|Home | About | Journals | Submit | Contact Us | Français|
The S. cerevisiae MCD1 (mitotic chromosome determinant) gene was identified in genetic screens for genes important for chromosome structure. MCD1 is essential for viability and homologs are found from yeast to humans. Analysis of the mcd1 mutant and cell cycle–dependent expression pattern of Mcd1p suggest that this protein functions in chromosome morphogenesis from S phase through mitosis. The mcd1 mutant is defective in sister chromatid cohesion and chromosome condensation. The physical association between Mcd1p and Smc1p, one of the SMC family of chromosomal proteins, further suggests that Mcd1p functions directly on chromosomes. These data implicate Mcd1p as a nexus between cohesion and condensation. We present a model for mitotic chromosome structure that incorporates this previously unsuspected link.
During mitosis, sister chromatids are paired and condensed. They form bipolar attachments to spindle microtubules emanating from opposite poles. At the onset of anaphase, commonly termed the metaphase-to-anaphase transition, a series of closely timed events ensue. The cohesion between sister chromatids is synchronously dissolved on all chromosomes. Sister chromatids move away from each other by microtubule-dependent movement toward the poles. At telophase, the segregated chromosomes decondense. Both sister chromatid cohesion and chromosome condensation are essential for proper chromosome segregation.
Insights into the timing, distribution, and mechanism of sister chromatid cohesion have been provided through studies of mammalian and yeast cells. Classical cytological analyses and fluorescent in situ hybridization (FISH) revealed that in yeast and mammals sister chromatids are associated along their lengths from the time of replication until anaphase (Wilson, 1925; Selig et al., 1992; Guacci et al., 1994). In many eukaryotes, the most persistent cohesion occurs at heterochromatin associated with large blocks of repetitive DNA (Lica et al., 1986; Cooke et al., 1987; Sumner, 1991). In budding yeast, chromatid cohesion occurs despite the absence of repetitive DNA. Yet, recent studies suggest that budding yeast has heterochromatin (reviewed in Grunstein, 1997). Therefore, regions of heterochromatin, and not simply the presence of repetitive DNA, may be important sites for cohesion. Two general mechanisms for cohesion have been proposed: a persistent catenation of sister DNA molecules or a protein-mediated scaffold (Murray and Szostak, 1985; Cooke et al., 1987; Rattner et al., 1988). In budding yeast, sister chromatids of mini-chromosomes remain associated in the absence of catenation, supporting the idea that proteins mediate sister chromatid cohesion (Guacci et al., 1994).
The proteins comprising the structural components responsible for cohesion during mitosis have not yet been identified. INCENP and CLiP proteins are intriguing candidates since they localize between paired sister chromatids prior to anaphase in mammalian cells (Cooke et al., 1987; Rattner et al., 1988). However, their importance to chromatid cohesion is not known. The Cut2 and Pds1 proteins of fission and budding yeast, respectively, have been implicated in cohesion (Cohen-Fix et al., 1996; Funabiki et al., 1996; Yamamoto et al., 1996b). However, they are not likely to be chromosomal components that play a structural role in cohesion. For example, the Cut2 protein localizes to the mitotic spindle, not to chromosomes, and Cut2 mutants do not exhibit precocious sister separation (Funabiki et al., 1996). While pds1 mutants do precociously separate sister chromatids, additional experiments suggest that Pds1p plays a more general role in regulating the metaphase-to-anaphase transition as well as other aspects of cell cycle progression (Yamamoto et al., 1996b).
Mammalian and yeast chromosomes undergo cell cycle–dependent chromosome condensation (Wilson, 1925; Lawrence et al., 1988; Guacci et al., 1994). A breakthrough in elucidating the mechanism of condensation was the discovery of the SMC protein family (structural maintenance of chromosomes). This family is conserved from bacteria to humans with multiple members present in each eukaryote (reviewed in Koshland and Strunnikov, 1996). SMC proteins localize to chromosomes, and a subset are essential for chromosome condensation in vitro and in vivo (Chuang et al., 1994; Hirano and Mitchison, 1994; Saitoh et al., 1994; Saka et al., 1994; Strunnikov et al., 1995). Recent data from yeast, Xenopus, and mammals suggest that SMC proteins are components of higher order complexes (Castano et al. 1996; Jessberger et al., 1996; Hirano et al., 1997). The biochemical function of SMC or SMC-associated proteins has not been determined, but it is likely that SMC complexes possess multiple activities by analogy to replication complexes. Interestingly, SMC proteins may serve a more global role in DNA metabolism since they have been implicated in dosage compensation and mitotic recombination repair (Chuang et al., 1994; Jessberger et al., 1996).
