STRUCTURAL BIOLOGY

40 ESRF

as genome replication, repair or chromosome segregation, to sense if they are acting on the same or different DNA molecules.

Hi-C experiments have shown that genomes are organised in Topologically Associating Domains (TAD), regions of the genome that physically interact with each other more frequently than with sequences outside the TAD. TADs are typically flanked by DNA binding sites for CTCF (Figure 25a). Cohesin is located at the boundaries of TADs. It is thought that CTCF acts as a barrier to the loop extrusion carried out by cohesin. Stalling of cohesin at CTCF sites thus leads to stable DNA loops, a feature that leads to TAD corner peaks in Hi-C matrices.

Intriguingly, only convergently oriented pairs of CTCF sites form CTCF-anchored loops. Inversion of specific CTCF sites by genome editing alters the TAD boundary, moving it to the next CTCF site with the appropriate orientation. This reveals a fascinating conundrum: How can the orientation of two DNA elements, potentially ~1Mb apart, determine whether a TAD forms between them? Cohesin must somehow be able to sense appropriately oriented CTCF to allow productive DNA looping between such sites.

To understand how cohesin and CTCF contribute to such large-scale genome organisation, the interaction of these proteins was investigated. Cohesin is a ring-shaped complex (Figure 25b)

comprised of four core subunits: SMC1, SMC3, SCC1 and SA2, as well as a number of regulatory factors. An N-terminal segment of CTCF that directly engages a cohesin subcomplex containing the SA2 and SCC1 subunits was identified. The SA2-SCC1-CTCF complex was crystallised and its structure determined to 2.7 Å resolution using data collected at beamline ID30A-1/MASSIF-1. The structure revealed that CTCF binds to a conserved surface at the interface of the SA2 and SCC1 subunits (Figure 25c). Binding is mediated by a conserved YxF motif of CTCF.

Previous data indicate that the cohesin release factor WAPL also interacts with SA2-SCC1. It was found that CTCF competes with WAPL for binding to SA2- SCC1, thus suggesting that CTCF prevents WAPL-mediated cohesin release by direct competition. To test this model, the YxF motif of CTCF in cells was mutated. Fluorescence recovery after photobleaching (FRAP) experiments showed that cohesin is more mobile in a YxF mutant. The interaction with CTCF thus stabilises cohesin on chromatin. Intriguingly, Hi-C experiments showed that the

Fig. 26: a) Hi-C contact matrices. Genes and CTCF sites are depicted above. b) Genome-wide quantification of loops. An example of loops for a chromosomal region. c) Aggregate peak analysis of DNA loops. The Hi-C signal is averaged across these locations for both cell lines. d,e) Cohesin-mediated looping is stalled by engaging the N-terminal end of CTCF.