Report

Phosphorylation of Sic1p by G1 Cdk Required for Its Degradation and Entry into S Phase

See allHide authors and affiliations

Science  17 Oct 1997:
Vol. 278, Issue 5337, pp. 455-460
DOI: 10.1126/science.278.5337.455

Abstract

G1 cyclin-dependent kinase (Cdk)–triggered degradation of the S-phase Cdk inhibitor Sic1p has been implicated in the transition from G1 to S phase in the cell cycle of budding yeast. A multidimensional electrospray mass spectrometry technique was used to map G1 Cdk phosphorylation sites in Sic1p both in vitro and in vivo. A Sic1p mutant lacking three Cdk phosphorylation sites did not serve as a substrate for Cdc34p-dependent ubiquitination in vitro, was stable in vivo, and blocked DNA replication. Moreover, purified phosphoSic1p was ubiquitinated in cyclin-depleted G1 extract, indicating that a primary function of G1 cyclins is to tag Sic1p for destruction. These data suggest a molecular model of how phosphorylation and proteolysis cooperate to bring about the G1/S transition in budding yeast.

Exit from the G1phase and the initiation of DNA synthesis in the cell cycle of budding yeast require the activities of CDC34, CDC4,CDC53, SKP1, one member of a set of G1 cyclin genes (CLN1-3), and one member of a set of B-type cyclin genes (CLB1-6) (1-4). A key insight into the molecular switch that drives cells from G1 to the S phase was the observation that the G1/S cell cycle arrest ofcdc34 ts, cdc4 ts,cdc53 ts, and skp1 tsmutants is suppressed by deletion of SIC1, which encodes an inhibitor of the protein kinase activity of a set of S phase–promoting Clb/Cdc28p complexes (1, 3). Biochemical reconstitution experiments have revealed that Cdc4p, Cdc53p, and Skp1p constitute a ubiquitin ligase complex (SCFCdc4) that collaborates with the ubiquitin-conjugating enzyme Cdc34p and the G1-specific Cdk Cln2p/Cdc28p (G1 Cdkl) to promote the ubiquitination of Sic1p (5, 6). Taken together, these data suggest that the destruction of Sic1p via the SCFCdc4 ubiquitination pathway might trigger S phase entry in wild-type budding yeast cells.

G1 Cdk activity is thought to be required for entry into S phase in all eukaryotic cells, but its exact targets have remained elusive. Sic1p is a potential key substrate of the budding yeast G1 Cdk, because Cln proteins are required for Sic1p degradation in vivo and ubiquitination in vitro (2,5), and Cln function is dispensible in cells lacking Sic1p (2, 7). Thus, Cln proteins might trigger Sic1p destruction and entry into S phase by modulating the activity of the Sic1p degradation machinery or by phosphorylating Sic1p directly, thereby allowing it to be recognized as a substrate for proteolysis.

There are nine (Ser or Thr) Pro candidate G1 Cdk phosphoacceptor sites in Sic1p, seven of which (Fig.1A) are clustered in the first 105 NH2-terminal residues. This segment of Sic1p contains sequences that are both necessary and sufficient to specify Cdc34p-dependent ubiquitination (5). To test whether Sic1p serves as a substrate for G1 Cdk, we mixed a purified maltose-binding protein–Sic1p chimera (MBP-Sic1pmycHis6; myc and His6 refer to a bipartite epitope tag appended to the COOH-terminus of the hybrid protein) (8) with Cln2p/Cdc28pHA/Cks1p complexes that were immunoaffinity-purified from insect cell lysates by virtue of the hemagglutinin (HA) tag appended to Cdc28p (9). MBP-Sic1pmycHis6 was efficiently phosphorylated by eluted G1 Cdk complexes (Fig. 1B) but not by control eluates prepared from Sf9 cells expressing only Cln2p or Cdc28p. Unfused MBP was not phosphorylated by G1 Cdk complexes despite its having two potential Cdk phosphorylation sites.

