A Molecular Determinant for the Establishment of Sister Chromatid Cohesion

See allHide authors and affiliations

Science  25 Jul 2008:
Vol. 321, Issue 5888, pp. 566-569
DOI: 10.1126/science.1157880


Chromosome segregation, transcriptional regulation, and repair of DNA double-strand breaks require the cohesin protein complex. Cohesin holds the replicated chromosomes (sister chromatids) together to mediate sister chromatid cohesion. The mechanism of how cohesion is established is unknown. We found that in budding yeast, the head domain of the Smc3p subunit of cohesin is acetylated by the Eco1p acetyltransferase at two evolutionarily conserved residues, promoting the chromatin-bound cohesin to tether sister chromatids. Smc3p acetylation is induced in S phase after the chromatin loading of cohesin and is suppressed in G1 and G2/M. Smc3 head acetylation and its cell cycle regulation provide important insights into the biology and mechanism of cohesion establishment.

Sister chromatid cohesion is required for faithful chromosome segregation and for efficient DNA double-strand break (DSB) repair and is mediated by the cohesin protein complex (Fig. 1A) (14). Chromatin loading per se is not sufficient for cohesin to tether sister chromatids (3, 5, 6). Eco1p (also known as Ctf7p) must act on the chromatin-bound cohesin to promote the establishment of sister chromatid cohesion both during S phase and in response to DSBs in G2/M phase (710).

Fig. 1.

Eco1p acetylates the Smc3p subunit of the cohesin complex. See (27) for experimental details and strains used for each experiment. (A) Cohesin architecture. (B) An acetylated protein coimmunoprecipitates with Mcd1p. WB, Western blot; IP, immunoprecipitation; HA, hemagglutinin. (C) Identification of the acetylated protein as Smc3p. α-AcK, antibody to acetylated lysine. (D) Smc3p acetylation in eco1-203. Ac-Smc3, acetylated Smc3p; CEN, centromeric plasmid. (E) Recombinant Eco1 acetylates Smc3p in vitro. Ac-CoA, acetyl–coenzyme A. (F) Alignment of the Smc3 orthologs using ClustalW. The amino acid numbers above the alignment correspond to the budding yeast Smc3p; # indicates the acetylated lysine residues. Abbreviations: A, Ala; C, Cys; D, Asp; E, Glu; F, Phe; G, Gly; H, His; I, Ile; K, Lys; L, Leu; M, Met; N, Asn; P, Pro; Q, Gln; R, Arg; S, Ser; T, Thr; V, Val; W, Trp; Y, Tyr.

Eco1p possesses acetyltransferase activity (1113). A mutant form of Eco1p, eco1p (R222G, K223G) (fig. S1), purified from bacteria has almost no detectable catalytic activity in vitro (11). In yeast, eco1 (R222G, K223G) cells are defective only in DSB-induced cohesion in G2/M (9, 14). Thus, Eco1p acetylation of cohesin seemed required for DNA damage–induced cohesion but not S-phase cohesion (9). However, the S-phase conclusion was challenged by two observations. First, we found that eco1p (R222G, K223G) purified from yeast extracts has auto-acetyltransferase activity in vitro (fig. S2A). Second, eco1 mutants lacking the acetyltransferase domain are inviable, a phenotype of cells defective in S-phase cohesion (fig. S2B). Together, these results suggest that the Eco1p acetyltransferase activity is required for establishing sister chromatid cohesion during S phase, and that the eco1 (R222G, K223G) protein must have sufficient acetyltransferase in vivo to carry out this function. The eco1p (R222G, K223G) protein may be unable to promote DSB-induced cohesion in G2/M for several reasons. For instance, its acetyltransferase activity might be reduced such that it is unable to overcome an antagonizing activity (like the activity of a deacetylase), which accrues after exit from S phase. Alternatively, DSB-induced cohesion may require acetylation of DNA damage–specific targets in G2/M that are recognized poorly by the mutant protein.

