The cell of origin and the development of breast cancer are not fully elucidated in BRCA1 mutation carriers, especially for estrogen receptor (ER)–positive breast cancers. Here, we performed single-cell RNA sequencing (RNA-seq) on 82,122 cells isolated from the breast cancer tissues and adjacent or prophylactic normal breast tissues from four BRCA1 mutation carriers and three noncarriers. Whole-exome sequencing was performed on breast tumors from the four BRCA1 mutation carriers; for validation, bulk RNA-seq was performed on adjacent normal breast tissues from eight additional BRCA1 mutation carriers and 14 noncarriers. Correlation analyses suggested that breast cancers in BRCA1 mutation carriers might originate from luminal cells. The aberrant luminal progenitor cells with impaired differentiation were significantly increased in normal breast tissues in BRCA1 mutation carriers compared with noncarriers. These observations were further validated by the bulk RNA-seq data from additional BRCA1 mutation carriers. These data suggest that the cell of origin of basal-like breast tumors (ERneg) in BRCA1 mutation carriers might be luminal progenitor cells. The expression of TP53 and BRCA1 was decreased in luminal progenitor cells from normal breast tissue in BRCA1 mutation carriers, which might trigger the basal/mesenchymal transition of luminal progenitors and might result in basal-like tumor development. Furthermore, ERhigh luminal tumors might originate from mature luminal cells. Our study provides in-depth evidence regarding the cells of origin of different breast cancer subtypes in BRCA1 mutation carriers.

Significance:

Single-cell RNA-seq data indicate that basal-like breast cancer (ERneg) might originate from luminal progenitors, and ERhigh luminal breast cancer might originate from mature luminal cells in BRCA1 mutation carriers.

Women carrying a germline mutation in the BRCA1 gene confer an increased risk of developing breast cancers, especially aggressive basal-like breast tumors (1), and nearly one-third of breast cancers exhibit a luminal phenotype [namely, estrogen receptor (ER) positive] in BRCA1 mutation carriers (2). However, the cell of origin and the course of breast cancer development in BRCA1 mutation carriers remain largely obscure.

Multiple lines of evidence support a crucial role for BRCA1 gene in mammary epithelial cell differentiation (3–9). Recent studies have shown that normal breast tissues from BRCA1 mutation carriers display increased luminal progenitor cells based on FACS analysis (10–12), and targeted deletion of BRCA1 in luminal progenitor cells results in basal-like tumorigenesis in mouse models (13, 14), indicating that basal-like tumors in BRCA1 mutation carriers originate from expanded luminal progenitor cells. However, the underlying mechanism(s) of the increase of luminal progenitor cells and the transformation from luminal progenitor cells to basal-like tumors in BRCA1 mutation carriers is largely unknown. In addition, it is also worth investigating the cell of origin of the nearly 30% luminal breast cancers in BRCA1 mutation carriers, which was never explored in previous studies.

Single-cell RNA sequencing (scRNA-seq) is a revolutionary tool to profile the transcriptional state of single cells, enabling an unbiased analysis of the spectrum of complex subpopulations within normal breast or breast cancer tissues (15–17). In this study, we performed scRNA-seq on breast cancer tissues, as well as normal breast tissues obtained from adjacent and/or contralateral breast mastectomy from four BRCA1 mutation carriers and three noncarriers. We further performed whole-exome sequencing (WES) on breast tumors from the four BRCA1 mutation carriers and bulk RNA sequencing (RNA-seq) on adjacent normal breast tissues from eight additional BRCA1 mutation carriers and 15 noncarriers for validation. We aimed to comprehensively investigate the differences of breast epithelial cells between BRCA1 mutation carriers and noncarriers in normal breast tissues; and to identify the cell of origin of breast tumorigenesis in BRCA1 mutation carriers. Finally, we explored the underlying mechanisms that the aberrant breast epithelial cells lead to breast tumorigenesis in BRCA1 mutation carriers.

Sample collection

In this study, four BRCA1 germline mutation carriers (cases 1–4) and three noncarriers (cases 5–7) were recruited from Peking University Cancer Hospital (Beijing, P.R. China) and Peking University International Hospital (Beijing, P.R. China; Supplementary Table S1). The BRCA1 mutation carriers were matched with noncarriers in age and parity status (Supplementary Table S1). All BRCA1 mutation carriers were diagnosed with invasive breast cancer and underwent mastectomy with/without contralateral prophylactic mastectomy. For BRCA1 mutation carriers, breast cancer tissue, adjacent normal mammary (ANM) from the ipsilateral breast and contralateral normal mammary (CNM) were collected (Supplementary Table S2). For noncarriers who underwent mastectomy or lumpectomy (including 1 patient with invasive breast cancer and 2 patients with benign fibroadenoma), normal breast tissue from the ipsilateral breast was collected as a normal control(Supplementary Table S2). A total of 13 samples from the 7 patients were collected and immediately dissected into single-cell suspensions after surgical resection. This study was carried out in accordance with the ethical principles of the Declaration of Helsinki and was approved by the Research and Ethics Committee of Peking University Cancer Hospital (Beijing, P.R. China) and Peking University International Hospital (Beijing, P.R. China). Written informed consent was obtained from all participants.

Sample processing and scRNA-seq

Breast cancer tissues and normal mammary tissues obtained from surgery were immediately stored in Tissue Storage Solution (Miltenyi Biotec) and processed within an hour. Samples were washed in PBS and mechanically dissociated using a razor blade. Dissociated samples were digested in RPMI Medium 1640 (Gibco) with collagenase/hyaluronidase (20%, Stemcell Technologies) and BSA (20 mg/mL, Solarbio) at 37°C for an hour to generate cell suspensions. Then the cells were washed in cold Hank's balanced salt mixture (Solarbio) with 0.04% BSA and filtered using a 30-μmol/L cell strainer to generate single-cell suspensions. The single-cell suspensions were resuspended at a concentration of approximately 1,000 cells/μL. Library preparation was performed according to the instructions in the 10x Chromium single-cell 5′ Library construction kit. Then, the libraries were pooled and sequenced on the NovaSeq 6000 platform to achieve an average of 100,000 reads per cell.

scRNA-seq data processing and quality control

Alignment of 5′ end counting libraries from scRNA-seq analyses was completed utilizing the 10× Genomics Cell Ranger software suite. Each library was aligned to the GRCh38 reference genome. The “Cell Ranger Aggr” function was used to normalize the number of confidently mapped reads per cell across the libraries from different samples. In total, the Cell Ranger software identified 99,538 barcodes that contained enough unique molecules to be considered as cells. Poor-quality cells were identified as having the following features: (i) a total number of unique molecular identifiers (UMI) <1,000; (ii) a number of genes detected <500 or >6,000; and (iii) a percentage of molecules mapped to mitochondrial genes ≥20%. The cells met any of the above rules were excluded from further analysis, which left us with a total of 82,122 cells.