Figure 1

G1Cdk–dependent phosphorylation of Sic1p in vitro and in vivo. (A) Amino acid sequence of the NH2-terminal 105 residues of Sic1p (17). Potential Cdk phosphoacceptor sites are marked with an asterisk. This fragment is sufficient to serve as a substrate for Cln2p- and Cdc34p-dependent ubiquitination (5). The complementary COOH-terminal domain, which is dispensible for ubiquitination, has two additional consensus sites at Thr173 and Ser191. (B) Phosphorylation of MBP-Sic1pmycHis6 by G1 Cdk complexes. Sf9 insect cells were either singly or doubly infected with recombinant baculoviruses encoding Cdc28pHA or Cln2p, as indicated. Lysates prepared from infected (lanes 1 through 6) or uninfected (lane 7) Sf9 cells were activated by addition of Cks1p and adsorbed to 12CA5 beads (protein A–Sepharose beads cross-linked to an α-HA monoclonal antibody). The activated kinase complex was eluted with the HA peptide (28) and assayed for MBP-Sic1pmycHis6 protein kinase activity (25). MBP-Sic1pmycHis6 was purified from E. coli (29). (C) Coomassie blue–stained SDS-polyacrylamide gel of 0.2 and 0.4 μg of unmodified MBP-Sic1pmycHis6 (lanes 1 and 2) and phosphorylated MBP-Sic1pmycHis6 (lane 3) used for phosphopeptide mapping. MBP-Sic1pmycHis6 was phosphorylated with G1 Cdk complexes as described (28). (D) HPLC profile of typsin-digested phospho–MBP-Sic1pmycHis6. Trypsin digests were loaded on a reversed-phase column; and after the UV detector, the column eluent was split, with 90% collected as fractions and 10% sent directly to the mass spectrometer, which was operated in the negative-ion mode and optimized to detect m/z79 product ions (PO3 ) generated by collision-induced dissociation before the first quadrupole (27). Material absorbing at 214 nm (clear trace) was overlain with the m/z 79 product ion trace (gray trace) to identify phosphopeptide-containing fractions. (E) Sic1p is phosphorylated in vivo by a Cln-dependent kinase. An exponential culture of cdc34-2 cln1,2,3-Δ GAL-CLN3 (RJD768) cells at 24°C was transferred from phosphate-free YP-galactose (gal) medium to dextrose (dex) medium to effect Cln depletion and G1 arrest. Arrested cells were shifted to 37°C for 1 hour, after which the culture was split into two halves, which were harvested and resuspended in galactose or dextrose medium. After a further 1-hour incubation at 37°C, cells were labeled with carrier-free phosphate (1 mCi per 25 OD units). Cell extracts were prepared and immunoprecipitated with antiserum to Sic1p as described (30) , and immunoprecipitates were evaluated by autoradiography (IP, top panel) and immunoblotting with antiserum to Sic1p (IB, bottom panel).

To map the sites at which MBP-Sic1pmycHis6was phosphorylated by G1 Cdk complexes, we used a multidimensional electrospray mass spectrometry (ESMS) technique (10-12). Tryptic phosphopeptides derived from purified phosphorylated MBP-Sic1pmycHis6 (Fig.1C) were isolated by high-performance liquid chromatography (HPLC) followed by ESMS in the negative ion mode through single-ion monitoring of PO3 , which has a mass-to-charge ratio (m/z) of 79. The m/z 79 ion is a specific marker for phosphopeptides; the distribution of this ion (Fig. 1D, gray) is superimposed on the 214-nm ultraviolet (UV) chromatogram (Fig. 1D). HPLC fractions shown to contain phosphopeptides were reanalyzed by nanoelectrospray (nanoES) MS (13), with a precursor ion scan of m/z 79 to define the molecular weights of the phosphopeptides. Locations of modified residues within phosphopeptides were then determined by sequencing that used collision-induced dissociation tandem MS. All eight potential Cdk phosphorylation sites within MBP-Sic1pmycHis6[the ninth site at Thr2 was changed to Ala during cloning into the expression vectors (8)] appeared to be quantitatively phosphorylated in vitro (Table1). The presence of phosphate was directly confirmed by tandem MS–based sequencing for residues 5, 33, 45, 173, and 191 and was inferred (by mass and partial sequence data) for residues 69, 76, and 80. Two-dimensional thin-layer chromatography of tryptic peptides derived from in vitro–phosphorylated wild-type and Ser76 → Ala76 (S76A) mutant Sic1pHA confirmed that Ser76 was phosphorylated (14). Modification was also observed at the nonconsensus sites Thr22 and Ser23 through Ser25, which were phosphorylated poorly, and Thr48, which was phosphorylated quantitatively (Table 1). The long incubation of kinase with substrate may have compromised the specificity of the kinase reaction. However, no phosphopeptides were recovered from the MBP domain of the hybrid protein.

