Eukaryotic cell proliferation requires proper coordination of growth with key cell cycle events such as the G1–S cell cycle transition, which is also coined START in yeast [1]. Initiation of and commitment to a new round of cell division at START depends on the G1 cyclin-dependent protein kinase (Cln-CDK; Cln1/2-Cdc28) that phosphorylates critical target proteins [2]. One of these is the B-type cyclin (Clb)-CDK inhibitor (CDKI) Sic1 [3], which, following its CDK-dependent phosphorylation at multiple residues [4], associates with the F-box protein Cdc4 of the SCFCdc4 ubiquitin ligase complex that directs it for ubiquitin-dependent proteolysis (for reviews, see Deshaies et al. [5] and Barberis [6]). The phosphorylation of Sic1 in late G1 follows a temporal and spatial order that involves an initial Cln-CDK-dependent phosphorylation of Thr5 and Thr33 within its N-terminal region [7]. This priming event creates docking sites for the phospho-acceptor subunit Cks1 of Cln-CDK-Cks1 and emerging Clb5/6-CDK-Cks1 complexes, which subsequently phosphorylate critical pairs of neighboring residues (Thr45/Thr48 and Ser76/Ser80; also coined phosphodegrons [8]) that engage Cdc4 to mark Sic1 for destruction [7, 9, 10]. Both the synergistic effect between Cln- and Clb-CDKs in semi-processive, multisite-phosphorylation of Sic1 and the positive feedback loop involving liberated Clb5/6-CDK-Cks1 complexes render the destruction of Sic1 switch-like and are therefore key for launching the G1–S cell cycle transition in yeast [11].

Consistent with the important role of Sic1 degradation for G1–S transition, this process is tightly coupled to growth- and nutrient-related signals that are mediated by TORC1, the central controller of growth in eukaryotes [12, 13]. Accordingly, TORC1 stimulates the formation of Cln-CDK complexes that prime Sic1 for multisite-phosphorylation by activating the transcription and translation of G1 cyclins (Cln1–3) [14, 15]. In addition, TORC1 also antagonizes the phosphorylation of a specific residue (Thr173) within the C-terminal part of Sic1 that critically contributes to its stabilization [16]. TORC1 controls the phosphorylation status of this residue via two circuits, of which one involves the indirect inhibition of the mitogen-activated protein kinase Slt2/Mpk1 that phosphorylates Thr173 in Sic1 [17]. In parallel, TORC1 warrants effective dephosphorylation of pThr173 in Sic1 by inhibiting a signaling branch, which restrains the activity of the Sic1-pThr173-targeting PP2ACdc55 protein phosphatase [17]. The core of this highly conserved signaling branch consists of the greatwall kinase (Gwl) Rim15, which phosphorylates a residue within the endosulfines Igo1/2 to convert them to potent inhibitors of PP2ACdc55 (or B55-PP2A in higher eukaryotes) [1820]. TORC1 contributes to the silencing of Rim15 by promoting indirectly the phosphorylation of two residues (that is, Ser1061 and Thr1075), which cooperatively mediate tandem 14-3-3 protein binding to sequestrate inactive Rim15 in the cytoplasm [2124]. Downregulation of TORC1 therefore results in the release of Rim15 from its cytoplasmic anchor, enabling it to accumulate in the nucleus to activate the PP2ACdc55 inhibitory Igo1/2 proteins. Combined with the coordinated activation of Mpk1, the latter process ensures the phosphorylation of Sic1 at Thr173 under conditions that inhibit TORC1 such as nutrient limitation or rapamycin treatment. How this particular phosphorylation event stabilizes Sic1 remains currently mysterious.

Here we show that the phosphorylation of Thr173 in Sic1 creates a docking site for the phospho-acceptor subunit Cks1 that tethers Clb5-CDK-Cks1 complexes away from the respective N-terminal docking sites on Sic1 where they could otherwise favor multi-phosphorylation and degradation of Sic1. The spatial sequestration of the pThr173-associated Clb5-CDK-Cks1 complexes also promotes their inhibition via the additional association of Clb5 with the RXL motif within the CDK-inhibitory C-terminus (RXLCDKI) of Sic1. Thus TORC1 inactivation (and consequently Mpk1 and Gwl pathway activation) triggers the phosphorylation of Thr173 in Sic1, thereby converting Sic1 from a target to a potent inhibitor of Clb5-CDK-Cks1 complexes. Interestingly, this switch of Sic1 also occurs in proliferating cells in G1 and is rapidly reversed at START as a result of the increase of Cln-CDK, which, cooperatively with TORC1, inhibits Rim15 by stimulating its confinement to the cytoplasm. Thus, our data pinpoint the Gwl pathway as a central node that interprets both growth and cell cycle cues to coordinate G1–S transition.

