Epithelial cell-based multi-omics integration identifies SPINK5 and SRI as novel diagnostic biomarkers for ulcerative...

Ulcerative colitis (UC) is a chronic inflammatory bowel disease characterized by epithelial barrier dysfunction, yet reliable epithelial-derived diagnostic biomarkers remain elusive.

We employed an integrative multi-omics approach, combining single-cell RNA sequencing (scRNA-seq) and bulk transcriptomics from public datasets. Epithelial cell-specific signatures were identified and intersected with differentially expressed genes to pinpoint candidate markers. Four machine learning algorithms-LASSO, Random Forest, Support Vector Machine and K-Nearest Neighbor (KNN)-were applied for robust feature selection. The diagnostic model was validated in external cohorts, and its association with the immune microenvironment was assessed using CIBERSORT. Finally, key findings were experimentally confirmed in clinical UC tissues via qPCR, Western blot, and immunohistochemistry.

scRNA-seq revealed significant transcriptional remodeling of epithelial cells in UC. Cross-algorithm analysis consistently identified SPINK5 and SRI as the most robust diagnostic biomarkers. A model built on these two genes demonstrated exceptional performance, with an area under the curve (AUC) > 0.90 in both training and external validation sets. Immune infiltration analysis revealed a pro-inflammatory shift in UC and delineated distinct correlations of SPINK5 and SRI with specific immune cell subsets. Experimental validation confirmed significant upregulation of SPINK5 and downregulation of SRI in UC patient tissues at both the mRNA and protein levels.

Our study underscores the pivotal role of epithelial dysfunction in UC pathogenesis. By integrating multi-omics and machine learning, we have established SPINK5 and SRI as promising candidate diagnostic biomarkers, providing candidate biomarkers with potential utility for UC diagnosis and a basis for future studies on disease stratification.