Table 1

Summary of ESMS analysis of phosphopeptides from in vivo– and in vitro–phosphorylated Sic1p (17).

View this table:

Mutant cdc34 ts cells arrest with high Cln/Cdc28p protein kinase activity (15) and accumulate phosphorylated Sic1p (1, 2). To test directly if phosphorylation of Sic1p in vivo was dependent on Cln/Cdc28p kinase activity, cdc34 ts cln1,2,3-Δ GAL-CLN3 cells were first arrested in G1 at 24°C by extinguishing CLN3 expression in dextrose medium, and then shifted to 37°C to inactivate Cdc34p. After 1 hour of incubation, the culture was split in two: one half was resuspended in galactose medium (CLN3 expression on) and the other in dextrose medium (CLN3 expression off). After a further incubation at 37°C for 1 hour, cells were labeled with32P. As shown in Fig. 1E, Sic1p isolated from cells expressing CLN3 contained at least 12 times more phosphate (as determined by PhosphorImager analysis) than did Sic1p isolated from cells depleted of CLN3. Thus, Sic1p phosphorylation in vivo was largely dependent on G1 Cdk activity.

To identify the sites at which Sic1p was phosphorylated in vivo, we purified Sic1pHAHis6from cdc34 ts cells, digested it with trypsin, and analyzed the resulting peptides by multidimensional ESMS (16). Precursor ion scanning and sequencing by nanoES tandem MS of phosphopeptide-containing fractions (Table 1) revealed the following. (i) Peptide 2–10 (numbers refer to the amino acids that the peptide spans) was phosphorylated quantitatively on Thr5. (ii) Several distinct phosphopeptides encompassing residues Thr33 and Thr45 each contained 1 to 2 mol of phosphate, and direct MS-based sequencing of singly phosphorylated 14–53 peptide (doubly phosphorylated peptide could not be recovered for sequencing) excluded all S and T residues as being phosphorylated, except for Thr33. (iii) A peptide containing Ser76 (54–84) was identified in singly, doubly, and triply phosphorylated forms, and direct MS-based sequencing of the singly phosphorylated peptide (once again, multiply phosphorylated peptides could not be recovered for sequencing) excluded all Ser and Thr residues as being phosphorylated, except for Thr75 and Ser76. Given the in vitro mapping data above and the mutagenesis results described below, we conclude that Ser76was phosphorylated in vivo.

The detection of six phosphates among the tryptic phosphopeptides (Table 1) agrees well with the predicted phosphorylation state of the intact starting material, as determined by positive ion ESMS: >88% of the Sic1pHAHis6purified from cdc34 ts cells had one to six phosphates per molecule (14). Although we obtained no direct evidence for in vivo phosphorylation of Thr22, Ser23 to Ser25, Thr45, Thr48, Ser69, and Ser80, a subset of these sites may have been modified on the doubly and triply phosphorylated 14–50 and 54–84 peptides that we failed to recover for sequencing. Moreover, to obtain sufficient quantities of Sic1pHAHis6 for analysis, the protein was transiently overexpressed from the GAL promoter. This overproduction may have taxed the ability of G1 Cdk complexes to maintain Sic1pHAHis6 in a maximally phosphorylated state.

