslinked using 0.1N HCl and TE Buffer (pH 9.0) to release RNA trapped by formalin. The slides were then incubated with a Human whole transcriptome probe panel and transferred to Cytassist (10X Genomics). The Human whole transcriptome probe panel, consisting of three pairs of specific probes (5’ containing Small RNA Read 2S and 3’ containing poly-A), mostly hybridized to RNA. These probe pairs were then ligated to seal the junctions between them, forming single-stranded ligation products. The samples underwent RNase treatment and permeabilization to release the ligation products. The poly-A segment of these products was captured by the poly(dT) regions of the capture probes precoated on the Visium slide, which also included an Illumina Read 1, spatial barcode, and unique molecular identifier (UMI). The probes were extended to produce spatially barcoded, ligated probe products, which were then released from the slide for indexing via Sample Index PCR, followed by final library construction and sequencing. The Visium Spatial Gene Expression libraries consisted of Illumina paired-end sequences flanked with P5/P7 adaptors. The 16-bp Spatial Barcode and 12-bp UMI were encoded in Read 1, while Read 2S was used to sequence the ligated probe insert.

scRNA-seq data pre-processing

The raw data was analyzed using the Cell Ranger (Version 7.1.0) pipeline, with gene count data generated using the default and recommended parameters. Align the FASTQ output obtained from sequencing data with the GRCh38 reference genome using the STAR algorithm⁴⁹. We used Seurat (Version 5.0.1) R package to do all subsequent analyses. To eliminate the influence of low-quality cells such as empty droplets and multiplets, cells with expressed genes < 300 or > 10000 were excluded.

We use in Seurat:: NormalizeData, Seurat:: FindVariableFeatures and Seurat:: ScaleData functions to normalize all data. Seurat:: IntegrateLayers function was used to remove technical variability (method = RPCAIntegration). The top 50 PCs were calculated using the Seurat::RunTSNE function. A tSNE dimensional reduction was performed on the scaled matrix using the top 50 PCA components to obtain a two-dimensional representation.

Identification of malignant cancer cells

To distinguish between malignant and non-malignant cells in MBC, we first identified malignant epithelial cells in MBC using specific marker genes EPCAM, KRT8, and KRT18. Subsequently, we employed the Numbat method to classify all cells as malignant based on their patient origin.

Relative Meta-module Score

We first used the Seurat::FindAllMarkers function to calculate the differential expression marker genes for each CAFs subtype and filtered out the genes with an Adjusted P-value < 0.05 as the candidate gene set. Then, based on the different molecular types, we filtered out the top100 genes (sorted by fold-change) as the meta-gene list. The meta-module score for each molecular type was calculated using Seurat::AddModuleScore. Finally, the relative meta-module score was calculated and visualized using the scrabble::plot_hierarchy (Version 1.0.0) function⁵⁰.

Pathway enrichment analyses

Gene ontology enrichment analysis (GO)⁵¹. To identify enriched Gene Ontology (GO) biological processes based on differentially expressed genes (DEGs), analysis was performed on DEGs from each cell type using clusterProfiler::gseGO (Version 4.8.1) R package. Using org.Hs.eg.db annotation Package (Version 3.17.0). Gene set enrichment analysis (GSEA). We used GseaVis::gseaNb (Version 0.0.8) R package (https://github.com/junjunlab/). For input, we used pre-ranked gene lists by avg_log2FC calculated from Searat::FindAllMarkers. Kyoto Encyclopedia of Genes and Genomes (KEGG)⁵². We used ClusterGVis::enrichCluster (Version 0.1.0) R package. Using org.Hs.eg.db annotation Package (Version 3.17.0). The ssGSEA analysis was conducted using the R singleseqgset package (Version 0.1.2.9000). We used the human hallmark gene sets from the msigdbr package⁵³ (Version 7.5.1) as the gene sets of interest for enrichment analysis. Differences were considered statistically significant at P < 0.05.

