Identification of RCN3-positive cancer-associated fibroblasts as novel drivers of colorectal cancer through a TGF-β pathway-based risk signature

作者：Weichun Tang, Linlin Zhu, Qiang Li, Biao Gu, Pengpeng Xu, Nannan Wang, Shangshang Hu · 发表于：International Journal of Surgery · 年份：2025 · DOI：10.1097/js9.0000000000003825 · 被引用次数：2 · 研究领域：TGF-β signaling in diseases、Colorectal Cancer Treatments and Studies、Cancer Cells and Metastasis

BACKGROUND: Transforming growth factor-β (TGF-β) signaling shapes tumor biology and immunity, yet its clinical impact in colorectal cancer (CRC) is unclear. METHODS: 3000 CRC specimens integrating bulk, single-cell, and spatial transcriptomes plus proteomics were profiled. Single-sample gene-set enrichment scored TGF-β activity; consensus clustering defined subtypes. A 12-gene TGF-β-related risk signature (TGFRS) was built by machine learning and interpreted with SHapley Additive exPlanations. Key genes were validated in vitro and in vivo . Drug-sensitivity screening, molecular docking, and molecular-dynamics simulations nominated candidate inhibitors. RESULTS: High TGF-β scores or subtype C2 independently predicted poor survival and resistance to immune-checkpoint blockade. TGFRS outperformed 139 published signatures for prognosis and immunotherapy response and remained significant in multivariable models. Reticulocalbin-3 (RCN3) carried the greatest TGFRS weight and was restricted to cancer-associated fibroblasts (CAFs). RCN3-positive CAFs (RCN3⁺Fib) exhibited hyper-activated TGF-β signaling, neighbored tumor cells and M2 macrophages, and enhanced epithelial-mesenchymal transition, invasion, and M2-like polarization in coculture and xenografts. TGF-β induced RCN3 through RELB, creating a positive feedback loop. Among screened compounds, dasatinib showed the greatest sensitivity in RCN3⁺Fib. CONCLUSIONS: Comprehensive multi-omics mapping links heightened TGF-β activity and a...