INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Proteolytic degradation plays an important role in cell cycle progression. The initiation of S phase, the onset of anaphase, and the final exit from mitosis are three cell cycle transitions for which involvement of proteolysis has been studied in some detail. Ubiquitin-dependent destruction of the Cdk inhibitor Sic1 is essential for the initiation of S phase. The ubiquitination of phospho-Sic1 is catalyzed by SCF^Cdc4, a ubiquitin-ligase (E3) complex, comprising Skp1, Cdc53 (or cullin), and F-box-containing Cdc4 (12, 27, 39, 44, 48). These subunits appear to have distinct functions: Cdc34 acts as an E2 enzyme; Cdc53 is a scaffold protein for SCF (34, 51); and Cdc4, a WD40 repeat-containing protein, is responsible for docking specific substrates into this E3 complex (12, 44). Cdc4 is also required for Cdc6 proteolysis in late G₁, S, and M phases (11, 35). The target specificity of SCF appears to be determined by the F-box-containing adapter proteins such as Cdc4, Grr1, and Met30. The SCF complex which contains Grr1 instead of Cdc4 targets G₁ cyclins Cln1 and Cln2 (3) and the Cdc42 effector Gic2 (18), whereas the complex with Met30 ubiquitinates the tyrosine kinase Swe1 (20).

Proteolysis is also essential for the initiation of nuclear division, which requires another E3 enzyme called anaphase-promoting complex (APC) (reviewed in reference 15), also known as cyclosome (46). This is supported by the observation that temperature-sensitive (ts) mutants carrying mutations in three genes encoding APC components, CDC16, CDC23, and CDC27 (17, 23, 24, 47, 53), are arrested prior to nuclear division (14). The APC targets a number of proteins for proteolytic degradation in mitosis and G₁: the anaphase inhibitors Pds1 (6) and Scc1 (29), mitotic cyclins like Clb2 (16, 17), and mitotic spindle-associated protein Ase1 (19). However, it is not clear what activates or regulates the initiation of their proteolysis at specific stages of mitosis.

Cdc20, which belongs to a family of WD40 (also known as -transducin) proteins (31), has been implicated in the regulation of proteolysis and is essential for the chromosome segregation at the onset of anaphase. The cdc20-1 mutant is arrested with duplicated DNA, a short spindle, and an undivided nucleus, a phenotype very similar to that of cdc16, cdc23, and cdc27 mutants (14, 26, 40). For some time, CDC20 was thought to be involved in the regulation of microtubule depolymerization because the cdc20-1 ts mutant accumulates thickened spindles at the restrictive temperature (33, 40). Recent evidence, however, suggests that CDC20 function may not be required for spindle elongation (26); instead, it is essential for the regulation of the APC-dependent proteolysis (26, 49). Cdc20-related proteins have now been identified in other organisms: fizzy and fizzy-related in Drosophila melanogaster, p55CDC in mammalian cells (50), and slp1⁺ in fission yeast (28). fizzy is required for the degradation of both cyclins A and B during mitosis (9, 42), while fizzy-related is involved in down-regulating mitotic cyclins in interphase (43). Hct1/Cdh1, the Cdc20 homolog in budding yeast, has been implicated in Clb2 degradation in G₁, but it is not essential for viability (38, 49).

To identify genes that functionally interact with CDC20, we conducted a second-site suppressor screen and identified three alleles of CDC4 which alleviate the nuclear division defect of the cdc20-1 mutant. We find that the previously characterized cdc4-1 mutant, the cdc4 mutant, and additional cdc4 ts mutants, isolated by PCR mutagenesis, all exhibit both G₁/S and G₂/M defects. The inability of the new cdc4 mutants to properly initiate anaphase is not due to the accumulation of Sic1, resulting from incomplete degradation during G₁/S transition. Instead, the G₂/M arrest in one of the newly isolated cdc4 mutants is relieved by the deletion of the PDS1 gene, allowing the mutant to undergo anaphase. Thus, we have uncovered an essential function of CDC4 during G₂/M transition, which may be linked to the APC activity required for the onset of anaphase.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Yeast media and reagents. All strains used in this study were haploid (unless otherwise stated) and were derived from the wild-type strain W303. Cells were grown in standard yeast extract-peptone or selective medium supplemented with 2% glucose, raffinose, or raffinose-galactose. To obtain synchronized cultures in G₁ or early S phase, cells were treated with -factor (1 µg/ml for bar1 strains) or hydroxyurea (HU; 15 mg/ml), respectively.

