Single-Cell RNA Sequencing of a Postmenopausal Normal Breast Tissue Identifies Multiple Cell Types That Contribute to Breast Cancer.

Abstract

Simple SummaryThe human body is composed of multiple cell types that form structures and carry out the functions of specific tissues. The human breast is mainly known for the milk ducts organized by epithelial cells, but also contains many other cell types of little-known identity. In this study, we employed the single-cell sequencing technology to ascertain the various cell types present in the normal breast. The results showed 10 distinct cell types that included three epithelial and other novel cell types. The gene signatures of five cell types (three epithelial, one fibroblast subset, and immune cells) matched to the gene expression profiles of >85% breast tumors cataloged in The Cancer Gene Atlas dataset, suggesting their significant contribution to breast cancer. These findings provide a framework for the better mapping of the cellular composition in the breast and its relationship to breast disease.The human breast is composed of diverse cell types. Studies have delineated mammary epithelial cells, but the other cell types in the breast have scarcely been characterized. In order to gain insight into the cellular composition of the tissue, we performed droplet-mediated RNA sequencing of 3193 single cells isolated from a postmenopausal breast tissue without enriching for epithelial cells. Unbiased clustering analysis identified 10 distinct cell clusters, seven of which were nonepithelial devoid of cytokeratin expression. The remaining three cell clusters expressed cytokeratins (CKs), representing breast epithelial cells; Cluster 2 and Cluster 7 cells expressed luminal and basal CKs, respectively, whereas Cluster 9 cells expressed both luminal and basal CKs, as well as other CKs of unknown specificity. To assess which cell type(s) potentially contributes to breast cancer, we used the differential gene expression signature of each cell cluster to derive gene set variation analysis (GSVA) scores and classified breast tumors in The Cancer Gene Atlas (TGGA) dataset (n = 1100) by assigning the highest GSVA scoring cell cluster number for each tumor. The results showed that five clusters (Clusters 2, 3, 7, 8, and 9) could categorize >85% of breast tumors collectively. Notably, Cluster 2 (luminal epithelial) and Cluster 3 (fibroblast) tumors were equally prevalent in the luminal breast cancer subtypes, whereas Cluster 7 (basal epithelial) and Cluster 9 (other epithelial) tumors were present primarily in the triple-negative breast cancer (TNBC) subtype. Cluster 8 (immune) tumors were present in all subtypes, indicating that immune cells may contribute to breast cancer regardless of the subtypes. Cluster 9 tumors were significantly associated with poor patient survival in TNBC, suggesting that this epithelial cell type may give rise to an aggressive TNBC subset.