Molecular basis of CTCF binding polarity in genome folding

Elphège P Nora,Leonid A Mirny,Katherine S Pollard,Antoine Coulon,Kevin So,Benoit G Bruneau,Bassam Hajj,Geoffrey Fudenberg,Laura Caccianini,Vasumathi Kameswaran,Abigail Nagle,Agnès Le Saux,Alec Uebersohn,Maxime Dahan

doi:10.1038/s41467-020-19283-x

Abstract

Current models propose that boundaries of mammalian topologically associating domains (TADs) arise from the ability of the CTCF protein to stop extrusion of chromatin loops by cohesin. While the orientation of CTCF motifs determines which pairs of CTCF sites preferentially stabilize loops, the molecular basis of this polarity remains unclear. By combining ChIP-seq and single molecule live imaging we report that CTCF positions cohesin, but does not control its overall binding dynamics on chromatin. Using an inducible complementation system, we find that CTCF mutants lacking the N-terminus cannot insulate TADs properly. Cohesin remains at CTCF sites in this mutant, albeit with reduced enrichment. Given the orientation of CTCF motifs presents the N-terminus towards cohesin as it translocates from the interior of TADs, these observations explain how the orientation of CTCF binding sites translates into genome folding patterns.

Highlights

Current models propose that boundaries of mammalian topologically associating domains (TADs) arise from the ability of the CTCF protein to stop extrusion of chromatin loops by cohesin
Using an inducible complementation system, we found that CTCF mutants lacking the N terminus are unable to insulate TADs properly, in spite of normal binding to cognate CTCF sites
Using a mouse embryonic stem cell line in which CTCF can be degraded by the auxin-inducible degron (AID) system[3], we observed near-complete disappearance of the cohesin ring subunit RAD21 by ChIP-seq from its initial position at CTCF peaks (Fig. 1a, b)

Summary

Introduction

Current models propose that boundaries of mammalian topologically associating domains (TADs) arise from the ability of the CTCF protein to stop extrusion of chromatin loops by cohesin. Given the orientation of CTCF motifs presents the N-terminus towards cohesin as it translocates from the interior of TADs, these observations explain how the orientation of CTCF binding sites translates into genome folding patterns. The PDS5Ainteracting region of CTCF is distinct from the N-terminal region recently reported to interact with RAD21–SA2 in vitro[19] and required for cohesin enrichment at CTCF sites[19,20] This CTCF motif displays homology with the PDS5-binding domain of both WAPL and its competitors SORORIN and HASPIN. Given that the orientation of the CTCF DNA motif presents the CTCF N terminus toward cohesin as it translocates from the interior of TADs, these observations provide a molecular explanation for how the polarity of CTCF-binding sites determines the genomic distribution of chromatin loops

Methods

Results

Conclusion