Strains and plasmids. The detailed genotypes of various strains used in this study are shown in Table 1. CDC4 was cloned by transforming a low-copy plasmid library into the cdc4-10 mutant and isolating plasmids that rescued the ts phenotype at 37°C. The shortest DNA fragment from a plasmid which rescued both cdc4-10 and cdc4-1 contained the full-length CDC4 gene. A fragment of about 3.2 kb, consisting of an 0.75-kb 5' untranslated region, a 2.3-kb coding sequence, and an 0.2-kb 3' untranslated region, was cloned into a CEN plasmid to obtain pUS456.

Isolation of ts cdc4 mutants. Low-fidelity PCR was used to mutagenize the WD40 region with primers USPY25 (5'GGGCTGATGACAAAATGATCA3') and USPY26 (5'CCGTAGATTATAGATGTTGAA3') in four reactions, each with one of the deoxynucleoside triphosphates reduced in a 1:5:5:5 ratio. The PCR products of all four reactions were pooled. The gap plasmid was obtained by dropping out a 0.85-kb fragment between NcoI and XbaI from pUS456. The PCR products and the gap plasmid overlap by about 100 bp at either end. The pooled PCR products and the gap plasmid were cotransformed into US893 and plated on Leu Gal⁺ plates at a density of about 1,000 colonies per plate. The transformants were then replica plated on Leu Glu⁺ plates at 23 and 37°C. Plasmids from the ts candidates were isolated and retransformed into US893 to confirm that they carried a ts cdc4 allele. Approximately 31,000 clones were screened, and 20 ts mutants were isolated. The ts alleles were maintained on a CEN plasmid in the cdc4 strain. Integrating cdc4-12 into the genome caused it to be predominantly defective in G₁/S. It is unclear why this was so; the G₂/M phenotype could be exhibited only when the cells carried several copies of the mutant gene. cdc4-12, cdc4-14, and cdc4-15 were not dominant because plasmids carrying these alleles, when transformed into wild-type cells, did not give rise to a ts phenotype.

Northern blot hybridization, flow cytometry, and immunofluorescence. RNA extraction was done according to the method of Cross and Tinkelenberg (8), and Northern blot analysis was performed as described by Price et al. (36). Cells were prepared for FACScan analysis (flow cytometry) as in the work of Lim et al. (25). Cells were prepared for immunofluorescence as described in the work of Kilmartin and Adams (22), and microtubules and DNA were visualized with antitubulin monoclonal antibody YOL1/34 (gift from J. V. Kilmartin) and diamidinophenylindole, respectively.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Mutations in CDC4 suppress the nuclear division defect in the cdc20-1 mutant. To identify genes that functionally interact with CDC20, we screened for second-site suppressors of the ts cdc20-1 mutant. Since cdc20-1 cells lose viability rapidly at 37°C, the screens were performed at 35°C, the temperature at which these cells still are arrested prior to nuclear division but remain viable. Two suppressors were obtained from a spontaneous revertant screen, and an additional one was identified in a genetic screen in which cells were mutagenized with ethyl methanesulfonate. In all three cases, the suppression was due to a single, recessive mutation. When segregated away from the cdc20-1 mutation, the suppressor mutations were found to render the segregants for growth at 37°C. To clone the corresponding genes, the mutants were transformed with a genomic library on a CEN vector and the plasmids were retrieved from the clones that grew well at 37°C. The complementing plasmids from all three suppressor mutants contained the CDC4 gene, suggesting that the suppressor mutations were in CDC4. This was supported by the observation that the suppressor mutations failed to complement the canonical cdc4-1 allele at 37°C when tested in a heteroallelic combination in a diploid. To further confirm that the suppressor mutations were indeed in the CDC4 gene, the wild-type CDC4 gene was first cloned into a URA3-selectable integrative vector. The integration of the entire plasmid was targeted to the CDC4 locus in the strain carrying one of the ts suppressor mutations (later named cdc4-10). Southern blot analysis confirmed that the integration had occurred at the CDC4 locus (data not shown). The resultant integrant was crossed to a wild-type strain; the diploid was allowed to sporulate, and in all, 21 tetrads were dissected. As expected, the URA3 marker showed a 2:2 segregation pattern in all tetrads. All segregants in 20 tetrads grew well at 37°C; the remaining tetrad showed an inexplicably abnormal segregation pattern (data not shown). The absence of any ts segregant in this cross, together with the complementation studies described above, suggest that the suppressor mutations are most likely at the CDC4 locus. Therefore, these mutations were named cdc4-10 (Fig. 1), cdc4-11, and cdc4-16. cdc4-10 was used for further studies.