Single-cell level metabolic analysis

We grouped cells by different subtypes and characterized cellular metabolic states at the single-cell level using the Compass algorithm on scRNA-seq data. Given that the Compass algorithm requires substantial computational resources, we utilized the Micropooling technique to reduce computational costs. The size of the Micropooling varied across different groups, ensuring that each group had data for at least 50 clusters. The final classification of each cluster was determined based on the proportion of cells originating from the same source within the group; a cluster was deemed to originate from a particular cell type if cells from that source constituted over 70% of the cluster. We then performed log2 normalization on the Compass result data and conducted differential analysis of the metabolic expression matrix using the Wilcoxon rank-sum test. Differences were considered statistically significant at P < 0.05, and the tendency of differential expression was assessed using Cohen's d index.

Single-cell level somatic mutations Calling

1) Cell level: The single-cell level somatic mutations analysis we performed based on Numbat R package. Used cellsnp-lite for generating SNP pileup data and eagle2 for phasing and the high polymorphic regions detected using WGS data from the 1000 Genomes Project⁵⁴ mapped to the GRCh38 build of the human reference genome. Next, we ran Numbat get the VCF files for downstream analysis. All parameters and other config were set as described in the Numbat documentation. All output VCF files were annotated using the snpEff⁵⁵ software, which uses the GRCh38 reference genome file.

2) Cell type level: The cell type level somatic mutations analysis we performed based on the protocol of SComatic. First, we annotate the cell type of each MBC cancer cells and output it as a list (1st column is cell-barcode, 2nd column is cell identity). We then split the BAM files from scRNA-seq into separate files based on cell identity according to the list. Next, we ran SComatic to get the VCF files for downstream analysis. All parameters, configurations, and reference data use the default values or default files provided in the SComatic. All output VCF files were annotated using the snpEff software, which uses the GRCh38 reference genome file.

Cell-cell communication analysis

Cell–cell communication analysis was conducted with the scRNA-seq data by using the cellchat⁵⁶ software (Version 2.1.2). Specifically, only receptors and ligands expressed in > 1% of cells were further evaluated, while a cell–cell communication was considered nonexistent if the ligand or the receptor was unmeasurable.

Pseudo-evolution tree

We performed pseudo-time analysis on malignant epithelial cells using the R package dyno⁵⁷ (Version 0.1.2). Due to the large volume of data, in order to reduce computational cost and memory usage, we randomly selected 20,000 cells for analysis. All parameters and settings were kept at their default values. Finally, we utilized PAGA-TREE⁵⁸ for pseudo-time analysis and constructing the pseudo-evolution tree.

TCGA data analysis

We utilized the R package easyTCGA (Version 0.0.4.200.0) to download the TCGA pan-cancer gene expression matrix⁵⁹ and sample clinical information. Functional module scoring was performed for different functional modules using sets of genes associated with specific functions, which include: Neurotrophic (NGF, BDNF, NTF3, CNTFR, VEGFA, NTRK2, NTRK3, SEMA3B, SEMA3C), Angiogenesis (VEGFA, FGF13, PDGFB, ANGPT1, ANGPT2, HIF1A, MMP2, MMP9, EPHB4, EFNB2, TIE1), Glycolysis (SLC2A1, HK2, PFKM, LDHA, PKM, GAPDH, ENO1, PGK1, ALDOA, TPI1), Fatty acid metabolism (ACACA, FASN, SCD, CPT1A, PPARA, PPARG, SREBF1, DGAT1, LPL, LIPE, PNPLA2). Kaplan–Meier survival curves were plotted using ggsurvplot (Version 0.4.9) function in the R package Survminer (Version 0.4.9).

Data and Code availability

All dataset, data processing codes, and configuration files are stored on GitHub (https://github.com/yuansh3354/scMBC). All scRNA-seq resource in this study are available in the China National Center for Bioinformation (BioProject ID PRJCA025336, accession number HRA010393).