The development of FISH to monitor mitotic chromosome structure and the identification of proteins such as Pds1p and the SMC family have established budding yeast as a model system to study sister chromatid cohesion and condensation (Guacci et al., 1993, 1994; Strunnikov et al., 1995; Cohen-Fix et al., 1996; Yamamoto et al., 1996a, 1996b). The development of GFP-tagged chromosomal loci to follow cohesion has enhanced this system (Straight et al., 1996). The identification of additional components is crucial to elucidate the mechanisms of sister chromatid cohesion and condensation. Here we identify and analyze the mcd1–1 mutant in budding yeast and show that Mcd1p is a chromosomal protein required for sister chromatid cohesion and condensation. Our analyses provide novel insights into a previously unsuspected interrelationship between cohesion and condensation.
The mcd1 (mitotic chromosome determinant) mutant was identified in a screen to isolate mutants defective in sister chromatid cohesion (Guacci et al., 1993). To enrich for mutants defective in mitotic functions, mutants temperature-sensitive for growth (Ts−) were screened for enhanced inviability after arrest in M phase as compared to G1 phase. mcd1–1 was one mutant with these phenotypes (Figures 1A and 1B). Similar mitotic lethality has been observed for cells defective in the mitotic checkpoint (Hoyt et al., 1991; Li and Murray, 1991). However, mcd1 cells have a functional mitotic checkpoint since nocodazole-treated mcd1 cells do not undergo new rounds of DNA replication or new bud formation (data not shown). Therefore, the mitotic lethality of mcd1–1 cells is not due to a defect in cell cycle regulation but reflects a potential role for Mcd1p in chromosome cohesion. The mcd1 mutant was subsequently shown to exhibit precocious dissociation of sister chromatids (see below). The MCD1 gene was cloned by complementation of the Ts− and mitotic lethal phenotypes of mcd1 cells.
MCD1 was also identified in a screen for proteins that interact with the Smc1 protein (Smc1p), a member of the SMC family. To this end, high copy suppressors of the Ts− phenotype of an smc1 mutant (smc1–2) were isolated. One plasmid suppressed the Ts− phenotype and associated morphological defects (Figure 1C). The suppressor gene was found to be MCD1 (Experimental Procedures). While multiple copies of MCD1 were required to suppress the smc1–2 defects, they did not suppress the Ts− phenotype of either an smc1 deletion mutant or an smc2 mutant (data not shown). These results indicate that MCD1 suppression is specific and occurs by augmenting smc1–2 function rather than by replacing it.
The genetic interaction between MCD1 and SMC1 suggested an in vivo physical interaction between the Mcd1 protein (Mcd1p) and Smc1p. To test this possibility, a functional full-length Mcd1p tagged with the T7 epitope (Novagen) was overexpressed in an otherwise wild-type strain. Overexpression of Mcd1p recapitulates the conditions under which MCD1 and SMC1 genetically interact (see above). Mcd1p and Smc1p were found to coimmunoprecipitate in a chromatin-independent manner since large chromatin fragments were removed by high-speed centrifugation (Figure 1D) and coimmunoprecipitation was unaffected by DNase treatment. Polyclonal anti-Mcd1p antibodies failed to immunoprecipitate Mcd1p when expressed under its own promoter, but in preliminary experiments Mcd1p and Smc1p co-fractionate through several steps of a biochemical purification (data not shown). No immunoprecipitation of Mcd1p or Smc1p was observed in strains lacking a T7 epitope tag. Taken together, these data suggest that Mcd1p and Smc1p are in a common complex.
MCD1 was shown to be essential by two approaches. First, a diploid strain heterozygous for a complete deletion of MCD1 was constructed and sporulated. All tetrads contained two viable and two inviable spores. None of the viable spores contained the deleted MCD1 gene. Second, strains were constructed in which the sole source of Mcd1p was from an MCD1 gene under control of an inducible GAL1 promoter. These strains were inviable in the absence of inducing agent (galactose) for MCD1 (data not shown). Thus, loss of MCD1 is a lethal event.