View larger version (63K): [in this window] [in a new window]   FIG. 1.   cdc4-10 suppresses the growth defect of the cdc20-1 mutation. cdc4-10 is one of the three cdc4 alleles isolated as second-site suppressors of the cdc20-1 ts mutation. Various strains were streaked on yeast extract-peptone-dextrose plates, incubated for 2 (35°C) or 3 (24°C) days, and then photographed. The viability of the cdc20-1 strain at 35°C is enhanced by the presence of cdc4-10 but not by that of cdc4-1, cdc4-12, or cdc34-2. The cdc4-10 and cdc34-2 mutants are ts at 37°C but can grow at 35°C.

View larger version (55K): [in this window] [in a new window]   FIG. 2.   The cdc4-10 strain is defective in G1/S transition at 37°C. Exponentially growing cultures of cdc4-10 and cdc4-1 cells were shifted to 37°C for 4 h. The cdc4-10 strain is arrested like the cdc4-1 strain at G1/S with an elongated (and often deformed) bud and a single nucleus prior to DNA synthesis and bipolar spindle formation. Differential interference contrast (DIC) micrographs of undigested cells in a different field show the bud morphology. Antitubulin antibodies (YOL1/34; -tub) stain a single aster of microtubules in a single nucleus stained with diamidinophenylindole (DAPI). FACScan analysis shows that the cdc4-1 strain is arrested with 1N DNA content while the DNA profile for the cdc4-10 strain contains a smaller 2N peak because it is slightly leaky at 37°C. Bar, 1 µm.

The suppression activity of cdc4-10 is independent of Sic1 accumulation. It has been previously reported that the cdc20-1 mutation can be suppressed by overexpression of SIC1 (38). Moreover, the inhibition of the mitotic kinase by Sic1 can lead to the activation of cyclosome-mediated proteolysis (1). Therefore, it is possible that the suppression of the cdc20-1 mutation by CDC4 alleles isolated in this study is a result of Sic1 accumulation due to a defect in Cdc4-mediated proteolytic degradation in G₁. To test this possibility, we used a version of the SIC1 gene that had been tagged with three tandem copies of a sequence encoding the c-myc epitope. That the epitope-tagged SIC1 gene was functional was indicated by its ability to allow sic1 cells to grow as well as the wild-type cells (Fig. 3B). The cdc4-1, cdc4-10, cdc4-12, and cdc34-2 mutants and the wild-type cells carrying the native promoter-driven SIC1-myc3 gene (on a CEN plasmid) were released from HU arrest (a stage beyond the G₁/S execution point of Cdc4) at 35°C. The cell cycle progression and Sic1 protein levels were monitored upon release from HU arrest (Fig. 3A). The wild-type and cdc4-10 cells traverse through S phase and mitosis at about the same rate. The degradation kinetics of the Sic1 protein are also quite similar in these two strains. Upon release, Sic1 is degraded at 60 min, coinciding with the completion of S phase, and increases again as cells undergo mitosis (10). However, in cdc4-1, cdc4-12, and cdc34-2 mutants, the Sic1 protein is not efficiently degraded upon release from HU arrest (Fig. 3A). This is expected, since Cdc4 and Cdc34 are required for the ubiquitination and hence for the proteolysis of Sic1 (12, 44). In all five strains, the small amount of Sic1 present in HU block could be due to the small percentage of cells that have not initiated S phase, possibly represented by the cells that have not formed a visible bipolar spindle (about 20%).