Potential acetyltransferase targets include the four cohesin subunits Smc1p, Smc3p, Mcd1p (also known as Scc1p or Rad21p), and Scc3p (Fig. 1A), as well as the cohesin-associated factor Pds5p (7, 1519). Mcd1p, Pds5p, and Scc3p are acetylated by Eco1p in vitro (11). We immunoprecipitated cohesin from extracts of asynchronous wild-type and eco1 (R222G, K223G) cells and observed a single acetylated band of ∼150 kD, which we demonstrated to be Smc3p (Fig. 1, B and C, and fig. S2C). In eco1-203 cells at the nonpermissive temperature, acetylated Smc3p was barely detectable (Fig. 1D and fig. S3) but was restored to wild-type levels in the presence of a plasmid-borne ECO1 (Fig. 1D). In addition, Smc3p immunoprecipitated from yeast was acetylated in vitro by a recombinant Eco1p (Fig. 1E). Therefore, Smc3p is a bona fide substrate of the Eco1p acetyltransferase.

Liquid chromatography–mass spectrometry (LC-MS) revealed eight acetylated lysine residues in Smc3p (Fig. 1F and fig. S4). We initially focused on the conserved lysine (K) residues and mutated them either as a pair (K112, K113) or individually (K931) to arginine (R), a structurally similar amino acid that cannot undergo acetylation. The K112R, K113R mutation, but not the K931R mutation, failed to complement smc3 and support growth (Fig. 2A), which suggests that sister chromatid cohesion requires K112, K113 acetylation. Moreover, in the smc3 (K112R, K113R) mutant, as in the eco1 mutant, cohesin associated with chromatin (Fig. 2, B and C) (5, 7, 8, 20, 21) but failed to establish cohesion (Fig. 2D) (7, 8). This phenotypic similarity between smc3 (K112R, K113R) and eco1 strongly suggests that K112, K113 acetylation by Eco1p promotes chromatin-bound cohesin to become cohesive [also reported in (22)]. Because both the Eco1p acetyltransferase and its functionally relevant target sites on Smc3p are conserved [(1113) and this study], this mechanism is likely to be conserved between yeast and the other eukaryotes.

Fig. 2.

Smc3p acetylation is necessary for the establishment of sister chromatid cohesion. (A) Spot test for growth. Triangles indicate decreasing concentration of cells in spots. 5-FOA, 5-fluoroorotic acid. (B) smc3p (K112R, K113R) binding to chromosomes. Each inset is a magnified view. Asterisk indicates the magnified field. DAPI, 4′,6′-diamidino-2-phenylindole. (C) smc3p (K112R, K113R) binding to cohesin-associated regions. (D) Cohesion establishment defect in smc3 (K112R, K113R). Violet area corresponds to S phase. (E) Suppression of the cohesion establishment defect in eco1-203. Error bars indicate SD; n = 3.

If acetylation of K112 and K113 are the only functionally important targets of Eco1p in S phase, then changing them to an acetyl-mimic glutamine (Q) (23) should allow cells to establish cohesion without Eco1. The presence of an ectopic copy of smc3 (K112Q, K113Q) restored sister chromatid cohesion in the eco1 mutant (Fig. 2E). K113 seems to be the more critical target of the Eco1 acetyl-transferase, because smc3 (K113Q) suppresses the cohesion defect of eco1 to the same extent as smc3 (K112Q, K113Q). However, in both smc3 (K113Q) and smc3 (K112Q, K113Q), cohesion and growth were only partially rescued (Fig. 2E and fig. S5A). This partial suppression might reflect the presence of additional targets or the incomplete mimic of the acetylated state by glutamine substitution. Alternatively, forcing Smc3p to be acetylated constitutively could compromise sister chromatid cohesion because cohesin becomes active at the wrong time or in the wrong context.

