Smt3/SUMO and Ubc9 are required for efficient APC/C-mediated proteolysis in budding yeast

Authors


E-mail sirnige@gwdg.de; Tel. (+49) 551 393 818; Fax (+49) 551 393 820.

Summary

Ubiquitin-mediated proteolysis triggered by the anaphase-promoting complex/cyclosome (APC/C) is essential for sister chromatid separation and the mitotic exit. Like ubiquitylation, protein modification with the small ubiquitin-related modifier SUMO appears to be important during mitosis, because yeast cells impaired in the SUMO-conjugating enzyme Ubc9 were found to be blocked in mitosis and defective in cyclin degradation. Here, we analysed the role of SUMOylation in the metaphase/anaphase transition and in APC/C-mediated proteolysis in Saccharomyces cerevisiae. We show that cells depleted of Ubc9 or Smt3, the yeast SUMO protein, mostly arrested with undivided nuclei and with high levels of securin Pds1. This metaphase block was partially relieved by a deletion of PDS1. The absence of Ubc9 or Smt3 also resulted in defects in chromosome segregation. Temperature-sensitive ubc9-2 mutants were delayed in proteolysis of Pds1 and of cyclin Clb2 during mitosis. The requirement of SUMOylation for APC/C-mediated degradation was tested more directly in G1-arrested cells. Both ubc9-2 and smt3-331 mutants were defective in efficient degradation of Pds1 and mitotic cyclins, whereas proteolysis of unstable proteins that are not APC/C substrates was unaffected. We conclude that SUMOylation is needed for efficient proteolysis mediated by APC/C in budding yeast.

Introduction

Important cell cycle transitions, such as the initiation of DNA replication, sister chromatid separation and the exit from mitosis, require the proteolytic destruction of specific regulatory proteins. Degradation is initiated by the attachment of chains of ubiquitin molecules to these target proteins (reviewed by Hershko and Ciechanover, 1998; Kornitzer and Ciechanover, 2000). Ubiquitin-mediated proteolysis ensures that proteins are degraded by the 26S proteasome only after they are tagged with ubiquitin.

Ubiquitylation requires the activity of three enzymes, termed E1, E2 and E3 (reviewed by Hershko and Ciechanover, 1998; Glickman and Ciechanover, 2002). Ubiquitin, a 76 kDa protein, is first bound to E1, a ubiquitin-activating enzyme. From E1, it is transferred to one of several E2 enzymes, called ubiquitin-conjugating enzymes. Subsequently, ubiquitin is attached to lysine residues of the substrate. This final step is catalysed by ubiquitin ligases, the E3 enzymes. Polyubiquitinated substrates are recognized and degraded by the 26S proteasome.

Two ubiquitin ligases are known to play fundamental roles during the cell cycle, the Skp1–cullin–F-box complex (SCF) and the anaphase-promoting complex/cyclosome (APC/C) (Jackson et al., 2000; Peters, 2002). In budding yeast, the major cell cycle role of SCF is the regulation of the G1/S transition, whereas APC/C is required for sister chromatid separation and the exit from mitosis. APC/C is a large complex consisting of at least 13 subunits and is highly conserved among eukaryotes (Harper et al., 2002; Peters, 2002). This ubiquitin ligase is activated at the metaphase/anaphase transition and is turned off at the end of the subsequent G1 phase. Important regulatory proteins are Cdc20 and Cdh1, which contain WD40 repeat motifs and are implicated in substrate recognition (Pfleger et al., 2001; Vodermaier, 2001). A key role of APC/CCdc20 is the ubiquitylation of securins, known as Pds1 in budding yeast (Cohen-Fix et al., 1996). Securins are inhibitors of separases that trigger the separation of sister chromatids by cleavage of a cohesin subunit (Nasmyth, 2002). APC/CCdh1 is activated in late anaphase and triggers the complete destruction of mitotic cyclins (Harper et al., 2002; Peters, 2002).

Previously, Seufert et al. (1995) found that degradation of the budding yeast B-type cyclins Clb2 and Clb5 requires the Ubc9 protein. UBC9 encodes a protein homologous to ubiquitin-conjugating enzymes and was initially proposed to act as an E2 enzyme for the APC/C ubiquitin ligase. Later, it was shown that Ubc9 does not transfer ubiquitin to its target proteins, but instead modifies its targets with a protein related to ubiquitin, termed small ubiquitin-related modifier or SUMO (Johnson and Blobel, 1997; Schwarz et al., 1998).

The sequence identity between SUMO and ubiquitin is only 18%, but these proteins share a similar three-dimensional structure (Bayer et al., 1998). Like ubiquitin, SUMO is highly conserved in eukaryotes. Budding yeast has a single SUMO gene, known as the essential SMT3 gene, whereas three members of the SUMO family have been identified in vertebrates (reviewed by Melchior, 2000; Müller et al., 2001; Kim et al., 2002). The pathway of SUMO conjugation, termed SUMOylation, is similar to ubiquitylation, but requires different enzymes. SUMO is first bound to a heterodimeric E1 composed of Aos1 and Uba2. It is then transferred to Ubc9, the only known SUMO-conjugating enzyme, which in turn catalyses the formation of an isopeptide bond between the C-terminus of SUMO and an ∈-lysine residue of a target protein. Ubc9 shares structural similarities with ubiquitin-conjugating enzymes but, unlike these E2s, Ubc9 contains a positively charged surface. By analogy with ubiquitylation, E3 SUMO ligases were recently identified in yeast and mammals (Johnson and Gupta, 2001; Takahashi et al., 2001; Pichler et al., 2002).