View larger version (135K): [in this window] [in a new window]   FIG. 3.   Suppression of cdc20-1 is not due to Sic1 protein accumulation. (A) Sic1 protein levels in the wild-type strain and the cdc4-1, cdc4-10, cdc4-12, and cdc34-2 mutants. The cells transformed with a plasmid carrying SIC1-myc3 were synchronized by HU treatment for 3 h at 24°C, shifted to 35°C for 1 h, and then released at 35°C. The DNA content, spindle morphology and Sic1 protein levels were analyzed. The progression through the cell cycle and the Sic1-degradation kinetics are very similar in the wild type and in cdc4-10 cells. cdc4-1 and cdc4-12 mutants are slower in transiting from G2/M to G1 and are arrested with a mixture of G1/S and G2/M phenotypes, while the cdc34-2 strain does not show a G2/M defect and is arrested quite uniformly in G1/S. cdc4-1, cdc4-12, and cdc34-2 strains are defective in Sic1 degradation upon release from HU. For spindle morphology, cells were classified into three categories: those with single asters (no spindle), those with short spindles (undivided nucleus), and those with anaphase spindles (spindles extended between the two well-separated nuclei). Numbers to the right of each FACScan graph and above the lanes of each gel are times (in minutes). (B) The SIC1-myc3 construct is functional in vivo. Equal numbers of cells from the sic1 strain, the sic1 strain carrying the native promoter-driven SIC1-myc3 (on a CEN vector), and the wild-type (wt) strains were plated on glucose plates and incubated at 24°C for 1 day. The photomicrographs show a representative section from each plate. SIC1-myc3 restores the growth of sic1 cells to the wild-type level. (C) The cdc4-10 cdc20-1 sic1 triple mutant grows better than the cdc20 sic1 double mutant at 30°C, indicating that the suppression is not due to accumulation of the Sic1 protein. A deletion of the SIC1 gene results in generally poor growth so that cdc4-10 cdc20-1 sic1 and cdc20-1 sic1 mutants are inviable at 35°C. cdc4-10 does not suppress sic1, as sic1 and cdc4-10 sic1 mutants both grow at about the same rate.

sic1 cdc4-10 cdc20-1

sic1 cdc20-1

sic1 cdc4-10 cdc20-1

sic1 cdc20-1

CDC4 serves a function in G₂/M. The ability of CDC4 alleles to suppress the defect of the cdc20-1 mutant was unexpected, since CDC4 is known to have a function only in the G₁/S transition (13, 39) while CDC20 serves a function only in G₂/M. That CDC4 may have a second role in addition to its G₁/S function was first indicated by our observation that about 10% of cdc4-1 cells are arrested with a short mitotic spindle when an asynchronous culture is shifted to 37°C. To demonstrate the G₂/M phenotype of the cdc4-1 strain more clearly, cdc4-1 cells were first synchronized in early S phase by HU treatment for 3 h before they were allowed to resume cell cycle progression at 37°C. In comparison to the wild-type strain, cdc4-1 cells undergo S phase about 15 min later but are delayed in G₂/M transition from 60 to 150 min after release. At 60 min after the release, a large proportion of cdc4-1 cells had a short mitotic spindle (about 70% [Fig. 4]) while the wild-type cells had already undergone nuclear division. At 150 min, 50% of mutant cells were arrested with a short preanaphase spindle while the remainder had progressed to the next cycle and were arrested at the G₁/S transition. Wild-type cells at this point were dividing asynchronously (Fig. 4). Approximately 40% of cdc4-1 cells still remained arrested with a short spindle at the end of the experiment.

