Budding yeast RSI1/APC2, a novel gene necessary for initiation of anaphase, encodes an APC subunit

Kate M. Kramer, Didier Fesquet, Anthony L. Johnson, Leland H. Johnston

Author Affiliations

  1. Kate M. Kramer1,
  2. Didier Fesquet2,
  3. Anthony L. Johnson1 and
  4. Leland H. Johnston*,1
  1. 1 Division of Yeast Genetics, National Institute for Medical Research, The Ridgeway, Mill Hill, London, NW7 1AA, GB
  2. 2 Centre de Recherches de Biochimie Macromoleculaire, CNRS ERS 155, 1919 Route de Mende, 34293, Montpellier, France
  1. *Corresponding author. E-mail: ljohnst{at}
View Full Text


SIC1 is a non‐essential gene encoding a CDK inhibitor of Cdc28‐Clb kinase activity. Sic1p is involved in both mitotic exit and the timing of DNA synthesis. To identify other genes involved in controlling Clb‐kinase activity, we have undertaken a genetic screen for mutations which render SIC1 essential. Here we describe a gene we have identified by this means, RSI1/APC2. Temperature‐sensitive rsi1/apc2 mutants arrest in metaphase and are unable to degrade Clb2p, suggesting that Rsi1p/Apc2p is associated with the anaphase promoting complex (APC). This is an E3 ubiquitin‐ligase that controls anaphase initiation through degradation of Pds1p and mitotic exit via degradation of Clb cyclins. Indeed, the anaphase block in rsi1/apc2 temperature‐sensitive mutants is overcome by removal of PDS1, consistent with Rsi1p/Apc2p being part of the APC. In addition, like our rsi1/apc2 mutations, cdc23‐1, encoding a known APC subunit, is also lethal with sic1Δ. Thus SIC1 clearly becomes essential when APC function is compromised. Finally, we find that Rsi1p/Apc2p co‐immunoprecipitates with Cdc23p. Taken together, our results suggest that RSI1/APC2 is a subunit of APC.


Orderly progression through the eukaryotic cell cycle is controlled by the regulated association of specific cyclins with a CDK (cyclin‐dependent kinase). In the budding yeast, Saccharomyces cerevisiae, Cdc28 is the major CDK and is largely responsible for controlling cell cycle progression (reviewed in Nasmyth, 1996). G1 cyclins Clns 1, 2 and 3 are active during G1 up until S phase while B‐type cyclins Clbs 1–6 control DNA synthesis (Schwob et al., 1994) and mitosis (Surana et al., 1991). The specific association of the appropriate cyclin with Cdc28 is achieved by cell cycle‐controlled synthesis as well as controlled degradation of the cyclin at key stages of the cycle (reviewed in King et al., 1996; Deshaies, 1997). The mitotic B‐type cyclin Clb2p is active from late S phase until the end of mitosis when it is rapidly degraded by ubiquitin‐mediated proteolysis (Surana et al., 1991; Irniger et al., 1995; Amon, 1997; Irniger and Nasmyth, 1997). Exit from mitosis and entry into the G1 phase of the next cell cycle requires the inactivation of the Clb2 protein; over‐production of Clb2p which has been stabilized by removal of its destruction box causes cells to arrest in telophase with divided chromatin and an elongated spindle (Surana et al., 1993).

E3 ubiquitin‐protein ligases play an important role in specifying the selection of proteins for ubiquitin‐mediated degradation (Ciechanover, 1994). Recently, studies in human, Xenopus and yeast systems have converged to reveal the E3 ubiquitin‐protein ligase activity which confers specificity for cell cycle‐regulated degradation of mitotic cyclin by the ubiquitin pathway (Irniger et al., 1995; King et al., 1995; Tugendreich et al., 1995). This E3 is a large complex of at least seven proteins, and has been called the cyclosome or anaphase promoting complex (APC). The APC is part of the essential cell cycle machinery whose components are evolutionarily conserved (Irniger et al., 1995; King et al., 1995; Tugendreich et al., 1995; Peters et al., 1996; Zachariae et al., 1996). In yeast CDC16, CDC23 CDC26, CDC27 and APC1 have been identified as genes coding for some of these components (Lamb et al., 1994; Irniger et al., 1995; Zachariae et al., 1996). As its name suggests, an active APC is required for cells to initiate anaphase; APC‐defective yeast strains arrest in metaphase with replicated DNA but with unseparated sister chromatids (Lamb et al., 1994; Zachariae et al., 1996).

While Clb2p degradation is normally initiated at anaphase onset (Surana et al., 1991; Irniger et al., 1995), its degradation is not required for the metaphase to anaphase transition (Surana et al., 1993) and some Clb2 protein persists until telophase (Irniger et al., 1995). Thus, the APC clearly has other critical targets. One of the early events required for the metaphase to anaphase transition is sister chromatid separation. Several proteins whose proteolysis is required for sister chromatid separation have recently been identified; the pimples product in Drosophila (Stratmann and Lehner, 1996), Cut2 in fission yeast (Funabiki et al., 1996) and Pds1p in budding yeast (Cohen Fix et al., 1996; Yamamoto et al., 1996a, b). Interestingly, yeast strains lacking the PDS1 gene are viable but in APC‐deficient strains show a mixed terminal arrest phenotype with a large proportion of the cells in telophase, that is, with divided chromatin (Yamamoto et al., 1996a). This suggests that the critical requirement for the APC at the metaphase to anaphase transition is to ensure the proper separation of sister chromatids. Other known APC targets include Ase1p, a budding yeast protein involved in spindle elongation (Juang et al., 1997) whose destruction is required for proper spindle disassembly after the metaphase to anaphase transition. The final essential role of the APC in mitotic exit is the degradation of Clb2p (Surana et al., 1993; Irniger et al., 1995). However, there are clearly other factors which contribute to the control of Clb2p inactivation since the APC becomes active at metaphase but Clb2p levels persist until telophase.

