Molecular subtyping of colorectal cancer based on aging-related gene network perturbation analysis and functional validation of MAP1B

Molecular heterogeneity of colorectal cancer (CRC) remains a major barrier to precision therapy. In this study, we integrated eight public aging-related resources to construct an aging-related gene set and performed network perturbation analysis on the TCGA CRC cohort using a Reactome-based reference network. Unsupervised consensus clustering identified three molecular subtypes (C1, C2, C3). Nearest Template Prediction (NTP) validation across five independent GEO cohorts confirmed cross-cohort stability. Multi-dimensional biological characterization revealed that C1 is enriched for EMT activation, stromal infiltration, and immune exclusion; C2 for metabolic reprogramming dominated by lipid and glycerophospholipid handling; and C3 for an active cell cycle, high tumor mutational burden, and the lowest predicted immune dysfunction and exclusion. Survival analysis demonstrated that both C1 and C2 exhibit poor overall prognosis, while C3 has the most favorable outcome, with statistically significant inter-subtype differences. Through multi-cohort Cox regression cross-screening, we identified MAP1B as a core candidate driver gene of the C1 subtype. Its high expression was significantly associated with distant metastasis, lymph node metastasis, and poor prognosis, and was further validated at the protein level by immunohistochemistry. Functional experiments confirmed that MAP1B silencing significantly inhibited CRC cell proliferation, invasion, and migration, reversed EMT-related protein expression patterns, and suppressed tumor growth in nude mouse xenograft models. GDSC2 drug sensitivity analysis suggested that MAP1B-high CRC is more sensitive to dasatinib, and combining MAP1B silencing with dasatinib produced the largest reduction in invasion, migration and xenograft growth, although the added effect over MAP1B silencing alone did not reach significance in vivo. This study establishes a CRC subtyping framework based on aging-related gene network perturbation and provides preclinical evidence for MAP1B as an intervention target in EMT-activated CRC.