Unsupervised cell clustering

We performed principal component analysis (PCA) on all 82,122 single-cell transcriptomes by using genes expressed in more than two cells (33,694 genes). We used 25 principal components for PCA. We then applied the k-means algorithm to cluster cells based on PCA results, and the parameter k was chosen from 2 to 10 iteratively. The cluster-specific genes enabled us to empirically determine the optimal k. We found that the clusters across all four BRCA1 mutation carriers had the best consistency when k was set as 9.0. With this setting, we identified a total of nine main clusters. The t-distributed statistical neighbor embedding (t-SNE) method was used for visualization of cell distances in the reduced two-dimensional space.

Next, we used a series of cell specific markers to identify the cell type of each cluster based on previous studies (Supplementary Fig. S1A; refs. 10, 11, 18–20). We found that the atlas mainly comprised immune cells (C6, C9), stromal cells (C2, C3), and normal epithelial cells (C1, C4, C5, C7). C6 and C9 were tagged as immune cells based on the expression of PTPRC (Supplementary Fig. S1B), C2 as fibroblasts based on the expression of DCN (Supplementary Fig. S1C) and C3 as endothelial cells based on the expression of PECAM1 (Supplementary Fig. S1D). The epithelial cells were further divided into luminal lineage (C1, C4, KRT8/18+, CD24+; Supplementary Fig. S1E) and basal/myoepithelial lineage (C5, C7, ATAC2+, CNN1+, MME+; Supplementary Fig. S1F), which constitute the inner layer and outer layer of the breast ductal system, respectively. Within the luminal lineage, C4 was tagged as luminal progenitors based on the expression of GABRP, KIT, and ALDH1A3 (Supplementary Fig. S1E), and C1 was tagged as mature luminal cells based on the expression of AGR2, ANKRD30A, ESR1, FOXA1, and PGR (Supplementary Fig. S1E). Within the basal/myoepithelium lineage, C7 was tagged as basal progenitor cells based on the expression of ITGA6, KRT5/6B, KRT14, and TP63 (Supplementary Fig. S1F and S1G), and C5 was tagged as myoepithelium. In addition, Monocle also reconstructed the differentiation trajectories from upstream basal progenitor cells (C7) to downstream mature myoepithelial cells (C5). A small part of cells in C5 expressed PDGFRβ and NG2 (marker genes for pericytes; Supplementary Fig. S1H; ref. 21); thus, we cannot exclude the possibility that a small fraction of the myoepithelial cluster might be pericytes.

To profile the transcriptome of various cell types from the breast tissues of BRCA1 mutation carriers, we performed differential gene expression analysis on all cells from four BRCA1 mutation carriers to identify cell type–specific marker genes. For pairwise comparisons, genes with a mean expression level below 0.1 were removed from the analysis. The cell type–specific marker genes were identified as: (i) log2 fold change ≥ 2.0 and (ii) P value of <0.01. The R package “pROC” was used to calculate the area under the ROC curve (AUC) of each cell type–specific marker gene. The lists of marker genes of various cell types within the breast tissues of BRCA1 mutation carriers are presented in Supplementary Table S3.

Identification of the cancer cell populations

Tumor tissues may contain normal cell contamination. In our study, the adjacent normal breast sample served as a perfect control to filter various types of normal cells from tumor samples. In each BRCA1 mutation carrier, the normal cells from the tumor sample were merged with the same types of normal cells from adjacent normal breast samples, leaving one cell cluster only generalized by tumor sample. These tumor-specific clusters were considered as cancer cell populations. We further found that the cancer cell population showed higher UMI counts than the normal cell populations, consistent with active metabolism in the tumor cells. More importantly, ER, progesterone receptor (PR), and HER2 expression of the tumor cells in each case were exactly matched with the clinical records, and the PAM50 subtypes of the tumor cells in each case were also matched with the clinical records (Supplementary Table S1).

Reconstruction of differentiation trajectories

Cell fate decisions and differentiation trajectories were reconstructed with the R package Monocle 2 with standard settings (22). First, we selected four types of breast epithelial cells (luminal progenitor, mature luminal, basal progenitors, and myoepithelium) from BRCA1 mutation carriers to construct differentiation trajectories of premalignant stage. The differentiation trajectories of three noncarriers (cases 5–7) were also analyzed. Second, we selected luminal lineage cells and tumor cells within each BRCA1 mutation carrier to identify the original cell of breast cancer and to construct tumorigenesis trajectories for each BRCA1 mutation carrier.

Enrichment analysis

Gene set enrichment analysis (GSEA; RRID:SCR_003199; ref. 23) based on Kyoto Encyclopedia of Genes and Genomes terms (RRID:SCR_012773) and a self-defined basal/mesenchymal gene set (Supplementary Table S4) was used to identify the biological functional differences of epithelial cells between each BRCA1 mutation carriers and noncarriers. The basal/mesenchymal signature gene set was defined on the basis of previous studies (14, 24, 25). In addition, scRNA-seq data of tumor cells were compared with original epithelial cells within each BRCA1 mutation carrier. Genes with a log2 fold change > 1.0 and an Padjusted < 0.05 were regarded as upregulated differential genes, and genes with a log2 fold change <1.0 and an Padjusted < 0.05 were regarded as downregulated differential genes. These differentially expressed genes were subjected to Gene Ontology analysis using the R package “clusterProfiler” (26).

Similarity analysis between tumor cells and normal epithelial cells

The scRNA-seq data of epithelial cell types in the normal breast tissues from the four BRCA1 mutation carriers were locally compared with the top 200 upregulated genes, which were sorted by log2 fold change in each cell type. The top 200 genes were defined as cell type–specific marker genes and were used to construct a heatmap. Tumor cells and epithelial cells in the normal breast tissues from the four BRCA1 mutation carriers were compared by Spearman correlation analysis based on the expression of the cell type–specific marker genes in each case. Our aim was to identify the potential origin cells of tumors in each BRCA1 mutation carrier.

Bulk RNA-seq

We performed bulk RNA-seq on adjacent normal breast tissues from 8 additional patients with breast cancer who carrying a pathogenic BRCA1 germline mutation and 14 age-matched patients with breast cancer without BRCA1 germline mutation (Supplementary Table S5). RNA library preparation was performed with total RNA according to the literature (27). The libraries were sequenced on the Illumina HiSeq platform to generate fastq files. The fragments per kilobase of transcript per million mapped reads values were obtained using the tophat-cufflinks pipeline, where the sequencing reads were mapped to the GRCh38 reference genome. In addition, the bulk RNA-seq data and clinicopathologic information of breast cancers from 41 BRCA mutation carriers were downloaded from The Cancer Genome Atlas (TCGA) database (https://portal.gdc.cancer.gov/projects; Supplementary Table S6).

WES

We performed WES on frozen-fresh tumor tissues and normal breast tissues of the four BRCA1 carriers that were subjected to scRNA-seq. A total of 300 ng of DNA for each sample was prepared to construct a paired-end DNA library, and the library was subjected to whole-exome capture by using the Agilent SureSelect Human ALL Exon V5/V6 Kit. The products were sequenced on an Illumina Hiseq PE150 sequencing platform to achieve an average depth of 300× on tumor samples and 100× on matched normal samples. Sequenced reads were aligned to the human reference genome (GRCh37). Indels and single-nucleotide variations (SNV) were called by GATK (version 3.1, RRID:SCR_001876). Somatic indels were called with Strelka (RRID:SCR_005109), and somatic SNVs were detected by MuTect (RRID:SCR_000559).