Another protein controlling Clb2p activity is Sic1p, a CDK inhibitor of both S phase and mitotic Clb–Cdc28 kinases (Donovan et al., 1994; Schwob et al., 1994; Amon, 1997). Sic1 protein regulates the timing of DNA synthesis by binding to and inhibiting S phase kinase Cdc28–Clb5 early in the cell cycle (Schwob et al., 1994). Low Clb‐kinase activity at this stage of the cell cycle is critical for enabling replication proteins such as Cdc6 and Mcm2‐7 to be loaded onto replication origins making the DNA competent for replication (Donovan et al., 1997), a process that is inhibited by Clb kinase activity (Dahmann et al., 1995; Tanaka et al., 1997). Shortly before S phase, Sic1p is degraded by the Cdc4/Cdc34/Cdc53‐dependent ubiquitin‐mediated proteolytic pathway, thereby activating Cdc28–Clb5 kinase. Sic1p is then resynthesized in late mitosis (Donovan et al., 1994; Toyn et al., 1997).

Three separate lines of evidence suggest a role for Sic1p in late mitosis. The first is the synthetic lethality between sic1Δ and hct1Δ. HCT1 is a recently described non‐essential gene that directly controls Clb2p proteolysis (Schwab et al., 1997); hct1Δ mutants have high levels of Clb2p throughout the cell cycle and overexpression of HCT1 induces ectopic activation of Clb2 proteolysis. Second, SIC1 interacts genetically with a variety of late mitotic arrest mutants (for review see Deshaies, 1997) and multiple copies of SIC1 rescue temperature‐sensitive mutants of the DBF2 group of kinases which all arrest in telophase with large budded cells, divided chromatin, an elongated spindle and high mitotic kinase activity (Toyn and Johnston, 1994; Deshaies, 1997). Third, sic1Δ dbf2Δ is a lethal combination and the double mutant has a late mitotic arrest terminal phenotype (Toyn et al., 1997). Taken together, these observations argue strongly that SIC1 and various other genes such as DBF2 and HCT1 are involved in functionally overlapping pathways controlling exit from mitosis.

Because control of mitotic exit clearly involves redundant pathways, isolating mutations that are synthetically lethal with sic1Δ should identify other genes involved in mitotic exit. Accordingly, we carried out a screen for such mutants and describe here the isolation of a gene we call RSI1 (Requires Sic1 CDK Inhibitor). Although RSI1 is essential, we have characterized a particular allele which is viable but becomes lethal in a sic1Δ background. We show that Rsi1p is required for the metaphase to anaphase transition and also for Clb2p degradation at key cell cycle stages. Finally we show that Rsi1p interacts in vivo with the APC component Cdc23p. We conclude that RSI1 codes for an APC subunit and propose alternative models accounting for the rsi1 sic1Δ synthetic lethality.


A genetic screen for mutants requiring SIC1

Mutagenesis was carried out in strain JD100 (Table I), a sic1Δ::TRP1 haploid strain containing SIC1 on a low copy number URA3 plasmid (pSIC1U). We exploited the toxicity of 5‐fluoroorotic acid (FOA) in URA3 strains (see Materials and methods; Ota and Varshavsky, 1992) to isolate mutants requiring SIC1 for growth; such mutants will not be able to lose pSIC1U and therefore will be unable to grow on FOA. For the mutagenesis, we used a yeast genomic library, generously provided by Dr M.Snyder, Yale University (Burns et al., 1994), in which wild‐type sequences are randomly disrupted by LEU2 insertions. Disrupted yeast sequences are cut out of the library vector and integrated into the target genome by transplacement. This effectively replaces wild‐type genomic DNA with LEU2 disrupted sequences. We felt disruption to be an appropriate mutagenesis since late mitotic pathways are clearly redundant (see Introduction), hence, at least some genes are expected to be dispensable in a wild‐type background. We screened 80 000 LEU+ transformants and only a single mutation lethal with sic1Δ was identified. This lay in a previously uncharacterized gene we are calling RSI1 (see above). We refer to this mutant allele as rsi1‐1.

View this table:
Table 1. Yeast strainsa

sic1Δ rsi1‐1 synthetic lethality

The rsi1‐1 mutant was isolated by its inability to grow on FOA plates in a sic1Δ pSIC1U genetic background (KTM100U, Table I and Figure 1A). Providing wild‐type SIC1 on a plasmid vector with HIS7 as a selectable marker (KTM100H, Table I) allowed growth on FOA whereas sic1Δ on the same plasmid vector (KTM100sΔH, Table I) did not. To confirm the synthetic lethality of rsi1‐1 and sic1Δ, we crossed KTM100U with wild‐type CG378 (Table I). Out of eight tetrads dissected, all LEU+ (rsi1‐1) TRP+ (sic1Δ) spore clones were also URA+ (pSIC1U) and unable to grow on FOA. Furthermore, a subsequent cross between an rsi1‐1 SIC1 haploid (KTM110, Table I) and JD100, the original RSI1 sic1Δ strain used for the mutagenesis, yielded no LEU+ (rsi1‐1) TRP+ (sic1Δ) viable spore clones from 18 tetrads dissected, confirming that rsi1‐1 requires a wild‐type copy of SIC1 for viability. Non‐viable spores had germinated but arrested as large budded cells after one or several divisions.

Figure 1.

