CLDN8 and ABCA12 define a shared molecular signature in ulcerative colitis-psoriasis comorbidity.

Background: Ulcerative colitis (UC) and psoriasis (PsO) are two common immune-mediated inflammatory diseases with significant comorbidity, yet whether one directly causes the other or they share common genetic and immunopathological backgrounds remains unclear. This study integrated bioinformatics, animal experiments, and bidirectional Mendelian randomization to systematically identify potential common molecular targets mediating the comorbidity. Methods: Transcriptomic data for UC and PsO were obtained from the Gene Expression Omnibus (GEO) database. Differential expression genes (DEGs) and WGCNA were performed to identify common transcriptomic alterations. Then, machine learning algorithms (SVM, RF, XGBoost, GLM) were used to screen for shared signature genes. The nomogram diagnostic models were constructed, and immune infiltration landscape were analyzed. An acute UC-PsO comorbidity mouse model was established by simultaneous administration of dextran sulfate sodium and imiquimod, and the expression levels of key genes were examined. Finally, bidirectional Mendelian randomization (MR) was performed to assess the genetic causal relationship between UC and PsO. Results: 3,840 DEGs in UC and 4,208 in PsO, with 619 overlapping DEGs, were identified. WGCNA further screened 17 up-regulated and 16 down-regulated co-expressed genes common to both diseases. Cross-validation using four machine learning algorithms finally identified CLDN8 and ABCA12 as the core signature genes shared by UC and PsO. Diagnostic models based on these two genes performed well for both diseases (specific AUC values can be added). Immune infiltration analysis revealed similar inflammatory pathway characteristics between the two diseases. In the UC-PsO comorbidity animal model, the expression changes of CLDN8 and ABCA12 were consistent with those in clinical samples, confirming their functional relevance. Bidirectional MR analysis showed no significant causal effect of UC on PsO nor of PsO on UC, which does not support the direct causation hypothesis but instead supports the shared mechanism hypothesis. Conclusion: CLDN8 and ABCA12 shared molecular signature in UC-PsO comorbidity, however, no direct genetic causal relationship between UC and PsO. Clinically, each disease should be managed according to its own pathology, while targeting common inflammatory pathways involving CLDN8/ABCA12 to enable personalized treatment.