To test their role in Sic1p destruction, we mutagenized phosphoacceptor sites of Sic1p detected by ESMS either singly or in various combinations. The mutants were then tested for their ability to serve as ubiquitination substrates with the use of a Cln2p- and Cdc34p-dependent in vitro ubiquitination assay (5). The effects of mutations in Thr173 and Ser191 were not examined, because an NH2-terminal 160–amino acid segment of Sic1p is both necessary and sufficient to direct its Cdc34p- and Cln2p-dependent ubiquitination (5). Whereas the single point mutants in which Thr2 (14), Ser69, and Ser80 were changed to Ala were as efficiently ubiquitinated as was wild-type Sic1pHAHis6, T2A/T5G, T33A, and S76A mutants (17) exhibited a modest decrease in ubiquitination, and the T45A mutant was poorly ubiquitinated (Fig. 2A). Because none of the point mutations eliminated Sic1pHAHis6 ubiquitination, we analyzed the effect of the T2A/T5G, T33A, T45A, and S76A mutations in various combinations. Double mutants (T2A/T5G was considered as a single mutant) invariably were poorer ubiquitination substrates than single mutants. All triple mutants analyzed, including one lacking three in vivo phosphoacceptor sites identified by ESMS (Sic1p-Δ3P: T5G, T33A, and S76A), had a severe ubiquitination defect (Fig. 2, A and B). Mutating these residues did not result in global misfolding of Sic1p-Δ3P, as assayed by its ability to bind Clb5pHA(Fig. 2B) and inhibit Clb5p-associated kinase activity (14). Although the Sic1p-Δ3P mutant was impaired in its ability to serve as a ubiquitination substrate, it was still phosphorylated in vitro and in vivo, although to a lesser extent than wild-type Sic1p (14). In sum, the mutational analysis suggested that Sic1p must be phosphorylated on a subset of sites by G1 Cdk complexes to serve as a substrate for the Cdc34p pathway, but that phosphorylation of no single site within this subset was either sufficient or absolutely necessary to target Sic1p for Cdc34p-dependent ubiquitination.

Figure 2

In vitro ubiquitination of Sic1p mutants lacking sites of phosphorylation. (A) Single and combinatorial point mutants of Sic1pHA were generated by PCR and transcribed and translated in wheat germ extract supplemented with [35S]methionine (31). Wild-type and mutant Sic1pHA translation products were assayed for ubiquitination as described (5,26). In vitro ubiquitination reactions were resolved on 10% SDS-polyacrylamide gels and quantitated with a PhosphorImager (Molecular Dynamics) (32). (B) Sic1p-Δ3P (Sic1pHA carrying the T2A/T5G, T33A, and S76A substitutions) was poorly ubiquitinated but nonetheless bound tightly to Clb5p/Cdc28p. (Left panel) Wild-type and mutant Sic1pHAwere assayed for ubiquitination as described in (A). Incubations proceeded for 0, 10, or 22 min in the presence or absence of glutathione S-transferase (GST)–Cln2p (5) as indicated. Ub-Sic1p, PP-Sic1p, and Sic1p refer to ubiquitinated, phosphorylated, and unmodified Sic1pHA, respectively. Wild-type and mutant Sic1pHA translation products were mixed with extracts prepared from either CLB5 orCLB5 HA cells, as indicated (right panel). One-third of the total input protein is depicted in lanes 9 and 12, and 12CA5 precipitates are depicted in lanes 10, 11, 13, and 14. Cell extracts and immunoprecipitates with 12CA5 were prepared as described (5).