(A) sic1Δ rsi1‐1 synthetic lethality. Cells were streaked onto plates containing either no FOA or 1mg/ml FOA and incubated for 3 days at 30°C. The strain sic1Δ rsi1‐1 pSIC1U carries SIC1 on a URA3 marked plasmid (Table I). Strains sic1Δ rsi1‐1 pSIC1H and sic1Δ rsi1‐1 psic1ΔH are identical to sic1Δ rsi1‐1 pSIC1U but carry either SIC1 or sic1Δ on a HIS7 marked plasmid. (B) The heavy line indicates the RSI1 2.5 kb open reading frame located on the left arm of CHR XII. The rsi1‐1 mutation is a LEU2 ‐transposon insertion 425 bp upstream of the initiating ATG. The XhoI–NdeI fragment was replaced by the HIS7 gene in rsi1Δ strains. (C) Sequence alignment showing the homologous region between Rsi1 and cullin family members. DDBJ/EMBL/GenBank accession numbers are indicated in brackets. Sc, S.cerevisiae Rsi1; Sp, S.pombe hypothetical protein; Dm, D.melanogaster Lin‐19; Hs, Homo sapiens Cul‐1.

The sequences around the LEU2 insertion site in rsi1‐1 were cloned by integrating a single copy of a bacterial/yeast shuttle vector at that site in KTM100H (see Materials and methods). Sequence analysis revealed that the LEU2 insertion in rsi1‐1 is in the promoter of a gene on chromosome XII (Figure 1B) we have named RSI1 (see above). RSI1 encodes a 853 amino acid protein of 99 kDa (DDBJ/EMBL/GenBank Z73299; YPD YLR127C) containing several potential Cdc28 phosphorylation sites. Database searches reveal a domain of weak homology (23–28% identity, 49–55% similarity) to Drosophila and and human genes belonging to the cullin family (Jackson, 1996; Kipreos et al., 1996) as well as to a Schizosaccharomyces pombe hypothetical protein (Figure 1C). Cullin family members appear to have a role in cell cycle regulation and include S.cerevisiae CDC53, which controls G1 cyclin degradation (Jackson, 1996; Willems et al., 1996).

A null allele of RSI1 was created by replacing one copy of the RSI1 coding sequences with the HIS7 gene (Figure 1B) in diploid strain CG378×CG379 (Table I). Sporulation followed by dissection of 18 tetrads gave a segregation pattern of 2 viable:2 dead in all cases and the viable spores were all his (RSI1). RSI1 is therefore an essential gene. The non‐viable spores had germinated, entered the cell cycle and arrested as large budded cells after only a few divisions.

The rsi1‐1 strain is hypersensitive to Clb2p levels

Strains defective in mitotic exit, including sic1Δ strains, are hypersensitive to increased levels of Clb2p (Shirayama et al., 1994; Toyn et al., 1997); a single copy of GAL‐CLB2 prohibits growth on galactose. We reasoned that if the sic1Δ rsi1‐1 synthetic lethality is due to a defect in Clb2p kinase inactivation, then the rsi1‐1 mutation on its own might be sensitive to increased levels of Clb2p. To examine this, a single copy of GAL‐CLB2 was integrated at the URA3 locus in the rsi1‐1 mutant and also in congenic wild‐type cells. The copy number of the integrated GAL‐CLB2::URA3 was confirmed by Southern analysis (data not shown). Figure 2 shows that while all strains can grow on 2% glucose, the rsi1‐1 mutation, like sic1Δ, prohibits growth on 2% galactose. Furthermore, sensitivity to GAL‐CLB2 expression in the rsi1‐1 strain is overcome by providing either RSI1 or SIC1 on a low copy plasmid. These results taken together with the sic1Δ rsi1‐1 synthetic lethality strongly suggest that Rsi1p and Sic1p function in parallel pathways controlling the inactivation/degradation of Clb2p.

Figure 2.

The rsi1‐1 mutant is hypersensitive to CLB2 levels. A single copy of GAL‐CLB2 was integrated at the URA3 locus in the rsi1‐1 mutant as well as in congenic sic1Δ and wild‐type strains. Cells were streaked onto plates containing either 2% glucose or 2% galactose and incubated for 3 days at 30°C.

Cell cycle phenotype of rsi1 temperature‐sensitive mutants

To characterize the function of RSI1 in more detail, we generated a bank of temperature‐sensitive alleles using mutagenic PCR and in vivo ‘gap repair’ (Connelly and Hieter, 1996). All of the mutant alleles showed a similar terminal arrest phenotype as exemplified by the rsi1‐8 allele in strain KTM208 (Table I) described below. When a log phase culture of rsi1‐8 cells was shifted from growth at 25°C to 37°C for 3 h, the cells displayed a cell cycle phenotype, arresting with large buds, an undivided nucleus positioned at the neck of the bud and a short mitotic spindle (Figure 3A). FACS analysis showed them to contain fully replicated DNA (Figure 3B). This phenotype is identical to that reported for cdc23‐1, a temperature‐sensitive allele of CDC23 which encodes an APC subunit (Sikorski et al., 1993).

Figure 3.

Terminal arrest phenotype of rsi1‐8 temperature‐sensitive mutants. rsi1‐8 cells were grown to log phase at 25°C then shifted to 37°C for 3 h. (A) Chromatin was visualized by DAPI staining and spindles were visualized by anti‐tubulin staining. (B) FACS analysis. Cells were stained with propidium iodide and 20 000 were analyzed for each sample.