IHC

To validate the results from the scRNA-seq, IHC assay was performed on normal breast tissues from seven BRCA1 mutation carriers (four scRNA-seq cases and three additional cases) and 10 matched noncarriers (Supplementary Table S7). Antibodies for ALDH1A3 (ab129815, rabbit polyclonal antibody; Abcam); AGR2 (clone EPR3278, ab76473, rabbit mAb, Abcam); and WIF1 (YM0648, mouse mAb, Immunoway) were used.

A single-cell atlas of normal and breast cancer tissues in BRCA1 mutation carriers

A total of 13 samples from four BRCA1 mutation carriers (cases 1–4) and three noncarriers (cases 5–7) were subjected to scRNA-seq (Fig. 1; Supplementary Tables S1 and S2). After removing low-quality cells, a total of 82,122 cells (including 46,766 epithelial cells) were retained for subsequent analysis (Supplementary Table S2), which yielded a median of 1,752 detected genes per cell. Then, we performed dimensionality reduction and unsupervised cell clustering on all 82,122 cells and finally identified a total of nine main cell clusters (Fig. 2A). Except for the tumor cell populations that were detected only in each breast cancer sample, the other eight clusters were generalizable across all samples (Fig. 2B). On the basis of the expression of a series of known cell type–specific markers (details described in Materials and Methods), four clusters (C1, C4, C5, and C7) were identified as normal breast epithelial cells (Fig. 2C and D; Supplementary Fig. S1). Among these, C1 resembled mature luminal cells (marked as AGR2, AUC = 0.90), C4 corresponded to luminal progenitor cells (marked as ALDH1A3, AUC = 0.89), C5 matched with myoepithelial epithelial cells (marked as TAGLN, AUC = 0.92), and C7 matched with basal progenitor cells (marked as KRT14, AUC = 0.91; Fig. 2C).

Figure 1.

Overview of the study design. ANM, adjacent normal mammary tissue; BC, breast cancer tissue; CNM, contralateral normal mammary tissue; NC, normal breast tissues from noncarriers served as normal control. The BRCA1 carriers were matched with noncarriers in age and parity status (see also Supplementary Table S1).

Figure 1.

Overview of the study design. ANM, adjacent normal mammary tissue; BC, breast cancer tissue; CNM, contralateral normal mammary tissue; NC, normal breast tissues from noncarriers served as normal control. The BRCA1 carriers were matched with noncarriers in age and parity status (see also Supplementary Table S1).

Close modal
Figure 2.

A single-cell atlas of normal breast and breast cancer tissues in BRCA1 mutation carriers. A, t-SNE plot of 82,122 high-quality cells from all samples. The cells are colored by K-means clusters (C1–C9). The cell type of each cluster was annotated on the basis of the expression of known marker genes; the detailed cell type–specific markers used to define clusters are listed in Supplementary Fig. S1. B, The cells are colored by BRCA1 germline status (top left), by case (top right), by sample (bottom left), and by unique UMI counts (bottom right). ANM, adjacent normal mammary tissue; BC, breast cancer; CNM, contralateral normal mammary tissue; NC, normal mammary tissue in noncarriers. C, Heatmap showing the expression patterns of multiple cell types in normal breast tissue from BRCA1 mutation carriers. D, Violin plots displaying the distribution of expression of KRT8 (a known marker gene for the luminal epithelium) and ACTA2 (a known marker gene for the basal/myoepithelium) across cell types. BP, basal progenitor cells; endo, endothelial cells; fibro, fibroblasts; LP, luminal progenitor cells; macro, macrophages; ML, mature luminal cells; myo, myoepithelial cells.

Figure 2.

A single-cell atlas of normal breast and breast cancer tissues in BRCA1 mutation carriers. A, t-SNE plot of 82,122 high-quality cells from all samples. The cells are colored by K-means clusters (C1–C9). The cell type of each cluster was annotated on the basis of the expression of known marker genes; the detailed cell type–specific markers used to define clusters are listed in Supplementary Fig. S1. B, The cells are colored by BRCA1 germline status (top left), by case (top right), by sample (bottom left), and by unique UMI counts (bottom right). ANM, adjacent normal mammary tissue; BC, breast cancer; CNM, contralateral normal mammary tissue; NC, normal mammary tissue in noncarriers. C, Heatmap showing the expression patterns of multiple cell types in normal breast tissue from BRCA1 mutation carriers. D, Violin plots displaying the distribution of expression of KRT8 (a known marker gene for the luminal epithelium) and ACTA2 (a known marker gene for the basal/myoepithelium) across cell types. BP, basal progenitor cells; endo, endothelial cells; fibro, fibroblasts; LP, luminal progenitor cells; macro, macrophages; ML, mature luminal cells; myo, myoepithelial cells.

Close modal

Aberrant proportions of epithelial cells in normal breast tissues from BRCA1 mutation carriers

To understand the hierarchical relationships among the four distinct epithelial cell types (Fig. 3A), we next applied Monocle to reconstruct differentiation trajectories of the epithelial cells from normal breast tissues of the BRCA1 mutation carriers and noncarriers, respectively. Clear segregation between the luminal lineage and basal/myoepithelial lineage was observed, supporting the previous observations that the two lineages originate from different classes of unipotent stem cells (Fig. 3B; refs. 16, 28–30). The left arm of the two differentiation trajectories showed that the basal progenitor cells were upstream of myoepithelial cells, and the right arm showed the luminal differentiation process starting with luminal progenitor cells and moving to mature luminal cells (Fig. 3B).

Figure 3.

Comparison of the epithelial proportions in normal breast tissues between BRCA1 mutation carriers and noncarriers. A, t-SNE plot to visualize distinct breast epithelial cell types for BRCA1 mutation carriers (left) and noncarriers (right). B, Monocle-generated differentiation trajectories of breast epithelium in normal breast tissue from four BRCA1 mutation carriers and three noncarriers. C, Proportions of luminal progenitor cells and mature luminal cells across all normal breast samples. D, Proportions of basal progenitor and myoepithelial cells across all normal breast samples. ANM, adjacent normal mammary tissue; CNM, contralateral normal mammary tissue; NC, normal mammary tissue in noncarriers.

Figure 3.

Comparison of the epithelial proportions in normal breast tissues between BRCA1 mutation carriers and noncarriers. A, t-SNE plot to visualize distinct breast epithelial cell types for BRCA1 mutation carriers (left) and noncarriers (right). B, Monocle-generated differentiation trajectories of breast epithelium in normal breast tissue from four BRCA1 mutation carriers and three noncarriers. C, Proportions of luminal progenitor cells and mature luminal cells across all normal breast samples. D, Proportions of basal progenitor and myoepithelial cells across all normal breast samples. ANM, adjacent normal mammary tissue; CNM, contralateral normal mammary tissue; NC, normal mammary tissue in noncarriers.