View larger version (41K): [in this window] [in a new window]   FIG. 4.   The cdc4-1 mutant exhibits a delay in G2 when released from HU-induced arrest. cdc4-1 and wild-type cells were arrested by HU treatment for 3 h at 24°C, shifted to 37°C for an hour, and then released at 37°C in HU-free medium. Samples were taken for immunofluorescence and FACScan analysis. Photomicrographs show cells at 150 min after the release. At this point, about 50% of cdc4-10 cells remain blocked in G2/M with a short mitotic spindle while the remainder show a single aster of microtubules (arrows). The wild-type cells are dividing asynchronously at this time point. Differential interference contrast (DIC) micrographs of undigested cells from a different field show the bud morphology. -tub, antitubulin. DAPI, diamidinophenylindole. Bar, 1 µm.

View larger version (77K): [in this window] [in a new window]   FIG. 5.   cdc4 cells are defective in both G1/S and G2/M transitions, whereas the cdc4 sic1 double mutant is arrested mainly in G2/M when depleted of the Cdc4 protein. cdc4 and cdc4 sic1 cells kept alive by GAL-CDC4 on a CEN vector were shifted from galactose to glucose medium for 16 or 12 h, respectively, at 24°C to repress the GAL1 promoter. cdc4 cells contain a mixture of cells arrested at either G1/S or G2/M. The G1/S phenotype, seen in about 34% of cells, is identical to that of cdc4-1 cells shifted to 37°C (arrows), while 65% of the cells are arrested prior to anaphase with a short spindle. The FACScan analysis of cdc4 cells shows a 1N peak and a broad peak at a position slightly greater than 2N DNA content, perhaps because the cells are enlarged and some appear to be undergoing lysis. The cdc4 sic1 cells are round budded, and more than 85% of the cells contain short spindles. This predominant G2/M phenotype correlates with the 2N peak shown by FACScan analysis. Differential interference contrast (DIC) micrographs of undigested cells in a different field show the bud morphology. -tub, antitubulin. DAPI, diamidinophenylindole. Bar, 1 µm.

cdc4 sic1

cdc4 sic1

Isolation of cdc4 ts mutants with a G₂/M phenotype. The notion that CDC4 may serve an additional function in mitosis prompted us to isolate mutations in the CDC4 gene which would render it exclusively defective in its G₂/M function. The CDC4 gene was mutated by error-prone PCR and gap-repair methods (see Materials and Methods) (30). The ts alleles were selected in a cdc4 strain kept viable by GAL-CDC4 on a CEN plasmid. We specifically targeted the region that encodes the WD40 repeats for mutagenesis, since these motifs are thought to facilitate multiprotein complex formation (reviewed by Neer et al. [31]). The repaired plasmids were isolated and retransformed into the parental strain to confirm that they conferred a ts phenotype. None of the ts mutants obtained exhibited exclusively a G₂/M arrest. Four of the eleven ts mutants showed a predominant G₁/S phenotype, and the remaining seven showed a mixture of G₁/S and G₂/M phenotypes at 37°C when shifted from the permissive temperature as cycling cultures to 37°C. Attempts to mutate regions outside the WD40 repeats yielded five additional mutants, which arrested mainly in G₁/S at 37°C. It appears that mutations that affect the G₂/M function also affect G₁/S function.