The rsi1‐8 strain is defective in Clb2p degradation

cdc23‐1 mutants are defective for Clb2p degradation, therefore we assessed the Clb2‐Cdc28‐specific H1 kinase activity as well as Clb2p stability in the rsi1‐8 mutant. The H1 kinase activity was measured using anti‐HA immunoprecipitated Clb2HA3. First, we examined the kinase activity from a log phase culture of rsi1‐8 cells containing an integrated copy of CLB2HA3, that had been shifted from growth at 25°C to 37°C for 3 h. In this experiment the tagged CLB2 replaced the endogenous gene and was fully functional. This was compared with the same activity from congenic wild‐type cells also containing CLB2HA3 and treated for 3 h either in nocodazole (a microtubule‐depolymerizing drug), to induce an M phase arrest where Clb2–Cdc28‐specific H1 kinase activity is high (Amon et al., 1994), or with α‐factor to induce an early G1 arrest where the kinase activity is undetectable. FACS analysis confirmed that the predicted arrests had occurred (Figure 4A). As expected, the α‐factor arrested wild‐type cells had no detectable Clb2–Cdc28‐specific H1 kinase activity (Figure 4A, lane 1). In marked contrast, the extracts from rsi1‐8 mutant cells had a high Clb2–Cdc28‐specific H1 kinase activity at 37°C (Figure 4A, lane 3), comparable with the nocodazole arrested wild‐type cells (Figure 4A, lane 2). Moreover, Clb2p was readily detectable in the rsi1‐8 cells at 37°C (Figure 4A, lane 3) suggesting that it was stabilized.

Figure 4.

rsi1‐8 temperature‐sensitive mutants arrest with high Clb2p kinase activity and have a defect in Clb2p degradation at the restrictive temperature. (A) Strain KTM 208CH, rsi1‐8, congenic wild‐type CG378 and strain CG378CH containing integrated CLB2HA3 were grown to log phase at 25°C. The wild‐type cells were treated for 3 h with α‐factor or with nocodazole. The rsi1‐8 cells were shifted to 37°C for 3 h. Extracts were prepared and Clb2p levels and associated H1 kinase activity was measured. Lanes 1 and 2, wild‐type CLB2HA3 cells arrested with α‐factor and nocodazole, respectively; lane 3, 37°C arrested rsi1‐8 CLB2HA3 cells; lane C, nocodazole‐arrested wild‐type cells that do not contain CLB2HA3. (B) Strain KTM208GCH, rsi1‐8, and a congenic wild‐type strain, CG378GCH, both containing GAL–CLB2HA3 on a low copy number plasmid, were grown to log phase at 25°C in 2% raffinose medium. α‐factor was added and after 4 h galactose was added to 2% and the culture was simultaneously shifted to 37°C. Extracts were prepared from cells after 0, 25, 60 and 100 min growth at 37°C. Additional α‐factor was added at intervals to ensure continued G1 arrest. Equal amounts of protein were analyzed by SDS–PAGE and Western blotting using a monoclonal anti‐HA antibody (12CA5, upper panel). The nitrocellulose membrane was stained with Ponceau S and is shown as a loading control (lower panel). At the same time points, cells were also removed for FACS analysis.

To confirm this we examined Clb2p stability in rsi1‐8 cells that had been arrested in G1 with α‐factor. Clb2p normally has a half‐life of <1 min in G1 (Amon et al., 1994) but is stabilized in G1‐arrested cdc23‐1 cells (Irniger et al., 1995). We used a GAL‐CLB2 fusion construct with a triple HA epitope tag, pGAL‐CLB2HA3, to examine Clb2p stability in G1‐arrested rsi1‐8 and isogenic wild‐type cells. Log phase cultures were exposed to the mating pheromone α‐factor during growth in raffinose at 25°C. Once the cells were arrested in G1, as determined by the absence of buds and the formation of mating projections, galactose was added to 2% to induce high levels of ectopic Clb2p expression and simultaneously the cultures were shifted to 37°C. In the strain carrying the rsi1‐8 mutation, Clb2 protein was detectable by Western blotting after 25 min with Clb2p levels increasing throughout the 100 min induction (Figure 4B). In contrast, in the wild‐type strain, we observed no such Clb2 protein accumulation. Thus, in α‐factor‐arrested cells, Rsi1p is required for Clb2p degradation.

It is significant that in the rsi1‐8 mutant after 100 min of galactose‐induced CLB2 expression at 37°C, FACS analysis showed that the cells had entered S phase (Figure 4). In contrast, the wild‐type cells maintained a G1 DNA content throughout the experiment. It is important to note that the rsi1‐8 cells had not initiated DNA synthesis due to escape from the α‐factor holding; they retained the mating projection induced by α‐factor and remained unbudded. Clb kinase is required for entry into S phase (Schwob et al., 1994) but is kept low during G1 by the presence of the Sic1p CDK inhibitor and persistent APC function (Irniger and Nasmyth, 1997). In the mutant, however, sufficient levels of Clb kinase had accumulated to initiate S phase despite the presence of α‐factor. This phenomenon has been previously observed in mutants of APC components (Zachariae et al., 1996; Irniger and Nasmyth, 1997) and supports a direct association of Rsi1p with the APC.

cdc23‐1 is also synthetically lethal with sic1Δ

The phenotypic similarities between rsi1‐8 and cdc23‐1 led us to look for a possible genetic interaction between CDC23 and SIC1. We crossed a cdc23‐1 mutant with a sic1Δ::TRP1 strain and examined segregation of temperature sensitivity (cdc23‐1) with the TRP1 (sic1Δ) marker. Out of 27 tetrads dissected, no spore clones were temperature‐sensitive (cdc23‐1) and TRP+ (sic1Δ), indicating that the sic1Δ cdc23‐1 combination is lethal in haploids. We also examined spores where the sic1Δ cdc23‐1 combination could be predicted based on the genotype of the viable spore clones. Of 16 such spores, all but one had germinated and arrested as large budded cells after one or a few divisions. Because this is identical to the phenotype observed in the sic1Δ rsi1‐1 spores described above we considered the possibility that Rsi1p is an APC subunit and thus shares a common role in cell cycle progression with Cdc23p.

