Spatial and single-cell transcriptomics decipher butyrate metabolic reprogramming and its impact on colorectal cancer progression and tumor microenvironment remodeling

Background Colorectal cancer (CRC) is closely linked to dietary and microbial factors. Butyrate, a major gut microbial metabolite with histone deacetylase-inhibitory and immunomodulatory effects, contributes to epithelial homeostasis, but the spatial heterogeneity and clinical relevance of butyrate metabolism in CRC remain unclear. Methods Multi-algorithm single-cell metabolic scoring was used to quantify butyrate metabolism and define tumor epithelial subgroups, which were mapped spatially using cell2location. CellChat, Monocle, pySCENIC, inferCNV, PROGENy, and stLearn were applied to characterize lineage states, transcriptional regulation, signaling activity, and intercellular crosstalk. Ensemble machine learning was used for prognostic modeling, followed by molecular and functional validation. Results We identified malignant epithelial cells with low butyrate metabolism (LowBTAepis), which were associated with poor prognosis, early pseudotime states, increased stemness and proliferation, and distinct copy-number variation (CNV) patterns. Spatial analysis revealed preferential interactions between LowBTAepis and fibroblasts. HOXB8 emerged as a key transcriptional regulator of LowBTAepis. Sodium butyrate (NaB) suppressed CRC cell proliferation, migration, and invasion in a dose-dependent manner, accompanied by reduced HOXB8 expression, inhibition of AKT/mTOR signaling, and changes in apoptosis- and metastasis-related proteins. A butyrate metabolism-related model showed strong prognostic performance and was associated with an immunosuppressive tumor microenvironment. Conclusions LowBTAepis represents an aggressive CRC epithelial state characterized by stromal crosstalk and malignant transcriptional features. Butyrate may suppress CRC progression partly through the HOXB8/AKT/mTOR axis, supporting its potential relevance to metabolic intervention and patient stratification.