Multi-omics integration analysis based on plasma circulating proteins reveals potential therapeutic targets for ulcerative colitis

Jihai Zhou¹Wenwen Zhao¹Yiping Lin²Bo Yang³Dongjie Sun⁴Zhu Liu^1*

• ¹Department of Gastroenterology, Shandong Provincial Maternal and Child Health Care Hospital Affiliated to Qingdao University, Jinan, China
• ²Department of Health Medicine, 900th Hospital of PLA Joint Logistic Support Force, Fuzong Clinical Medical College of Fujian Medical University Fuzhou, Fuzhou, China
• ³Department of Gastroenterology and Hepatology, Guizhou Aerospace Hospital, Zunyi, China
• ⁴Department of Digestive Diseases, 900th Hospital of PLA Joint Logistic Support Force, Fuzong Clinical Medical College of Fujian Medical University Fuzhou, Fuzhou, China

Background: Ulcerative colitis (UC) is a complex inflammatory bowel disease with unclear etiology and challenging molecular mechanisms. This study aims to identify potential diagnostic and therapeutic biomarkers for UC through multi-omics integrative analysis, providing new insights into its precise diagnosis and treatment. Methods: Data samples from the Gene Expression Omnibus database and protein quantitative trait loci data from genome-wide association studies were integrated to identify overlapping genes. Three machine learning (ML) algorithms were employed to screen core hub genes from these overlapping genes, followed by the construction and external validation of a diagnostic model. Single-cell sequencing data were used to explore the expression profiles of core hub genes across different cell types. Additionally, immune infiltration, functional enrichment, and regulatory networks were analyzed. Finally, the expression trends of the core hub genes were validated in a dextran sulfate sodium (DSS)-induced UC mouse model using RT-qPCR. Results: Mendelian randomization (MR) analysis identified 168 plasma proteins causally associated with UC. Differential expression analysis revealed 1,011 DEGs, and the intersection of DEGs and MR results yielded 12 overlapping genes. Four core hub genes, including EIF5A2, IDO1, CDH5, and MYL5, were identified using three ML algorithms. The nomogram model constructed with these four genes demonstrated strong predictive performance, which was further confirmed in an external This is a provisional file, not the final typeset article validation dataset. GSEA analysis revealed that these genes are involved in various biological processes, including immune response, signal transduction, metabolism, and cellular stress. CIBERSORT immune infiltration analysis showed significant differences in immune cell infiltration between UC and normal tissues. Furthermore, a comprehensive mRNA-miRNA-lncRNA regulatory network was constructed, identifying key molecular interactions potentially driving UC pathogenesis. Single-cell RNA sequencing analysis revealed that CDH5 is primarily expressed in endothelial cells, EIF5A2 is enriched in stem cells/T cells, IDO1 is expressed in monocytes, and MYL5 is found in epithelial and endothelial cells. Finally, RT-qPCR validation in the DSS-induced UC mouse model confirmed that the expression changes of core hub genes were consistent with bioinformatics predictions. Conclusion: This study systematically identified core diagnostic genes and their regulatory networks for UC through multi-omics integration.