Acknowledgments

Author Contributions

Y.S. conceived and designed the project. X.Z. collected the human samples and clinical information with help from S.C., B.X., H.W., J.C. Z.Y. S.C. Z.Z S.Z. Y.C. Y.Z. L.Z. and J.W.. Y.S. performed experiments and bioinformatic analyses. Y.S. discussed and interpreted the data. Y.S. wrote the manuscript.

Brinton, L. A. et al. Prediagnostic Sex Steroid Hormones in Relation to Male Breast Cancer Risk. J Clin Oncol 33, 2041–2050 (2015).
AlFehaid, M. Male Breast Cancer (MBC) - A Review. Pol Przegl Chir 95, 24–30 (2023).
Giordano, S. H. Breast Cancer in Men. N Engl J Med 378, 2311–2320 (2018).
Massarweh, S. A. et al. Molecular Characterization and Mortality From Breast Cancer in Men. JCO 36, 1396–1404 (2018).
Wang, F. et al. Overall Mortality After Diagnosis of Breast Cancer in Men vs Women. JAMA Oncol 5, 1589–1596 (2019).
Dmukauskas, M. et al. DISP-10. A SYSTEMATIC REVIEW OF SEX DIFFERENCES OF INCIDENCE AND OVERALL SURVIVAL IN 15 NON-REPRODUCTIVE CANCERS. Neuro-Oncology 24, vii129–vii129 (2022).
Haupt, S., Caramia, F., Klein, S. L., Rubin, J. B. & Haupt, Y. Sex disparities matter in cancer development and therapy. Nat Rev Cancer 21, 393–407 (2021).
Graham, F. Daily briefing: Y chromosome affects cancer risk. Nature d41586-023-02086–7 (2023) doi:10.1038/d41586-023-02086-7.
Severson, T. M. et al. Characterizing steroid hormone receptor chromatin binding landscapes in male and female breast cancer. Nat Commun 9, 482 (2018).
Hondermarck, H. Neurotrophins and their receptors in breast cancer. Cytokine & Growth Factor Reviews 23, 357–365 (2012).
Renz, B. W. et al. β2 Adrenergic-Neurotrophin Feedforward Loop Promotes Pancreatic Cancer. Cancer Cell 33, 75–90.e7 (2018).
Albo, D. et al. Neurogenesis in colorectal cancer is a marker of aggressive tumor behavior and poor outcomes. Cancer 117, 4834–4845 (2011).
Ayala, G. E. et al. Cancer-related axonogenesis and neurogenesis in prostate cancer. Clin Cancer Res 14, 7593–7603 (2008).
Hanahan, D. & Monje, M. Cancer hallmarks intersect with neuroscience in the tumor microenvironment. Cancer Cell 41, 573–580 (2023).
Winkler, F. et al. Cancer neuroscience: State of the field, emerging directions. Cell 186, 1689–1707 (2023).
Narayan, P. et al. Perineural invasion as a risk factor for locoregional recurrence of invasive breast cancer. Sci Rep 11, 12781 (2021).
Reszka, E., Przybek, M., Muurlink, O. & Pepłonska, B. Circadian gene variants and breast cancer. Cancer Letters 390, 137–145 (2017).
Hadadi, E. et al. Chronic circadian disruption modulates breast cancer stemness and immune microenvironment to drive metastasis in mice. Nat Commun 11, 3193 (2020).
Mohammadpour, H. et al. β2 adrenergic receptor-mediated signaling regulates the immunosuppressive potential of myeloid-derived suppressor cells. J Clin Invest 129, 5537–5552 (2019).
Peters, M. A. M. et al. Dopamine and serotonin regulate tumor behavior by affecting angiogenesis. Drug Resist Updat 17, 96–104 (2014).
Tang, F. et al. A pan-cancer single-cell panorama of human natural killer cells. Cell 186, 4235–4251.e20 (2023).
Sh, Y. et al. CaSee: A lightning transfer-learning model directly used to discriminate cancer/normal cells from scRNA-seq. Oncogene 41, 4866–4876 (2022).
Hao, Y. et al. Dictionary learning for integrative, multimodal and scalable single-cell analysis. Nat Biotechnol 42, 293–304 (2024).
Gao, T. et al. Haplotype-aware analysis of somatic copy number variations from single-cell transcriptomes. Nat Biotechnol 41, 417–426 (2023).
Esseghir, S. et al. A role for glial cell derived neurotrophic factor induced expression by inflammatory cytokines and RET/GFR alpha 1 receptor up-regulation in breast cancer. Cancer Res 67, 11732–11741 (2007).
Treanor, J. J. et al. Characterization of a multicomponent receptor for GDNF. Nature 382, 80–83 (1996).
Deshpande, K. et al. Neuronal exposure induces neurotransmitter signaling and synaptic mediators in tumors early in brain metastasis. Neuro Oncol 24, 914–924 (2022).
Ottini, L. et al. HER2-positive male breast cancer: an update. Breast Cancer (Dove Med Press) 2, 45–58 (2010).
