Cryo-EM structure of the human cohesin-NIPBL-DNA complex

Abstract

As a ring-shaped adenosine triphosphatase (ATPase) machine, cohesin organizes the eukaryotic genome by extruding DNA loops and mediates sister chromatid cohesion by topologically entrapping DNA. How cohesin executes these fundamental DNA transactions is not understood. Using cryo–electron microscopy (cryo-EM), we determined the structure of human cohesin bound to its loader NIPBL and DNA at medium resolution. Cohesin and NIPBL interact extensively and together form a central tunnel to entrap a 72–base pair DNA. NIPBL and DNA promote the engagement of cohesin’s ATPase head domains and ATP binding. The hinge domains of cohesin adopt an “open washer” conformation and dock onto the STAG1 subunit. Our structure explains the synergistic activation of cohesin by NIPBL and DNA and provides insight into DNA entrapment by cohesin.

Condensin complexes: understanding loop extrusion one conformational change at a time

“…This bend, referred to as the 'elbow', has also been observed in EM analysis and crosslinking data of S. cerevisiae and S. pombe cohesin and E. coli MukBEF SMC-Kleisin, and results in the hinge bending to contact the SMC arms near the ATPase domains [40,60]. In cohesin, crosslinking and FRET based studies support the hypothesis that the hinge may fold to contact S. pombe Pds5, Psc3 or Mis4 [49,61,62] and cryo-EM structures of H. sapiens, S. cerevisiae and S. pombe suggest that the hinge folds to contact the HAWKs [39][40][41]. Folding of the SMC arms is observed in recent AFM data of S. cerevisiae condensin, showing that the hinge can fold towards the ATPase domains, however, does so with the SMC arms open, transitioning from an open O to a B shaped conformation [63].…”

“…In single-molecule experiments, while condensin remains bound to DNA after ATP is washed out [17], loss of ATP or NIPBL/MAU2 results in human cohesin loop release [28], suggesting that cohesin is less stably tethered to DNA. This could be explained by the discovery of the 'gripping state' where Scc2/NIPBL within a cohesin complex firmly grips DNA in the presence of ATP [39][40][41]. In the case of human cohesin, lower DNA association stability might be compensated for by additional DNA binding factors, such as CTCF, which binds directly at the Scc3/Scc1 interface [42].…”

“…Furthermore, Scc2 plays a role in regulating the ATP hydrolysis rate and is required for maximal ATPase activity of cohesin [46]. However, the cryo-EM structures of H. sapiens and S. pombe cohesin in an ATP-bound state show that Scc2 homologous subunit contacts SMC1 even in the presence of engaged heads [39,40]. This could reflect differences between the cohesin and condensin complexes, or simply a difference in how subunits behave in the presence or absence of DNA.…”

“…However, cryo-EM structures of H. sapiens and S. pombe cohesin with engaged ATPase heads clearly show the N-terminus of the Kleisin, Rad21 binding to SMC3. This suggests either a conformational change of coiledcoils results in the temporary release of Rad21 or that the presence of DNA and/or the loader NIBPL/Mis4 prevents release or contribute to rebinding of Rad21 to SMC3/Psc3 [39,40]. Based on similarity, the release of N-terminal Brn1 could contribute to loss of condensin DNA loops, and if ATP binding to Smc2 promotes release, it might explain why condensin has evolved such that Smc2 has low ATP binding affinity.…”

Cohesin mutations are synthetic lethal with stimulation of WNT signaling

“…Three subunits, SMC1A, SMC3 and RAD21, form the core ring-shaped structure of human cohesin [1,2]. A fourth subunit of either STAG1 or STAG2 binds to cohesin by contacting RAD21 and SMC subunits [3], and is required for the association of cohesin with DNA [1][2][3]. The STAG subunits of cohesin are also capable of binding RNA in the nucleus [4].…”