In contrast to ubiquitylation, the modification of proteins with SUMO does not result in their proteolytic degradation. Instead, multiple other effects of SUMOylation have been described, for example the modulation of the subcellular localization of proteins, of protein–protein interactions or of the activity of transcription factors (Müller et al., 2001; Wilson and Rangasamy, 2001). Intensively studied SUMO substrates are, for example, mammalian RanGAP1, a factor required for nucleocytoplasmic transport, and the tumour suppressor p53. SUMO targets RanGAP1 to nuclear pores and stimulates the transcriptional and apoptotic activities of p53 (Gostissa et al., 1999; Rodriguez et al., 1999). Previous studies have also revealed a functional link between SUMOylation and ubiquitylation. In the case of IκBα, an inhibitor of the NF-κb transcription factor, SUMOylation was shown to antagonize ubiquitylation, thus preventing degradation of the protein by the ubiquitin pathway (Desterro et al., 1998).

The defects of yeast ubc9-1 mutants in the degradation of B-type cyclins may also indicate a link between SUMOylation and ubiquitin-mediated proteolysis (Seufert et al., 1995). Furthermore, several reports have implicated important functions for SUMO during M-phase of the cell cycle. Yeast cells depleted of UBC9 were impaired in mitosis, and temperature-sensitive smt3 mutants, defective in the yeast SUMO gene, were identified in a screen for mutants defective in chromosome segregation (Biggins et al., 2001). Schizosaccharomyces pombe cells lacking the SUMO gene pmt3 and the UBC9 homologue hus5 strains also displayed defects in chromosome segregation (al-Khodairy et al., 1995; Tanaka et al., 1999).

Here, we found that yeast cells lacking functional Smt3 and Ubc9 proteins fail to degrade Pds1 in mitosis and that the metaphase block of these cells can be partially bypassed by a PDS1 deletion. Furthermore, we show that mutant strains defective in SUMOylation are impaired in proper proteolysis of securin and other APC/C target proteins in mitosis and in G1-arrested cells.

Results

Yeast cells depleted of Ubc9 and Smt3 arrest with short spindles and high levels of Pds1

It was reported previously that yeast ubc9Δ cells, which were kept alive by UBC9 expressed from the galactose-inducible GAL1-10 promoter, arrested in mitosis upon promoter shut-off in glucose medium (Seufert et al., 1995). Despite this cell cycle defect, we found that most cells of a ubc9Δ GAL-UBC9 strain were viable and able to form colonies on plates containing glucose (data not shown). These findings prompted us to reinvestigate the requirement of UBC9 for viability, by analysing cells completely depleted of Ubc9. Haploid segregants containing the ubc9Δ allele were obtained from a sporulating heterozygous diploid UBC9/ubc9Δ strain. The analysis of about 30 haploid ubc9Δ segregants revealed that cells lacking UBC9 were able to produce microcolonies (Fig. 1A), mostly consisting of 30–50 cells, before cells ceased cell division and finally lysed. Thus, ubc9Δ segregants were able to divide on average four to six times, implying that Ubc9 protein inherited from the parental diploid strain is sufficient for ubc9Δ cells to undergo several cell divisions. A depletion of the SUMO protein Smt3 resulted in a similar phenotype. smt3Δ segregants obtained from a SMT3/smt3Δ diploid strain were able to undergo three to five cell divisions on average. Thus, low amounts of Ubc9 and Smt3 are sufficient for viability. The dissection of ubc9Δ and smt3Δ cell microcolonies showed that at least 80% of these cells were large budded, suggesting that they terminally arrested in G2/M phase. These findings underline the essential role of Ubc9 and Smt3 in G2/M phase.

Figure 1.

Yeast cells depleted of Ubc9 and Smt3 mostly arrest in metaphase with high Pds1 levels, but this arrest can be bypassed by a deletion of PDS1.
A. Diploid UBC9/ubc9Δ and diploid SMT3/smt3Δ yeast strains were sporulated, and tetrads were dissected. Spores were grown on YEPD plates at 30°C. Representative segregants containing the ubc9Δ and smt3Δ alleles were photographed after 36 h. Microcolonies from segregants containing the deletion alleles were dissected with a micromanipulator to determine the phenotype of individual cells.
B and C. The following yeast strains were used for these experiments: (i) a strain containing SMT3 expressed from the GAL1-10 promoter instead of the wild-type SMT3 gene (smt3Δ GAL-SMT3, S550); (ii) a strain containing in addition a deletion of the MAD2 gene (smt3Δ mad2Δ, GAL-SMT3, S566); (iii) a strain with a deletion of the UBC9 gene, but kept alive by the UBC9 gene from the GAL1-10 promoter (ubc9Δ GAL-UBC9, S552); and (iv) a strain containing in addition a deletion of the PDS1 gene (ubc9Δ GAL-UBC9 pds1Δ, S483). Strains (i)–(iii) also contained a Myc18-tagged version of Pds1. These strains were pregrown in YEP medium containing 2% raffinose and 2% galactose. Cells were then transferred to YEP medium containing 2% glucose (YEPD) and incubated for 15 h. Subsequently, cells were collected for indirect immunofluorescence microscopy. Large-budded cells were analysed using DAPI, anti-tubulin antibodies and anti-Myc antibodies to visualize nuclei, spindles and Pds1-Myc respectively.
B. Analysis of large-budded cells. Black columns, percentage of cells with short spindles and a single nucleus; grey columns, percentage of cells with elongated spindles and dispersed or barely separated nuclei; white columns, percentage of cells with elongated spindles and distinctly separated nuclei. At least 200 cells from each strain were analysed.
C. DAPI, anti-tubulin and anti-Myc staining of representative cells The arrow marks a cell containing a dispersed nucleus and an elongated spindle.