MCD1 encodes a protein with a predicted molecular mass of 63 kDa. Its sequence was compared to other proteins in the database. Mcd1p shares homology with the S. pombe Rad21 protein and with putative proteins encoded by ORFs from human, mouse, and C. elegans. These proteins are 25% identical over their entire length except for the human and mouse sequences, which are 97% identical. Three blocks of >50% similarity are shared, and their spacing and relative position is conserved (Figure 1E). The rest of the predicted protein sequence is not conserved, but in the central region there are numerous potential PEST sequences, which have been proposed to target polypeptides for rapid degradation by the 26S proteasome (Coux et al., 1996; Rechsteiner and Rogers, 1996).
The mitotic lethality of mcd1–1 cells suggested a mitotic function for Mcd1p. To determine the role of MCD1 in cell cycle progression, wild-type and mcd1–1 haploid cells growing at 23°C were shifted to 37°C and cell, spindle, and DNA morphologies as well as DNA content were scored. At 23°C, mcd1 cells were indistinguishable from wild-type cells at 23°C or 37°C (Figure 2A). However, at 37°C, the mcd1 culture was enriched for cells with 2C DNA content (Figure 2A) and showed a 3-fold increase in the frequency of large budded cells with short or partially elongated spindles.
mcd1–1 cells were further characterized using synchronous populations obtained after release at 37°C from S phase arrest. While arrested in S phase, wild-type and mcd1 cells had a 1C DNA content, large bud, undivided DNA mass, and short spindle (Figure 2C). After release (1 hr), most wild-type cells completed both DNA replication and chromosome segregation since cells had a 2C DNA content, elongated spindle, and segregated DNA masses (Figure 2B). In contrast, most mcd1 cells had a stretched nuclear DNA mass and partially elongated spindle indicative of a chromosome segregation defect (Figure 2B). By 2 hr, 50% of wild-type cells exited mitosis (unbudded and small budded cells), compared to only 20% for mcd1 cells (Figure 2C). The majority of mcd1 cells remained in mitosis with cell and DNA morphologies similar to that at 1 hr (Figure 2C). Similar results were obtained when Mcd1p was depleted in strains in which the sole source of MCD1 was under control of an inducible promoter (data not shown). These data suggest that mcd1 inactivation causes a mitotic defect that disrupts chromosome segregation and delays, but does not prevent, exit from mitosis. Finally, mcd1 cells exhibited a 10 min delay in early S phase, indicating a requirement for Mcd1p at this time (data not shown).
To establish that mcd1–1 cells are defective in sister chromatid cohesion, synchronized populations of wild-type and mcd1 haploid cells were arrested in mid-M phase by either of two regimens and subjected to FISH. For regimen 1, cells were at the nonpermissive temperature for the mcd1–1 mutation through S and mid-M phases, which is the time sister chromatid cohesion is both established and maintained (Experimental Procedures). Cells were processed for FISH using either a chromosome XVI centromere-proximal (CEN-proximal) or distal probe to assay cohesion at different sites along chromosome XVI.
Consistent with our previous studies, in mid-M phase most wild-type cells had one FISH signal per nuclear DNA mass, demonstrating that sister chromatids were paired (Figures 3A and 3B). In a small number of DNA masses, two FISH signals were detected due to either a low level of precocious sister chromatid dissociation or spurious background (Figure 3B). In contrast, most mcd1 cells in mid-M phase at 37°C had two FISH signals per DNA mass, indicating that sister chromatids had precociously dissociated (Figures 3A and 3B). The cohesion defect is not restricted to nocodazole arrested cells since inactivation of Mcd1p also caused precocious sister chromatid dissociation in cells arrested in G2/M by a mutation in an anaphase promoting complex (APC) subunit (Cdc16p). For example, 60% of mcd1 cdc16 cells exhibited precocious sister separation at CEN-proximal and distal loci compared to only 15% in cdc16 cells. The second FISH signal in mid-M phase mcd1 cells was not due to preexisting aneuploidy since in G1 phase, most cells had only one FISH signal (Figure 3B). The few cells that had two FISH signals in G1 in wild-type and mcd1 cells were likely due to spurious background. Similar results were obtained using probes from CEN-proximal regions of chromosomes I and IV and a more CEN-distal chromosome XVI region (data not shown). Thus, Mcd1p is required for sister chromatid cohesion at CEN-proximal and distal chromosomal regions in yeast cells.