Sun, H. et al. Single-cell transcriptome analysis indicates fatty acid metabolism-mediated metastasis and immunosuppression in male breast cancer. Nat Commun 14, 5590 (2023).
Diamantopoulou, Z. et al. The metastatic spread of breast cancer accelerates during sleep. Nature 607, 156–162 (2022).
Liu, C., Guo, L. & Fu, C. Circadian-rhythm-regulating hormones: Key factors to regulate breast cancer metastasis via circulating tumor cells. MedComm (2020) 3, e189 (2022).
Wang, J. et al. Circadian protein BMAL1 promotes breast cancer cell invasion and metastasis by up-regulating matrix metalloproteinase9 expression. Cancer Cell Int 19, 182 (2019).
Kuol, N., Stojanovska, L., Apostolopoulos, V. & Nurgali, K. Role of the Nervous System in Tumor Angiogenesis. Cancer Microenviron 11, 1–11 (2018).
Kermani, P. & Hempstead, B. Brain-derived neurotrophic factor: a newly described mediator of angiogenesis. Trends Cardiovasc Med 17, 140–143 (2007).
Muyas, F. et al. De novo detection of somatic mutations in high-throughput single-cell profiling data sets. Nat Biotechnol (2023) doi:10.1038/s41587-023-01863-z.
Müller, P. et al. Why loss of Y? A pan-cancer genome analysis of tumors with loss of Y chromosome. Comput Struct Biotechnol J 21, 1573–1583 (2023).
Brownmiller, T. et al. Y Chromosome LncRNA Are Involved in Radiation Response of Male Non-Small Cell Lung Cancer Cells. Cancer Res 80, 4046–4057 (2020).
Wagner, A. et al. Metabolic modeling of single Th17 cells reveals regulators of autoimmunity. Cell 184, 4168–4185.e21 (2021).
Zheng, L. et al. Pan-cancer single-cell landscape of tumor-infiltrating T cells. Science 374, abe6474 (2021).
Pittet, M. J., Michielin, O. & Migliorini, D. Clinical relevance of tumour-associated macrophages. Nat Rev Clin Oncol 19, 402–421 (2022).
Yang, L. & Zhang, Y. Tumor-associated macrophages: from basic research to clinical application. J Hematol Oncol 10, 58 (2017).
Bao, X. et al. An immunometabolism subtyping system identifies S100A9 + macrophage as an immune therapeutic target in colorectal cancer based on multiomics analysis. Cell Rep Med 4, 100987 (2023).
Bill, R. et al. CXCL9:SPP1 macrophage polarity identifies a network of cellular programs that control human cancers. Science 381, 515–524 (2023).
Chen, X. & Song, E. Turning foes to friends: targeting cancer-associated fibroblasts. Nat Rev Drug Discov 18, 99–115 (2019).
Park, D., Sahai, E. & Rullan, A. SnapShot: Cancer-Associated Fibroblasts. Cell 181, 486–486.e1 (2020).
Sun, H. et al. Single-cell transcriptome analysis indicates fatty acid metabolism-mediated metastasis and immunosuppression in male breast cancer. Nat Commun 14, 5590 (2023).
Abdel-Hafiz, H. A. et al. Y chromosome loss in cancer drives growth by evasion of adaptive immunity. Nature 619, 624–631 (2023).
Li, J. et al. Histone demethylase KDM5D upregulation drives sex differences in colon cancer. Nature 619, 632–639 (2023).
Dobin, A. et al. STAR: ultrafast universal RNA-seq aligner. Bioinformatics 29, 15–21 (2013).
Neftel, C. et al. An Integrative Model of Cellular States, Plasticity, and Genetics for Glioblastoma. Cell 178, 835–849.e21 (2019).
Wu, T. et al. clusterProfiler 4.0: A universal enrichment tool for interpreting omics data. Innovation (Camb) 2, 100141 (2021).
Kanehisa, M. & Goto, S. KEGG: kyoto encyclopedia of genes and genomes. Nucleic Acids Res 28, 27–30 (2000).
Liberzon, A. et al. The Molecular Signatures Database (MSigDB) hallmark gene set collection. Cell Syst 1, 417–425 (2015).
1000 Genomes Project Consortium et al. A map of human genome variation from population-scale sequencing. Nature 467, 1061–1073 (2010).
Cingolani, P. Variant Annotation and Functional Prediction: SnpEff. Methods Mol Biol 2493, 289–314 (2022).
Jin, S. et al. Inference and analysis of cell-cell communication using CellChat. Nat Commun 12, 1088 (2021).
Saelens, W., Cannoodt, R., Todorov, H. & Saeys, Y. A comparison of single-cell trajectory inference methods. Nat Biotechnol 37, 547–554 (2019).
Wolf, F. A. et al. PAGA: graph abstraction reconciles clustering with trajectory inference through a topology preserving map of single cells. Genome Biol 20, 59 (2019).
Goldman, M. J. et al. Visualizing and interpreting cancer genomics data via the Xena platform. Nat Biotechnol 38, 675–678 (2020).