To characterize further the mitotic arrest of cells defective in SUMOylation, we introduced a Myc-tagged version of Pds1 into a strain in which the SMT3 gene was replaced by a GAL-SMT3 construct (Biggins et al., 2001) and into a ubc9Δ GAL-UBC9 strain. These strains were pregrown in galactose medium and then transferred to glucose medium. After a 15 h incubation period, most of the cells were large budded. Nuclei, spindles and Pds1-Myc were analysed by indirect immunofluorescence. About 70–80% of these large-budded cells had a single nucleus and a short mitotic spindle, and virtually all these cells contained Pds1-Myc (Fig. 1B and C). In contrast, 20–30% of large-budded cells had elongated anaphase spindles, and Pds1-Myc signals were only rarely detectable. Only a few cells had clearly separated masses of DNA. Instead, the DAPI signal was often dispersed, and chromosomes appeared to be lagging along the spindles (Fig. 1C, arrow), indicating that these cells failed to segregate chromosomes to opposite poles properly.

We then asked whether the metaphase arrest of Smt3-depleted cells and the failure to degrade Pds1 may be caused by the activation of the spindle checkpoint (Wassmann and Benezra, 2001). To test this, we constructed a GAL-SMT3 strain lacking MAD2, a gene required for the checkpoint arrest (Li and Murray, 1991). Upon a shift to glucose medium, the GAL-SMT3 mad2Δ strain arrested similarly to the GAL-SMT3 strain (Fig. 1B and C). The percentage of cells containing short spindles and the Pds1-Myc signal was not reduced, implying that the metaphase block of Smt3-depleted cells is independent of Mad2.

These results show that SUMOylation is needed for Pds1 degradation and the onset of anaphase, independently of the spindle checkpoint.

Deletion of PDS1 alleviates anaphase onset in Ubc9-depleted cells

We next addressed the question whether the metaphase arrest of cells defective in SUMOylation may be caused by a failure to degrade Pds1. We argued that a deletion of the PDS1 gene would then allow these cells to elongate their spindles and segregate their chromosomes, as shown previously for other mutants defective in metaphase, such as apc or cdc20 mutants (Yamamoto et al., 1996; Lim et al., 1998). To test this, we crossed pds1Δ mutants with a ubc9Δ GAL-UBC9 strain and received a ubc9Δ GAL-UBC9 pds1Δ strain that was reasonably viable at 25°C. This strain and a ubc9Δ GAL-UBC9 control strain were pregrown at 25°C in galactose medium and then transferred to glucose medium. A large fraction of ubc9Δ GAL-UBC9 pds1Δ cells were large budded after incubation in glucose medium, and these were analysed by immunofluorescence microscopy (Fig. 1B and C). In contrast to ubc9Δ GAL-UBC9 cells containing the PDS1 gene, about 50% of pds1Δ cells had elongated spindles and separated or dispersed DNA masses. DNA masses were only poorly segregating in many cells containing elongated spindles, indicating defects in chromosome segregation.

Thus, a deletion of PDS1 partially suppresses the metaphase arrest of Ubc9-depleted cells, implying that the metaphase block of cells defective in SUMOylation is caused, at least in part, by a failure to degrade Pds1.

Temperature-sensitive ubc9-2 mutants are delayed in degradation of securin Pds1 and cyclin Clb2 during mitosis

To characterize further the role of Ubc9 in mitosis, we used a temperature-sensitive ubc9-2 mutant. The cell cycle arrest of this mutant at the restrictive temperature was less distinct than for Ubc9-depleted cells. After a shift for to 37°C 4 h, both large-budded and unbudded cells accumulated, but the number of small-budded cells was reduced compared with wild-type cultures (data not shown). A similar phenotype was described previously for a smt3-331 temperature-sensitive mutant, which only moderately accumulated G2/M cells at the non-permissive temperature (Biggins et al., 2001). By analysing large-budded cells of ubc9-2 mutants containing a Myc-tagged version of Pds1 by immunofluorescence microscopy, we found that about 70% of these cells contained short spindles, whereas about 30% of cells had elongated spindles (Fig. 2A). Virtually every cell with a short spindle contained a Pds1-Myc signal (data not shown). ubc9-2 mutants containing long spindles frequently failed to segregate chromosomes to opposite poles properly. In a ubc9-2 mad2Δ mutant strain, cells with short spindles and Pds1-Myc accumulated similarly, as in ubc9-2 mutants, implying that the observed mitotic delay occurs independently of the spindle checkpoint. This phenotype of ubc9-2 mutants was similar to that described previously for smt3-331 mutants (Biggins et al., 2001). Thus, both ubc9-2 and smt3-331 mutants are characterized by the accumulation of cells with undivided nuclei and by obstructions in chromosome segregation.

Figure 2.

Temperature-sensitive ubc9-2 mutants are delayed in degradation of Pds1 and Clb2.
A. ubc9-2 (S546) and ubc9-2 mad2Δ (S568) mutants, both containing an Myc18-tagged version of Pds1, were pregrown at 25°C and then shifted to 37°C for 4 h. Then, cells were collected for indirect immunofluorescence microscopy. Large-budded cells were analysed by DAPI, anti-tubulin antibodies and anti-Myc antibodies to visualize nuclei, spindles and Pds1-Myc respectively. Black columns, percentage of cells with short spindles and a single nucleus; grey columns, percentage of cells with elongated spindles and dispersed or barely separated nuclei; white columns, percentage of cells with elongated spindles and distinctly separated nuclei. At least 200 cells from each strain were analysed.
B and C. Log-phase cultures of a wild-type strain (S185) and a ubc9-2 mutant strain (S406), both containing PDS1-HA, were pregrown in YEPD medium at 25°C and then treated with the pheromone α-factor for 2.5 h. Then, cultures were shifted to a restrictive temperature, 36°C, and incubated for an additional 30 min in the presence of α-factor. To remove the pheromone, cells were filtered, washed and transferred to fresh YEPD medium. Synchronized cultures were incubated further at 36°C. At the indicated time points after the release, samples were collected and analysed by immunoblotting, using the HA antibody to detect Pds1-HA and Clb2 antibodies to visualize Clb2 (B). Cdc28 was used as a loading control. Pds1-HA and Clb2 protein levels were quantified using a densitometer (C).