Close modal

We then compared the proportions of distinct epithelial cell types in normal breast tissues between the BRCA1 mutation carriers and noncarriers. Within the luminal lineage, the luminal progenitor cells were significantly increased (39% vs. 11%, P = 0.02) and the mature luminal cells were significantly decreased (61% vs. 89%, P = 0.02) in BRCA1 mutation carriers when compared with noncarriers (Fig. 3C). Within the basal/myoepithelial lineage, the basal progenitor cells were also significantly increased in BRCA1 mutation carriers compared with noncarriers (51% vs. 17%, P = 0.004; Fig. 3D). Notably, the abnormal composition of the epithelial cells was observed not only in normal breast tissues adjacent to cancers, but also in contralateral normal breast tissues from the same BRCA1 mutation carriers, indicating that the enrichment of luminal progenitor cells and basal progenitor cells in normal breast tissues was a common phenomenon in BRCA1 mutation carriers.

To validate the findings based on the scRNA-seq, we performed IHC assay and bulk RNA-seq on additional BRCA1 mutation carriers and noncarriers. IHC staining for ALDH1A3 (a marker of luminal progenitor cells), WIF1 (a marker of basal progenitor cells) and AGR2 (a marker of mature luminal cells), was carried out on normal breast tissues from seven BRCA1 mutation carriers and 10 noncarriers (details in Materials and Methods). The normal tissues from BRCA1 mutation carriers showed more positive cells for ALDH1A3 and WIF1, and fewer positive cells for AGR2 than those from noncarriers (Supplementary Fig. S2A). In addition, bulk RNA-seq of normal breast tissues from eight additional BRCA1 carriers and 14 matched noncarriers was performed. Normal breast tissues from BRCA1 mutation carriers had a trend to express higher levels of ALDH1A3 and WIF1 but lower levels of AGR2 than those from noncarriers (Supplementary Fig. S2B). These data further indicated that the expanded luminal progenitors and basal progenitors existed in the normal breast tissues in BRCA1 mutation carriers.

We performed GSEA to explore the mechanism underlying the increased luminal progenitor cells and basal progenitor cells. We found that signaling pathways involved in breast gland differentiation (the wnt, notch, and hedgehog pathways; refs. 31–35) were significantly downregulated in luminal progenitors from BRCA1 mutation carriers compared with those from noncarriers (Supplementary Fig. S3A–S3C), and the mRNA level of key genes of these pathways were also downregulated in luminal progenitors from BRCA1 mutation carriers compared with those from noncarriers (Supplementary Fig. S3D). In addition, TP63, an essential regulator for basal/myoepithelial lineage differentiation (32), was downregulated in basal progenitor cells from BRCA1 mutation carriers compared with those from noncarriers (Supplementary Fig. S3E), indicating that the differentiation of the luminal and basal/myoepithelial lineages might be hindered in BRCA1 mutation carriers and led to the accumulation of the upstream cells.

The putative cell of origin of breast cancer in BRCA1 mutation carriers

We compared the expression profiles of the tumor cells with that of each distinct type of epithelial cell from normal breast tissues at single-cell level to identify the putative cell of origin of the tumors. In case 1, the tumor cells (ERhigh) displayed a transcriptomic signature of mature luminal cells (Supplementary Fig. S4A and S4B), and Spearman correlation analysis showed that the tumor cells had the highest similarity to mature luminal cells (Fig. 4A), suggesting that the tumor cells in case 1 were most likely to originate from mature luminal cells. The tumor cells in case 2 (ERlow) showed higher similarity to luminal cells than basal and myoepithelial cells, while the expression profiles of this tumor cells showed patterns of both luminal progenitor cells and mature luminal cells (Fig. 4A; Supplementary Fig. S4C and S4D), thus we cannot infer the exact cell of origin of the tumor in case 2. The tumor cells in cases 3 and 4 (basal-like subtype, ERneg) displayed a transcriptomic signature of luminal progenitor cells (Supplementary Fig. S4E–S4H), and Spearman correlation analysis showed that the tumor cells had the highest similarity to luminal progenitor cells (Fig. 4A), suggesting that the tumor cells in cases 3 and 4 were most likely to originate from luminal progenitor cells. Collectively, all of the four breast cancers from BRCA1 mutation carriers might originate from luminal lineage cells. We further utilized the scRNA-seq data of luminal lineage cells from the normal tissues and the tumor cells to construct a map of carcinogenesis for each BRCA1 mutation carrier (Fig. 4B). The pseudotime maps again showed that mature luminal cells (ERhigh) were the putative cellular origin of the case 1 tumor (ERhigh) and luminal progenitors (ERneg) were the putative cellular origin of the case 3 and 4 tumors (basal-like subtype, ERneg; Fig. 4B).

Figure 4.

Identification of the putative cell of origin of breast cancer in BRCA1 mutation carriers via scRNA-seq. A, Spearman correlation analysis of tumor cells and distinct breast epithelial cell types in each case (the tumor cells and compared normal cells were from the same patients). B, Monocle-generated cell trajectories of tumor cells and luminal cells in each BRCA1 mutation carrier (the tumor cells and compared normal cells were from the same patients). BP, basal progenitor cells; LP, luminal progenitor cells; ML, mature luminal cells; myo, myoepithelial cells.

Figure 4.

Identification of the putative cell of origin of breast cancer in BRCA1 mutation carriers via scRNA-seq. A, Spearman correlation analysis of tumor cells and distinct breast epithelial cell types in each case (the tumor cells and compared normal cells were from the same patients). B, Monocle-generated cell trajectories of tumor cells and luminal cells in each BRCA1 mutation carrier (the tumor cells and compared normal cells were from the same patients). BP, basal progenitor cells; LP, luminal progenitor cells; ML, mature luminal cells; myo, myoepithelial cells.

Close modal

To validate these results, we analyzed the bulk RNA-seq data of 41 breast cancers from BRCA mutation carriers from TCGA dataset. Of these 41 tumors, 18 were basal-like tumors (ERneg), five ERlow tumors, and 18 ERhigh tumors (Fig. 5A). Hierarchical clustering analysis showed that all basal-like tumors (ERneg) were clustered as a subgroup that shared the expression signature of luminal progenitor cells, and ERhigh tumors were clustered as another subgroup that shared the expression signature of mature luminal cells (Fig. 5B and C). While ERlow tumors clustered as a subgroup that shared the expression signature of both luminal progenitor cells and mature luminal cells (Fig. 5B and C). These findings were further validated by Spearman correlation analysis (Fig. 5D). Furthermore, we performed WES in breast tumors from the four BRCA1 mutation carriers (Supplementary Table S8). The case 1 tumor (ERhigh) was driven by GATA3 mutation and both case 3 and case 4 (basal-like, ERneg) were driven by TP53 mutation (Supplementary Table S9). We could not define the exact driver mutation in case 2. The mutations in the driver genes (case 1, case 3, and case 4) were concordant with the ER expression.

Figure 5.

