Comprehensive analysis of monocyte palmitoylation-related genes IFITM3 and CCL3-CCR5 axis in crohn's disease...

Crohn's disease (CD) is a chronic inflammatory bowel disease marked by immune imbalance and monocyte dysfunction. IFITM3, a palmitoylation-related immune protein, may play a role in this process, but its involvement in CD remains unclear. This study aimed to explore the causal role of IFITM3 and related proteins in CD using Mendelian randomization, multi-omics analysis, and machine learning, to identify potential diagnostic markers and therapeutic targets.

We performed Mendelian randomization (MR) analysis using pQTL data to identify causal associations between monocyte palmitoylation-related proteins and CD. Multi-omics data were integrated to construct a diagnostic model. A total of 12 machine learning algorithms across 110 combinations-including Enet, GBM, glmBoost, Lasso, LDA, NaiveBayes, plsRglm, RF, Ridge, Stepglm, SVM, and XGBoost-were evaluated. Models were ranked by area under the curve (AUC), and the top-performing model (random forest) was selected. SHAP analysis was applied to interpret feature contributions. Single-cell transcriptomics further explored monocyte communication in CD tissues.

MR analysis provided genetic evidence consistent with a causal role for IFITM3 in CD, suggesting its potential role in driving monocyte-related immune dysregulation. Additionally, four CCL chemokines-CCL3, CCL7, CCL20, and CCL26-were found to be causally associated with CD pathogenesis.The seven core genes (C2, IFITM3, FXYD3, LAX1, CD63, CKAP4, and SPPL2A) identified by Lasso regression achieved a minimum single-gene diagnostic AUC of ≥ 0.79 in the training cohort, showing robust diagnostic and predictive performance. The final random forest model incorporating these seven genes demonstrated strong predictive performance in three independent external validation cohorts (GSE75214, GSE16879, and GSE24287), with AUCs of 0.980, 0.961, and 0.902, respectively. SHAP analysis revealed that C2 had the highest predictive contribution in the model, whereas IFITM3-despite showing moderate predictive importance-had a significant causal association with CD, underscoring its biological relevance. Finally,single-cell transcriptomic analysis suggested enhanced monocyte communication in CD tissues, particularly through the CCL3-CCR5 axis, providing a computational inference that this pathway may contribute to chronic inflammation and disease progression.

Our findings highlight IFITM3 as a key immune-related macromolecule contributing to CD via monocyte-driven inflammation. The identified CCL chemokines and the CCL3-CCR5 axis further elucidate the inflammatory landscape in CD. The diagnostic gene model presents potential for clinical translation, and IFITM3 may serve as a candidate biomarker and a putative therapeutic target for CD, although experimental and prospective validation are required before clinical application.