Rsi1p interacts with the APC component Cdc23p in vivo

As a conclusive indication that RSI1 encodes a subunit of APC we looked for a direct association of Rsi1p with Cdc23p in vivo. A triple Myc epitope‐tagged version of RSI1, RSI1Myc3, was integrated in wild‐type haploid CG378 at the RSI1 locus. In this strain the only copy of RSI1 with a promoter is the RSI1Myc3 version of the gene; hence this is the only source of Rsi1p. RsiMyc3p was detected as a 100 kDa protein by Western blotting of crude extracts from log phase cells (Figure 5B). The RSI1Myc3 strain and the corresponding isogenic parental strain were transformed with a low copy plasmid containing a double HA epitope‐tagged CDC23, CDC23HA2. Extracts prepared from cells expressing various combinations of these tagged proteins were subjected to immunoprecipitation with anti‐Myc antibodies and probed with an anti‐HA antibody to look for co‐precipitation of Cdc23p with Rsi1p (Figure 5A). As expected, Cdc23HA2p was specifically co‐precipitated with Rsi1Myc3p. This in vivo interaction between Cdc23p and Rsi1p strongly supports our evidence that Rsi1p is indeed a newly identified component of APC. Rsi1p is likely to be the previously detected 100 kDa component of APC (Zachariae et al., 1996).

Figure 5.

The in vivo interaction between Rsi1p and Cdc23p. Isogenic strains containing either RSI1Myc3 or CDC23HA2 alone or with both constructs together were grown to log phase and cell extracts were prepared. (A) Protein from crude extracts (left panel) and from extracts immunoprecipitated using a monoclonal anti‐Myc antibody (9E10, right panel) were electrophoresed and probed with a monoclonal anti‐HA antibody (12CA5) to detect Cdc23HA2p. (B) The anti‐HA antibody was stripped from the blot shown in (A) and it was re‐probed with monoclonal anti‐Myc antibody to detect Rsi1Myc3p. *indicates remaining traces of the previous HA probing.

The pre‐anaphase block in rsi1‐8 mutants can be overcome in the absence of Pds1p

A unique and characteristic phenotype of mutations in APC components is manifested in strains lacking Pds1p. Because APC‐mediated proteolysis of Pds1p is essential for sister chromatid separation (Cohen Fix et al., 1996; Yamamoto et al., 1996a, b), a high proportion of APC‐deficient pds1Δ cells enter anaphase and arrest in telophase with divided chromatin (Yamamoto et al., 1996a). Accordingly, we examined the terminal arrest phenotype in a rsi1‐8 pds1Δ double mutant compared with a rsi1‐8 PDS1 control strain. Log phase cultures of both strains growing at 25°C were shifted to growth at 37°C for 3 h. FACS analysis showed that after 3 h both strains had stopped growing and had arrested with fully replicated DNA (Figure 6). However, examination of chromosome segregation in DAPI stained cells showed a dramatic difference in the arrest phenotype. The number of cells with divided chromatin in the rsi1‐8 pds1Δ double mutant had increased from 12% in the log phase culture (0 h, Figure 6) to 51% after 3 h at 37°C. In contrast, the number of cells with divided chromatin in the rsi1‐8 PDS1 control strain showed a significant decrease from 12% in log phase to 5% after 3 h at 37°C, these cells having accumulated in metaphase. The percentage of cells with divided chromatin in the rsi1‐8 pds1Δ double mutant after 3 h at 37°C is in good agreement with the previously observed percentage in terminally arrested cdc16 pds1 and cdc23 pds1 double mutants (Yamamoto et al., 1996a). Thus, rsi1 behaves in the same way as cdc16 and cdc23 in a pds1Δ genetic background. The high proportion of cells with divided chromatin in the rsi1‐8 pds1Δ double mutant shows that those cells which were able to initiate anaphase had subsequently arrested in telophase demonstrating, in common with APC, a critical requirement for Rsi1p at two cell‐cycle transition points, the metaphase to anaphase transition and mitotic exit.

Figure 6.

Deletion of PDS1 relieves the anaphase block in rsi1‐8 arrested cells. rsi1‐8 and rsi1‐8 pds1Δ isogenic strains were grown to log phase at 25°C and then shifted to 37°C for 3 h. Cells harvested at 0 and 3 h were prepared for FACS analysis and stained with DAPI to visualize the chromatin. The DAPI‐stained cells were examined for the presence of divided chromatin and the figures are presented in Table I.