We next monitored the levels of Pds1 and cyclin Clb2 in synchronized wild-type and ubc9-2 cultures. For this purpose, cultures were pregrown at 25°C and then arrested in G1 phase with α-factor. Subsequently, cells were shifted to 36°C and released from the pheromone arrest. Both wild-type and ubc9-2 strains synchronously entered the cell cycle, as monitored by the appearance of small buds. The ubc9-2 mutation affected neither budding nor the initiation of DNA replication (data not shown). Immunoblot analysis revealed that haemagglutinin (HA)-tagged Pds1 and Clb2 accumulated with similar kinetics in wild-type cells and ubc9-2 mutants (Fig. 2B and C). However, ubc9-2 cells were delayed in the degradation of Pds1 and Clb2. At the 75 min time point, Pds1 levels were decreased in wild-type cells, but remained at high levels in the mutant strain, before they dropped with a 15 min delay (Fig. 2B and C). The Clb2 protein levels decreased in the wild-type culture after 90 min, whereas only a slow decrease occurred in ubc9-2 mutants at later time points.

These results show that proteolysis of the two APC/C substrates Pds1 and Clb2 is delayed in ubc9-2 mutants, suggesting that proper proteolysis of these substrates during mitosis requires a functional Ubc9 protein.

Proteolysis of securin Pds1 depends on UBC9 and SMT3

To test more directly whether Ubc9 and Smt3 are required for Pds1 proteolysis, we tested its stability in α-factor-arrested G1 cells. In these G1 cells, APC/C is fully active, and Pds1 is highly unstable (Amon et al., 1994; Cohen-Fix et al., 1996). The stability of Pds1 in cells arrested in G1 phase was determined by promoter shut-off experiments. Wild-type cells, ubc9-2 and smt3-331 mutants were arrested with α-factor at 25°C, and then PDS1-HA was transiently expressed by galactose addition. After a temperature shift to 36°C, PDS1-HA expression was turned off by transferring cells to glucose medium. In wild-type cells, Pds1 was rapidly degraded under these conditions (Fig. 3). In ubc9-2 and smt3-331 mutants, Pds1 proteolysis was inefficient, and its half-life increased to more than 10 min, compared with less than 5 min in wild-type cells. The analysis of the DNA content by fluorescence-activated cell sorting (FACS) analysis confirmed that cultures remained arrested in G1 phase during the course of the experiment. Thus, proteolysis of Pds1 is impaired in the absence of functional Ubc9 and Smt3 proteins, suggesting that SUMOylation is required for efficient securin degradation during G1 phase.

Figure 3.

Proteolysis of securin Pds1 is impaired in yeast ubc9-2 and smt3-331 mutants.
A. A wild-type strain (S206) and a ubc9-2 (S365) mutant strain, both containing bar1 deletions and GAL-PDS1-HA constructs, were pregrown in YEP + raffinose medium at 25°C to log phase. α-factor was added to arrest cells in G1 phase. After 2.5 h incubation with α-factor, 2% galactose was added to induce PDS1-HA expression. After 30 min, cultures were shifted to the restrictive temperature, 36°C, and incubated for another 30 min. To turn off the GAL1-10 promoter, the cultures were filtered, transferred to YEPD medium containing α-factor and incubated at 36°C. Samples were collected at the indicated time points and analysed by immunoblotting. The HA antibody was used to detect Pds1-HA. Cdc28 served as a loading control. Pds1-HA protein levels were quantified using a densitometer. Samples at 0 and 90 min time points were collected for determining the DNA content by FACS analysis to confirm the G1 arrest (right). Microscopic analysis showed that more than 90% of cells were unbudded and displayed a shmoo-like phenotype.
B. A wild-type strain (S206) and a smt3-331 (S535) mutant strain, both containing bar1 deletions and GAL-PDS1-HA constructs, were treated as described in (A). Samples at the 0 and 60 min time points were analysed by FACS (right). Microscopic analysis confirmed that more than 90% of cells were unbudded.

Proteolysis of cyclin and non-cyclin APC/C substrates is impaired in ubc9-2 and smt3-331 mutants

Previous results described a role for Ubc9 in the degradation of cyclins Clb2 and Clb5 (Seufert et al., 1995). These and our results indicate that SUMOylation may generally be required for proteolysis mediated by APC/C. To address these assumptions, we tested whether ubc9-2 and smt3-331 mutants are defective in degradation of cyclin and non-cyclin substrates. Consistent with the defects observed previously with ubc9-1 mutants (Seufert et al., 1995), Clb2 was partially stabilized in smt3-331 mutants (Fig. 4A). To test whether a further mitotic cyclin, Clb3, is stabilized in ubc9-2 and smt3-331 mutants, we performed promoter shut-off experiments. We found that proteolysis of Clb3 was delayed in smt3-331 and ubc9-2 mutants (Fig. 4B and C). The stabilization of Clb2 and Clb3 in G1-arrested cells suggests that Ubc9 and Smt3 are required for efficient proteolysis of mitotic cyclins.