Individual substitution mutants were generated for K112, K113, and the remainder of the Smc3 acetylation sites to assess their contribution to cohesion establishment. Only the K113R mutation failed to support growth and establish cohesion (fig. S5, B and C, and fig. S6A). Surprisingly, unlike cohesin in the eco1 or smc3 (K112R, K113R) mutants, cohesin in the smc3 (K113R) mutant failed to associate stably with chromosomes (fig. S6, B and C). One explanation for this instability is that K113R continues to be acetylated at K112. Indeed, the monoacetylated K112 was recovered from wild-type cells (fig. S4). This monoacetylated form may normally be generated from the diacetylated K112, K113 as part of the Wapl-dependent mechanism for dissociating cohesin from chromatin in interphase and prophase (2426). By mutating K113R, the monoacetylated K112 would be generated inappropriately, causing precocious cohesin removal. Consistent with this, deletion of RAD61, the budding yeast homolog of WAPL, restores viability and (by inference) cohesion to the K113R mutant (fig. S7). Changes in expression or Rad61p in different genetic backgrounds may explain the phenotypic difference between the K113R mutants in this and the accompanying study (22).

Modeling indicates that K112 and K113 are near the adenosine triphosphate (ATP)–binding pocket (fig. S8) (6) such that Eco1p-mediated acetylation could modulate Smc3p adenosine triphosphatase activity (28). This could stabilize the cohesin ring once it has embraced the sister chromatids, or could induce a conformational change to stimulate interactions within or between cohesin complex(es) (29). Alternatively, Smc3p acetylation could promote the dissociation of a negative cohesin regulator such as Wapl (25, 26). The modification-defective and modification-mimic alleles of Smc3 provide powerful tools to test these models.

The Smc3 mutations also allowed us to identify the acetylated residue recognized by the Calbiochem antibody as K112 (fig. S9). Using this antibody, we began to assess the temporal and spatial regulation of Smc3 acetylation. Cells were sampled for K112 acetylation at regular intervals after release from arrest in G1 or S (Fig. 3, A and B). Acetylated K112 is undetectable in G1, accumulates during S phase, is relatively constant until G2/M, and then diminishes when Mcd1p is degraded at the onset of anaphase to dissolve cohesion. Analysis of its spatial regulation shows that K112 acetylation also requires the Scc2p/Scc4p (30)–dependent loading of cohesin on chromatin (Fig. 3C). This requirement for loading might have explained the low K112 acetylation in G1, because cohesin normally does not load at this time in budding yeast. However, even when we induced the premature loading of cohesin onto chromosomes in G1 (Fig. 3D), K112 acetylation still remained greatly reduced relative to S-phase cells (Fig. 3E). Thus, Eco1p-dependent acetylation of Smc3p, minimally at K112 and likely at K113, occurs only upon entry into S phase and after chromatin loading of cohesin.

Fig. 3.

Smc3p acetylation is cell cycle–regulated. (A and B) K112 acetylation during and after G1 (A) or S phase (B). (C) K112 acetylation in scc2; asyn, asynchronous. (D) Induction of cohesin loading during G1. (E) K112 acetylation status in G1-loaded cohesin.

This regulation of Smc3 acetylation may serve several biological functions. It may ensure robust cohesion by ensuring that cohesin becomes cohesive only after binding chromatin and only in the presence of an emerging sister chromatid. In addition, it may allow cells to mark a subset of cohesins to generate two functional cohesin pools. Indeed, evidence for two pools of cohesin with different chromatin-binding properties has been found in mammalian cells (31). We suggest that cohesins with acetylated Smc3p may be stably bound to chromosomes and locked in their cohesive state, thereby maintaining cohesion for chromosome segregation. Cohesins lacking Smc3p acetylation may be an uncommitted reservoir that can be targeted to de novo regions of the genome to respond to dynamic processes like transcription or DNA repair.

Supporting Online Material

Materials and Methods

Figs. S1 to S9

Table S1


References and Notes

View Abstract

Navigate This Article