Mammalian enhancers and GWASs act proximally and seldom skip active genes

Summary Enhancers play a critical role in regulating transcription. Nearly 90% of human genetic variants identified in genome-wide association studies (GWAS) are located in distal regions, underscoring the importance of enhancers in human development, diseases, and traits. It is widely suggested that mammalian enhancers frequently skip active genes, and thus, linear proximity is a poor predictor of their targets. A key unresolved question is how often mammalian enhancers skip proximal active genes to specifically target distal genes. Genome-wide enhancer-promoter mapping shows that enhancers frequently bypass active genes, while ultra-deep locus-specific analyses reveal extensive multi-way interactions between enhancers and promoters, forming nested microcompartments. The functional significance of these seemingly contrasting phenomena remains unclear. Here, we compared hundreds of enhancer-target gene pairs identified using enhancer-promoter chromatin contact maps, enhancer-promoter RNA interaction data, and genome-scale CRISPR interference (CRISPRi) perturbations. Our findings reveal limited overlap between active gene-skipping enhancer-gene pairs identified through physical interaction mapping and CRISPRi. Additionally, promoters involved in multi-way enhancer interactions are not co-regulated by shared coactivators. Notably, gene-skipping and non-skipping enhancers identified via CRISPRi differ fundamentally in chromatin features, gene activation strength, false discovery rates, target gene distance, coactivator requirements, and cell-type specificity of target genes. These results suggest that gene-skipping enhancer-promoter interactions observed in chromatin and RNA-based analyses do not reliably predict functional enhancer-gene relationships. We propose that linear enhancer-promoter proximity and coactivator dependency offer a simple, scalable, and cost-effective method for genome-wide prediction of enhancer and GWAS targets, with accuracy comparable to state-of-the-art experimental techniques. While enhancers can skip active genes, such deliberate skipping appears to be the exception rather than the rule.