Single-cell CRISPR perturbation in human adipocytes reveals an omnigenic core remodeling network linking CAD loci to cardiometabolic risk

Background Genome- and transcriptome-wide association studies have identified numerous coronary artery disease (CAD) loci enriched in adipose tissue, yet their functional convergence remains unclear. The omnigenic model predicts that peripheral genes influence disease through shared core regulatory networks. Purpose We aim to determine whether genetically prioritized CAD adipose genes converge on common transcriptional programs and core genes relevant to cardiometabolic disease. Methods We established an arrayed dual-guide CRISPR-Cas9 Perturb-seq platform in human Simpson-Golabi-Behmel Syndrom (SGBS) preadipocytes. 12 CAD-prioritized genes and two positive controls, in total 14 were individually targeted. After adipogenic differentiation, single-cell RNA sequencing with gene-specific-sgRNA capture was performed. Differential expression, pathway enrichment, core gene identification, drug interaction annotation, and phenome-wide association (PheWAS) analyses were conducted to define convergent transcriptional programs and their cardiometabolic relevance. Results Eleven of 14 perturbation lines converged on a dominant transcriptional axis marked by suppression of adipogenesis and activation of epithelial mesenchymal transition (EMT). These lines showed coordinated downregulation of mature adipocyte markers (ADIPOQ, FABP4, PLIN1) and robust induction of extracellular matrix and epithelial–mesenchymal transition genes (COL1A1, COL1A2, FN1, LOX), together with inflammatory pathways. Core gene analysis identified 44 genes differentially expressed in at least 11 perturbation groups, defining a shared adipocyte ‘core’ network. Core genes displayed coherent directionality across independent knockouts, supporting structured convergence consistent with the omnigenic framework. Compared with the originally targeted genes, core genes exhibited a markedly higher number of curated and known drug–gene interactions, indicating greater translational tractability at the network level. PheWAS integration demonstrated that 28 of the core genes were associated with cardiometabolic traits, and 18 showed consistent protective or detrimental effects across metabolic phenotypes. Thus, the experimentally defined remodeling program aligns with genetically inferred disease directionality. Conclusion Most genetically prioritized adipose CAD genes converge on an omnigenic core transcriptional network in human adipocytes. In 11 of 14 perturbations, diverse loci induced a shared shift from metabolic adipocyte programs toward a pro-fibrotic, EMT-like state. Enrichment of druggable interactions and coherent cardiometabolic PheWAS signals among core genes highlights network-level regulators as key mediators of polygenic CAD risk and potential therapeutic targets.