Figure 4.

Efficient proteolysis of APC/C substrates is impaired in ubc9-2 and smt3-331 mutants.
A. A wild-type strain (S057) and a smt3-331 mutant strain (S543), both containing bar1 deletions and the GAL-CLB2-HA construct, were pregrown in YEP medium containing 2% raffinose at 25°C. α-factor was added to arrest cells in G1 phase. After 3 h incubation with α-factor, 2% galactose was added to induce CLB2-HA expression, and then cultures were incubated for 30 min. Then, they were shifted to the restrictive temperature, 36°C, for another 30 min. To turn off the GAL1-10 promoter, the cultures were filtered, transferred to YEPD medium containing α-factor and incubated at 36°C. Microscopic analysis confirmed that at least 90% of cells were unbudded and displayed a shmoo-like phenotype, implying that these cells were arrested in G1 phase (not shown). Samples were collected at the indicated time points and analysed by immunoblotting. The HA antibody was used to detect Clb2-HA. Cdc28 served as a loading control.
B. A wild-type strain (S056) and a smt3-331 mutant strain (S544), both containing bar1 deletions and GAL-CLB3-HA constructs, were treated as described in (A). The HA antibody was used to detect Clb3-HA.
C. A wild-type strain (S056) and a ubc9-2 mutant strain (S390), both containing bar1 deletions and GAL-CLB3-HA constructs, were treated as described in (A). The HA antibody was used to detect Clb3-HA.
D. A wild-type strain (S088) and a ubc9-2 mutant strain (S487), both containing bar1 deletions and GAL-CDC5-HA constructs, were pregrown at 25°C overnight in YEP + Raf medium. α-factor was added to arrest cells in G1 phase. After 2.5 h, galactose (2%) was added to induce CDC5-HA expression. At the same time, cells were shifted to 36°C and incubated for 120 min in the presence of α-factor. Microscopic analysis confirmed that at least 90% of cells were unbudded. Samples were collected at 0, 60 and 120 min and analysed by immunoblotting. Cyc, sample of cycling cultures at the 120 min time point. Cdc5-HA levels were analysed by immunoblotting using the HA antibody.

To test the requirement of Ubc9 for degradation of another non-cyclin APC/C substrate, we analysed the accumulation of the polo-like kinase Cdc5 in G1-arrested cells. Upon expression of CDC5-HA from the GAL1-10 promoter, Cdc5 accumulates only to low levels in wild-type G1 cells, because of its instability (Shirayama et al., 1998). We found that Cdc5 accumulates to higher levels in ubc9-2 mutants than in wild-type cells at the restrictive temperature, indicating that its rapid degradation is impaired (Fig. 4D).

In summary, these results indicate that Ubc9 and Smt3 are required for efficient proteolysis of various APC/C substrates, implying that SUMOylation is generally important for the proper function of this ubiquitin ligase.

To elucidate further the involvement of Ubc9 in APC/C-mediated proteolysis, we tested whether the ubc9-2 mutation displays genetic interactions with mutations in the APC/C subunit genes CDC16 and APC10. Double mutants containing the ubc9-2 mutation in combination with either the cdc16-123 or apc10-22 mutations were constructed. Both ubc9-2 apc10-22 and ubc9-2 cdc16-123 were non-viable at 30°C, whereas each of the single mutants was viable (data not shown). These synthetic phenotypes reveal at least moderate genetic interactions between UBC9 and genes encoding APC/C subunits.

Other unstable proteins are normally degraded in ubc9-2 and smt3-331 mutant strains

It may be possible that SUMOylation does not specifically affect APC/C-mediated proteolysis, but may instead be needed for proteolysis in general, for example for the proper function of the 26S proteasome. We therefore tested whether ubc9-2 and smt3-331 mutants are generally impaired in the rapid degradation of unstable proteins. To test this possibility, we compared the stability of the transcription activator Gcn4 in wild-type cells and in ubc9-2 and smt3-331 mutants. Gcn4 proteolysis is not dependent on APC/C, but is instead a substrate of the SCFCdc4 ubiquitin ligase (Meimoun et al., 2000). Wild-type and mutant strains were transformed with a centromeric plasmid containing a GAL-GCN4-MYC construct. GCN4-MYC was transiently expressed by galactose addition. The temperature was shifted to 36°C, and Gcn4 stability was determined by transferring cells to glucose medium. We found that Gcn4 degradation occurred with a similar efficiency in wild-type cells and in ubc9-2 and smt3-331 mutants (Fig. 5A). Ime2 is a further unstable protein, the proteolysis of which appears to be independent of the SCF and APC/C ubiquitin ligases (Bolte et al., 2002). To test whether Ime2 is stabilized in ubc9-2 mutants, IME2-HA was transiently expressed in wild-type and mutant cells. We found that Ime2 is similarly degraded in both strains at the restrictive temperature (Fig. 5B).

