Separase is a caspase-family protease required for the metaphase–anaphase transition in eukaryotes. In budding yeast, the separase ortholog, Esp1, has been shown to cleave a subunit of cohesin, Mcd1 (Scc1), thereby releasing sister chromatids from cohesion and allowing anaphase. However, whether Esp1 has other substrates required for anaphase has been controversial. Whereas it has been reported that cleavage of Mcd1 is sufficient to trigger anaphase in the absence of Esp1 activation, another study using a temperature-sensitive esp1 mutant concluded that depletion of Mcd1 was not sufficient for anaphase in the absence of Esp1 function. Here we revisit the issue and demonstrate that neither depletion of Mcd1 nor ectopic cleavage of Mcd1 by Tev1 protease is sufficient to support anaphase in an esp1 temperature-sensitive mutant. Furthermore, we demonstrate that the catalytic activity of the Esp1 protease is required for this Mcd1-independent anaphase function. These data suggest that another protein, possibly a spindle-associated protein, is cleaved by Esp1 to allow anaphase. Such a function is consistent with the previous observation that Esp1 localizes to the mitotic spindle during anaphase.
THE metaphase–anaphase transition is regulated in part by the activity of protein-ubiquitin ligase known as the anaphase-promoting complex (APC), in conjunction with its cofactor Cdc20. In budding yeast the principal target of the APCCdc20 that enables anaphase is Pds1, known generically as securin (Cohen-Fix et al. 1996; Shirayama et al. 1999). The activity of APCCdc20 toward Pds1 and other substrates is regulated in several ways: periodic transcription of CDC20 prior to anaphase (Prinz et al. 1998; Zhu et al. 2000); inhibition of APCCdc20 in an inactive complex with spindle checkpoint proteins Mad2, Mad3, and Bub3, which are part of a larger signaling pathway that ensures bipolar attachment of spindle microtubules to kinetochores (reviewed in Tan et al. (2005); degradation of Cdc20 in response to mitotic checkpoints (Pan and Chen 2004); and ubiquitin-mediated proteolysis of Cdc20 upon mitotic exit by a different form of the APC, activated by the cofactor Cdh1 (Huang et al. 2001). Pds1 forms a complex with and inhibits a caspase-like protease, Esp1, also known as separase. It is the ubiquitin-mediated proteolysis of Pds1 that triggers anaphase by promoting activation of Esp1 (Ciosk et al. 1998). The prevalent model posits that the critical anaphase restraining target of the Esp1 protease is the Mcd1 (Scc1) subunit of cohesin, which binds sister chromatids together (Uhlmann et al. 1999). The cohesin complex, composed of the two structural-maintenance-of-chromosomes (SMC) family proteins, Smc1 and Smc3, and several non-SMC components, including Mcd1, has been proposed to form rings around the chromatid pairs spaced at intervals along the arms thereby preventing their separation (Gruber et al. 2003). Once the integrity of the cohesin ring has been breached by cleavage of Mcd1, sister chromatids are then free to separate and be drawn to opposite poles by the mitotic spindle. This model is for the most part supported by genetic and biochemical analysis. esp1 mutants fail to lose sister chromatid cohesion and undergo an abortive mitosis (McGrew et al. 1992) and Esp1 can cleave Mcd1, in vivo and in vitro, in a sequence-specific manner (Uhlmann et al. 1999, 2000). pds1 mutants, on the other hand, are unable to maintain sister chromatid cohesion, even under conditions of APC inhibition imposed by the spindle checkpoint (Yamamoto et al. 1996; Ciosk et al. 1998). Mcd1 mutants fail to establish sister chromatid cohesion (Guacci et al. 1997; Severin et al. 2001). It has been proposed that sister chromatid cohesion mediated by intact cohesin is the only force blocking anaphase spindle elongation in metaphase cells and that Mcd1 cleavage by Esp1 is sufficient for anaphase (Uhlmann et al. 2000).
However, there is not yet a consensus concerning whether Mcd1 is the only critical target of Esp1 for anaphase or whether one or more other Esp1-dependent events are required for anaphase. The Mcd1-only model is based on an experiment where Mcd1 was engineered to have a site cleavable by an ectopically expressed protease, Tev, and Esp1 was inhibited indirectly by depleting cells of the essential mitotic cofactor of the APC, Cdc20 (Uhlmann et al. 2000). Presumably under conditions of Cdc20 depletion, APC would be inactive, Pds1 stabilized, and Esp1 kept inactive. Under conditions of Tev induction and Cdc20 depletion, anaphase occurred. This result is supported by anaylsis of two different esp1 temperature-sensitive mutants in the absence of functional Mcd1 (Severin et al. 2001; Stegmeier et al. 2002). However, a different result was obtained when Esp1 was inactivated directly using a more stringent temperature-sensitive esp1 allele (Jensen et al. 2001). Under conditions of Mcd1 depletion and Esp1 thermal inactivation, anaphase did not occur, even though loss of sister chromatid cohesion could be observed. This suggests that Esp1 has at least one function, in addition to Mcd1 cleavage, required for anaphase. To resolve this discrepancy and further elucidate the mitotic functions of Esp1, we have investigated the role of Esp1 in anaphase in greater detail. We report here that multiple temperature-sensitive mutations directed to different regions of the Esp1 polypeptide are blocked at metaphase even if Mcd1 is eliminated and sister chromatid cohesion lost and that active Esp1 protease is specifically required for spindle elongation under such circumstances.