The G1/S arrest phenotype of skp1 ts,cdc4 ts, cdc53 ts, andcdc34 ts mutants can be suppressed by deletingSIC1, which suggests that these mutants are unable to replicate because of accumulation of Sic1p, which in turn inhibits the S phase–promoting Clb5p/Cdc28p protein kinase (1,3). This interpretation is clouded by the fact that multiple proteins [including the G1 Cdk inhibitor Far1p (18)] accumulate in these mutants. Furthermore, the cell cycle defect of Cdc34p pathway mutants might be ameliorated by eliminating a major target such as Sic1p because the temperature-sensitive growth phenotype of these mutants is sensitive to the dosage of substrates (3, 18, 19). Thus, we tested whether Sic1p-Δ3P was stable in vivo and whether expression of nondegradable Sic1p-Δ3P would block DNA replication in wild-type cells. Regulated expression of Sic1p-Δ3P from theGAL promoter strongly inhibited cell proliferation in the presence of galactose (Fig. 3A). To evaluate the stability of Sic1p-Δ3P in vivo and the consequences of Sic1p-Δ3P expression on cell cycle progression, we arrestedGAL-SIC1-Δ3P and GAL-SIC1 strains in G1 phase with α-factor, transiently activated expression of the GAL promoter for 45 min, and released the induced cells into dextrose medium lacking α-factor (Fig. 3B).SIC1 transcripts expressed from the GAL promoter declined rapidly upon transfer to dextrose medium (14) . Immunoblotting with antiserum to Sic1p revealed that transient activation of the GAL promoter directed the accumulation of both Sic1p and Sic1p-Δ3P to amounts 1.5 times times that expressed from the SIC1 HAHis6 allele residing at the natural SIC1 locus (Fig. 3B). After release from cell cycle arrest, the amounts of Sic1p expressed from the GAL promoter and that of endogenous Sic1pHAHis6 declined in parallel at ∼45 min, concomitant with the initiation of S phase as judged by flow cytometry (Fig. 3C). In contrast, Sic1p-Δ3P persisted for at least 3 hours (Fig. 3B), and most of the cells containing theGAL-SIC1-Δ3P allele failed to replicate their DNA during this time (Fig. 3C). In addition, Clb5p/Cdc28p kinase inhibitory activity (1) persisted indefinitely in cells expressing Sic1p-Δ3P but disappeared at ∼45 min in cells expressing Sic1p (14).

Figure 3

Sic1p-Δ3P is stable in vivo and blocks DNA replication. (A) Constitutive expression ofSIC1Δ3P is toxic. RJD1025 (sector 1; W303 plus sic1::SIC1HAHis6::TRP1, ura3::GAL-SIC1::URA3), RJD1026 (sector 2; W303 plus sic1::SIC1HAHis6::TRP1, ura3::GAL-SIC1-Δ3P::URA3), and RJD730 (sector 3; W303 plus sic1::SIC1HAHis6::TRP1, ura3::GAL::URA3) cells were streaked onto YP dextrose (dex; left) or YP galactose (gal; right) plates and incubated at 30°C for 3 days. (B) RJD1025 and RJD1026 cells grown in YP raffinose medium were arrested with a-factor (50 ng/ml) for 2.5 hours, and SIC1 expression was induced transiently for 45 min by the addition of galactose to 2%. Cells were then transferred to fresh YP medium containing 2% dextrose, and portions were withdrawn every 15 min and processed for immunoblotting with antiserum to Sic1p, followed by secondary detection with [125]I-labeled donkey antibodies to a rabbit immunoglobulin F(ab')2 fragment. Antiserum to Sic1p detected both untagged Sic1p expressed from the GAL promoter andSic1pHAHis6 expressed from the natural SIC1promoter. Exp, a-f, and sic1Δ refer to exponentially growing cells, α-factor–arrested cells, and RJD1021sic1Δ mutants, respectively. (C) The same cells used in (B) were stained with propidium iodide and analyzed by flow cytometry to evaluate cellular DNA contents (33).

A distinctive phenotype of Cdc34p pathway mutants is the gradual appearance of multiply budded cells in arrested cultures. Pheromone-synchronized cells that transiently expressed Sic1p-Δ3P likewise exhibited a multiply budded phenotype: Three hours after release from α-factor arrest, 69% of Sic1p-Δ3P–expressing cells were singly budded and 26% were multiply budded. Although Sic1p-Δ3P expression mimics the major characteristics of Cdc34p/SCFCdc4 pathway mutants, it does not appear to inhibit the Cdc34p/SCFCdc4 pathway, because the bulk of wild-type Sic1pHAHis6 was degraded on schedule in Sic1p-Δ3P–expressing cells (Fig. 3B).

