By revealing the scope of cohesin looping in gene regulation, and by disentangling CTCF´s looping-dependent and -independent roles of CTCF, the work offers a unified model of how Cohesin and CTCF impact gene expression.

At the promoter of CTCF-regulated genes, CTCF functions as an activator or repressor, depending on the promoter binding position and strand orientation. As an activator, it promotes Pol2 recruitment, as a repressor it occludes Pol2 binding/elongation.

The CTCF paralog CTCFL (BORIS) cannot anchor cohesin loops but can function as transcription activator to substitute for CTCF at promoter-regulated targets. This means it can disrupt 3D genome organization, but without hampering expression of CTCF-dependent essential genes.

By endogenously mutating CTCF to disable cohesin anchoring, we show that promoter-proximal CTCF activates genes independently of enhancers and looping. Instead CTCF acts as an orientation-dependent transcription activator, primarily regulating essential housekeeping genes.

CTCF, however behaves differently. Acute CTCF loss also alters hundreds of genes, but the pattern is mixed (both up and down) and only some require CBP/p300 dependent enhancers.

We profiled the impact of acute CTCF or RAD21 removal on nascent transcription and compared it to gene regulation by CBP/p300. Cohesin is specifically important for CBP/p300-dependent enhancers, regulates hundreds of genes, but with a subtle impact previously overlooked.

We propose that gene-skipping by enhancers are the exception, not the rule. Linear enhancer-promoter distance and coactivator requirement of genes can predict enhancer targets with accuracy rivalling state-of-the-art experimental approaches.

Multi-way enhancer-promoter interactions in ultra-deep chromatin contact mapping analyses don't indicate co-regulation of multiple promoters. Within multi-way interacting promoters, enhancers regulate proximal promoter but not distal ones.

Attributes of enhancer targets identified by mapping eRNA-PROMPT RNA interactions by RIC-seq are fundamentally different from enhancer targets identified by CRISPRi.

Gene-skipping and non-skipping enhancers differ in chromatin features, target gene distance, activation strength, and coactivator requirements. High H2BNTac strongly predict functional non-skipping enhancers, while skipping enhancers show weaker regulation and higher FDRs.

Gene-skipping enhancer-promoter contacts are rarely confirmed as being functional by CRISPRi screening. Conversely, CRISPRi identified gene-skipping enhancers also seldom show physical interactions.

Our findings resolve several long-standing conundrums regarding the involvement of cohesin and CTCF in gene regulation, and we present a unified model for their combined and individual roles acting directly near promoters or in concert with p300/CBP-dependent enhancers.

For the genes regulated by CTCF at the promoter, CTCF directly impacts DNA accessibility and downstream recruitment of PolII.

The CTCF-like protein BORIS is also unable to anchor cohesin. Re-analysis of published data with this new perspective show, that unlike CTCF, BORIS has no impact on enhancer-regulated genes. In contrast, BORIS can selectively rescue expression of promoter-activated genes.

Using a cohesin-loop-deficient CTCF mutant, we reveal that CTCF is a position- and orientation-dependent transcription activator at promoters. It directly activates hundreds of essential housekeeping genes—independently of cohesin and enhancers—even in plasmid reporter assays.

Among diverse types of posited enhancers, cohesin seems to uniquely facilitate enhancer-mediated gene regulation through p300/CBP-dependent enhancers.

Acute depletion of cohesin and CTCF dysregulate several hundreds of genes – far more than previously appreciated. Many of these are subtle, but as confirmed through several repetitions, they are consistent.