MATERIALS AND METHODS
Yeast strains and methods:
The yeast strains applicable to this study are listed in Table 1. These strains are isogenic derivatives of BF264-15DU: MATa ura3Δns ade1 his2 leu2-3,112 trp1-1 (Richardson et al. 1989). Esp1-1 was backcrossed to the BF264-15DU background five times. Genetic procedures and yeast media were formulated according to (Ausubel et al. 2002). For enhanced expression of the TET repressor GFP fusion protein under the CUP1 promoter, CuSO4 was added to a final concentration of 250 μm. The disruption of genes was carried out by the PCR-based targeting technique (Wach et al. 1994).
For analysis of cells depleted of Mcd1, overnight cultures of yeast bearing the Mcd1Δ GAL1-MCD1 constructs were allowed to reenter the cell cycle by diluting the culture to OD600 0.15 in room temperature YEP galactose for 2 hr and then were centrifuged to remove the medium and placed in YEP dextrose (YEPD) containing 12.5 ng/ml α-factor. Cells were allowed to synchronize in G1 for 2 hr, washed two times in YEPD, and released into 34° YEPD. Samples of the synchronized culture were collected at 15-min intervals and fixed for 30 min in 10% formalin. Cells were sonicated briefly to break up clumps, centrifuged at low speed to remove medium, and washed with phosphate-buffered saline (PBS). Cells were stored at 4° in PBS containing 25% glycerol until microscopic analysis could be performed.
For analysis of cells in which Mcd1 had been cleaved with TEV protease, overnight cultures of yeast bearing MCD1(TEV)-myc6 as their sole copy of MCD1 were passaged overnight in YEPD. The next morning the culture was transferred to fresh YEP raffinose (YEPR) at an OD600 of 0.3 and allowed to reenter the cell cycle for 2 hr. Cells were then arrested in G1 for 1.75 hr by the addition of α-factor at a concentration of 12.5 ng/ml. The GAL1-HA-TEV-CMV(NLS) construct was then induced for 1 hr by the addition of galactose to 2% (w/v) prior to release from G1. Cells were released from G1 by centrifuging at low speed, removing the medium, washing two times in YEPR containing 2% galactose (gal), and releasing into YEPR + 2% gal at 34°. Samples of the synchronized culture were collected at regular intervals and either fixed in 10% formalin (1 ml) for microscopic analysis or centrifuged and frozen at −80° (25 ml) for biochemical analysis.
To compare temperature sensitivity of mutant Esp1 proteins, overnight cultures grown at room temperature in YEPD were diluted to OD600 = 0.2 in YEPD + 12.5 ng/ml α-factor. Cells were arrested for 3 hr, and then collected, washed twice in YEPD, and released into 34° YEPD. Sixty minutes after release, α-factor was added back to the cultures to prevent reentry into S phase and new synthesis of Mcd1. Aliquots of cells were collected at 0, 60, and 120 min after the initial release from α-factor. Approximately 50 μg of protein were loaded into each lane of an 8% SDS–PAGE gel. To assess Mcd1 cleavage, blots were probed with anti-myc antibody (9E10). Blots were also probed with anti-α-tubulin (YOL-1) for normalization.
To carry plating assays to compare temperature sensitivity of esp1 alleles, serial threefold dilutions were prepared in YEPD liquid medium and spotted onto YEPD plates so that the most concentrated drop contained 5000 cells. Triplicate spotting was carried out with one plate incubated at 22°, a second at 30°, and a third at 33.5°. Plates were incubated until significant colony formation was observed for the wild-type control.
