Single-Cell Multiomics Integration Reveals Cell Trajectories and Regulatory Networks in Breast Cancer Progression

Abstract

Background

Breast cancer is a molecularly heterogeneous disease encompassing distinct subtypes with divergent prognoses and therapeutic responses. Despite extensive bulk genomic characterization, the single-cell transcriptional trajectories, chromatin regulatory landscapes, and intercellular communication networks that drive breast cancer progression from luminal to mesenchymal states remain incompletely understood. Emerging evidence implicates gut microbiome dysbiosis as a systemic modulator of breast cancer risk, immune regulation, and treatment response; however, its coordinated characterization alongside high-resolution single-cell profiling has not been undertaken. Integrating single-cell RNA sequencing (scRNA-seq) with chromatin accessibility profiling (scATAC-seq), cell-cell communication inference, and gut microbiome 16S rRNA analysis offers a comprehensive multilayered approach to dissecting both the intratumoral regulatory logic and the systemic microbial landscape of breast cancer progression.

Methods

We performed integrated multiomics analysis of 42,037 high-quality single cells from breast cancer patient samples obtained from the GEO database (GSE161529), encompassing tumor and adjacent normal tissue. scRNA-seq data were processed using Seurat v4.3.0 and Scanpy v1.9.1. Dimensionality reduction, Leiden clustering, differential expression analysis, pseudotime trajectory inference (Monocle3, scVelo, Slingshot), SCENIC regulon analysis, chromatin accessibility profiling, cell-cell communication inference (CellChat), and functional enrichment analysis were applied. Key marker genes were validated by quantitative RT-PCR in MCF-7 and MDA-MB-231 cell lines. In parallel, gut microbiome 16S rRNA amplicon sequencing data from 599 matched breast cancer–healthy control pairs (266 genera, four BioProjects) were analyzed for alpha/beta diversity, differential genus abundance (Mann–Whitney U with Benjamini–Hochberg correction), genus co-occurrence networks, and random forest breast cancer classification.

Results

Unbiased clustering identified 12 observed cell populations spanning luminal (Luminal A, Luminal B), basal, HER2+, cancer-associated fibroblast (CAF), tumor-associated macrophage (TAM), CD8 T cell, regulatory T cell (Treg), NK cell, B cell, endothelial, and pericyte compartments. Quality control confirmed robust data quality with median 2400–3100 genes per cell and mitochondrial content below 20%. Pseudotime trajectory analysis revealed a continuous epithelial-to-mesenchymal transition (EMT) spectrum with three divergent lineages: EMT, senescence, and immune activation branches. SCENIC identified SNAI1(+), ZEB1(+), and FOXM1(+) as the most differentially activated regulons in basal/mesenchymal cells. ATAC-seq chromatin accessibility profiling confirmed EMT transcription factor access through distal enhancers. CellChat identified TGF-β1–TGFBR1 (CAF → tumor) and PD-L1–PD-1 (tumor → CD8 T cell) as dominant intercellular communication axes. Multiomics integration enhanced cell-type classification AUC from 0.76 (scRNA-seq alone) to 0.91 (full multiomics). EMT-high tumor status was associated with significantly inferior overall survival (HR = 2.14, p = 0.001). qRT-PCR validation confirmed ERBB2 enrichment in HER2+ lines and KRT5/VIM enrichment in basal/TNBC lines (all p < 0.001). Gut microbiome analysis revealed significant dysbiosis in breast cancer patients versus healthy controls, with reduced Shannon diversity (1.94 vs. 2.06, p = 6.1 × 10⁻⁴), significant community-level separation (Bray–Curtis PERMANOVA F = 30.2, p = 0.001), and nine differentially abundant genera including enriched Faecalitalea (log₂FC = 1.44), Parasutterella (log₂FC = 1.21), and Bacteroides (log₂FC = 1.19). A random forest microbiome classifier achieved a cross-validated AUC of 0.79 (95% CI: 0.74–0.82), supporting the gut–breast axis as a systemic codeterminant of cancer state.

Conclusions

This integrated single-cell multiomics and gut microbiome atlas of breast cancer progression characterizes the transcriptional regulatory architecture driving cellular trajectory diversification, highlights SNAI1–ZEB1–FOXM1 coregulatory networks as candidate therapeutic targets, suggests that multiomics integration can enhance molecular stratification of breast cancer subtypes with clinical prognostic relevance, and identifies gut microbiome dysbiosis as a potential systemic dimension—characterized by loss of microbial diversity and enrichment of specific genera—as a systemic dimension of breast cancer pathobiology that warrants integration with intratumoral molecular profiling.