Spatial genetic mapping links shared inflammatory bowel disease liability to adult immune-epithelial lesion contexts

Inflammatory bowel disease (IBD), including Crohn's disease (CD) and ulcerative colitis (UC), shows marked clinical heterogeneity despite a shared immune-genetic background. The adult spatial contexts through which inherited IBD susceptibility is expressed remain unclear.

We integrated GWAS summary statistics for overall IBD, CD, and UC with LDSC, stratified LDSC, LDSC-SEG, MAGMA, PoPS, and genetically informed spatial mapping (gsMap). Human genetic signals were projected onto the E16.5 mouse single-cell spatial atlas as an exploratory developmental reference, and adult disease-tissue spatial support was assessed across SCP2959 CD spatial sections and GSE189184 idiopathic UC inflamed Visium sections. PoPS-independent MAGMA-only module-score and FUSION-TWAS sensitivity analyses, together with targeted RT-qPCR in NCM460 epithelial cells and THP-1-derived macrophage-like cells were performed.

LDSC showed strong positive genetic correlations among overall IBD, CD, and UC, including overall IBD versus CD (r_g = 0.9458, P = 1.79 × 10^-8), overall IBD versus UC (estimated r_g = 1.0907, P = 3.48 × 10^-6), and CD versus UC (r_g = 0.9535, P = 0.0133). MAGMA and PoPS prioritized immune-inflammatory candidates, including IL23R, JAK2, STAT3, CCL2, NOD2, and HLA-region genes. Exploratory developmental gsMap showed nominal signals in gastrointestinal, liver, smooth-muscle, epidermal, and neural regions. In adult disease-tissue gsMap, overall IBD signals showed FDR-significant enrichment in SCP2959 CD immune regions (ACAT P = 3.52 × 10^-7, q = 1.41 × 10^-6), lamina propria (ACAT P = 2.75 × 10^-5, q = 4.27 × 10^-5), follicular clusters (ACAT P = 7.31 × 10^-7, q = 1.10 × 10^-5), and myeloid clusters (ACAT P = 6.67 × 10^-6, q = 3.34 × 10^-5). In GSE189184 idiopathic UC inflamed tissue, enrichment was observed in GWAS-independent immune-rich (ACAT P = 1.72 × 10^-5, q = 1.38 × 10^-4), structural/barrier (ACAT P = 7.52 × 10^-5, q = 3.01 × 10^-4), epithelial-mucosal (ACAT P = 9.08 × 10^-4, q = 0.00182), inflammation-repair (ACAT P = 0.00140, q = 0.00224), and stromal-fibrotic domains (ACAT P = 0.00268, q = 0.00357). MAGMA-only module-score and FUSION-TWAS sensitivity analyses provided PoPS-independent support for the adult lesion-context interpretation. RT-qPCR showed that JAK2 knockdown reduced cytokine-induced CCL2 and CXCL8 by 52.6% and 36.9% and partially restored OCLN expression, while LPS induced IL1B, TNF, CCL2, and PYCARD in macrophage-like cells.

Shared IBD genetic liability was most consistently linked to an adult immune-epithelial inflammatory lesion program involving immune-rich, epithelial-inflammatory, myeloid/follicular, lamina propria, structural/barrier, and remodeling-associated contexts. Developmental and subtype-weighted spatial signals, including neural-related signals in the embryonic reference, should be viewed as hypothesis-generating clues to developmental and neuroimmune programs rather than definitive subtype-specific mechanisms.