Correlation between the cell of origin and breast cancer subtypes revealed by TCGA dataset and scRNA-seq. A, Breast cancers with BRCA1/2 germline mutation in TCGA dataset were stratified into basal-like (ERneg), ERlow, and ERhigh subtypes. B, Hierarchical clustering heatmap of the expression patterns of 41 breast cancers with BRCA1/2 germline mutation from TCGA database and the expression patterns of normal epithelial cell types in this study. C, The average levels of ER mRNA in breast cancers with BRCA1/2 germline mutation in TCGA dataset. The cancers were divided into three groups according the putative cell of origin. Mann–Whitney test was used for statistical analysis. D, Spearman correlation analysis showing the similarities in expression profiles between breast cancer cells from BRCA1 mutation carriers in TCGA dataset and the normal epithelial cell types in this study. BP, basal progenitor cells; LP, luminal progenitor cells; ML, mature luminal cells; myo, myoepithelial cells.

Figure 5.

Correlation between the cell of origin and breast cancer subtypes revealed by TCGA dataset and scRNA-seq. A, Breast cancers with BRCA1/2 germline mutation in TCGA dataset were stratified into basal-like (ERneg), ERlow, and ERhigh subtypes. B, Hierarchical clustering heatmap of the expression patterns of 41 breast cancers with BRCA1/2 germline mutation from TCGA database and the expression patterns of normal epithelial cell types in this study. C, The average levels of ER mRNA in breast cancers with BRCA1/2 germline mutation in TCGA dataset. The cancers were divided into three groups according the putative cell of origin. Mann–Whitney test was used for statistical analysis. D, Spearman correlation analysis showing the similarities in expression profiles between breast cancer cells from BRCA1 mutation carriers in TCGA dataset and the normal epithelial cell types in this study. BP, basal progenitor cells; LP, luminal progenitor cells; ML, mature luminal cells; myo, myoepithelial cells.

Close modal

Abnormal epithelial cells in normal breast tissues from BRCA1 mutation carriers trigger breast tumorigenesis

Our correlation analyses suggested that all the breast tumors from BRCA1 mutation carriers were likely to originate from luminal lineage cells, and the breast gland differentiation processes were downregulated in luminal lineage cells in normal breast tissues from BRCA1 mutation carriers compared with noncarriers. Intriguingly, we found that these differentiation processes were further downregulated in tumor cells compared with their parallel normal cells of origin in four BRCA1 mutation carriers (Supplementary Fig. S5A–S5F), indicating that the impaired differentiation process of normal luminal cells in BRCA1 mutation carriers might contribute to tumorigenesis.

The correlative analysis suggested that basal-like tumors in BRCA1 mutation carriers were likely to originate from luminal progenitor cells. We surprisingly found that levels of BRCA1 mRNA in luminal progenitor cells in normal breast tissues were the highest among all the cell types in both BRCA1 mutation carriers and noncarriers, but the luminal progenitors from BRCA1 mutation carriers showed significantly lower BRCA1 expression compared with those from noncarriers (Fig. 6A). Furthermore, TP53, the most common driver gene in basal-like tumors (36), was also significantly downregulated in the luminal progenitors (Fig. 6B). It is well documented that the knockout of BRCA1 and TP53 genes could cause basal-like tumorigenesis in mouse models (14, 37), and our data showed that the expression of BRCA1 and TP53 were already decreased in luminal progenitors in normal breast tissues from BRCA1 mutation carriers prior to tumorigenesis. In addition, previous studies suggested that “basal cytokeratins” (such as KRT14/17) are expressed not only in the basal cells of human breast ducts but also in the luminal cells of the lobules (38, 39). Our scRNA-seq data showed that KRT14/17 were expressed in the luminal progenitor cells in both BRCA1 mutation carriers and noncarriers (Supplementary Fig. S6A and S6B), but the expression of KRT14/17 in luminal progenitors in BRCA1 mutation carriers was higher than that in noncarriers (Supplementary Fig. S6A and S6B). Examination of the basal/mesenchymal gene set further revealed that the basal/mesenchymal features of luminal progenitors in normal breast tissues were inconsistent across BRCA1 mutation carriers. Luminal progenitor cells from two basal-like cancers (cases 3 and 4) showed significantly upregulated basal/mesenchymal features and epithelial–mesenchymal transition (EMT) transcription factors compared with noncarriers, while the basal/mesenchymal features were not significantly different between ER-positive cancers (cases 1 and 2) and noncarriers (Supplementary Fig. S7A and S7B). Bulk RNA-seq also exhibited that basal/mesenchymal features and EMT transcription factors were significantly upregulated in the normal breast tissues from additional BRCA1 mutation carriers with basal-like tumors compared with that in the normal breast tissues from matched noncarriers; this observation was not found in the normal breast tissues from BRCA1 mutation carriers with ER-positive tumors (Supplementary Fig. S8A and S8B). In addition, the basal/mesenchymal features in the basal-like cancer cells (cases 3 and 4) were further upregulated compared with the luminal progenitors in normal tissues from the same case (Fig. 6C and D). Our data suggested that the luminal progenitors transitioned to a basal/mesenchymal state during basal-like tumorigenesis, and that this transition might have started in normal breast tissue from BRCA1 mutation carriers before tumorigenesis.

Figure 6.

Transcriptome differences between the normal epithelium of BRCA1 mutation carriers and the normal epithelium of noncarriers. A, Comparison of BRCA1 mRNA level in distinct epithelial cell types in normal breast tissue from BRCA1 mutation carriers and noncarriers. B, Comparison of average counts of TP53 mRNA in distinct epithelial cell types in normal breast tissue from BRCA1 mutation carriers and noncarriers. C and D, Basal-like tumor cells from BRCA1 mutation carriers (case 3 and case 4) showing a decrease in luminal features (top) and an increase in basal/mesenchymal features (bottom) compared with the origin cells of the tumor (luminal progenitor cells). E, Comparison of average counts of GATA3 mRNA in distinct epithelial cell types in normal breast tissue from BRCA1 mutation carriers and noncarriers. BP, basal progenitor cells; LP, luminal progenitor cells; ML, mature luminal cells; myo, myoepithelial cells. Number of cells from normal breast tissue in BRCA1 mutation carriers (cases 1–4): 4,485 BP cells; 4,293 myo cells; 5,035 LP cells; and 7,036 ML cells. Number of cells from normal breast tissue in noncarriers (cases 5–7): 353 BP cells; 775 myo cells; 119 LP cells; and 6,624 ML cells. Data represent the mean ± SEM. P values less than 1e-5 were considered to be statistically significant due to a large number of cells (Mann–Whitney test). *, P < 1e-5; **, P < 1e-10; ***, P < 1e-15; ****, P < 1e-20 (Mann–Whitney test). No asterisk indicates no significant difference (ns).

Figure 6.