Rsi1p is an APC component

We have used a wide range of criteria including genetic interactions, cytology, cell physiology and biochemistry to demonstrate that RSI1 codes for a previously uncharacterized component of the APC. Our first indication that Rsi1p might be an APC component was that the rsi1‐8 temperature‐sensitive mutant showed the same metaphase arrest phenotype as a temperature‐sensitive allele of CDC23 (Sikorski et al., 1993). All of our subsequent results support this; RSI1 is an essential gene and the rsi1‐8 mutant shows a metaphase arrest; Rsi1p controls proteolysis of two known APC targets, namely Clb2p and Pds1p; Rsi1p interacts with Cdc23p in vivo. While it could be argued that Rsi1p is merely involved in controlling APC function and is not itself an APC subunit, there are several lines of evidence against this argument. First, we have shown that Rsi1p interacts physically with Cdc23p, a known APC component. Second, Rsi1p controls both Clb2p and Pds1p degradation. We showed this directly in the case of Clb2p; in the rsi1‐8 mutant Clb2p is stabilized at the cell‐cycle arrest point and when expressed in G1. Regarding Pds1p we find that in the rsi1‐8 mutant, the metaphase arrest is dependent on the presence of the anaphase inhibitor Pds1p. In the absence of Pds1p, a high proportion of rsi1‐8 cells are able to enter anaphase but subsequently arrest in telophase, a phenotypic characteristic that has been reported for other APC mutants in a pds1Δ background (Cohen Fix et al., 1996; Yamamoto et al., 1996a). We can contrast this with Hct1p which is a regulator specifically of Clb2p proteolysis (Schwab et al., 1997). Although hct1 mutants share some phenotypic characteristics with rsi1 mutants, one of the most important features of Hct1p is that it acts specifically in Clb2p degradation but has no apparent effect on Pds1p.

The APC isolated from yeast contains at least seven proteins; Cdc16p, Cdc23p, Cdc26p, Cdc27p, Apc1p, a 100 kDa protein and an 80 kDa protein (Zachariae et al., 1996). In all probability, Rsi1p corresponds to the 100 kDa protein which was previously detected in an immunoprecipitate from a CDC16‐myc6 strain (Zachariae et al., 1996). Although it is known that the APC functions as an E3 ubiquitin‐protein ligase (see Introduction), it is not clear what the function of each component might be or how these functions are co‐ordinated. Cdc16p, Cdc23p and Cdc27p all contain TPR repeat protein–protein interaction domains (Lamb et al., 1994; Zachariae et al., 1996). On the other hand, Rsi1p has no recognized protein–protein interaction domains. The significance of the two to four potential Cdc28 phosphorylation sites in Rsi1p is unclear but they may be important for the regulation of APC. Rsi1p function is not controlled by cell cycle‐regulated RSI1 expression according to our Northern analysis (data not shown). On the other hand, it is thought that Clb–Cdc28 kinase controls the APC (Irniger et al., 1995; Amon, 1997).

We have recently learned that, in agreement with our results, Rsi1p has been identified as an APC component under the name Apc2 in other laboratories (K.Nasmyth, personal communication). As the nomenclature APC2 reflects the cross species conservation of APC members we will henceforth refer to RSI1 as APC2.

Rsi1p/Apc2p may be a cullin family member

The homology between Rsi1p and members of the cullin family of proteins is intriguing. Cullins were first defined as a family of proteins by homology to Caenorhabditis elegans cul‐1 (Kipreos et al., 1996) which was identified as a mutant failing to properly exit the cell cycle during C.elegans post‐embryonic development. Yeast database searches identify several related cullin proteins, including Cdc53p. It has been proposed that Cdc53p is a component of an E3 ligase which mediates G1 cyclin as well as Sic1p degradation (Willems et al., 1996). The functional similarity between Cdc53p, Rsi1p/Apc2p and (perhaps) Cul‐1 suggests that Rsi1p/Apc2p may be a cullin family member, despite the rather limited homology with other cullin family members. It can hardly be coincidental that structurally related proteins occur in two yeast E3 ubiquitin ligases. The role of individual proteins within these complexes is at present unknown but it is tempting to speculate that Rsi1p/Apc2p and Cdc53p may perform similar functions.

The role of Sic1p in cell‐cycle control

The genetic interaction between RSI1 and SIC1 is a striking result. Sic1p clearly has no role in anaphase initiation since the protein is only present in cells from the end of mitosis into G1. Sic1p is known to act late in the cell cycle, presumably facilitating mitotic exit by inhibition of Clb2p kinase activity (Donovan et al., 1994; Toyn et al., 1997) whilst during G1 Sic1p binds to and inhibits Clb5 allowing the formation of functional replication complexes before S phase initiation (Schwob et al., 1994; Dahmann et al., 1995; Tanaka et al., 1997). Since the only known function of Sic1p is as a CDK inhibitor of Clb kinases and since Clb2p is an APC target, we assume that in the sic1Δ rsi1‐1 double mutant there is a lethal excess of Clb2p at some stage of the cell cycle. This is supported by the sensitivity of the rsi1‐1 mutation to elevated levels of Clb2p and by the ability of SIC1 in low copy number to suppress this sensitivity. As mentioned above (see Introduction) cells overexpressing non‐degradable Clb2p are unable to exit mitosis and arrest in telophase. Thus, one possible explanation for the sic1Δ rsi1‐1 synthetic lethality is that in an APC‐compromised strain lacking Sic1p there is an accumulation of Clb2p which causes a telophase arrest. Supporting this model is the fact that rsi1‐8 pds1Δ cells are able to initiate anaphase but cannot exit mitosis. This implies an essential function for APC in mitotic exit. We assume that this function is the targeting of Clb2p for proteolysis; however, it is possible that another currently unknown protein also needs to be degraded before cells can exit mitosis. A second possibility is that even with elevated levels of Clb2p, the sic1Δ rsi1‐1 double mutant is able to exit mitosis but cannot then form functional replication complexes because of high levels of inhibitory Clb2p and Clb5p kinase activity. This would result in a late G1 cell cycle block. Clearly, it will be critical to examine the terminal phenotype displayed by the sic1Δ rsi1‐1 double mutant. We have attempted this by shutting off SIC1 expression in the double mutant using a variety of regulatory promoters but have been unable to reproduce the lethal phenotype. Evidently, even the very low basal levels of Sic1p expressed from these promoters is sufficient to drive the cell cycle in the rsi1‐1 mutant.