View larger version (40K): [in this window] [in a new window]   FIG. 6.   The nuclear division defect in the cdc4-12 mutant can be alleviated by the deletion of the PDS1 gene. (A) Positions of mutations in some cdc4 alleles that cause a proportion of cells to be arrested in G2/M. The region that was amplified by error-prone PCR (arrows indicating the primers) in cdc4-12, cdc4-14, and cdc4-15 was sequenced, and the changes that produced amino acid substitutions (vertical bars) are indicated. All the amino acid changes map between the first and second WD40 repeats (black boxes). (B) The cdc4-12 mutant is defective in both G1/S and G2/M transitions. The exponentially growing mutant cells at 24°C were filtered and resuspended in fresh medium prewarmed at 37°C. Samples for immunofluorescence and FACScan analysis were withdrawn at various times. The photomicrographs show the phenotype of cells 4 h after the temperature shift. About 55% of the cells have a single aster of microtubules, while 44% are arrested with a short spindle (arrows). Differential interference contrast (DIC) photomicrographs showing bud morphology are of undigested cells in a different field. (C) The wild-type (data identical to that in Fig. 4), cdc4-12, and cdc4-12 pds1 cells were arrested in HU block for 3 h 15 min at 24°C, shifted to 37°C for 45 min, and then released at 37°C. Samples were withdrawn every 30 min for immunofluorescence and FACScan analysis. While the percentage of wild-type cells with a short spindle drops dramatically 1 h after the release (Fig. 4), 70 to 75% of cdc4-12 cells remain arrested with a short spindle (left panels and graphs). The photomicrographs show phenotypes of cdc4-12 cells (left panels) 150 min after the release from HU block. The cdc4-12 pds1 cells (right panels) undergo anaphase at about 60 min after release from HU, with approximately 40% of cells showing anaphase spindles at 2 to 2.5 h after the release (graphs). Photomicrographs show cells at 150 min after HU release. The mid-region of anaphase spindles tends to become thin. Arrows indicate cells in which nuclear division occurred within one cell. Anaphase spindles are defined as those extended between two well-separated masses of nuclear DNA. The number of G1 cells with single asters of microtubules increases as cells exit from mitosis. Differential interference contrast photomicrographs showing bud morphology are of undigested cells in a different field. Bar, 1 µm. For panels B and C, -tub means antitubulin and DAPI means diamidinophenylindole.

Deletion of PDS1 in the cdc4-12 strain relieves its preanaphase block. Recent evidence (26, 49) suggests that the essential function of Cdc20 in the metaphase-to-anaphase transition is to mediate proteolytic destruction of the anaphase inhibitor Pds1. Consistent with this proposal, deletion of the PDS1 gene allows the cdc20-1 mutant to undergo anaphase (26, 52). Since CDC4 interacts genetically with CDC20, we tested if the preanaphase arrest in the cdc4-12 mutant could also be relieved by deletion of PDS1. We monitored the progression of the double mutant cdc4-12 pds1 through the cell cycle after the release from HU-induced arrest. While the control strain, the cdc4-12 strain, remained largely arrested with a short mitotic spindle throughout the course of the experiment, the cdc4-12 pds1 strain underwent anaphase (Fig. 6B). At 2 to 2.5 h after the release, up to 40% of cdc4-12 pds1 cells (four times as many as the number of cdc4-12 cells) contained anaphase spindles. A small proportion of cells succeeded in completing anaphase normally, so that the nuclei were segregated into mother and daughter cells. However, many cells elongated the spindle within the mother cell. These cells may have been defective in spindle orientation, which may explain why they could not fully extend their spindle into the daughter cell. Another striking feature is the thin appearance of the mid-region of the anaphase spindles, which is also seen in the cdc20-1 pds1 double mutant (26). Unlike the cdc20-1 pds1 mutant, which remains arrested in telophase with high Clb2-associated kinase (26), however, the cdc4-12 pds1 strain progressed through anaphase and then exited mitosis, giving rise to many G₁ cells (represented by cells with single asters of microtubules) by 3 h after the release. These results may imply that the role of Cdc4 during G₂/M transition is related to the proteolytic destruction mechanism operative during the onset of anaphase (but see Discussion).

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

The mechanisms responsible for the proteolytic destruction of mitotic regulators have become subjects of considerable investigation for their now-obvious importance in the cell cycle progression. The E3 enzyme complexes SCF and APC are of particular interest as they control the initiation of S phase and the onset of anaphase, respectively. Cdc4 and Cdc20, both WD40 repeat-containing proteins, are crucial for the activities of these proteolytic systems: while Cdc4 is a component of SCF^Cdc4 (12, 39, 44, 48), Cdc20 is an activator of the APC-dependent proteolysis (26, 49). Thus far, no functional overlap between the two proteolytic systems has been reported. It is therefore surprising that the three mutations that we have isolated in this study as suppressors of the cdc20-1 mutation are all in the CDC4 gene. Consistent with the role of CDC4 in G₁/S transition, all three alleles exhibit a G₁/S defect. However, they are also able to suppress the nuclear division defect of cdc20-1 cells. Furthermore, reexamination of the original cdc4-1 allele and the characterization of cdc4-12 and cdc4 mutants show that they are also defective in the onset of anaphase. The evidence presented in this report strongly suggests that Cdc4 serves an essential function during G₂/M transition.

