Deciphering the tumor ecosystem dynamics undergoing immunochemotherapy therapy across multiple cancer types unveils the immunosuppressive role of S100A4 in fibroblasts by promoting PD-L1 expression in tumor cells.

BACKGROUND: Neoadjuvant therapy (NAT) has transformed cancer treatment by improving surgical outcomes and survival rates, yet resistance mechanisms across multiple cancer types remain unclear. This study aimed to decipher tumor ecosystem dynamics during NAT using cross-cancer single-cell sequencing data, focusing on identifying key mediators of immunosuppression and treatment resistance. METHODS: Single-cell RNA-sequencing (scRNA-seq) datasets from five solid tumors (esophageal squamous cell carcinoma, esophagogastric junction carcinoma, colorectal cancer, cervical cancer, and triple-negative breast cancer) were integrated. The data from these five cancer types underwent a rigorous process to standardize cell types across all datasets. Cell-cell communication analysis, Meta-Programs (MPs) via non-negative matrix factorization, and functional enrichment were performed. Immunohistochemistry (IHC) and Western blot validated S100A4 expression and PD-L1 induction in vitro. RESULTS: We constructed a single-cell map across cancer types and systematically characterized dynamic changes in tumor cells and diverse microenvironmental cell populations following neoadjuvant therapy, along with thier gene expression and pathway alterations. Our findings highlight that crosstalk between cancer-associated fibroblasts (CAFs) and tumor cells represents a critical determinant of neoadjuvant therapy resistance. Fibroblasts underwent significant state transitions post-treatment, marked by hypoxia-associated gene upregulation (e.g., S100A4) and immunosuppressive pathways. Meta-Programs (MPs) analysis identified a hypoxia-driven fibroblast state (MP5) containing S100A4 that correlated with treatment resistance. In vitro experiments, S100A4 co-localized with α-SMA + fibroblasts and directly induced PD-L1 expression in tumor cells, linking CAFs secreted S100A4 to immunosuppressive PD-L1 upregulation. CONCLUSION: This cross-cancer single-cell atlas reveals S100A4, secreted by CAFs, as a conserved mediator of PD-L1 upregulation in tumor cells, driving immunosuppression and resistance to nICT. The atlas and mechanistic findings provide a rationale for targeting S100A4 to enhance treatment efficacy, pending validation in larger cohorts and mechanistic studies. This resource also supports the development of personalized, cross-cancer neoadjuvant strategies.