Transcriptome differences between the normal epithelium of BRCA1 mutation carriers and the normal epithelium of noncarriers. A, Comparison of BRCA1 mRNA level in distinct epithelial cell types in normal breast tissue from BRCA1 mutation carriers and noncarriers. B, Comparison of average counts of TP53 mRNA in distinct epithelial cell types in normal breast tissue from BRCA1 mutation carriers and noncarriers. C and D, Basal-like tumor cells from BRCA1 mutation carriers (case 3 and case 4) showing a decrease in luminal features (top) and an increase in basal/mesenchymal features (bottom) compared with the origin cells of the tumor (luminal progenitor cells). E, Comparison of average counts of GATA3 mRNA in distinct epithelial cell types in normal breast tissue from BRCA1 mutation carriers and noncarriers. BP, basal progenitor cells; LP, luminal progenitor cells; ML, mature luminal cells; myo, myoepithelial cells. Number of cells from normal breast tissue in BRCA1 mutation carriers (cases 1–4): 4,485 BP cells; 4,293 myo cells; 5,035 LP cells; and 7,036 ML cells. Number of cells from normal breast tissue in noncarriers (cases 5–7): 353 BP cells; 775 myo cells; 119 LP cells; and 6,624 ML cells. Data represent the mean ± SEM. P values less than 1e-5 were considered to be statistically significant due to a large number of cells (Mann–Whitney test). *, P < 1e-5; **, P < 1e-10; ***, P < 1e-15; ****, P < 1e-20 (Mann–Whitney test). No asterisk indicates no significant difference (ns).

Close modal

ERhigh tumors in BRCA1 mutation carriers were likely to originate from mature luminal cells. GATA3, a crucial gene for the maintenance of mature luminal cells (40, 41), was significantly downregulated in the mature luminal cells of BRCA1 mutation carriers (Fig. 6E). Given that GATA3 is a frequently mutated gene in the luminal subtypes of breast cancers (36), GATA3 downregulation might be helpful for the tumorigenesis of mature luminal cells.

In this study, we found that luminal cells were the potential origin cells of breast cancers in BRCA1 mutation carriers and that luminal progenitor cells were increased in the normal breast tissues from BRCA1 mutation carriers due to impaired differentiation processes in luminal lineages (Fig. 7). Our correlation analyses suggested that basal-like breast cancers (ERneg) might originate from luminal progenitor cells, while ERhigh luminal breast cancers might originate from mature luminal cells (Fig. 7). More importantly, the decreased BRCA1/TP53 expression and increased basal/mesenchymal transition in luminal progenitor cells from normal breast tissues might contribute to basal-like breast cancer development in BRCA1 mutation carriers (Fig. 7).

Figure 7.

Schematic summary of abnormalities in breast epithelial cells and the putative cell of origin for basal-like (ERneg) and ERhigh luminal breast cancer in BRCA1 mutation carriers.

Figure 7.

Schematic summary of abnormalities in breast epithelial cells and the putative cell of origin for basal-like (ERneg) and ERhigh luminal breast cancer in BRCA1 mutation carriers.

Close modal

Our scRNA-seq data in combination with our bulk RNA-seq data and IHC results suggested that both luminal progenitor cells and basal progenitor cells were increased in BRCA1 mutation carriers compared with noncarriers. This finding might be useful to clarify the conflicting results in the previous studies based on FACS analysis (10–12). Our study revealed that the potential mechanism of the expanded luminal progenitors in BRCA1 mutation carriers might be the downregulation of wnt/notch/hedgehog signaling pathways in luminal cells, which may halt luminal differentiation and cause the accumulation of luminal progenitor cells. These notions are in line with previous findings that notch pathways are upregulated by BRCA1 gene (42).

Several previous studies have suggested that the origin cells of BRCA1-related basal-like cancers (ERneg) probably originate from luminal progenitors (10, 11, 13, 14, 43), but the origin cells of ER-positive breast cancer in BRCA1 mutation carriers have never been explored in previous studies. Our scRNA-seq data validated and extended previous findings at a higher resolution, in which the cell development trajectory could be traced from single-cell level expression profiles. Therefore, we were able to reveal a correlation between cell of origin of ERneg/ERhigh tumors and luminal progenitor/mature luminal cell in BRCA1 mutation carriers. Our findings are consistent with a model revealed by previous study in which BRCA1-related basal-like breast tumors are originated from luminal progenitors (13), although we cannot exclude the possibility that the expression profiles of epithelial cells might switch during the tumorigenesis process.

In this study, an interesting finding was that the mRNA levels of BRCA1 and TP53 genes in luminal progenitors in the normal breast tissues from BRCA1 mutation carriers were markedly lower than those in normal breast tissues from noncarriers. Indeed, the luminal progenitor cells in normal breast tissues expressed the highest levels of BRCA1 and TP53 mRNA among all epithelial cell types in noncarriers, suggesting that this highly proliferative cell population might require enough BRCA1 and TP53 protein to maintain the genome stability during DNA replication. In BRCA1 mutation carriers, heterozygous germline mutations decrease BRCA1 mRNA levels in luminal progenitors (namely, haploinsufficiency). It has been reported that DNA damage accumulates in luminal progenitor cells in normal breast tissues from BRCA1 mutation carriers (44), which in turn triggers p53-dependent checkpoint control and leads to apoptosis of cells (37). In contrast, TP53 mRNA expression in luminal progenitor cells was lower in BRCA1 mutation carriers than that in noncarriers; thus, the function of p53-dependent checkpoint control may be compromised. This was in line with the observation that luminal progenitor cells in BRCA1 mutation carriers indeed showed higher proliferation than those in noncarriers (10, 44). Moreover, a recent study suggested that BRCA1 and TP53 loss gradually triggers a luminal to basal/mesenchymal transition in mouse model (14). Intriguingly, our scRNA-seq data also showed the luminal progenitors in normal breast tissues from BRCA1 mutation carriers exhibited upregulated basal/mesenchymal features, suggesting that BRCA1 haploinsufficiency only occurred in luminal progenitors and might have gradually triggered a luminal progenitor to basal/mesenchymal transition before tumorigenesis. Of note, we observed that this luminal progenitor to basal/mesenchymal transition was further greatly enhanced during tumorigenesis and eventually basal-like cancers occurred in BRCA1 mutation carriers, suggesting that the increased basal/mesenchymal features of luminal progenitors might be the precursor of basal-like tumor.

Abnormalities of luminal progenitors in normal breast tissues may explain the predisposition of developing basal-like tumors in BRCA1 mutation carriers: (i) the original cells of basal-like tumors (luminal progenitors) were significantly increased in BRCA1 mutation carriers; (ii) luminal progenitors have a long lifetime and strong proliferation capacity (45); and (iii) BRCA1 germline mutations downregulated BRCA1 and TP53 mRNA levels in luminal progenitors, creating unique conditions for the rapid accumulation of mutations and triggering a luminal to basal/mesenchymal transition (37).

There are three limitations in this study. First, the sample size of scRNA-seq in this study was very small, and more samples for scRNA-seq were needed to confirm our findings. Therefore, interpretation of the findings in this study should be caution. Second, scRNA-seq data only revealed a correlation between the cell of origin of ERneg/ERhigh tumors and luminal progenitor/mature cells. Thus, in vivo lineage tracing or in vitro transformation assays are needed to fully illustrate the cell of origin of breast cancer in BRCA1 mutation carriers. Third, our data suggested that the expansion of luminal progenitors and basal progenitors in BRCA1 mutation carriers might be owing to the downregulation of wnt/notch/hedgehog signaling pathways in luminal cells and TP63 in basal cells, respectively. Taken together, independent studies and more samples are needed to validate our findings.

