Integrative microbiome-methylome analysis links bacterial taxa to colorectal cancer histological subtypes

Colorectal cancer (CRC) arises through distinct molecular and histological routes that may be shaped by interactions between the mucosa-associated microbiota and host epigenetic regulation. We analyzed paired tumor and adjacent non-tumor colorectal mucosa from CRC patients stratified by histological subtype (conventional vs serrated-pathway groups). Microbiota composition was profiled by Illumina sequencing, and host DNA methylation was assessed using genome-wide CpG arrays with targeted validation. Alpha diversity showed modest tumor-non-tumor differences, with variation by anatomical location driven by distal tumors. Differential abundance testing identified tumor-associated genera, and linear discriminant modeling highlighted taxa with high discriminatory power, including Bacteroides, Eubacterium, Fusobacterium, and Acinetobacter. Methylome ordination separated tumor from non-tumor samples and revealed prominent contributions of zinc finger (ZNF) loci. Integrative latent-variable modeling (PLS/sparse partial least squares and multi-block sparse partial least squares discriminant analysis) supported coordinated microbiome-methylome variation distinguishing tumor from non-tumor mucosa, prioritized a limited set of bacterial and methylation signals, and suggested partial discrimination between conventional and serrated tumor profiles, particularly in the methylome block. Focusing on Fusobacterium, methylation changes in selected host loci were associated with its abundance, and qPCR-based quantification of Fusobacterium nucleatum correlated with CpG methylation at ZNF788. F. nucleatum levels were also associated with serrated/microsatellite instability-related CRC features and more advanced disease. These findings identify subtype-aware microbiome-methylome signatures in CRC and highlight ZNF788 methylation as a candidate epigenetic correlate of intratumoral F. nucleatum burden.IMPORTANCEColorectal cancer does not develop in a single way, and different tumor types may interact differently with bacteria living on the bowel lining. This study examined both tissue-associated bacteria and DNA methylation, an epigenetic mark that helps regulate genes, in paired tumor and nearby non-tumor tissue from patients with colorectal cancer. By analyzing these two layers together, we identified microbial and host methylation patterns linked to tumor tissue and to serrated-pathway cancers. In particular, Fusobacterium nucleatum was associated with serrated/microsatellite instability-related features, more advanced disease, and methylation of the host gene ZNF788. These findings suggest that combining microbiome and epigenetic information may help explain why colorectal cancer subtypes behave differently and may support future subtype-aware biomarkers.