Cell biology protocols:
Fluorescence microscopy was performed on a Zeis Axioskop2 with a 63× objective. Images of cells were captured using Axiovision Rel. 4.3 software. Specifically, images for FITC and Cy3 filter sets (Carl Zeis, Thornwood, NY) were captured in seven planes of focus. These images were first analyzed for loss of cohesion (FITC) by selecting cells that exhibited two clearly separated chromosome IV dots and then scoring spindle length in the Cy3 channel. Cells that did not exhibit loss of cohesion were not scored for spindle length. Cells were scored for budding by counting at least 200 cells using transmission microscopy. Cells were scored for loss of cohesion by evaluating separation of chromosome IV dots in 100 cells at each time point. For time-lapse recordings of spindle dynamic behavior, strains CBY128 and CBY102 (esp1-C113 or ESP1, respectively, and containing Mcd1Δ GAL1:Myc-MCD1 trp1∷TET(o)∷TRP1 CUP1PTET(rep)-GFP HIS3:mCherry:TUB1) were grown overnight in selective galactose-containing medium at 25° to obtain cultures with no more than 20% budded cells. After collecting by centrifugation, cells were resuspended to a concentration of 3 × 106 cells/ml in synthetic glucose-containing medium (to repress GAL1:MCD1) and incubated for 1.5 hr at 32° before mounting on the same medium containing 25% w/v gelatin to perform time-lapse recordings at 32° using a Nikon Eclipse E800 with a CFI Plan Apochromat 100× N.A. 1.4 objective, Chroma Technology triple band filter set 82000v2 and a Coolsnap-HQ CCD camera (Roper Scientific). Five-plane Z-stacks were acquired at 30-sec or 1-min intervals and images were processed as previously described (Maddox et al. 1999; Huisman et al. 2004). To avoid excessive photobleaching of tagged microtubule-based structures, cells were monitored by direct observation to select cells that had undergone loss of sister chromatid cohesion and contained a short spindle at the start of the recording.
Frozen cell pellets were mixed with buffer containing 25 mm HEPES, 0.5 mm EGTA, 0.1 mm EDTA, 2 mm MgCl2, 1 mm NaHSO3, 0.02% NP-40, 20% glycerol, 50 mm NaF, 200 μm NaVO3, and complete protease inhibitor cocktail (catalog no. 1697498, Roche, Indianapolis) then transferred to 2 ml screw-cap minimally conical tubes. Glass beads (0.5 mm, Biospec) were added until their level was just under the meniscus. Cells were lysed in a FastPrep FP120 at power lever 4.5 (4 × 30 sec). Soluble lysates were quantitated for protein using the Lowry method (Bio-Rad, Hercules, CA).
Protein samples (50 μg) were resolved on 8% tris-glycine SDS–PAGE gels. Gels were then transferred to PVDF. Blots were blocked in 10% milk then probed with antibodies diluted in PBS containing 20 mg/ml BSA and 0.05% NaN3. To probe for the myc epitope, 9E10 was used at a dilution of 1:5000, and for detection of α-tubulin the rat monoclonal YOL1/34 was used. Blots were developed using antimouse or antirat, as appropriate.
The mCherry-TUB1 construct was made in a manner similar to the GFP-tubulin plasmid described previously by (Straight et al. 1997). Briefly, the mCherry-RFP sequence described by Shaner et al. (2004) was amplified by PCR and fused to the HIS3 promoter. The TUB1 gene was fused in-frame to the 3′ end of mCherry-RFP. The intergenic region between CLB1 and CLB6 was added 3′ of the TUB1 ORF for transcriptional termination. The HIS3(pr)-mCherry-TUB1 construct was then placed in pRS406 between XhoI and SacI for integration at the URA3 locus. For ESP1 complementation experiments ESP1 was cloned into this construct between XhoI and KpnI.
A genomic replacement of the region encoding the N terminus of the Esp1 cleavage site within Mcd1 with a TEV protease cleavage site was created using the pop-in/pop-out technique. A cassette of the sequence between Bst EII and Bst BI was created using primer sequences GATTTCGAACATAATAATTTGTCTAGTATG and GTTCGAAATCTTCATCTGGACTGAATCCTTGGAAGTATAGGTTTTCAGTATCCCATGGAGCAGCACC. This cassette changed the region encoding the Esp1 cleavage sequence WDTSLEVGRRFSP to one encoding the TEV-P cleavable site WDTENLYFQGFSP and placed a unique NcoI site within the MCD1 ORF. The plasmid pRS406 with the MCD1 gene containing this cassette was digested at the unique Bst EII site 5′ to the TEV-P cleavage site. Cells were first selected on uracil dropout medium then on medium containing 5-FOA to select for loss of the URA3 marker. Pop-outs were identified as being properly recombined by selecting for NcoI cleavage of the PCR product produced by primers TCTTCAATTGACCCTTCTCGCCCA and TCGGGCACTGTTGCCGTATATTCT using genomic DNA as a template.
