Integrating single-cell and spatial transcriptomes reveals COL4A1/2 facilitates the spatial organisation of stromal cells differentiation in breast phyllodes tumours

Background: Breast phyllodes tumours (PTs) are a unique type of fibroepithelial neoplasms with metastatic potential and recurrence tendency. However, the precise nature of heterogeneity in breast PTs remains poorly understood. This study aimed to elucidate the cell subpopulations composition and spatial structure and investigate diagnostic markers in the pathogenesis of PTs. Methods: We applied single-cell RNA sequencing and spatial transcriptomes on tumours and adjacent normal tissues for integration analysis. Immunofluorescence experiments were conducted to verify the tissue distribution of cells. Tumour cells from patients with PTs were cultured to validate the function of genes. To validate the heterogeneity, the epithelial and stromal components of tumour tissues were separated using laser capture microdissection, and microproteomics data were obtained using data-independent acquisition mass spectrometry. The diagnostic value of genes was assessed using immunohistochemistry staining. Results: Tumour stromal cells harboured seven subpopulations. Among them, a population of widely distributed cancer-associated fibroblast-like stroma cells exhibited strong communications with epithelial progenitors which underwent a mesenchymal transition. We identified two stromal subpopulations sharing epithelial progenitors and mesenchymal markers. They were inferred to further differentiate into transcriptionally active stromal subpopulations continuously expressing COL4A1/2. The binding of COL4A1/2 with ITGA1/B1 facilitated a growth pattern from the stroma towards the surrounding glands. Furthermore, we found consistent transcriptional changes between intratumoural heterogeneity and inter-patient heterogeneity by performing microproteomics studies on 30 samples from 11 PTs. The immunohistochemical assessment of 97 independent cohorts identified that COL4A1/2 and CSRP1 could aid in accurate diagnosis and grading. Conclusions: Our study demonstrates that COL4A1/2 shapes the spatial structure of stromal cell differentiation and has important clinical implications for accurate diagnosis of breast PTs.