Results and Discussion

Phosphorylation of Sic1-Thr173 is essential for Sic1 to escape CDK-mediated, N-terminal multi-phosphorylation upon TORC1 inactivation

Previous studies suggested that the phosphorylation of Thr173 in Sic1 may compromise its capacity to interact with components of the SCFCdc4 complex [16, 25]. As the respective interaction is mainly mediated by a set of defined N-terminal phosphodegron sites [4, 26], we considered the possibility that pThr173 may, rather than directly interfering with SCFCdc4 binding, indirectly affect the phosphorylation status of the N-terminal residues in Sic1. To address this possibility, we first compared on phos-tag gels the phosphorylation levels of Sic1 and Sic1T173A from rapamycin-treated cdc4-2ts strains, which were preincubated at the restrictive temperature (that is, 37 °C) to inactivate Cdc4 and consequently block SCFCdc4-mediated degradation of the Sic1 variants. Under these conditions, the slowly migrating, hyperphosphorylated wild-type (WT) Sic1, but not the Sic1T173A variant, rapidly collapsed into several fast-migrating, hypophosphorylated isoforms (Figure 1a). Inhibition of Cdc28 during the rapamycin treatment (using the ATP analog 1NM-PP1 and strains carrying the analog-sensitive Cdc28as allele) prevented the observed sustained hyperphosphorylation of Sic1T173A, indicating that Cln- and/or Clb-CDK complexes are not properly inhibited in Sic1T173A-expressing cells (Figure 1b). This was further supported by our finding that both the Sic1-Thr5 and Sic1-Thr33 residues, which are preferred Cln2-CDK and Clb5-CDK targets in Sic1, respectively [7], were swiftly dephosphorylated in rapamycin-treated cdc4-2ts cells, while this effect was significantly delayed in cdc4-2ts sic1T173A cells (Figure 1c). Using strains that carry the Cdc28as allele and 1NM-PP1 to inactivate Cdc28as during the rapamycin treatment as above, we confirmed that the delayed response in pThr5 and pThr33 dephosphorylation in rapamycin-treated, Sic1T173A-expressing cells was entirely dependent on Cdc28 (Figure 1d). Collectively, these results indicate that the instability of the Sic1T173A allele results primarily from its sustained phosphorylation by Cln- and/or Clb-CDK.

Figure 1
figure 1

Sic1-Thr173 mediates protection from cyclin-dependent kinase (CDK)-dependent multi-phosphorylation upon target of rapamycin complex 1 (TORC1) inactivation. (a) Sic1T173A is hyperphosphorylated in rapamycin-treated cdc4-2ts cells. cdc4-2ts and cdc4-2ts sic1T173A mutant cells were incubated for 3 h at 37 °C, then treated (RAP, 15–60 min), or not (RAP, 0 min), with rapamycin. Sic1 and Sic1T173A were analyzed using polyclonal anti-Sic1 antibodies and phos-tag phosphate affinity gel electrophoresis. The asterisk denotes unspecific bands that were also present in sic1Δ control cells. (b) Inactivation of Cdc28 prevents hyperphosphorylation of Sic1T173A in rapamycin-treated cdc4-2ts cells. Sic1 and Sic1T173A were analyzed as in a using cdc4-2ts cdc28as and cdc4-2ts sic1T173A cdc28as strains. 1NM-PP was added simultaneously with rapamycin. (c) Sic1T173A is hyperphosphorylated at Thr5 and Thr33 in rapamycin-treated cdc4-2ts cells. Strains and conditions were as in a. Sic1-pThr5 and Sic1-pThr33 levels were determined by immunoblot analysis using phospho-specific anti-Sic1-pThr5 and anti-Sic1-pThr33 antibodies, respectively (see Supplementary Figure S1). Sic1 was probed with polyclonal anti-Sic1 antibodies. Notably, hyperphosphorylation of Sic1T173A causes it to migrate over a larger region within standard 9% sodium dodecyl sulfate (SDS) gels, which is why the respective protein bands appear less sharp and intense when probed with polyclonal anti-Sic1 antibodies (third panel from top; lanes 6–10). (d) Inactivation of Cdc28 prevents hyperphosphorylation of Thr5 and Thr33 in Sic1T173A in rapamycin-treated cdc4-2ts cells. Strains and conditions were as in b. Sic1-pThr5, Sic1-pThr33 and Sic1 were determined as in c. As in c, hyperphophorylated Sic1T173 at time point 0 (third panel from the top; lane 6) migrates over a larger region within the SDS gel. Inactivation of Cdc28as (lanes 7–10), however, results in a rapid dephosphorylation of Sic1T173A, which is why the respective protein bands collapse into a sharper and intense band under these conditions. In all panels, Adh1 levels served as loading control.

