Chromosome Cohesion: Closing Time

Ever since the founding work of Walther Fleming more than a century ago, the process by which one cell divides to create two daughters has fascinated biologists. But only now are we beginning to understand the mechanisms that underlie the coordinated distribution of the duplicated chromosomes — sister chromatids — to the two daughters when a eukaryotic cell divides. At the heart of this process lies an ingenious trick: between the times of genome duplication in S phase and segregation in mitosis, the newly formed sister chromatids are held together by a protein complex called cohesin. Thus, only matching pairs of sisters line up in mitosis, ensuring an equal distribution of the genomic information to the two daughter cells. The multisubunit cohesin complex is conserved from yeast to man [1,2]. The core complex of the budding yeast Saccharomyces cerevisiae is formed by SMC1, SMC3, Scc1 and Scc3. The two structural maintenance of chromosome (SMC) proteins are well adapted to function in cohesion (Figure 1A): they have a striking 500 Å long, antiparallel coiled-coil domain which connects an ATP-binding cassette (ABC) ATPase domain at one end to a dimerization (or hinge) domain at the other [3,4]. The dimerization and ATPase domains are involved in higher-order cohesin architecture. The dimerization domains of SMC1 and SMC3 tightly associate to form a stable SMC1–SMC3 heterodimer [4]. Similarly, the ABC ATPase domains of SMC1 and SMC3 associate by mutual binding to the Scc1 subunit [4]. On the basis of these structural features, Nasmyth and colleagues suggested that cohesin forms a large ring, which in principle could embrace both sister chromatids [4,5] (Figure 1A). In this model, the coiledcoil domains of SMC1 and SMC3 each form one half of the ring. The ring is closed at one end by association of the SMC1 and SMC3 dimerization domains, and at the other by association of Scc1 with the SMC1 and SMC3 ATPase domains. The ring model provides an intriguing mechanism for cohesion. For instance, it might allow for the stable but laterally flexible attachment of sister chromatids. This could be important, for example to allow simultaneous cohesion and condensation processes on chromosomes. Loss of cohesion in mitosis is triggered by cleavage of the cohesin Scc1 subunit by the protease separase, which is activated by destruction of its inhibitor securin (reviewed in [6]). The ring model beautifully explains many aspects of cohesion, but it raises an important question: how do the chromosomes get into the cohesin ring? Two studies [7,8] published in this issue of Current Biology have shed new light on this crucial issue. Weitzer et al. [7] and Arumugam et al. [8] addressed the role of the SMC1/SMC3 ATPase domains, in particular their ATP binding/hydrolyzing motifs, in cohesin function. These new studies have revealed that a multilayered process involving both the binding and hydrolysis of ATP by SMC1 and SMC3 is necessary to establish sister chromatid cohesion. The new data reported in these two papers [7,8] lead to the following picture. ATP binding, but not ATP hydrolysis, by SMC1 is required for de novo formation of a closed SMC1–SMC3–Scc1/3 ring. Scc1 readily associates with SMC3, but needs ATP to additionally associate with SMC1. Once the ATP-bound cohesin ring is formed, ATP hydrolysis is required for DNA to be transported into the ring. For this process to work, ATP hydrolysis has to transiently open the ring structure. A possible mechanism is readily at hand: if ATP binding to SMC1 is necessary for its interaction with Scc1, ATP hydrolysis might disrupt the SMC1–Scc1 complex and open the ring. A possible mechanism for the ATP-driven modulation of the SMC1–Scc1 interaction is suggested by the observation that ATP binding induces a substantial conformational change in ABC domains [9,10]. Such a conformational change in SMC1 would seem to be well suited to modulate its interaction with Scc1. While ATP-driven modulation of the cohesin ring integrity is an elegant model to explain loading of cohesin onto chromosomes, it imposes a mechanistic dilemma to the cell. On the one hand, ATP hydrolysis on unbound cohesin should occur frequently to allow sufficiently rapid loading of cohesin on chromosomes. On the other hand, the ATP hydrolysis activity of DNAassociated cohesin should be low to ensure persistent cohesion. This problem could be resolved if the ATPase activities of SMC1 and SMC3 are either activated when cohesin encounters DNA or inhibited when the complex is not bound to DNA. The new data of Weitzer et al. [7] and Arumugam et al. [8] indicate that cohesin might employ both mechanisms. Before we look into this, however, we need briefly to consider how ABC ATPases work. They have a conserved ATP-driven engagement mechanism [10,11]: on binding of ATP, ABC domains undergo a conformational change and dimerise. ATP bound at the dimer interface tightly connects the two ABC domains by interacting with the A and B Walker motifs of one ABC domain and with the signature motif of the other domain (Figure 1B). ATP hydrolysis disengages the two ABC domains. This engagement–disengagement mechanism not only provides a conformational switch to drive the enzymatic function, but it also allows allosteric regulation. In fact, Current Biology, Vol. 13, R866–R868, November 11, 2003, ©2003 Elsevier Science Ltd. All rights reserved. DOI 10.1016/j.cub.2003.10.045