Figure 5.

ubc9-2 and smt3-331 mutants are not generally defective in proteolysis of unstable proteins.
A. A wild-type yeast strain (S001) and ubc9-2 (S099) and smt3-331 (S542) mutant strains were transformed with a centromeric plasmid containing GAL-GCN4-MYC. Transformants were pregrown at 25°C in minimal medium lacking uracil and containing 2% raffinose. Galactose was added to induce GCN4-MYC expression, and cells were incubated for 30 min. Then, cultures were shifted to a restrictive temperature, 36°C, and incubated for another 30 min. To turn off the GAL1-10 promoter, cultures were filtered, transferred to minimal medium containing 2% glucose and incubated at 36°C. Samples were collected at the indicated time points and analysed by immunoblotting, using the MYC antibody to detect Gcn4-Myc. Cdc28 served as a loading control. Protein levels were quantified using a densitometer.
B. A wild-type yeast strain (S396) and a ubc9-2 mutant strain (S534), both containing a GAL-IME2-HA construct, were pregrown in YEP medium containing 2% raffinose at 25°C. Galactose (2%) was added to induce IME2-HA expression. After 30 min, cultures were shifted to 36°C and incubated for another 30 min. To turn off the GAL1-10 promoter, the cultures were filtered, transferred to YEPD medium and incubated at 36°C. Samples were collected at the indicated time points and analysed by immunoblotting. The HA antibody was used to detect Ime2-HA. Cdc28 served as a loading control. Protein levels were quantified using a densitometer.

The undisturbed degradation of Gcn4 and Ime2 in these mutants suggests that Ubc9 and Smt3 are not generally required for proteolysis of unstable proteins.

APC/C and Pds1 are localized to the nucleus in ubc9-2 and smt3-331 mutant strains

Previous findings have implicated SUMO modification in the subcellular localization of proteins (Wilson and Rangasamy, 2001; Kim et al., 2002). APC/C is mainly localized to the nucleus (Tugendreich et al., 1995) and, most probably, this ubiquitin ligase is highly active predominantly in this cellular compartment. We asked whether impaired APC/C-mediated proteolysis in mutants defective in SUMOylation may primarily be caused by the mislocalization of the APC/C core complex, regulatory proteins or substrates.

To test this, we analysed the subcellular localization of the core subunit Cdc16, the regulatory protein Cdc20 and of Pds1 in ubc9-2 and smt3-331 mutants. Cells were shifted to the restrictive temperature and, after 4 h, samples were collected for indirect immunofluorescence. We found that tagged versions of Cdc16, Cdc20 and Pds1 were localized mainly in the nucleus in ubc9-2 and smt3-331 mutants, as found for wild-type cells (Fig. 6). Similar data were obtained with Clb2 (data not shown). Thus, ubc9-2 and smt3-331 mutants apparently do not affect the nuclear localization of either of these proteins.

Figure 6.

The localization of APC/C and substrates is similar in wild-type cells and ubc9-2 and smt3-331 mutants. A wild-type strain and ubc9-2 and smt3-331 mutant strains containing MYC-tagged versions of the APC/C subunit Cdc16 (Cdc16-Myc), the activator protein Cdc20 (Myc-Cdc20) or securin Pds1 (Pds1-Myc) were pregrown in YEPD medium at 25°C. Cultures were shifted to 36°C for 4 h. Then, cells were fixed, spheroplasted and analysed by indirect immunofluorescence using the MYC antibody. DAPI was used to visualize nuclei.

Discussion

A role for SUMO in the onset of anaphase and in chromosome segregation

In this report, we have analysed the roles of the budding yeast SUMO protein Smt3 and of the SUMO-conjugating enzyme Ubc9 in cell cycle progression and in proteolysis mediated by the anaphase-promoting complex/cyclososme (APC/C). Consistent with previous reports (al-Khodairy et al., 1995; Seufert et al., 1995; Tanaka et al., 1999; Biggins et al., 2001), our data showed that Smt3 and Ubc9 have pivotal functions during mitosis. A large proportion of budding yeast cells depleted of either Ubc9 or Smt3 accumulated as large-budded cells that contained short mitotic spindles, undivided nuclei and high levels of securin Pds1. These findings imply that SUMO modification is an essential process for the metaphase/anaphase transition.

The onset of anaphase requires proteolytic destruction of securins, which act as inhibitors of separases (Nasmyth, 2002). As Ubc9- or Smt3-depleted cells arrest with stable Pds1 protein, we addressed the question whether the metaphase arrest of cells impaired in SUMOylation is caused by a failure to degrade Pds1. We found that a deletion of PDS1 allowed a significant fraction of ubc9Δ cells to elongate their spindles and to start to segregate their chromosomes, suggesting that Pds1 proteolysis is indeed an important function of SUMO at the metaphase/anaphase transition. As a fraction of ubc9Δ pds1Δ cells still contained short spindles and a single mass of DNA, it is likely that SUMO has important roles other than securin proteolysis for the onset of anaphase. In contrast to the PDS1 deletion, cells impaired in SUMOylation were not affected by a deletion of the spindle checkpoint gene MAD2, suggesting that the mitotic delay is not a consequence of the checkpoint activation.

We have presented evidence that, in addition to the onset of anaphase, SUMOylation is also required for proper chromosome segregation. A fraction of Ubc9- or Smt3-depleted cells had elongated spindles, and these mostly contained DNA masses dispersed along the elongating spindle. Temperature-sensitive ubc9-2 and smt3-331 mutants displayed similar defects. Interestingly, the temperature-sensitive smt3-331 mutant was initially identified in a screen for mutants defective in chromosome segregation (Biggins et al., 2001). In Schizosaccharomyces pombe, cells lacking either the SUMO gene pmt3 or the UBC9 homologue hus5 were found to have abortive mitosis, and it was suggested that this resulted from defects in chromosome segregation (al-Khodairy et al., 1995; Tanaka et al., 1999). In conclusion, these and our data imply that SUMOylation plays an important role in sister chromatid separation and in chromosome segregation.