Since both Cdc4 and Cdc20 contain WD40 repeats and play important roles in proteolytic processes, it is conceivable that Cdc4 can substitute for the Cdc20 function in G₂. However, several observations argue against this possibility: (i) Cdc4 overexpression does not suppress cdc20-1; (ii) two of the cdc4 suppressor alleles tested cannot complement the cdc20 mutant at permissive or restrictive temperatures, suggesting that these are not bypass-suppressor alleles but instead require the presence of the cdc20-1 mutation for the suppression; and (iii) the suppressor activities of cdc4-10 and cdc4-11 alleles are recessive.

It can be argued that the suppression of cdc20-1 by the CDC4 alleles isolated in this study is due to indirect causes. We have considered and tested some of these possibilities. (i) Sic1 overexpression can both suppress the cdc20-1 mutation and activate anaphase (1, 38). Since CDC4 is required for Sic1 proteolysis, Sic1 may accumulate in the cdc4 alleles, which can then suppress the nuclear division defect of the cdc20-1 mutant. However, cdc4-1, cdc4-12, and cdc34-2 mutants accumulate Sic1 when released from HU arrest, and yet they are unable to suppress cdc20-1 (Fig. 1 and 3A). Furthermore, the suppressor mutation cdc4-10 can alleviate the defect of cdc20-1 even in the absence of the SIC1 gene (Fig. 3C). Hence, the accumulation of Sic1 does not contribute to the suppression by the CDC4 alleles. (ii) Cdc20 contains a destruction box in its N terminus and is rapidly degraded in G₁ (12a, 40). As Cdc4 is a component of an E3 complex, it may be responsible for the degradation of Cdc20. The suppressor activity of CDC4 alleles could be due to their failure to degrade the mutant Cdc20-1 protein, allowing its accumulation to levels sufficient to mediate nuclear division. This explanation is clearly untenable because we have found that Cdc20 proteolysis in G₁ is not dependent on Cdc4 function (data not shown). (iii) The suppression of the cdc20-1 mutation by cdc4-10 might be due to a slower progression through G₂/M, which may allow the mutant cdc20-1 protein to accumulate and partially perform its function in G₂/M. This is not likely since the cdc4-10 mutant undergoes mitosis at about the same rate as do wild-type cells at 35°C (Fig. 3). Moreover, cdc4-1 and cdc4-12 strains show a delay or an arrest in G₂/M but cannot suppress the nuclear division defect of cdc20-1 cells. Thus, the ability to suppress cdc20-1 and the G₂/M phenotype are distinctly the characteristics of cdc4 alleles themselves. We have detected no physical interaction between Cdc4 and Cdc20 in coimmunoprecipitation experiments and two-hybrid assays. Hence, the nature of this specific interaction between Cdc4 and Cdc20 remains largely unknown.

SCF is required for the degradation of Gcn4, Far1, Cln1, Cln2, Swe1, and Cdc6. It appears that the substrate specificity is conferred by adapters such as Cdc4, Grr1, and Met30. Although the proposed role of Cdc4 in G₁/S transition is to promote ubiquitination and degradation of phospho-Sic1 (12, 39, 44, 48), the activities of Cdc4 and SCF^Cdc4 are also required in both G₂ and M phases for the degradation of the Cdc6 protein (11, 35). Moreover, Skp1, another component of SCF^Cdc4, is involved in both G₁/S and G₂/M transition (2). The cdc34 sic1 double mutant also exhibits a G₂ delay (39). Collectively, these data imply that SCF^Cdc4 may be active through a large part of the cell cycle. Thus, the requirement for Cdc4 in G₂/M transition, as suggested by our results, is not entirely inconceivable. The constitutive transcription of the CDC4 gene and the presence of the Cdc4 protein in the nucleus throughout the cell cycle are consistent with this notion. The region between the first and second well-conserved WD repeats, in which the ts mutations (isolated in this study) were mapped, may define the part of the Cdc4 protein which is important for its function in both G₁/S and G₂/M transitions. It has been suggested previously that WD repeats form propeller-like structures that may confer a rigid scaffold for surface embellishments (32).