In summary, our scRNA-seq data provided in-depth evidence for the putative cellular origin and evolution of breast cancers in BRCA1 mutation carriers. We revealed that the aberrant luminal progenitors might contribute to basal-like breast cancers through upregulation of basal/mesenchymal features during and even before tumorigenesis, while ERhigh luminal tumors might originate from mature luminal cells, respectively. These findings may shed light on the current understanding of breast tumor susceptibility and development in BRCA1 mutation carriers and are useful for developing new targeted prevention strategies.

No disclosures were reported.

L. Hu: Data curation, formal analysis, validation, investigation, visualization, methodology, writing–original draft. L. Su: Investigation. H. Cheng: Formal analysis. C. Mo: Formal analysis. T. Ouyang: Resources. J. Li: Resources. T. Wang: Resources. Z. Fan: Resources. T. Fan: Resources. B. Lin: Resources. J. Zhang: Resources, formal analysis, methodology. Y. Xie: Conceptualization, resources, data curation, formal analysis, supervision, funding acquisition, investigation, writing–original draft, writing–review and editing.

This study was supported by grant numbers 81372832, 81974422, and 81772824 from the National Natural Science Foundation of China.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Turner
NC
,
Reis-Filho
JS
. 
Basal-like breast cancer and the BRCA1 phenotype
.
Oncogene
2006
;
25
:
5846
53
.
2.
Sun
J
,
Meng
H
,
Yao
L
,
Lv
M
,
Bai
J
,
Zhang
J
, et al
Germline mutations in cancer susceptibility genes in a large series of unselected breast cancer patients
.
Clin Cancer Res
2017
;
23
:
6113
9
.
3.
Smart
CE
,
Wronski
A
,
French
JD
,
Edwards
SL
,
Asselin-Labat
ML
,
Waddell
N
, et al
Analysis of Brca1-deficient mouse mammary glands reveals reciprocal regulation of Brca1 and c-kit
.
Oncogene
2010
;
30
:
1597
607
.
4.
Liu
S
,
Ginestier
C
,
Charafe-Jauffret
E
,
Foco
H
,
Kleer
CG
,
Merajver
SD
, et al
BRCA1 regulates human mammary stem/progenitor cell fate
.
Proc Natl Acad Sci U S A
2008
;
105
:
1680
5
.
5.
Kubista
M
,
Rosner
M
,
Kubista
E
,
Bernaschek
G
,
Hengstschläger
M
. 
Brca1 regulates in vitro differentiation of mammary epithelial cells
.
Oncogene
2002
;
21
:
4747
56
.
6.
Chiang
HC
,
Zhang
X
,
Li
J
,
Zhao
X
,
Chen
J
,
Wang
HTH
, et al
BRCA1-associated R-loop affects transcription and differentiation in breast luminal epithelial cells
.
Nucleic Acids Res
2019
;
47
:
5086
99
.
7.
Xu
X
,
Wagner
KU
,
Larson
D
,
Weaver
Z
,
Li
C
,
Ried
T
, et al
Conditional mutation of Brca1 in mammary epithelial cells results in blunted ductal morphogenesis and tumour formation
.
Nat Genet
1999
;
22
:
37
43
.
8.
Bai
F
,
Smith
MD
,
Chan
HL
,
Pei
XH
. 
Germline mutation of Brca1 alters the fate of mammary luminal cells and causes luminal-to-basal mammary tumor transformation
.
Oncogene
2012
;
32
:
2715
25
.
9.
Bai
F
,
Chan
HL
,
Scott
A
,
Smith
MD
,
Fan
C
,
Herschkowitz
JI
, et al
BRCA1 suppresses epithelial-to-mesenchymal transition and stem cell dedifferentiation during mammary and tumor development
.
Cancer Res
2014
;
74
:
6161
72
.
10.
Lim
E
,
Vaillant
F
,
Wu
D
,
Forrest
NC
,
Pal
B
,
Hart
AH
, et al
Aberrant luminal progenitors as the candidate target population for basal tumor development in BRCA1 mutation carriers
.
Nat Med
2009
;
15
:
907
13
.
11.
Proia
TA
,
Keller
PJ
,
Gupta
PB
,
Klebba
I
,
Jones
AD
,
Sedic
M
, et al
Genetic predisposition directs breast cancer phenotype by dictating progenitor cell fate
.
Cell Stem Cell
2011
;
8
:
149
63
.
12.
Heerma van Voss
MR
,
van der Groep
P
,
Bart
J
,
van der Wall
E
,
van Diest
PJ
. 
Expression of the stem cell marker ALDH1 in BRCA1 related breast cancer
.
Cell Oncol
2011
;
34
:
3
10
.
13.
Molyneux
G
,
Geyer
FC
,
Magnay
FA
,
McCarthy
A
,
Kendrick
H
,
Natrajan
R
, et al
BRCA1 basal-like breast cancers originate from luminal epithelial progenitors and not from basal stem cells
.
Cell Stem Cell
2010
;
7
:
403
17
.
14.
Wang
H
,
Xiang
D
,
Liu
B
,
He
A
,
Randle
HJ
,
Zhang
KX
, et al
Inadequate DNA damage repair promotes mammary transdifferentiation, leading to BRCA1 breast cancer
.
Cell
2019
;
178
:
135
51
.
15.
Karaayvaz
M
,
Cristea
S
,
Gillespie
SM
,
Patel
AP
,
Mylvaganam
R
,
Luo
CC
, et al
Unravelling subclonal heterogeneity and aggressive disease states in TNBC through single-cell RNA-seq
.
Nat Commun
2018
;
9
:
3588
.
16.
Bach
K
,
Pensa
S
,
Grzelak
M
,
Hadfield
J
,
Adams
DJ
,
Marioni
JC
, et al
Differentiation dynamics of mammary epithelial cells revealed by single-cell RNA sequencing
.
Nat Commun
2017
;
8
:
2128
.
17.
Pal
B
,
Chen
Y
,
Vaillant
F
,
Jamieson
P
,
Gordon
L
,
Rios
AC
, et al
Construction of developmental lineage relationships in the mouse mammary gland by single-cell RNA profiling
.
Nat Commun
2017
;
8
:
1627
.
18.
Fu
NY
,
Nolan
E
,
Lindeman
GJ
,
Visvader
JE
. 
Stem cells and the differentiation hierarchy in mammary gland development
.
Physiol Rev
2020
;
100
:
489
523
.
19.
Nguyen
QH
,
Pervolarakis
N
,
Blake
K
,
Ma
D
,
Davis
RT
,
James
N
, et al
Profiling human breast epithelial cells using single cell RNA sequencing identifies cell diversity