To place a C-terminal 6× Myc tag on Mcd1, a portion of the 3′ end of the MCD1 ORF was amplified using primer sequences GATGCGGCCGCGAGCATTGATAAACCTTTCAAATAGTGC and CGTCGACCAAACTGGCACAAGAAGGAACTC. This was fused in frame with a sequence encoding six repeats of the Myc epitope EQKLISEEDL followed by the CLB1/6 transcriptional terminator. This sequence was placed in the plasmid pAG34 with SalI and SacI linkers, respectively. Yeast that were resistant to hygromycin B were tested for proper integration of the Myc-tagging construct using primers TTTCTGTGTAGGCTAGCACCTGG and AGAAGCACCCGCAGGCAATATAGA.
The galactose-inducible TEV protease expression plasmid was created by amplifying the region encoding TEV-P from the tobacco etch virus genome using primer sequences ACTCGAGCCATGGCTGAAAGCTTGTTTAAG and GTCTAGATCAGTCGACTTGCGAGTACACCAATTCATTC. A cytomegalovirus nuclear localization signal sequence (CMV_NLS) was added to the pBluescript vector by using the primer sequences CACTCTAGAGTCGACTGTACTCCACCAAAGAAGAAGAGAAAGGTTGCCTAAGCGGCCGCCACCGCGGTG and TGGCCTTTTGCTGGCCTTTTGCTCACATGT. The TEV-P sequence was fused in frame 5′ to the CMV_NLS to make TEV-P-CMV_NLS. To make the expression of TEV-P-CMV_NLS inducible with galactose, the GAL1 promoter [GAL1(pr)] was amplified with primer sequences AGGGCCCTTGGATGGACGCAAAGAAGTTT and ACTCGAGCGCATAGTCAGGAACATCGTATGGGTAAGCCATGGTATAGTTT TTTCTCCTTGACGTTAAAGTATAGAG. Note that this version of the GAL1(pr) will express an HA epitope (YPYDVPDYA) tagged TEV-P-CMV_NLS (HA-TEV-P-CMV_NLS).
To introduce the HA-TEV-P-CMV_NLS under histidine selection, the HIS2 gene was amplified from yeast genomic DNA using primer sequences CTCAGCGATATCATTTTGATTTACTAAATGCTATTTATCC and CAGTGCAGATCTACAGCTTTTGTTTTTGATTTCTTTGCC. The backbone of pRS404 was amplified with primer sequences GCCAGTCAGGCCTATGCGGTGTGAAATACCGCAC and GTGCACTGATCATATGGTGCACTCTCAGTACAATC. The products of these PCRs were ligated together to form the HIS2 selectable integrating plasmid pRS40-HIS2. The GAL1(pr)-HA-TEV-P-CMV_NLS construct was then ligated into this plasmid between SacI and ApaI.
Mutations introducing substitutions throughout the Esp1 polypeptide prevent loss of sister chromatid cohesion and cause mitotic failure:
Separase homologs are identified on the basis of a region of sequence similarity in their caboxy-termini that corresponds to a CD clan-like protease motif. The aminotermini of this family of enzymes are poorly conserved. In the Saccharomyces cerevisiae separase homolog Esp1, the N-terminal region comprises approximately the first 850 amino acids (aa) of the 1630 aa open reading frame (ORF). This region is similar to several fungal separases but lacks sequence similarity to homologs in higher eukaryotes. To further elucidate the function(s) of the regions of Esp1 outside of the protease homology domain, we divided the ORF into three relatively equal regions that were independently mutagenized to create a library of temperature-sensitive lethal esp1 mutants. We were able to produce stringent temperature-sensitive alleles by independently targeting all three regions of the ORF, indicating that all are required for essential Esp1 function(s). In addition, on the basis of phenotypic analysis of a large number of mutant alleles in the three regions (n = 240), there were no domain-specific phenotypic differences (data not shown). In synchronized cultures progressing through mitosis, there was a failure to lose sister chromatid cohesion, a failure in anaphase spindle elongation, and a translocation of the undivided nucleus into the bud. Although this mutational analysis cannot be assumed to be saturating (only 240 alleles were analyzed), it appears that the entire Esp1 polypeptide contributes to the essential mitotic function of Esp1.