The previous regimen can not distinguish between defects in establishment or maintenance of cohesion. To assay for maintenance of cohesion, mcd1–1 and wild-type cells were allowed to establish cohesion at permissive temperature before mcd1–1 function was inactivated (Experimental Procedures, regimen 2). As expected, mid-M wild-type cells at 23°C and 37°C had one FISH signal per DNA mass (Figure 3C). Most mid-M mcd1 cells had one FISH signal at 23°C but two signals upon shift to 37°C. Similar results were obtained using probes from chromosomes I, IV, and XVI (data not shown). These results demonstrate that Mcd1p is required for maintenance of sister chromatid cohesion.
The stretched DNA mass in mcd1 mutants at 37°C is reminiscent of yeast mutants defective in condensation (Strunnikov et al., 1995). To test the role of Mcd1p in condensation, we used FISH to examine chromosome condensation at the rDNA locus, a 500 kb block of repetitive DNA (Guacci et al., 1994). We had shown that in G1 phase cells, an amorphous rDNA FISH signal characteristic of a decondensed chromosome is detected, while in mid-M phase haploid cells a single line-like FISH signal characteristic of condensed and paired sister chromatids is seen (Guacci et al., 1994). As expected, wild-type cells arrested in mid-M using regimen 2 had a single line-like FISH signal in 85% of the DNA masses (Figure 4A). In contrast, 74% of DNA masses from similarly treated mcd1 cells had an amorphous FISH signal (Figure 4A). An amorphous FISH signal is not expected for separation of sister chromatid, since in wild-type cells precociously separated sisters are seen as two line-like signals (Figure 4A, inset). Furthermore, the rDNA is line- like after sister chromatids have separated and segregated in anaphase nuclei cycling wild-type cells (Guacci et al., 1994). Thus, mcd1–1 cells in mid-M phase at 37°C have an aberrant rDNA morphology indicative of a defect in rDNA condensation.
To monitor chromosome condensation at unique chromosomal regions, synchronized populations of wild-type and mcd1 haploid cells arrested in mid-M phase at 37°C (regimen 1) were subjected to FISH using a mixture of four chromosome XVI probes or six chromosome VIII probes. The number and spacing of FISH signals from the mixture of chromosome XVI or VIII probes provides a qualitative measure of chromosome condensation. For example, when haploid cells are arrested in G1 phase and hybridized with chromosome XVI probes, up to four, often widely spaced, FISH signals are detected per DNA mass whereas in mid-M phase, one or two closely associated FISH signals are detected (Guacci et al., 1994). This change from dispersed in G1 phase to clustered in mid-M phase is characteristic of the change from a decondensed chromosome to a condensed chromosome with paired sisters.
As expected, in wild-type cells arrested in mid-M phase at 37°C and hybridized with either the chromosome XVI or VIII probe mixtures, a few closely associated FISH signals were detected in most DNA masses as expected for condensed and paired sister chromatids (Figure 4B). In contrast, in mcd1 cells hybridized with either the chromosome XVI or VIII probe mixtures, many, often widely spaced FISH signals were detected in many DNA masses (Figure 4B). Some of the increased numbers of FISH signals are expected due to sister chromatid dissociation. However, if sister chromatids remained condensed, there should be two tight clusters of FISH signals, one from each separated and condensed sister chromatid. Instead, the dispersed FISH signals are reminiscent of decondensed chromosomes (Guacci et al., 1994). Taken together, the results from FISH using rDNA, chromosome VIII, and chromosome XVI probes indicate that the mcd1–1 mutant exhibits defects in chromosome condensation as well as sister chromatid cohesion.
To determine if the in vivo localization of Mcd1p was consistent with its proposed role as a chromosomal structural protein, cells were processed for both Western blot analysis and indirect immunofluorescence using anti-Mcd1p antibodies. Initially, we examined a strain in which Mcd1p was overexpressed (Figure 5A). Many cells had prominent punctate nuclear staining, with the exception of cells undergoing anaphase, where Mcd1p was dispersed evenly throughout the cell, suggesting possible redistribution (Figure 5B). When expressed from its endogenous promoter in wild-type cells, Mcd1p staining was highly variable but nuclear when detected (data not shown). To examine whether this variability was cell–cycle dependent, wild-type cells were arrested in G1, S, or mid-M phase and processed for Western blotting and indirect immunofluorescence. Mcd1p was barely detectable in G1 phase, at high levels in S phase, and at lower levels in mid-M phase with a punctate nuclear localization in S and mid-M phases (Figures 5C and 5D). Similarly, the S. pombe Rad21 protein also localized to the nucleus (Birkenbihl and Subramani, 1995). The nuclear localization of Mcd1p during S phase and mitosis is consistent with its proposed role in chromosome structure.
