A unified model of gene expression control by cohesin and CTCF

Abstract

Cohesin and CTCF fold vertebrate genomes into loops and topologically associating domains (TADs). The genome folding by cohesin and CTCF is considered crucial for enhancer-promoter communication and gene regulation. However, the extent of cohesin and CTCF-dependent looping in global enhancer function and their potential non-architectural roles in gene activation have remained unclear. Here, we demonstrate that the acute removal of cohesin or CTCF in mouse cells dysregulates hundreds of genes, albeit subtly. Among various possible enhancer types, cohesin almost exclusively facilitates gene activation by one enhancer type--the CBP/p300-dependent enhancers. Interestingly, CTCF plays a dual role, operating both with and without cohesin, and promoting gene activation both in an enhancer-dependent and independent manner. By anchoring cohesin loops, CTCF directs enhancers to specific target genes and prevents their mistargeting. Independently of cohesin and enhancers, CTCF acts as a transcriptional activator or repressor, depending on its precise binding position near promoters. Acting as a canonical transcription activator, CTCF directly activates hundreds of housekeeping genes, including those essential for mammalian cell proliferation. Mechanistically, promoter-bound CTCF controls DNA accessibility and RNA polymerase II recruitment. The transcriptional activator function of CTCF appears unique to vertebrates and is shared by its vertebrate-specific paralog, CTCFL, despite CTCFLs inability to anchor cohesin loops. These findings reveal the scope of cohesin and CTCF in gene regulation, delineate their shared and unique functions, define a specific class of enhancers using cohesin-mediated looping, and establish a crucial function of CTCF in gene regulation independent of its architectural role. The work reconciles conflicting views and provides a unified model for gene regulation by cohesin and CTCF.