Comprehensive single-cell analysis reveals cellular heterogeneity and immune interactions in colorectal cancer

• A single-cell atlas of 18, 741 colorectal cancer cells delineates six major cell populations and reveals pronounced inter-patient heterogeneity. • T/NK cells segregate into four distinct clusters, with C3 exhibiting terminal differentiation and heightened inflammatory activity. • Bidirectional communication between T/NK cells and epithelial cells highlights active immune–tumor interactions shaping the tumor microenvironment. • A C3-based gene signature robustly predicts patient survival and associates with immune infiltration and tumor mutation burden. • LAT is identified as a key oncogenic regulator, promoting proliferation, migration, and invasion while suppressing apoptosis in colorectal cancer cells. Colorectal cancer (CRC) presents considerable therapeutic challenges due to its diverse cellular composition and intricate microenvironment. Our study utilized single-cell RNA sequencing (scRNA-seq) on CRC tissues, examining 18, 741 individual cells, which were grouped into six primary cell populations: epithelial, fibroblast, endothelial, T and NK, B, and myeloid. The epithelial cells exhibited notable variations in gene copy numbers. Within TNK cells, we identified four distinct subsets. CytoTRACE analysis indicated that subtype C3 exhibited lower differentiation potential, whereas subtypes C0 and C1 showed higher differentiation potential. Consistently, Monocle pseudotime trajectory analysis positioned C3 cells at the terminal stage of differentiation, while C0 cells were enriched at the early stage of the developmental trajectory, suggesting functional heterogeneity among T/NK subpopulations. Through functional analyses with GSVA and ssGSEA, subtype C3 displayed the highest inflammation-associated activity scores. Further exploration of transcription factors defined three unique regulatory clusters among TNK cells, illuminating their gene-regulation networks. We developed a prognostic signature using markers from subtype C3 TNK cells combined with age-associated genes, revealing a significant correlation with patient survival outcomes. This prognostic model proved effective in categorizing CRC patients according to risk. Additionally, immune profiling employing ESTIMATE, CIBERSORT, and Xcell algorithms underscored the complexity of immune cell populations within CRC tumors. Analysis of tumor mutational burden (TMB) highlighted differential patterns between patient groups and its relationship to prognostic risk levels. Collectively, these insights provide a detailed perspective on CRC cell diversity and immune dynamics, supporting the advancement of targeted and personalized therapeutic interventions.