To monitor the dynamics of cell-cycle dependent changes in Mcd1p levels, a synchronous population of wild-type cells was examined. Aliquots of cells were taken in G1 phase, as well as every 20 min after release from G1, then processed to monitor Mcd1p levels and scored for DNA content to assess cell cycle position. Mcd1p levels were barely detectable in G1 phase cells (T = 0), reached a peak in S phase (T = 40) for many cells, and decreased in G2/M phases (T = 60) (Figure 6A). To more precisely determine the time when Mcd1p levels decline, the relative levels of Mcd1p and Pds1p were compared. The decrease in Mcd1p levels occurred 20 min prior to the decrease in Pds1p levels (Figure 6B, arrows). Since Pds1p degradation is required for the metaphase-to-anaphase transition, the decrease in Mcd1p levels occurred before this cell cycle progression landmark (Cohen-Fix et al., 1996). Finally, Mcd1p levels were examined by Western at discrete cell cycle stages using cdc mutants or wild-type cells treated with α factor, HU, or Nz. Mcd1p was barely detectable in G1 (cdc28–4, α factor), peaked in early S (HU), and decreased to a lower but steady level in late S (cdc9), G2 (cdc28–1N), G2/M (cdc13 and cdc16), mid-M (cdc20, cdc23, and Nz), and telophase (cdc14 and cdc15) cells (data not shown). Taken together, these data show that Mcd1p levels peak in early S phase, are reduced by late S to a lower level that remains constant through telophase and decreases to nearly undetectable levels by G1.
Insights into the cell cycle–dependent changes in Mcd1p levels came from examination of MCD1 mRNA in synchronized cycling cells and Mcd1p levels in cdc mutants. MCD1 mRNA levels peak in early S phase then decrease 9-fold 20 min prior to the Mcd1p decrease (Figure 6B and 6C). This pattern of mRNA regulation is likely due to two MluI boxes in the MCD1 promoter region (data not shown; McIntosh et al., 1991). The decrease in mRNA coupled with possible PEST-mediated degradation is sufficient to account for the decrease in Mcd1p levels in late S. Mcd1p levels dropped 20 min prior to Pds1p degradation, a known landmark of anaphase initiation and the earliest known target of the anaphase promoting complex, APC (Figure 6B). Moreover, the decrease in Mcd1p levels observed in late S/G2 was not affected by mutations in APC components (cdc16 and cdc23) (data not shown). Thus, the change in Mcd1p levels in late S/G2 is independent of APC, although it may play a role in the subsequent decrease in Mcd1p levels in G1.
We identified the MCD1 gene and demonstrated that it encodes a conserved protein necessary for both sister chromatid cohesion and condensation in budding yeast. Mcd1p acts throughout the genome, as it is required for cohesion at CEN-proximal and distal (arm) sequences as well as for condensation at three different chromosomal regions, including both unique and repetitive DNA. A cohesion function for Mcd1p (Scc1p) has also been demonstrated by independent methods (Michaelis et al., 1997). We show that Mcd1p and Smc1p interact genetically and physically. Based on these interactions and previous observations that other SMC proteins are chromosomal, we suggest that Mcd1p functions in cohesion and condensation as a component of mitotic chromosomes. Our functional assays place Mcd1p as a chromosomal protein. Consistent with this conclusion, Mcd1p appears to localize to chromosomes in an Smc1p-dependent manner (Michaelis et al., 1997). Finally, we show that both the Mcd1p and MCD1 mRNA levels are cell cycle–regulated such that Mcd1p persists from the time cohesion is first established (S phase) until the time when chromosomes decondense (telophase), consistent with its role in chromosome cohesion and condensation.