A role for SUMO in APC/C-mediated proteolysis

We have demonstrated that Ubc9 and Smt3 are needed for efficient securin proteolysis in cells arrested in G1 phase, a period of the cell cycle in which APC/C is highly active (Amon et al., 1994). Similar to securin degradation, proteolysis of mitotic cyclins was also delayed in ubc9-2 and smt3-331 mutants. These data are consistent with the previously demonstrated defect of ubc9-1 mutants in the degradation of Clb2 and Clb5 (Seufert et al., 1995). The findings that APC/C-mediated proteolysis is perturbed in these mutants implies that SUMOylation is required for efficient function of the APC/C.

In contrast to mutants in APC/C subunit genes, we found that ubc9-2 and smt3-331 mutants display only partial defects in Pds1 degradation during mitosis of synchronous cultures and in G1-arrested cells. It is possible that SUMO modification plays only a minor role in proper APC/C function, for example for the fine tuning of its activity. Alternatively, a residual activity of SUMOylation in ubc9-2 and smt3-331 mutants at restrictive temperatures may be sufficient for degradation of APC/C substrates, at least with reduced kinetics. Consistent with this hypothesis, we found that low levels of Ubc9 and Smt3 are sufficient for the viability of yeast cells (Fig. 1A).

How may SUMO affect APC/C-mediated proteolysis? We have provided evidence that Ubc9 and Smt3 are not needed for proteolysis in general. Unstable proteins that are not APC/C substrates, such as the SCFCdc4 substrate Gcn4, were normally degraded in ubc9-2 and smt3-331 mutants, suggesting that the 26S proteasome is functional. These data favour the model in which specifically APC/C function may be affected in these mutants.

At present, it is unknown how SUMO modification influences this ubiquitin ligase. Previously, proteins with important roles in mitosis were found to be modified with SUMO. One of these is Top2, DNA topoisomerase II. It was shown that modification of Top2 is required for the control of chromosome cohesion at centromeric regions (Bachant et al., 2002). SUMO-modified Top2 was unable to promote centromeric cohesions, and this may indicate a regulatory role for SUMO in dissolving the cohesion between sister chromatids. However, there is no evidence that Top2 modification may affect securin stability or APC/C activity. Other recently identified yeast substrates known to be modified by SUMO are the septins Cdc3, Cdc11 and Sep7, which form a ring at the yeast bud neck (Johnson and Blobel, 1999; Takahashi et al., 1999). As yeast strains expressing septins lacking SUMO conjugation sites did not display defects in cell cycle progression, it is rather unlikely that a failure to modify these septins contributes to the mitotic defects of Ubc9- and Smt3-depleted cells. It remains to be shown whether factors directly involved in APC/C-mediated proteolysis are modified by SUMO. Using an HA-tagged version of SMT3, we did not detect modifications of immunoprecipitated APC/C subunits Cdc16 and Cdc23 or of the substrates Pds1 and Clb2 (data not shown).

Various different functions for SUMO have been identified in the last few years, and it is tempting to speculate that one of these is also required for promoting APC/C-mediated proteolysis. For example, SUMO has been shown to modulate nucleocytoplasmic transport, subcellular localization or protein–protein interactions (Müller et al., 2001; Wilson and Rangasamy, 2001). As functional nucleocytoplasmic transport is needed for proteolytic destruction of various APC/C substrates (Loeb et al., 1995; Bäumer et al., 2000; Hildebrandt and Hoyt, 2001), we tested whether the nuclear localization of APC/C or its substrates is defective when SUMOylation is disturbed. We found that the APC/C subunit Cdc16, the activator protein Cdc20 and the substrates Pds1 were localized in the nucleus in ubc9-2 and smt3-331 mutants similarly to wild-type strains, indicating that nuclear import of these factors is not distinctly affected. Furthermore, it is conceivable that defects in nucleocytoplasmic transport would also interfere with Gcn4 proteolysis in these mutants. SCFCdc4 activity is restricted to the nucleus (Blondel et al., 2000), and nuclear localization of Gcn4 is required for its efficient degradation (Pries et al., 2002). We suggest that SUMOylation rather affects APC/C inside the nucleus.

A possible function of SUMO may be the proper localization of APC/C or its substrates to specific subcellular elements. APC/C is known to be associated with centrosomes and mitotic spindles (Tugendreich et al., 1995). Substrates such as securins and cyclins were also localized to spindles (Jensen et al., 2001; Raff et al., 2002), and it is conceivable that the association of APC/C with the mitotic spindle is important for the destruction of spindle-associated substrates. As SUMOylation has been implicated in regulating protein–protein interactions, a further possibility could be that this modification is critical for the association of APC/C with regulatory proteins or for substrate recognition.

Proteomics may be a promising approach to reveal the SUMO targets essential for mitosis and APC/C function. Then, it will be interesting to elucidate whether SUMOylation is a novel mechanism for the regulation of this ubiquitin ligase.

Experimental procedures

Yeast strains and plasmids

All strains used in this study are derivatives of the Saccharomyces cerevisiae W303 strain and are listed in Table 1. The ubc9 deletion strain containing the temperature-sensitive ubc9-2 allele was created by C. Michaelis. Initially, a ubc9 deletion strain (ubc9Δ) was constructed by transformation of a deletion cassette containing the TRP1 gene inserted between a 500 bp fragment containing sequences immediately upstream of the UBC9 open reading frame (ORF) start codon and a 280 bp fragment containing sequences of the 3′ region. The temperature-sensitive ubc9-2 allele was created by polymerase chain reaction (PCR) mutagenesis of the UBC9 gene. Plasmids from colonies growing at 25°C, but not at 37°C were isolated and cloned into the integrative plasmid YIplac128 (containing the LEU2 marker).

