Cross-platform Hi-C meta-analysis identifies functional insulators that actively block enhancer-promoter interactions
Ontology highlight
ABSTRACT: Insulator protein CTCF controls genome architecture through forming thousands of cohesin-dependent structural loops. However, genome-wide studies only found mild transcriptional consequences upon acute CTCF-depletion, raising confusions about how CTCF regulates enhancer-promoter (E-P) interactions and gene expression. Here we reanalyze independent Hi-C, in situ Hi-C, and micro-C data in mouse embryonic stem cells upon acute CTCF-, RAD21-, and WAPL-depletion; DeepLoop is used to enable robust comparison of orthogonal Hi-C data at kb-resolution regardless of sequencing depth. All datasets show that most loops are lost upon CTCF depletion, but E-P interactions are enriched among the retained loops, and interestingly a small number of newly gained loops repressed by CTCF. From multiplatform Hi-C data, we identified several hundred recurrent events in which new E-P interactions form after the insulating CTCF loops disappear. We therefore define FINs (functional insulators) as CTCF sites that actively insulate their flanking sequences. In CTCF-depleted cells these newly gained E-P interactions require cohesin activity. WAPL-depletion causes relaxation of FIN loops and abolish insulator functions. Importantly, CTCF-repressed genes are enriched near FINs, but CTCF-dependent genes are enriched near TAD-boundaries. We also validated the transcription regulatory functions of several FINs with CTCF-blocking assays. Taken together, DeepLoop meta-analysis unifies multiplatform Hi-C data and demonstrated that FINs, but not TAD-boundaries, are bona fide insulators.
ORGANISM(S): Mus musculus
PROVIDER: GSE243728 | GEO | 2026/04/10
REPOSITORIES: GEO
ACCESS DATA