Stringent temperature-sensitive esp1 alleles confer a block to anaphase spindle elongation in the absence of Mcd1:
We have reported previously that a temperature-sensitive allele of ESP1, esp1-B3 was defective in anaphase spindle elongation, even under conditions of Mcd1 depletion (Jensen et al. 2001). This experiment was carried out by regulating the synthesis of the cohesin protein Mcd1 using a heterologous fusion construct of the GAL1 promoter with the MCD1 ORF. Growth of cells in dextrose medium where the GAL1 promoter is repressed leads to depletion of Mcd1. Under these conditions, even though sister chromatid cohesion was lost, on the basis of scoring of a fluorescent chromosome marker, spindle elongation did not occur. The esp1-B3 allele was derived from mutagenesis specifically in the central region of Esp1 and was characterized by a low restrictive temperature (31.5°), necessary technically for accurate scoring of spindle elongation. To determine whether cohesion-independent failure of spindle elongation is a characteristic of the esp1-B3 allele or whether it is a general characteristic of Esp1 loss of function, additional stringent temperature-sensitive alleles (restrictive temperature ≤33°) in different regions of the Esp1 ORF were chosen for analysis. All temperature-sensitive strains sequenced contained multiple mutations although we did not verify which mutations contributed to the phenotype. The specific combinations of mutations in these alleles are shown in Table 2. Mutant strains contained a construct expressing an mCherry-RFP-tubulin fusion protein for monitoring spindle length and GFP-Tet repressor with a Tet operator array integrated at the TRP1 locus on chromosome IV to determine loss of sister chromatid cohesion. Mcd1 expression was under control of the GAL1 promoter, as described previously (Jensen et al. 2001). In addition, the strains contained a mad2 deletion to prevent a mitotic delay that occurs in the absence of sister chromatid cohesion (Severin et al. 2001). Cells were synchronized by mating pheromone arrest, released into dextrose medium to terminate expression of Mcd1, and were harvested as a function of time after release at 15-min intervals. Populations were lightly fixed and then scored for spindle length and loss of chromatid cohesion. Representative images used for this analysis are shown in Figure 1. Only cells where two well-separated chromosome IV spots could be observed were measured for spindle length. For each population, >60% of cells fit this criterion at peak mitotic times (supplemental Figure 1A, data not shown). It has been reported that chromosomal regions that are centromere-proximal, can separate prior to anaphase (Pearson et al. 2001). However, there was no significant preanaphase separation at a site 12.7 kb from CEN11, more centromere-proximal than the Tet operator array used here (∼16 kb from CEN4). Consistent with these observations, the Tet operator array system has been shown previously not to exhibit preanaphase separation in the esp1 mutants (Jensen et al. 2001). Therefore, as expected, for mutants grown in the presence of galactose, so that Mcd1 was not depleted, <10% of cells showed separated chromosome IV spots, confirming that esp1 mutation prevents loss of cohesion under the experimental conditions employed (supplemental Figure 1A, data not shown). The spindle-length data are shown in Table 3 for the peak mitotic time point for the wild-type control (90 min). The peak average pole-to-pole spindle length for the wild-type strain is ∼6 μm at 90 min. Although fully extended spindles in S. cerevisiae are ∼8 μm, the smaller length of the average represents imperfect synchrony (see Figure 1). On the other hand, the three temperature-sensitive alleles analyzed, esp1-N120 (amino-terminal region), esp1-B120 (central region), and esp1-C113 (carboxy-terminal region) exhibited significantly less spindle elongation at 90 min and at later points. The entire time course for wild type and esp1-C113 is plotted in Figure 1. These data indicate that during the entire interval where wild-type cells undergo mitosis, esp1-C113 cells extend spindles to a maximum length of ∼4 μm with an average length of ∼3 μm. Therefore temperature-sensitive alleles of esp1 across the entire Esp1 ORF confer a defect in anaphase spindle elongation.
Time-lapse analysis of spindle elongation indicates that spindle length is not dynamic in esp1 mutants:
Even though on the basis of population counts of fixed cells it appears that esp1 spindles do not elongate sufficiently for anaphase, it is possible that these spindles might go through transient cycles of full elongation and collapse or that they might collapse during the fixation process. We therefore carried out time-lapse photomicroscopy of individual live wild-type and esp1-C113 cells progressing through mitosis. The same strains used in analysis of synchronized populations were subject to live imaging microscopy following a shift to dextrose medium to deplete Mcd1 and incubation at 32°. Cells that had undergone loss of sister chromatid cohesion and contained a short spindle were selected for recordings. Representative time-lapse sequences for two wild-type controls and two esp1-C113 mutants are shown in Figure 2, A and B, respectively. Both wild-type cells proceeded to elongate the mitotic spindle within ∼26.5 min. On the other hand, the mutant cells exhibited a marginal elongation (<3.5 μm) of the spindle by 47 min. The corresponding plots for the kinetics of spindle elongation in wild-type vs. esp1-C113 cells is shown in Figure 2C. These dynamic results are in good agreement with the still-image counts and measurements obtained from fixed cell populations (Figure 1).