The identification and characterization of Mcd1p has provided several new insights into the process of sister chromatid cohesion. Mcd1p is a novel conserved chromosomal protein shown to be required for cohesion during mitosis. This conservation implicates a common mechanism for mitotic sister chromatid cohesion among eukaryotes. The requirement of Mcd1p for cohesion at both centromeric and arm sequences indicates that these regions share a common underlying mechanism of cohesion. In many eukaryotes, a differential dissolution of cohesion at centromeres and arm regions has been observed in meiosis and mitosis. It will be interesting to determine how the common mechanism of cohesion is modulated to give region-specific responses. A candidate for a region-specific cohesion factor is the Drosophila MEI-S332 protein, which is essential for meiotic centromere cohesion (Kerrebrock et al., 1995).
Both APC-dependent proteolysis and protein phosphorylation have been implicated in the dissolution of sister chromatid cohesion at the metaphase-to-anaphase transition (Ohkura et al., 1989; Holloway et al., 1993; Cohen-Fix et al., 1996; Minshull et al., 1996; Yamamoto et al., 1996b). While the levels of Mcd1p are regulated during the cell cycle, the decrease in Mcd1p levels occurred prior to the metaphase-to-anaphase transition and was APC independent. Furthermore, significant amounts of Mcd1p persisted to telophase. Therefore, the degradation of Mcd1p, a chromosomal component of cohesion, is unlikely to play a role in the dissolution of cohesion. However, Mcd1p activity may be regulated by modification. Indeed, the Rad21 protein, the S. pombe homolog of Mcd1p, has multiple cell cycle–regulated forms due to phosphorylation, and Mcd1p migrates as a doublet, raising the possibility of its phosphorylation (Birkenbihl and Subramani, 1995; this study). Further analysis of Mcd1p and the identification of other chromosomal components of cohesion will be necessary to assess the relative contributions of proteolysis, phosphorylation, or other modifications to the dissolution of cohesion.
Mcd1p is a novel protein in budding yeast whose function has been shown to be important for condensation of both unique and repetitive sequences (this study). Other proteins like Smc2p, Trf4p, and Top1p have been shown to be important for condensation at a subset of chromosomal loci (Strunnikov et al., 1995; Castano et al., 1996). We suggest that Mcd1p is a key component of the general condensation machinery that may be acted upon by region-specific factors.
Mcd1p is a new conserved SMC-associated protein in addition to Trf4p, Top1p, and the non-SMC condensin components (Strunnikov et al., 1993, 1995; Castano et al., 1996; Hirano et al., 1997; this study). The addition of Mcd1p to a growing list of condensation factors indicates that condensation is biochemically complex, as might be expected for a highly determined process (reviewed in Koshland and Strunnikov, 1996). Finally, the conservation of Mcd1p further emphasizes that the mechanism of condensation is conserved.
Perhaps the most profound insight from this study is that Mcd1p functions in both sister chromatid cohesion and chromosome condensation, hence linking these two processes previously thought to be independent. A model to explain this link must incorporate the facts that cohesion can exist without condensation (during S and G2 phases), cohesion and condensation must coexist (prophase through metaphase), and condensation can exist without cohesion (anaphase to telophase). We propose a simple model that explains these observations and the link between cohesion and condensation (Figure 7). Sites of cohesion (rectangles) are distributed along the length of the chromosome. Soon after replication, the sister chromatids are held together at these sites. To condense the chromosome, cohesion sites are brought together, and consequently, the chromosomal regions between these sites are looped out (Figure 7). Proteins that mediate cohesion (hatched circles) and condensation (closed circles) are distinct but interact with common chromosomal core components (open circles) such as Mcd1p. These processes can be regulated independently by controlling the association of factors involved exclusively in either cohesion or condensation.
This model explains many observations about the structure of mitotic chromosomes. First, loops have been observed both in partially denatured mitotic chromosomes and in lampbrush chromosomes (Adolph et al., 1977; Paulson and Laemmli, 1977). Second, condensed sister chromatids exhibit mirror symmetry (Boy de la Tour and Laemmli, 1988; Baumgartner et al., 1991). It was previously thought that condensation occurred by a mirror symmetric mechanism which allowed sister chromatid cohesion to persist (Boy de la Tour and Laemmli, 1988). In contrast, our model suggests that cohesion is the cause of mirror symmetric condensation. Third, defects in some mitotic structural proteins give rise specifically to condensation defects. Mutants in smc2 are defective in condensation but not cohesion (Strunnikov et al., 1995; E. Hogan and D. K., unpublished data). Candidate mutants defective for cohesion but not condensation of repetitive DNA also exist (V. G. and D. K., unpublished data).