Our assertion that Cdc4 serves an essential function in G₂/M cannot be accounted for by the requirement of SCF^Cdc4 for Cdc6 degradation, since overexpression of wild-type or a nondestructible Cdc6 does not cause any discernible phenotype (11). Incidentally, Skp1, a component of the centromere binding complex (7, 45), was isolated as a high-dosage suppressor of cdc4-1 and interacts with Cdc4 via its F-box. Hence, it is possible that the G₂/M delay exhibited by the cdc4-12 allele may be due to a defective interaction between Cdc4-12 and Skp1, giving rise to a faulty kinetochore function. Alternatively, a checkpoint control may be induced by such a defective interaction (such as incomplete attachment of chromosomes to the spindle). However, we find that the overexpression of SKP1 cannot suppress the G₂/M defect of the cdc4-12 strain when released from HU arrest at 37°C (data not shown).

The alleviation of preanaphase arrest in the cdc4-12 strain by the deletion of PDS1 (Fig. 6B) suggests that the Cdc4 function in G₂/M may be linked to the degradation of Pds1. Pds1, an anaphase inhibitor, is a nonessential protein whose overexpression delays the metaphase-to-anaphase transition (6). Deletion of PDS1 allows cdc16, cdc23, cdc27, and cdc20 mutants to undergo nuclear division at the nonpermissive temperature (26, 52). This is expected, since these four genes are involved in the APC-dependent degradation of Pds1. Cdc23, Cdc16, and Cdc27 are the components of APC (17, 53), and Cdc20 acts either as a substrate-specific activator of Pds1 proteolysis (49) or as a general activator of APC (26). The facts that CDC4 and CDC20 interact genetically and that the PDS1 deletion relieves the cdc4-12 mutant of its nuclear division defect strongly implicate Cdc4, directly or indirectly, in the regulation of the proteolytic mechanism during the metaphase-to-anaphase transition. Since the cdc4-12 pds1 double mutant undergoes nuclear division and subsequently exits mitosis, Cdc4 may be involved in Pds1 degradation as well as Clb2 proteolysis. This is unlike the cdc20 pds1 double mutant, which undergoes nuclear division but is arrested in telophase, most probably due to Clb2 accumulation (26), which prevents exit from mitosis. Although it is tempting to suggest that Cdc4 function in G₂ may be associated with the proteolytic mechanism operative during the onset of anaphase, it must be noted that PDS1 deletion can also allow a number of other mutants to escape their G₂/M arrest. Moreover, it has recently been suggested that Pds1 is an essential component of an S-phase checkpoint control system (5). Therefore, it may be argued that the cdc4-12 mutant cells are arrested in G₂ not because they are defective in the proteolytic mechanism required for the onset of anaphase but due to a defect in progression through S phase. The deletion of PDS1 may "force" these cells to undergo anaphase prematurely, leading to mitotic spindles with thin mid-regions and somewhat defectively segregated nuclear masses (Fig. 6). Since this possibility cannot be discounted at present, the question of the role of Cdc4 in nuclear division remains largely open. We are currently conducting genetic screens for the suppressors of G₂/M-defective cdc4 mutations in the hope of identifying genes that may mediate the G₂/M function of CDC4.

The mitotic spindles in the cdc4-12 pds1 double mutant are often misoriented so that spindle elongation occurs within one cell (presumably the mother cell) (Fig. 6B) and chromosomes fail to segregate into the daughter cell. The failure to orient the spindle is not due to the loss of Pds1 function, since other cdc20 pds1 double mutants are able to elongate their spindles normally through the mother-daughter neck. Cdc4 function might be important in the proper orientation of mitotic spindles. However, it is difficult to envision the nature of its involvement, given that the only known function of Cdc4 to date is to recruit substrates for ubiquitination by SCF^Cdc4. Clearly, further investigations are required to elucidate the role of Cdc4 in the metaphase-to-anaphase transition.