An ectopically cleavable Mcd1 does not rescue the spindle elongation defect of esp1-C113:
To eliminate the possibility that the loss of sister chromatid cohesion at the TRP1 locus in Mcd1-depleted cells was not indicative of complete loss of cohesion throughout the entire length of all chromosomes because of residual Mcd1 protein, we created a strain in which the chromosomal MCD1 gene was replaced by a mutant gene encoding an Mcd1 protein in which the Esp1 cleavage site at amino acid 180 was replaced by a TEV protease cleavage site (Mcd1-Tev). This leaves one intact Esp1 cleavage site at amino acid 268. To monitor Mcd1 protein we added six copies of the Myc epitope tag at the carboxy terminus. Western blot analysis of lysates from these cells using anti-Myc antibody revealed a band that migrates at 105 kDa. This size is consistent with the mobility of Mcd1 reported by Uhlmann et al. (2000). The strain also contains the Tev protease ORF fused to the GAL1 promoter. Growth of these cells in galactose-containing medium resulted in the induction of the TEV protease and cleavage of Mcd1, as observed by the disappearance of the 105-kDa band and the appearance of a 77-kDa band (Figure 3C), and ultimately, lethality (data not shown). To assess the requirement of Esp1 for spindle elongation in cells where Mcd1 is ectopically cleaved by Tev protease, we synchronized cells by mating pheromone arrest–release and shifted cultures into galactose (to induce Tev protease expression) or dextrose (to maintain repression of Tev protease). Once again spindle length was visualized in cells that had clearly shown chromatid separation indicating a lack of cohesion. Wild-type cells exhibited mitotic spindle elongation whereas esp1-C113 mutants again demonstrated poor spindle elongation (Figure 3B). For each population, loss of sister chromatid cohesion was determined to be >60% at peak mitotic times on the basis of scoring chromosome IV dots (supplemental Figure 1B). Therefore, expression of Tev protease led to loss of cohesion in cells containing Mcd1-Tev, even in the absence of Esp1 function. To confirm that the Mcd1 was efficiently cleaved in this experiment, extracts were prepared and subjected to SDS–PAGE followed by Western blotting (Figure 3C). In both wild-type and esp1-C113 cells, induction of Tev protease by growth in galactose-containing medium led to an almost complete loss of full-length Mcd1 by 90 min after release from mating pheromone arrest. Note that this is the time at which wild-type cells exhibit spindle elongation. At the same time, the 77-kDa cleavage product was apparent during the entire time course for both wild type and mutant. Under conditions of growth in dextrose-containing medium, even when Mcd1 is not completely cleaved by Esp1 (Figure 3C), cells progress through mitosis normally (data not shown). Therefore complete loss of Mcd1 is not necessary for loss of cohesion and mitotic progression. The mitotic spindle elongation does not appear as robust in terms of spindle length for the wild-type control in this experiment compared to the experiments in Table 3 and Figure 1. The reason for this is that the strains used for this experiment did not contain a mad2 deletion, resulting in poorer synchrony of mitotic progression. This results in reduced average spindle length at any particular time point. Nevertheless, statistical analysis comparing the mean spindle lengths at every time point after 75 min indicated that the differences were highly significant (P ≤ 0.001) (Table 4). Importantly, the mean spindle lengths for the esp1-C113 mutant were not significantly different in this experiment as compared to those in Figure 1, consistent with the conclusion that these spindles do not undergo anaphase spindle elongation.
The protease activity of Esp1 is required for anaphase spindle elongation in Mcd1-depleted cells:
Whereas cleavage of Mcd1 requires Esp1 endoprotease activity, another role attributed to Esp1, participation in the FEAR (Cdc fourteen early anaphase release) mitotic exit pathway, does not require this activity of Esp1 (Stegmeier et al. 2002; Sullivan and Uhlmann 2003). To determine whether Mcd1-independent anaphase spindle elongation functions of Esp1 require endoprotease activity, an active site mutation was created (esp1-C1531S). Replacing this cysteine with serine removes the protease nucleophile. esp1-C1531S in parallel with wild-type ESP1 was introduced into the temperature-sensitive esp1-C113 mutant strain on a centromeric plasmid. Cells were synchronized using mating pheromone, Mcd1 was depleted by shift to dextrose, and loss of cohesion and spindle length were scored as a function of time (Figure 4). In esp1-C113 mutants expressing esp1-C1531S or ESP1, cohesion was lost, as expected (supplemental Figure 1C). However, whereas wild-type ESP1 could completely rescue the spindle elongation defect associated with esp1-C113, the catalytic site mutant (esp1-C1531S) could not. Therefore, like Mcd1 cleavage, anaphase spindle elongation functions of Esp1 require endoproteolytic activity.