Our screen for mutations lethal in a sic1Δ background was clearly not saturated since we did not recover other genes known to be synthetically lethal with sic1Δ such as DBF2 and HCT1. Nevertheless, given that RSI1 is associated with the APC, the question arises as to why we did not recover mutations in any other genes coding for APC components. The probable reason for this is that all known APC components are essential and therefore most LEU2 insertions in APC genes would be lethal events. We believe the RSI1 mutation we recovered, rsi1‐1, was fortuitous because the insertion of LEU2 in the promoter (Figure 1B) resulted in a compromised but not an inactive APC. In fact, a deletion of RSI1 promoter sequences just 125 bp closer to (but still 300 bp away from) the initiating ATG is lethal (data not shown). The sic1Δ cdc23‐1 synthetic lethality clearly indicates that the absence of other APC genes recovered from this screen is due to the nature of the screen itself and is not an indication that the RSI1 SIC1 genetic interaction is specific to Rsi1p as an APC component. Instead, our results suggest that SIC1 becomes essential whenever APC function is sufficiently compromised. According to this interpretation, elevated levels of Clb2p at inappropriate stages of the cell cycle augment the requirement for Sic1p. Support for this model comes from the sic1Δ hct1Δ as well as the sic1Δ cdc23‐1 synthetic lethality; both hct1Δ and cdc23‐1 are severely defective for Clb2p degradation, the cdc23‐1 defect being manifested even at the permissive temperature (Irniger et al., 1995; Schwab et al., 1997). On the other hand, in CDC16 temperature‐sensitive mutant cdc16‐1, Clb2p is degraded in α‐factor‐arrested cells even at the restrictive temperature (Irniger et al., 1995) and a cross between cdc16‐1 and sic1Δ haploids showed no synthetic lethality (data not shown). Thus, the synthetic lethality of sic1Δ with mutants of APC components clearly depends upon the severity of the mutant allele.

We are currently extending the screen for mutations that are lethal in a sic1Δ background using EMS mutagenesis. This will allow us to recover essential as well as non‐essential genes. We expect to recover other APC components, genes regulating APC activity and possibly genes involved in the G1 to S phase transition.

Materials and methods

Strains and media

Relevant yeast strain genotypes are indicated in Table I. All yeast strains were in a CG378 or congenic CG379 background except for cdc23‐1 strain 2531W (Table I). 2531W was generated by crossing cdc23‐1 haploid 2531 (Table I) to W303 to obtain the cdc23‐1 allele in a trp1 background. All plasmids were propagated in DH5α cells. Yeast transformations were as described (Schiestl and Gietz, 1989). Yeast was grown in rich broth supplemented with 2% glucose, YPD, (Sherman et al., 1986) or with the appropriate carbon source at 2%. Minimal medium was yeast nitrogen base (Difco), with appropriate supplements and a carbon source at 2%. FACS analysis was done using a FACStar Becton Dickinson flow cytometer. For cell cycle arrests, 10 μml α‐factor or 10 μg/ml nocodazole were used.

Plasmids and constructs

The RSI1 null allele was a deletion of the open reading frame made by removing the 2.5 kb internal XhoI–NdeI fragment and replacing it with the HIS7 gene. This construct retains only eight 5′ amino acids and eleven 3′ amino acids.

To measure sensitivity of CLB2 overexpression, CLB2 was expressed from YIpG7CLB2 (Shirayama et al., 1994) as previously described (Toyn et al., 1997). For CLB2 overexpression during α‐factor arrest we transformed pWS945 (a gift from B.Futcher, Cold Spring Harbor), a YCplac33 (Gietz and Sugino, 1988) ‐based plasmid containing CLB2HA3 under control of the GAL10 promoter, into KTM208 and CG378.

The rsi1‐8 temperature‐sensitive allele was generated using mutagenic PCR and in vivo gap repair (Connelly and Hieter, 1996). The mutagenic PCR product was synthesized from RSI1 containing Ycplac111 (Gietz and Sugino, 1988), a TRP1 marked vector. The oligonucleotides used, ACCGCCTCTCCCCGCGCG and TCTCTATGGTGAGACAGC were to opposite sides of the YCplac111 multiple cloning site and generated a product with ∼200 bp of vector sequences at either end. The PCR products were co‐transformed along with YCplac111 vector, that had been linearized by cutting with SmaI, into a rsi1Δ strain which was kept alive by an ARS‐based, RSI1 URA3 plasmid. The RSI1 URA3 plasmid was removed by growth on FOA and the remaining TRP+ transformant were tested for temperature sensitivity.

The pds1Δ construct was made by digesting pOC52 (Cohen Fix et al., 1996), a pUC19‐based plasmid carrying PDS1::HA and URA3, with BglII. This removed part of the promoter and the first 522 base pairs of the open reading frame. This construct was used to replace the wild‐type genomic PDS1 sequences by transplacement. The pds1Δ::URA3 disrupt was confirmed by Southern hybridization analysis.