.
Nat Commun
2018
;
9
:
2028
.
20.
Chung
W
,
Eum
HH
,
Lee
HO
,
Lee
KM
,
Lee
HB
,
Kim
KT
, et al
Single-cell RNA-seq enables comprehensive tumour and immune cell profiling in primary breast cancer
.
Nat Commun
2017
;
8
:
15081
.
21.
Attwell
D
,
Mishra
A
,
Hall
CN
,
O'Farrell
FM
,
Dalkara
T
. 
What is a pericyte?
J Cereb Blood Flow Metabolism
2015
;
36
:
451
5
.
22.
Qiu
X
,
Mao
Q
,
Tang
Y
,
Wang
L
,
Chawla
R
,
Pliner
HA
, et al
Reversed graph embedding resolves complex single-cell trajectories
.
Nat Methods
2017
;
14
:
979
82
.
23.
Subramanian
A
,
Tamayo
P
,
Mootha
VK
,
Mukherjee
S
,
Ebert
BL
,
Gillette
MA
, et al
Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles
.
Proc Natl Acad Sci U S A
2005
;
102
:
15545
50
.
24.
Kröger
C
,
Afeyan
A
,
Mraz
J
,
Eaton
EN
,
Reinhardt
F
,
Khodor
YL
, et al
Acquisition of a hybrid E/M state is essential for tumorigenicity of basal breast cancer cells
.
Proc Natl Acad Sci U S A
2019
;
116
:
7353
62
.
25.
Sengodan
SK
,
KH
S
,
Nadhan
R
,
Srinivas
P
. 
Regulation of epithelial to mesenchymal transition by BRCA1 in breast cancer
.
Crit Rev Oncol Hematol
2018
;
123
:
74
82
.
26.
Yu
G
,
Wang
LG
,
Han
Y
,
He
QY
. 
clusterProfiler: an R package for comparing biological themes among gene clusters
.
OMICS
2012
;
16
:
284
7
.
27.
Jiang
YZ
,
Ma
D
,
Suo
C
,
Shi
J
,
Xue
M
,
Hu
X
, et al
Genomic and transcriptomic landscape of triple-negative breast cancers: subtypes and treatment strategies
.
Cancer Cell
2019
;
35
:
428
40
.
e5
.
28.
Keymeulen
AV
,
Rocha
AS
,
Ousset
M
,
Beck
B
,
Bouvencourt
G
,
Rock
J
, et al
Distinct stem cells contribute to mammary gland development and maintenance
.
Nature
2011
;
479
:
189
93
.
29.
van Amerongen
R
,
Bowman
A
,
Nusse
R
. 
Developmental stage and time dictate the fate of Wnt/β-catenin-responsive stem cells in the mammary gland
.
Cell Stem Cell
2012
;
11
:
387
400
.
30.
Davis
FM
,
Lloyd-Lewis
B
,
Harris
OB
,
Kozar
S
,
Winton
DJ
,
Muresan
L
, et al
Single-cell lineage tracing in the mammary gland reveals stochastic clonal dispersion of stem/progenitor cell progeny
.
Nat Commun
2016
;
7
:
13053
.
31.
Gu
B
,
Watanabe
K
,
Sun
P
,
Fallahi
M
,
Dai
X
. 
Chromatin effector Pygo2 mediates wnt-notch crosstalk to suppress luminal/alveolar potential of mammary stem and basal cells
.
Cell Stem Cell
2013
;
13
:
48
61
.
32.
Yalcin-Ozuysal
Ö
,
Fiche
M
,
Guitierrez
M
,
Wagner
KU
,
Raffoul
W
,
Brisken
C
. 
Antagonistic roles of Notch and p63 in controlling mammary epithelial cell fates
.
Cell Death Differ
2010
;
17
:
1600
12
.
33.
Lafkas
D
,
Rodilla
V
,
Huyghe
M
,
Mourao
L
,
Kiaris
H
,
Fre
S
. 
Notch3 marks clonogenic mammary luminal progenitor cells in vivo
.
J Cell Biol
2013
;
203
:
47
56
.
34.
Bouras
T
,
Pal
B
,
Vaillant
F
,
Harburg
G
,
Asselin-Labat
ML
,
Oakes
SR
, et al
Notch signaling regulates mammary stem cell function and luminal cell-fate commitment
.
Cell Stem Cell
2008
;
3
:
429
41
.
35.
Liu
S
,
Dontu
G
,
Mantle
ID
,
Patel
S
,
Ahn
N
,
Jackson
KW
, et al
Hedgehog signaling and Bmi-1 regulate self-renewal of normal and malignant human mammary stem cells
.
Cancer Res
2006
;
66
:
6063
71
.
36.
Koboldt
DC
,
Fulton
RS
,
McLellan
MD
,
Schmidt
H
,
Kalicki-Veizer
J
,
McMichael
JF
, et al
Comprehensive molecular portraits of human breast tumours
.
Nature
2012
;
490
:
61
70
.
37.
Xu
X
,
Qiao
W
,
Linke
SP
,
Cao
L
,
Li
WM
,
Furth
PA
, et al
Genetic interactions between tumor suppressors Brca1 and p53 in apoptosis, cell cycle and tumorigenesis
.
Nat Genet
2001
;
28
:
266
71
.
38.
Gusterson
BA
,
Ross
DT
,
Heath
VJ
,
Stein
T
. 
Basal cytokeratins and their relationship to the cellular origin and functional classification of breast cancer
.
Breast Cancer Res
2005
;
7
:
143
8
.
39.
Dontu
G
,
Ince
TA
. 
Of mice and women: a comparative tissue biology perspective of breast stem cells and differentiation
.
J Mammary Gland Biol Neoplasia
2015
;
20
:
51
62
.
40.
Asselin-Labat
ML
,
Sutherland
KD
,
Barker
H
,
Thomas
R
,
Shackleton
M
,
Forrest
NC
, et al
Gata-3 is an essential regulator of mammary-gland morphogenesis and luminal-cell differentiation
.
Nat Cell Biol
2006
;
9
:
201
9
.
41.
Kouros-Mehr
H
,
Slorach
EM
,
Sternlicht
MD
,
Werb
Z
. 
GATA-3 maintains the differentiation of the luminal cell fate in the mammary gland
.
Cell
2006
;
127
:
1041
55
.
42.
Buckley
NE
,
Mullan
PB
. 
BRCA1 – conductor of the breast stem cell orchestra: the role of BRCA1 in mammary gland development and identification of cell of origin of BRCA1 mutant breast cancer
.
Stem Cell Rev Rep
2012
;
8
:
982
93
.
43.
Chaffer
CL
,
Weinberg
RA
. 
Cancer cell of origin: spotlight on luminal progenitors
.
Cell Stem Cell
2010
;
7
:
271
2
.
44.
Sau
A
,
Lau
R
,
Cabrita
MA
,
Nolan
E
,
Crooks
PA
,
Visvader
JE
, et al
Persistent activation of NF-κB in BRCA1-deficient mammary progenitors drives aberrant proliferation and accumulation of DNA damage
.
Cell Stem Cell
2016
;
19
:
52
65
.
45.
Villadsen
R
,
Fridriksdottir
AJ
,
Rønnov-Jessen
L
,
Gudjonsson
T
,
Rank
F
,
LaBarge
MA
, et al
Evidence for a stem cell hierarchy in the adult human breast
.
J Cell Biol
2007
;
177
:
87
101
.

Supplementary data