The temperature sensitivity of esp1 mutants correlates with severity of the protease defect regardless of where the mutation resides:
One possibility to explain why only highly temperature-sensitive esp1 mutants experience defects in spindle elongation even in the absence of Mcd1 is that spindle elongation requires only low levels of protease activity. On the basis of this idea, less temperature-sensitive strains would possess sufficient residual Esp1 protease activity for spindle elongation functions but not for Mcd1 cleavage at the restrictive temperature. We therefore compared temperature sensitivity on the basis of a plating assay carried out at several temperatures and ability to cleave Mcd1. In Figure 5A, serial dilutions of wild-type and several previously described esp1 mutants are plated at 22°, 30°, and 33.5°, respectively. While several of these have been reported to have spindle elongation defects in the absence of cohesion (esp1-B3, esp1-B120, and esp1-C113) (this study and Jensen et al. 2001), two (esp1-1 and esp1-N5) have been reported to support spindle elongation in the absence of cohesion (Severin et al. 2001; Stegmeier et al. 2002). Whereas all strains plated efficiently at 22°, varying degrees of growth defect were observed at the higher temperatures (Figure 5A). esp1-N5 and esp1-1 (Baum et al. 1988) were not temperature sensitive at 30°, while esp1-B3, esp1-B120, and esp1-C113, characterized in the current study or in our previous study (Jensen et al. 2001), were quite defective for growth at 30°. esp1-C113 showed slightly greater temperature sensitivity than the other mutants. None of the mutants grew at 33.5°. To measure the temperature sensitivity of the mutant Esp1 proteases, strains were constructed that contained the temperature-sensitive alleles and a 6xMyc-tagged endogenous allele of Mcd1. To compare efficiency of Mcd1 cleavage, strains were synchronized by α-factor arrest, released into medium at 34°, and then after bud emergence, α-factor was added back (60 min after release). Aliquots were harvested at the time of release from the α-factor block 60 min later, which corresponds to late S phase and G2 for the wild-type strain and 120 min later, when all cells that completed mitosis would be blocked in G1 due to pheromone treatment. The rationale for the second α-factor block is to prevent resynthesis of Mcd1 once cells have completed mitosis. Mcd1 levels were then determined by Western blotting and normalized to α-tubulin levels (Figure 5B). In wild-type cells, high levels of Mcd1 were present at 60 min but not at 0 or 120 min, indicating that Mcd1 is synthesized and then degraded. All esp1 mutants were defective at Mcd1 degradation to varying degrees. esp1-C113, the most temperature-sensitive allele appeared to be completely defective, in that Mcd1 levels were increased between 60 and 120 min. All other temperature-sensitive mutants tested decreased Mcd1 levels by a factor of two between 60 and 120 min. Therefore, on the basis of a combination of two criteria, growth at elevated temperature and ability of Esp1 to cleave Mcd1, esp1 temperature-sensitive mutants tested can be divided into three groups, with esp1-1 and esp1-N5 falling into the least sensitive group and esp1-C113 being the most temperature sensitive. esp1-B3 and esp1-B120 were in an intermediate category, on the basis of the plate test. However, esp1-B120 protease activity was not tested.
Cleavage of Mcd1 is not the only essential mitotic function of Esp1:
It has been previously suggested that tension exerted by the mitotic spindle is sufficient to pull apart sister chromatids once the counterforce of sister chromatid cohesion is neutralized (Uhlmann et al. 2000). This theory predicts that Mcd1 endoproteolysis relieves the single barrier to anaphase chromosome segregation. Experiments supporting this idea were conducted in cells that contained a repressible heterologous MET3-CDC20 expression construct and a TEV-cleavable mutant allele of Mcd1 (Uhlmann et al. 2000), similar to the one described here. The rationale for this experiment is that under conditions of MET3 promoter repression (high methionine), the depletion of Cdc20 should maintain Pds1-dependent inhibition of Esp1, which in this case is wild type. The validity of this experiment, however, depends on a number of uncertain assumptions. First, basal expression of the MET3-CDC20 construct might yield a low but significant amount of active Cdc20 leading to some Pds1 proteolysis. Second, it is not clear whether all Esp1 molecules are inhibited by Pds1 or whether Esp1 activity is completely inhibited by bound Pds1. In the context of the scenarios suggested above, it may be that only low levels of Esp1 activity or residual Esp1 activity characteristic of Pds1 inhibition are required for spindle elongation functions in contrast to high levels of activity required for complete loss of sister chromatid cohesion. In another study where the temperature-sensitive Mcd1-73 mutant was combined with a temperature-sensitive esp1 mutant (esp1-N5) spindle elongation proceeded in a manner comparable to the single Mcd1-73 mutant (Severin et al. 2001). However, the esp1-N5 mutation derived from our own work (Jensen et al. 2001) is a relatively leaky allele (Figure 5A) and is not defective in spindle elongation (data not shown). Therefore, this experiment is not informative with respect to Esp1 function. In a third study (Stegmeier et al. 2002), the esp1-1 mutant, also leaky (Figure 5A), was combined with an Mcd1 deletion, allowing spindle elongation, similarly to esp1-N5. Because these previous studies either employed less stringent alleles or used indirect methods to inactivate Esp1, we feel that they are inconclusive with respect to a direct role for Esp1 in spindle elongation and not in direct conflict with the data presented here.