Our model also predicts that a higher density of cohesion sites leads to smaller loops and less condensation, which can explain some differences between mammalian and budding yeast chromosome structure. FISH analysis of sister chromatids prior to and after condensation reveals two closely juxtaposed FISH signals in mammalian cells but only one signal in yeast (Selig et al., 1992; Guacci et al., 1993, 1994). We suggest that this difference can be explained by a higher density of cohesion sites in budding yeast leading to the observed 2.5- to 5- fold reduction in compaction compared to mammalian chromosomes (Lawrence et al., 1988; Guacci et al., 1994). Finally, we have candidates for cohesion sites in budding yeast (P. Megee and D. K., unpublished data), and manipulation of these sites should enable a direct test of their role in condensation.
In addition to the link between cohesion and condensation of mitotic chromosomes, other potential links are revealed by the properties of Mcd1p and the mcd1 mutant. Mcd1p levels change during S phase and the mcd1 mutant exhibits a G1/S delay (this study; V. G. and D. K., unpublished data). Since cohesion and condensation are thought to initiate in S phase (Rao and Adlakha, 1984; Selig et al., 1992; Guacci et al., 1994; Koshland and Strunnikov, 1996), this delay indicates that cells may use checkpoints to assure the proper assembly of chromosomes so that they are competent for subsequent segregation in mitosis. Consistent with this idea, G1/S delay has been observed in kinetochore mutants (Connelly and Hieter, 1996; Saitoh et al., 1997). Furthermore, the fact that both the mcd1 mutant and the S. pombe rad21 mutant are radiation sensitive (Birkenbihl and Subramani, 1992; V. G. and D. K., unpublished data) suggests a possible link between chromosome structure and the fidelity or efficiency of DNA damage repair.
Reagents were described (Yamamoto et al., 1996b; Guacci et al., 1997). Benomyl was a gift from Dupont. Standard media were made as described (Rose et al., 1990). YPRG and YPDG liquid contained 1% yeast extract, 2% peptone, 2% galactose, and either 2% raffi-nose or 2% dextrose, respectively. Yeast transformation, genetic manipulation, and plasmid isolation were described (Rose et al., 1990; Robzyk and Kassir, 1992). Yeast strains are listed in Table 1.
MCD1 was cloned in two ways. First, the mcd1–1 mutant was back-crossed and the Ts− and Mcl− phenotypes each segregated 2+:2− and cosegregated in all 27 tetrads. The Ts− phenotype was tightly linked to CEN4 (no recombination in 79 tetrads). Strain VG925–2A (mcd1) was transformed with a CEN vector-based yeast genomic library (provided by P. Hieter). Transformants (20,000) were replica plated to YPD at 37°C, and the four Ts+ transformants obtained had the same plasmid clone. A 2.8 kb XhoI/BglII fragment was inserted between the XhoI and BamHI sites of plasmid pRS316 (Sikorski and Hieter, 1989), and the resulting plasmid, pVG164, complemented the mcd1 Ts− and Mcl− phenotypes. Second, strain 3aAS273 (smc1–2) was transformed with a 2μ vector-based yeast genomic library (provided by J. Boeke). Several transformants complemented the smc1–2 Ts− phenotype and contained the same plasmid. A 2.8 kb XhoI/BglII fragment was inserted between the XhoI and BamHI sites of plasmid pRS426 (Sikorski and Hieter, 1989) and the resulting plasmid, pAS271/3, complemented the smc1–2 Ts− phenotype. Restriction mapping revealed that the 2.8 kb XhoI/BglII fragments from pVG164 and pAS271/3 were identical. The insert was sequenced (gene bank sequence U23759) and, when compared to the yeast genome sequence, mapped 3.3 kb from CEN4, consistent with the meiotic mapping data for mcd1–1 (see above).
Diploid strain (AS321) was made heterozygous for a complete deletion of MCD1 using a PCR deletion method (Baudin et al., 1993). MCD1 was also placed under control of an inducible GAL1 promoter (pGAL:MCD1) by inserting a BclI-HincII MCD1 fragment generated by PCR into plasmid pAS89 (Strunnikov et al., 1993) to form plasmid pAS339. This plasmid was linearized with BstEII, transformed into yeast to integrate pGAL:MCD1 at the MCD1 locus and to truncate the endogenous MCD1 gene to form strains YPH499b/pAS339 and BS334/pAS339. Integrations were confirmed by Southern blot.