There is NO Competing Interest.

SP1.pdf
Supplementary Figure 1. scRNA-seq feature plot group by sample. tSNE plot showing cell distribution among different samples.
SP2.pdf
Supplementary Figure 2. Numbat analyses genomic aberrations between different samples. Single-cell CNV landscape and reconstructed phylogeny among patients.
SP3.pdf
Supplementary Figure 3. Relationship between four meta-module scores and clinical prognosis. Relationship between four meta-module scores and overall survival of male malignant tumors in TCGA. Red represents high expression level, blue represents low expression level.
SP4.pdf
Supplementary Figure 4. Distribution of neurotrophic signal strength and angiogenesis signal strength in ST-seq data. Spatial feature plots show the distribution of neurotrophic signal strength and angiogenesis signal strength in each slice.
SP5.pdf
Supplementary Figure 5 Distribution of glycolysis signal strength and fatty acid metabolism signal strength in ST-seq data. Spatial feature plots show the distribution of glycolysis signal strength and fatty acid metabolism signal strength in each slice.
SP6.pdf
Supplementary Figure 6 . The relationship between neurotrophic signal strength and angiogenesis signal strength. Pearson correlation between the strength of neurotrophic signal and angiogenesis signal in TCGA male cancer patients.
SP7.pdf
Supplementary Figure 7. Relationship between the combined signal strength of neurotrophic and angiogenesis and prognosis of male malignant tumors. Survival plots showed the relationship between relative risk score and prognosis in male patients with TCGA malignancies. The relative risk is the combined signal strength of neurotrophic and angiogenesis.
SP8.pdf
Supplementary Figure 8. Spatial distribution of signal strength of cell-cell communications. Spatial feature plots show the signal strength of ANGPTL, FGF, NRG, and SEMA3 cell-cell communication among all slices.
Supplementarydata.xlsx
ED1.pdf
Extended Data Figure 1| scRNA-seq and ST-seq information. a, tSNE plots showing the distribution of patients. Dots represent individual cells. b, Stacked barplot showing proportions of major cell populations among different samples. Colors represent major cell populations. c, tSNE plots show the spatial distribution contour boundaries of major cell populations. d, tSNE plots showing the distribution of sampling areas. Dots represent individual cells. Pie charts show the proportion of cells in different areas. e, Bar chart shows the distribution of spots across all slices. f, Box plots show the Shannon Entropy of cell related spots in different slices.
ED2.pdf
Extended Data Figure 2| Cancer cell subtype information. a, Stack barplots show the distribution of malignant epithelial cells in different samples and sampling areas. b, Box plots show the Shannon diversity of all cancer cells, goup by subtypes. d, Box plots show the Shannon diversity of different cancer cell subtypes among all samples. e, Pearson correlation between the strength of neurotrophic signal and angiogenesis signal in all cancer cells. f, Pearson correlation between the relative strength of neurotrophic signal and total number of cancer cells. Dots represent individual samples.
ED3.pdf
Extended Data Figure 3| The genetic information of cancer cells. a, Boxplots showing the distribution of mutation burden among different cancer cells, group by sampling areas. b, Stack barplot shows the type of variations in top-10 mutated genes shared among different subtypes of cancer cells. c, Top-10 genes with the highest mutation burden in different chromosomes.
