A Ring for Holding Sister Chromatids Together?

Every time a cell divides, it is essential that both daughters receive the complete genetic information of their mother. In eukaryotes, the tight connection between the two copies of every chromosome generated by DNA replication, the sister chromatids, ensure the attachment of their kinetochores to spindle microtubules emanating from opposite poles so that the sister chromatids will be segregated into different daughter cells. The molecular basis for this "sister chromatid cohesion" was long unknown.

Genetic screens aimed to isolate mutants that lose cohesion identified, among others, four proteins named Smc1, Smc3, Scc1 (Mcd1), and Scc3 (1-3). All four localize to chromosomes and dissociate from them at the onset of anaphase when cohesion is lost. In addition, their chromosomal association is interdependent and they can be coimmunoprecipitated, which implies that they are subunits of a "cohesin" complex holding sisters together. Their orthologs also form a complex in vertebrate cells (1). Importantly, the proteolytic cleavage of cohesin's Scc1 subunit by a protease called separase is sufficient to release cohesin from chromosomes and to trigger the separation of sister chromatids at anaphase (5). However, how cohesin physically connects sister chromatids and how severing one of its subunits can destroy this connection was not known.

To shed light on these questions, I set out to determine the arrangement of the four subunits in the complex. Smc1 and Smc3, like all members of the Structural Maintenance of Chromosomes (SMC) protein family, contain N- and C-terminal globular domains separated by a long amphipathic helix that is interrupted by a third globular domain at its center. It was thought that two SMCs would dimerize by forming intermolecular antiparallel coiled coils along their helices, which could bend at the central domain to give rise to V-shaped molecules. Such an arrangement would bring Walker A and B motifs in the terminal domains of different protamers together to form complete ABC-ATPases at both apices (the heads) of the V-shaped dimers. When coexpressed in insect cells, yeast Smc1 and Smc3 formed a dimeric complex with biophysical properties and a V-shaped appearance in electron micrographs consistent with a 1:1 heterodimer (6). When expressed separately, Smc3 surprisingly formed monomers resembling one-half of the Smc1/Smc3 dimer. The only explanation was that each SMC protein folds back onto itself by forming intramolecular coiled coils, and then Smc1 and Smc3 heterodimerize at their central domains. A series of experiments where the central domains of Smc1 and Smc3 were swapped confirmed that this domain is sufficient for heterodimerization and a crystal structure of a central SMC domain finally proved intramolecular coiled-coil formation. Instead of intertwining by forming intermolecular coiled coils, SMCs in fact fold into individual protamers with their ATPase heads composed of different ends of the same protein chain.

Determining the arrangement of SMC dimers turned out to be essential for elucidating cohesin's architecture. The Smc1/Smc3 heterodimer bound Scc1 but lost its ability to do so when lacking both terminal ATPase head domains. Strikingly, a heterodimer lacking only the Smc1 head bound specifically to Scc1's N-terminal separase cleavage fragment, whereas one lacking only the Smc3 head bound specifically to Scc1's C-terminal fragment. Because only one copy of each subunit could be found stably associated per cohesin complex, Scc1 must connect the ATPase head domains of one Smc1/Smc3 heterodimer. Scc1 also recruits the Scc3 subunit to the complex. Cohesin therefore forms a large ring structure with a diameter of at least 30 nm. Would cohesin also form such a ring structure in vivo when it holds sister chromatids together? The answer is yes, as cohesin is released from chromosomes after separase cleavage with the N- and C-terminal Scc1 cleavage fragments associated with Smc3 and Smc1 heads, respectively (7).

How could such a huge ring structure be used to hold sister chromatids together? In principle, it could trap two 10-nm chromatid fibers, DNA wrapped around nucleosomes, inside its embrace (Fig. 1). Separase cleavage of Scc1 would open the ring to release sister chromatids from their entrapment. The revolutionary aspect of the model was the topological manner in which cohesin would bind DNA, in contrast to a purely physical interaction. We now had a testable idea of how cohesin may hold sister chromatids together.

Figure 1. Cohesin embrace model. Sister chromatids are symbolized by gray cylinders.

The embrace model predicts that opening the ring at any position should release cohesin from chromosomes. We therefore introduced target sites for a foreign protease into Smc3's coiled coil. Severing Smc3's coiled coil not only released cohesin from chromosomes without affecting the interactions between the ring subunits, but also led to a loss of cohesion in vivo (7). Another crucial aspect of the model is the strength of the connections between the ring subunits if they were to withstand the pulling forces of the mitotic spindle. The central domains of Smc1 and Smc3 indeed bind each other very strongly (6). The inability of one Scc1 molecule to exchange with another Scc1 in cohesin complexes holding sister chromatids together in vivo during extended periods of time after DNA replication implies that Scc1's interaction with Smc1/Smc3 must also be very stable (8). This stable connection is presumably due to the extensive hydrophobic contact between the winged helix domain in Scc1's C terminus and the Smc1 head domain as revealed by their crystal structure. Moreover, two Smc1 head domains had dimerized in the crystals by sandwiching a pair of ATP analogs in between their ABC-ATPase sites. A similar engagement between Smc1 and Smc3 head domains and their disengagement upon hydrolysis of bound ATP may regulate the interaction with Scc1 and could possibly open a gate for the passage of DNA into cohesin rings (9).

Strikingly, the N- and C-terminal domains of Scc1 binding Smc1/Smc3 heads are conserved in kleisin subunits of bacterial and other eukaryotic SMC protein complexes (10). All kleisins might therefore connect the head domains of their associated SMCs to form ring structures comparable to cohesin, which could then entrap chromosomes in a similar manner (11). If so, the topological mode set forth by cohesin's embrace model may be a universal solution to the mystery of how these ancient protein complexes organize chromosomes in every living organism.

References

1. C. Michaelis, R. Ciosk, K. Nasmyth, Cell 91, 35 (1997).
2. V. Guacci, D. Koshland, A. Strunnikov, Cell 91, 47 (1997).
3. A. Toth et al., Genes Dev. 13, 320 (1999).
4. A. Losada, M. Hirano, T. Hirano, Genes Dev. 12, 1986 (1998).
5. F. Uhlmann, D. Wernic, M. A. Poupart, E. V. Koonin, K. Nasmyth, Cell 103, 375 (2000).
6. C. H. Haering, J. Lowe, A. Hochwagen, K. Nasmyth, Mol. Cell 9, 773 (2002).
7. S. Gruber, C. H. Haering, K. Nasmyth, Cell 112, 765 (2003).
8. C. H. Haering et al., Mol. Cell 15, 951 (2004).
9. P. Arumugam et al., Curr. Biol. 13, 1941 (2003).
10. A. Schleiffer et al., Mol. Cell 11, 571 (2003).
11. C. H. Haering, K. Nasmyth, Bioessays 25, 1178 (2003).

To Advertise   |   Find Products