Screen for synthetic lethality with sic1Δ

JD100 (Table I) was transformed with pSIC1U, a YCplac33 (Gietz and Sugino, 1988) ‐based plasmid carrying SIC1 on a EcoRI–SspI fragment and the URA3 gene. The object of the screen was to find mutants unable to grow in the absence of SIC1 and we used sensitivity to FOA to indicate the inability to lose pSIC1U. The JD100 pSIC1U strain was transformed with DNA from a yeast genomic library whose inserts had been randomly disrupted with lacZ::LEU2 as described (Burns et al., 1994) (this library was a generous gift from M.Snyder, Yale University). LEU+ transformants were then screened for sensitivity to FOA. To test if the FOA sensitivity was SIC1‐dependent, we provided transformants with a second copy of SIC1. FOA‐sensitive mutants were thus transformed with either pSIC1H or psic1ΔH. Both of these plasmids are ARS‐based plasmids derived from YCplac33 with CEN4 deleted. pSIC1H carries SIC1 and the HIS7 gene whilst psic1ΔH is as pSIC1H but with the SIC1 ORF deleted from the SpeI site at the N‐terminus. Transformed strains were again tested for FOA sensitivity. Only one strain was found from ∼80 000 LEU+ transformants that was FOA‐sensitive in a SIC1‐dependent manner. The site of the LEU2 insertion was cloned as described (Burns et al., 1994). Briefly, a bacterial/yeast shuttle vector, YIp5 (Struhl et al., 1979), was integrated by transformation into the LEU2 insertion site by targeting it to bacterial vector sequences contained within the LEU2 insert. The genomic sequences surrounding the LEU2 insertion were recovered by digesting then self‐ligating the genomic DNA and recovering the LEU2 bearing plasmid by transformation into Escherichia coli. Sequencing was done as described (Burns et al., 1994) using a fmol Cycle Sequencing Kit (Promega).

Cloning RSI1 and epitope tagging

A DNA fragment containing the RSI1 ORF and flanking promoter sequences was generated by PCR using the following oligonucleotides: ATTGAATTGTTTGCCTGACTAGTG and CGGGATCCTCTGATATACGTACGACC. The resulting PCR product was digested with BamHI and SpeI and cloned into a URA3 bearing integrative plasmid vector. The plasmid was cut at a unique NruI site and integrated upstream of RSI1. Genomic DNA from the transformant was digested with HindIII, self‐ligated and transformed into E.coli. The resulting URA3 plasmid also carries 4.5 kb of RSI1 sequences including the complete open reading frame and 1.4 kb of upstream sequences. This construct was able to rescue a lethal open reading frame deletion.

The triple Myc epitope tagged RSI1Myc3 was constructed as follows: the RSI1 ORF was amplified using PCR with the oligonucleotides RSI1‐5 and RSI1‐3 (5′ GGCTCTAGATTTGGTATTGTTCAAAATTTTC 3′ and 5′ GGCTCTAGAATGTCATTTCAGATTACCCCA 3′) and Pfu Taq polymerase (Stratagene). The PCR product was digested with XbaI and cloned into XbaI digested pRS306‐3Myc vector (unpublished). In frame insertion was confirmed by sequencing the resulting pRS306‐RSI1‐3Myc construct. For integration at the RSI1 locus, this construct was cut at a unique SpeI site 240 bp downstream from the initiating ATG.


Chromatin and spindles were visualized as described (Toyn and Johnston, 1994) and references therein.

Preparation of yeast crude extracts and protein analysis

Strains were grown to mid‐log phase, cells were pelleted, washed once with saline buffer and frozen. Cell pellets (2×108 cells) were resuspended in 150 μl breaking buffer containing 50 mM Tris–HCl pH 7.5, 150 mM NaCl, 15 mM MgCl2, 5 mM EDTA, 1% NP‐40, 10 μg/ml each protease inhibitors (leupeptin, pepstatin, chymotrypsin, aprotinin, antipain) and 1 volume of acid‐washed glass beads was added. For the co‐immunoprecipitation experiment, extracts were prepared as described (Lamb et al., 1994). Lysis was achieved by vortexing four times (30 s vortexing and 30 s cooling on ice). A volume of 100 μl breaking buffer was added and the lysate was separated from the beads by centrifugation. The lysate was collected and protein concentration was determined using the Bio‐Rad Bradford protein assay. For Western blot analysis, 25 μg of protein in cracking buffer (Printen and Sprague, 1994) was resolved by PAGE–SDS, transferred to Protran Nitrocellulose membrane (Schleicher & Schuell, Dassel, Germany) and detected using chemiluminescence detection (ECL, Amersham) with primary antibodies 12CA5 (HA) (NIMR, London) or 9E10 (MYC) (ICRF, London).

For the immunoprecipitation, 200 μg (H1 kinase assay) or 1 mg (co‐immunoprecipitation experiment) of protein extract in 50 μl breaking buffer was immunoprecipitated with 2 μg of anti‐HA or anti‐Myc antibody for 1 h at 4°C. Protein A/G beads (15 μl) (50% slurry) was added and incubated on a rotating wheel for 30 min at 4°C. For the H1 kinase assay, protein A/G beads–Immune complex were collected by centrifugation and washed three times with breaking buffer, twice with kinase buffer (25 mM MOPS pH 7.2, 10 mM MgCl2, 1 mM EGTA) and incubated for 20 min at room temperature with 15 μl kinase buffer containing 1 mg/ml histone H1 and 0.1 μl [γ‐32P]ATP (10 mCi/ml). The reaction was stopped by adding 1 volume of cracking buffer. Phosphorylated histone H1 were analysed by PAGE–SDS and autoradiography. For the co‐immunoprecipitation experiment, protein A/G beads–immune complex were collected, washed three times with breaking buffer and co‐precipitating proteins were released by boiling in cracking buffer. The supernatant was processed for Western blotting analysis as detailed above.


We wish to thank A.Page, P.Hieter, D.Koshland, W.Zachariae, and K.Nasmyth for reagents; S.Tournier for help with microscopy; L.Frenz for sequencing the RSI1Myc3 construct; N.Bouquin, J.Millar and E.Schwob for valuable suggestions and discussions. K.M.K. was supported by a fellowship from the Human Frontiers Science Program (LT225/95). D.F. was supported by a grant from Fondation pour la recherche Medicale.


View Abstract