Mid-log cultures (YPD, 23°C) were treated as follows:
α factor (10−8M), HU (100 mM), or Nz (15 μg/ml″) was added and cells incubated 3 hr at 23°C (or 30°C when appropriate) to arrest in G1, S, or mid-M phase, respectively. When required, cultures were transferred to 37°C for 2 hr while arrested. Percent cell viability was determined as described (Yamamoto et al., 1996b).
G1 arrested cells (see above) were washed with either: YPD + Nz + pronase E (0.1 mg/ml) at 37°C, then incubated 2 hr at 37°C in YPD + Nz + pronase to arrest cells in mid-M (Regimen 1); or YPD + Nz + pronase at 23°C, then incubated 2 hr at 23°C in YPD + Nz + pronase to arrest cells in mid-M and incubated 1 hr at 37°C (Regimen 2).
G1 arrested cells at 30°C were released from G1 by washing and incubation in YPD + pronase at 30°C.
S phase arrested cells were shifted to 37°C for 1 hr while arrested. Cells were released from S by washing and incubation at 37°C in YPD.
Strains YPH499b/pAS339 and BS334/pAS339 were grown in YPRG and YPDG, respectively, at 30°C to induce MCD1 overexpression from pGAL:MCD1.
Flow cytometry was as described (Yamamoto et al., 1996a). Indirect immunofluorescence was as described (Kilmartin and Adams, 1984). Mcd1p was detected by affinity-purified anti-Mcd1p antibodies (Ab) diluted 1:1000 as described for microtubules, except cells were fixed 1 hr at 23°C. Images were collected using a Zeiss epifluorescence microscope and recorded digitally using a Princeton CCD camera with Signal Analytics processing software, which allowed image superimposition.
FISH was performed as described (Guacci et al., 1994) with the following minor changes. Spheroplasted cells were resuspended in 0.5 M sorbitol, 0.5% Triton X-100, added to slides, and incubated 10 min. Sorbitol/Triton was replaced with 1% SDS and treated as described. Probes were labeled with digoxigenin as described (Guacci et al., 1994). Probes from chromosomes I, IV, XII, and XVI were described (Guacci et al., 1994, 1997).
Six probes from the right arm of chromosome VIII were made from cosmids containing a yeast DNA insert. The insert sizes were 26 kb to 42 kb and spanned a 266 kb region that is 137 kb from CEN8. Cosmids were purchased from the American Tissue Culture Collection.
The NheI-Asp718 fragment from pAS271/3 was cloned into corresponding sites of pRSETa (Invitrogen) to express a 29 kDa Mcd1p fragment in bacteria. The resultant plasmid was transformed into the BL21(DE3) pLysS strain of E. coli (Novagen), and the Mcd1p fragment purified by IMAC (ProBond, Invitrogen) then PAGE and injected into NZW rabbits (Covance). Bleeds were affinity purified on CNBr sepharose columns (Pharmacia) with the coupled antigen. Anti-Mcd1p Ab was used in 1:2000 dilution for Western blots while 12CA5 monoclonal Ab (Boehringer) and YOL1/34 Ab (Harlan) were used to monitor Pds1p-HA and α-tubulin, respectively. For immunoprecipitations, monoclonal anti-T7 tag Ab (Novagen), rabbit polyclonal anti-Smc1p Ab and sera depleted of anti-Smc1p Ab were as described (Strunnikov et al., 1993, 1995).
For analytical purposes, cells were washed with 0.15 M NaCl with protease inhibitors, and total yeast protein extracted by breaking cells with glass beads in 2% SDS with Complete protease inhibitors (Boehringer) using a Multi-Tube Vortexer (VWR). Protein concentration in boiled samples was assessed using Pierce protein assay and SpectraMax340 microplate reader (Molecular Devices). Quantification of Western blot ECL signal was performed using a Molecular Dynamics densitometer.
We thank C. Fan, Y. Zheng, P. Megee, S. Laloraya, P. Meluh, R. Kamakaka, and L. Freeman for comments and C. Norman for help in manuscript preparation. V. G. and D. K. were supported by NIH grant number GM41718 and Howard Hughes Medical Institute. A. S. was supported by Intramural Program of NICHD.