Table 1.  Yeast strains used in this study.
StrainRelevant genotype
S001W303-1A wild-type strain
MATa, ade2-1, trp1-1, can1-100, leu2-3, 112, his3-11, 15, ura3, GAL
S056MATa, GAL-CLB3-HA3/URA3, bar1::HisG
S057MATa, GAL-CLB2-HA3/URA3, bar1::HisG
S088MATa, GAL-CDC5-HA3/URA3, bar1::HisG
S099MATα, ubc9::TRP1, ubc9-2/ΛΕΥ2
S185MATa, PDS1-HA3/URA3
S206MATa, GAL-PDS1-HA/URA3, bar1::HisG
S234MATa, CDC16-MYC6/URA3
S365MATa, ubc9::TRP1, ubc9-2/LEU2, GAL-PDS1-HA3/URA3, bar1::HisG
S390MATa, ubc9::TRP1, ubc9-2/LEU2, Gal-CLB3-HA3/URA3, bar1::HisG
S396MATa, GAL-IME2-HA3/TRP1
S406MATa, ubc9::TRP1, ubc9-2/LEU2, PDS1-HA3/URA3
S425MATa, ubc9::TRP1, pRS316-GAL-UBC9/URA3
S481MATa, MYC18-CDC20/TRP1
S483MATa, ubc9::TRP1, pRS316-GAL-UBC9/URA3, pds1::LEU2
S487MATa, ubc9::TRP1, ubc9-2/LEU2, GAL-CDC5-HA3/URA3, bar1::HisG
S527MATa, ubc9::TRP1, ubc9-2/LEU2, MYC18-CDC20/TRP1
S528MATa, ubc9::TRP1, ubc9-2/LEU2, CDC16-MYC6/URA3
S533MATa, GAL-HA3-SMT3/HIS3
S534MATa, ubc9::TRP1, ubc9-2/LEU2, GAL-IME2-HA3/TRP1
S535MATa, smt3-331, GAL-PDS1-HA3/URA3, bar1::HisG
S537MATa, PDS1-MYC18/LEU2
S542MATα, smt3-331
S543MATa, smt3-331, GAL-CLB2-HA3/URA3, bar1::HisG
S544MATa, smt3-331, GAL-CLB3-HA3/URA3, bar1::HisG
S545MATa, smt3-331, PDS1-MYC18/LEU2
S546MATa, ubc9::TRP1, ubc9-2/LEU2, PDS1-MYC18/LEU2
S547MATa, smt3-331, MYC18-CDC20/TRP1
S548MATa, smt3-331, CDC16-MYC6/URA3
S550MATα, GAL-HA3-SMT3/HIS3, PDS1-MYC18/LEU2
S552MATa, ubc9::TRP1, pRS316-GAL-UBC9/URA3, PDS1-MYC18/LEU2
S566MATa, GAL-HA3-SMT3/HIS3, mad2::URA3, PDS1-MYC18/LEU2
S568MATα, ubc9::TRP1, ubc9-2/LEU2, mad2::URA3, PDS1-MYC18/LEU2

ubc9 deletions strains kept alive by the GAL-UBC9 gene fusion were created by transformation of a UBC9/ubc9Δ heterozygous diploid with a centromeric plasmid containing GAL-UBC9 and URA3 as selection marker. ubc9Δ GAL-UBC9 segregants were obtained by sporulation and tetrad dissection. The GAL-UBC9 construct was isolated from a GAL-cDNA library (Liu et al., 1992) during a genetic screening. It contains the entire UBC9 ORF fused 24 bp upstream of the ATG start codon to the GAL1-10 promoter. The GAL-HA3-SMT3 strain, containing a replacement of the endogenous SMT3 promoter by the GAL1-10 promoter, and the smt3-331 mutant strain were received from S. Biggins and have been described previously (Biggins et al., 2001).

Growth conditions and cell cycle arrests

Cells were grown in YEP medium (2% bactopeptone, 1% yeast extract, 0.005% adenine sulphate) supplemented with either 2% glucose (YEPD) or 2% raffinose. Plasmid-carrying strains were grown in appropriate synthetic minimal medium containing 0.8% yeast nitrogen base, supplemented with amino acids and adenine. The GAL1-10 promoter was induced by the addition of galactose (2% final concentration). Before gene expression from the GAL1-10 promoter was induced, cells were grown in raffinose medium. To turn off the GAL1-10 promoter, cells were filtered and resuspended in medium containing 2% glucose. To arrest cells in G1 with α-factor pheromone, cultures were incubated for 2.5 h in the presence of 5 µg ml−1α-factor.

Immunoblotting

Preparation of yeast cell extracts and protein immunoblot analysis were performed as described previously (Surana et al., 1993). The enhanced chemiluminescence detection system was used. Antibodies were used in 1:1000 (Clb2), 1:2000 (Cdc28), 1:100 (HA, 12CA5) and 1:100 (MYC) dilutions.

Immunofluorescence and FACS analysis

For indirect immunofluorescence, cells were fixed in 3.7% formaldehyde, and spheroblasts were prepared as described previously (Pringle et al., 1991). Staining with 4,6-diamidino-2-phenylindole (DAPI) and anti-tubulin antibodies were used for visualization of nuclei and spindles respectively. The DNA content was determined by FACS analysis as described previously (Epstein and Cross, 1992).

Acknowledgements

We thank Christine Michaelis, Kim Nasmyth, Sue Biggins and Yoshiko Kikuchi for providing yeast strains. We are grateful to Christine Michaelis for comments on the manuscript, and Ingrid Bahr for help with the figures. This work was supported by the Deutsche Forschungsgemeinschaft (grant IR 36/1-3), the Fonds der Chemischen Industrie and the Volkswagen-Stiftung.

Ancillary