The entire Esp1 protein participates in its proteolytic functions:
Sequence alignments reveal that Esp1 contains a caspase-like sequence at its carboxy terminus. The proteolytic activity associated with this region is essential for Mcd1 cleavage and cohesin dissociation from sister chromatids (Uhlmann et al. 2000). However, Esp1 contains a large amino-terminal extension from the proteolytic domain that comprises the bulk of the protein. The specific molecular functions associated with this region have yet to be fully elucidated. However, highlighting the critical function of the separase amino terminus, insect separase is expressed from two essential genes: Separase (Sse) encoding the protease catalytic domain and three rows (Thr) encoding a positive regulatory domain (Jager et al. 2001). These two proteins are required to physically interact for full proteolytic activity of SSE. Indeed, interaction of the N-terminal and caspase-like domains is required for proteolytic activity of all separases tested including those synthesized as a single polypeptide. Structural predictions suggest that THR adopts an α–α superhelical structure characteristic of ARM/HEAT repeats. Similarly, structural predictions have also proposed α–α superhelical structures in the aminotermini of separases found in humans, Caenorhabditis elegans, Arabidopsis, Schizosaccharomyces pombe, and S. cerevisiae (Jager et al. 2004). The α–α superhelical structure is associated with proteins that assume scaffold or adapter roles, such as β-catenin, protein phosphatase 2A PR65/A subunit, and importin β. This suggests that the amino-terminal regions of separases are adapters that might recruit substrates to the catalytic site. Consistent with this idea, blockage of the active cleft of yeast Esp1 with a substrate-mimetic inhibitor does not inhibit association of Esp1 and Pds1, indicating that substrate recruitment sites are distal from the catalytic site (Hornig et al. 2002). In the current study, we carried out a phenotypic analysis of a large number of temperature-sensitive alleles targeted to three distinct regions encompassing the entire Esp1 protein. We do not know the precise locations of the mutations, and it is likely due to the degree of mutagenesis that each allele contains multiple point mutations. Nevertheless, in every case, the phenotype was identical to that defined on the basis of the original temperature-sensitive esp1 allele identified, esp1-1 (Baum et al. 1988; McGrew et al. 1992). Specifically at restrictive temperature, sister chromatid cohesion was not lost, spindles failed to elongate, and eventually the undivided nucleus translocated into the bud. We found no alleles that were permissive for sister chromatid separation but defective in spindle elongation. These data suggest that the entire length of the Esp1 protein contributes to an integral biochemical function that targets multiple substrates and argues against a substrate-specific allocation of domains. However, because the mutant screen, although intensive, was probably not saturating, this conclusion has to be taken as tentative.
The Esp1 target regulating spindle elongation:
To date, two endoproteolytic targets of Esp1 have been identified. Mcd1 is required for maintenance of sister chromatid cohesion (Uhlmann et al. 1999) and Slk19, a kinetochore protein is required for anaphase spindle durability as well as for early mitotic exit (FEAR) pathway (Sullivan et al. 2001; Stegmeier et al. 2002). However, a non-Esp1-cleavable version of Slk19 confers little of any overt phenotype (Sullivan et al. 2001). Certainly, there is no defect in spindle elongation, ruling out Slk19 as the Esp1 target whose cleavage is required for anaphase spindle elongation. Attempts to identify other Esp1 targets on the basis of the rather degenerate protease cleavage site consensus have failed (Sullivan et al. 2004). Therefore, the relevant target(s) remain(s) unknown. One untested possibility is the class of microtubule destabilizing motors that have been shown to antagonize spindle elongation. Kar3 and Kip3 have been shown to restrain the spindle elongating activities of Kip1 and Cin8, also microtubule motor proteins (Straight et al. 1998; Cottingham et al. 1999). It is conceivable that Esp1 is responsible for inactivating Kar3 and/or Kip3 (or positive regulators of these proteins) at the metaphase/anaphase transition, thereby potentiating spindle elongation. Alternatively, Esp1 might target negative regulators of Kip1 and/or Cin8, allowing them to promote spindle elongation. The targeting of spindle motors or associated proteins by Esp1 would be consistent with localization of Esp1 to spindle during anaphase (Jensen et al. 2001).
This work was supported by National Institutes of Health grant GM-38328 to S.I.R. and by grants from Cancer Research UK and The Isaac Newton Trust to M.S.
Communicating editor: T. Stearns
- Received November 30, 2007.
- Accepted January 19, 2008.
- Copyright © 2008 by the Genetics Society of America