ED4.pdf
Extended Data Figure 4| Metabolism and cell-cell communications of different subtypes of cancer cells. a, Volcano plots show metabolic levels with differential expression, with red representing high expression in the local area and blue vice versa, group by different cancer subtypes. b-f, circle diagrams showing the cell-cell communications within each cell types, each plots represent different signaling pathways. g, Spatial feature plots show the signal strength of ANGPTL, FGF, NRG, and SEMA3 cell-cell communication. The round-tailed arrows point to areas rich in matrix or immune cells, and the straight arrows point to areas rich in cancer cells. h-j, Density map shows the distance between SEMA3 and ANGPTL signal strength and cancer cells in all slices.
ED5.pdf
Extended Data Figure 5| Metabolism and cell-cell communications of different subtypes of T cells. a, Volcano plots show metabolic levels with differential expression, with red representing high expression in the local area and blue vice versa, group by different T cell subtypes. b-j, circle diagrams showing the cell-cell communications within each cell types, each plots represent different signaling pathways.
ED6.pdf
Extended Data Figure 6| Metabolism and cell-cell communications of different subtypes of TAMs. a, Volcano plots show metabolic levels with differential expression, with red representing high expression in the local area and blue vice versa, group by different TAMs subtypes. b-g, circle diagrams showing the cell-cell communications within each cell types, each plots represent different signaling pathways.
ED7.pdf
Extended Data Figure 7| Metabolism and cell-cell communications of different subtypes of CAFs. a, Volcano plots show metabolic levels with differential expression, with red representing high expression in the local area and blue vice versa, group by different CAFs subtypes. b-g, circle diagrams showing the cell-cell communications within each cell types, each plots represent different signaling pathways.

Download PDF

Version 1

posted

You are reading this latest preprint version

scRNA-seq and spatial transcriptomics reveal neuroendocrine-like cancer cells promote angiogenesis and EMT through neural signaling pathways in male breast cancer

Status:

Version 1

Abstract

Figures

Main

Result

Single-cell and spatial transcriptome characterization of male breast cancer

Subtype specific energy metabolism and neuronal signaling pathways in MBC cells

Pseudo-evolution tree of cancer cell Subtypes

Regulation of different subtypes of cancer cells by neuroendocrine-like cancer cells in MBC

Characterization of T cell subtypes in MBC and their cell-cell communications with cancer cells

Neuroendocrine-like cells work with TAMs to drive EMT and angiogenesis in MBC

Discussion

Methods

Patient sample collection

scRNA-seq library preparations

10X library construction and sequencing

Spatial sequencing library preparation

scRNA-seq data pre-processing

Identification of malignant cancer cells

Relative Meta-module Score

Pathway enrichment analyses

Single-cell level metabolic analysis

Cell-cell communication analysis

Pseudo-evolution tree

TCGA data analysis

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1