To address whether phosphorylation of Sic1p by G1 Cdk might be the primary cell cycle–regulated event that activates its destruction late in G1 phase, we performed a two-step in vitro ubiquitination reaction. Purified35S-labeled MBP-Sic1pmycHis6 was incubated with or without adenosine triphosphate (ATP) with G1 Cdk complexes immunoaffinity-purified from Sf9 cells. MBP-Sic1pmycHis6 was then re-isolated and added to yeast extract prepared from Cln-depleted cells. MBP-Sic1pmycHis6derived from kinase incubations lacking ATP was neither phosphorylated nor ubiquitinated in yeast extract lacking Cln2p (Fig. 4), which confirms that the repurified MBP-Sic1pmycHis6 was not contaminated with G1 Cdk complex. In contrast, phospho–MBP-Sic1pmycHis6 was ubiquitinated by yeast extract in a Cdc34p-dependent manner regardless of whether Cln2p was present. Ubiquitination of phospho–MBP-Sic1pmycHis6 was modestly enhanced, however, by the inclusion of Cln2p in the second step. Thus, whereas G1 Cdk might modestly augment the activity of Cdc34p or SCFCdc4, it is not absolutely required for the activity of this ubiquitination pathway.

Figure 4

Phosphorylation of Sic1p is sufficient to trigger Cdc34p-dependent ubiquitination. [35S]labeled MBP-Sic1pmycHis6 was isolated from E. coli (29) and incubated in the absence (lanes 1 and 2) or presence (lanes 3 through 7) of 1 mM ATP with G1 Cdk purified from baculovirus-infected Sf9 cells (stage I). After incubation, MBP-Sic1pmycHis6 was isolated on amylose resin and used as a ubiquitination substrate in stage II. Purified MBP-Sic1pmycHis6 was incubated with no further additions (lanes 1 and 3) or with Cdc34p- and Cln-depleted G1 yeast extract (lanes 2 and 4 through 7) supplemented with Cdc34p (lanes 2 and 6) or GST-Cln2p (lane 7) or both (lanes 4 and 5). Ub, ubiquitinated; PP, phosphorylated. All reactions were incubated at 25°C for 30 min except that shown in lane 4 (0 min). Ubiquitination-competent yeast extracts were prepared and ubiquitination reactions were assembled as described (5).

We envision the following sequence of events for the G1/S transition in budding yeast. In early G1cells, stable Sic1p quenches the activity of S phase–promoting Clb/Cdc28p complexes. In late G1 cells, G1 Cdk, which is insensitive to the inhibitory action of Sic1p, is activated and phosphorylates Sic1p on multiple residues. Such multisite phosphorylation may impart cooperativity to the G1/S transition (20). Whereas unmodified Sic1p is stable despite the presence of an active Cdc34p/SCFCdc4 pathway in G1cells, phospho-Sic1p is efficiently recognized by SCFCdc4 (6), ubiquitinated, and degraded by the 26S proteasome. Thus, G1Cdk–dependent phosphorylation of Sic1p is the key regulated event that sets in motion the transition from G1phase to S phase. An important question is, is S phase entry governed by a similar process in other eukaryotes? Whereas the Cdc4p homolog Pop1 is not required for DNA replication in fission yeast (21), progression into S phase in animal cells is restrained by the Cdc34p substrate p27Kip1 (22), and inhibition of Cdc34p activity blocks S phase in Xenopus extracts (23). Substrate phosphorylation has been implicated in the regulated destruction of multiple metazoan proteins, including key signaling proteins such as IκB and B-catenin (24). The extensive use of substrate phosphorylation to regulate protein stability may foreshadow a broadly conserved mechanism for regulated protein destruction in eukaryotic cells (6).

REFERENCES AND NOTES

View Abstract

Navigate This Article