The Sic1T173A allele exhibits normal specificity and intrinsic CDKI activity in vitro

To elucidate why rapamycin-treated sic1T173A cells exhibit enhanced Cln- and/or Clb-CDK activity, we next analyzed whether the mutation of Thr173 in Sic1 indirectly affected the levels of G1- (Cln1/2) or S-phase (Clb5) cyclins. As shown in Supplementary Figure S2, Cln1, Cln2 and Clb5 were all eliminated with similar kinetics in both rapamycin-treated cdc4-2ts and cdc4-2ts sic1T173A cells. Expression of the Sic1T173A allele does therefore not alter the dynamic regulation of G1/S cyclin levels in rapamycin-treated cells. Because Sic1T173A-expressing cells exhibited sustained phosphorylation of a preferred Cln-CDK target (that is, Thr5 in Sic1; see above), we next speculated that the phosphorylation of Thr173 in Sic1 may expand its inhibitory specificity in rapamycin-treated cells to include Cln-CDKs. However, this cannot be the case as Cln2 did not bind Sic1 or Sic1T173A under any conditions tested (Figure 2a), while Clb5 interacted readily with both forms in exponentially growing and rapamycin-treated cells (Figure 2b). Thus neither the phosphorylation (in rapamycin-treated WT cells) nor the mutation of Thr173 appeared to modify the specificity of Sic1 for binding Clb5-CDK in vivo. We therefore addressed yet another possibility, namely that the Thr173 residue may be important for the capacity of Sic1 to function as CDKI. To this end, we used bacterially expressed Sic1 and Sic1T173A that were exposed to the Thr173-targeting Mpk1 kinase prior to their use in CDKI assays. These revealed that both Sic1-pThr173 and Sic1T173A inhibited Clb5-Cdc28, but not Cln2-Cdc28, equally well in a concentration-dependent manner in vitro (Figure 2c). Unlike the Sic1R262A/L264A allele, which carries specific mutations in the RXLCDKI motif that likely tethers the hydrophobic patch (hp) docking site in Clb5 (Figure 2c; see also below) [2729], the recombinant Sic1T173A mutant is therefore not intrinsically deficient in Clb5-CDK inhibition when assayed in vitro.

Figure 2
figure 2

Sic1-pThr173 acts as a docking site for Cks1. (a, b) Sic1 and Sic1T173A bind Clb5 but not Cln2. HA3-tagged Cln2 (a) or Clb5 (b) were immunoprecipitated (IPed) from extracts of exponentially growing wild-type (WT; SIC1) and sic1T173A cells that were treated (RAP, +), or not (RAP, −), for 1 h with rapamycin. Cell lysates (Input) and anti-HA immunoprecipitates (IP: anti-HA) were analyzed by immunoblotting with anti-HA and anti-Sic1 antibodies. See also Supplementary Figure S3 for additional control IPs. (c) Sic1 and Sic1T173A inhibit Clb5-CDK, but not Cln2-CDK, in vitro. Cln2-HA3-CDK and Clb5-HA3-CDK immunocomplexes from exponentially growing yeast cells were used for in vitro kinase assays in which the bacterially purified, N-terminal part of Sic1 (GST-Sic11-100) served as substrate. Bacterially purified GST-Sic1, GST-Sic1T173A or GST-Sic1R262A/L264A (that were subjected to in vitro phosphorylation by Mpk1 before use) were titrated into the reactions at the indicated concentrations to assess their cyclin-dependent kinase (CDK) inhibitory activities. CDK activity was assessed by probing the levels of pThr5 in GST-Sic11-100 using anti-Sic1-pThr5 antibodies. GST-Sic11-100 levels were verified by immunoblot analysis using anti-GST antibodies. (d) Rapamycin-treated Sic1T173A-expressing cells are not defective in the clearance of Cks1. Exponentially growing WT and sic1T173A cells expressing genomically tagged Cks1-HA3 were treated (RAP, 20–120 min), or not (RAP, 0 min), with rapamycin. Cks1-HA3 levels were determined by immunoblot analysis (using monoclonal anti-HA antibodies). Ponceau staining served as loading control. (e, f) Cks1 binds Sic1, but not Sic1T173A, preferably in rapamycin-treated cells. Genomically HA3-tagged Cks1 was IPed from extracts of exponentially growing (RAP, −) or rapamycin-treated (1 h; RAP, +) WT or sic1T173A cells (e) or from respective WT cells that expressed plasmid-encoded myc13-tagged Sic1 fragments (Sic1150-285-myc13) carrying (T173A), or not (WT), the Thr173-to-Ala mutation (f). Cell lysates (Input) and anti-HA immunoprecipitates (IP: anti-HA) were analyzed by immunoblotting with anti-HA and anti-Sic1 (e) or anti-myc (f) antibodies.

Sic1-pThr173 acts as a docking site for the CDK phospho-adaptor subunit Cks1

An alternative model we entertained to explain the higher Clb/Cln-CDK activity in Sic1T173A expressing cells is that the phosphorylation of Thr173 in Sic1 indirectly contributes to CDK inhibition by sequestering a protein that otherwise favors the capacity of cells to stimulate CDKs. To identify such a protein(s), we purified Sic1 and Sic1T173A from rapamycin-treated cells and analyzed the coprecipitating proteins by mass spectrometry. Remarkably, when applying stringent criteria, this experiment pinpointed Cks1 as the only protein that interacted specifically and more robustly with Sic1 when compared with Sic1T173A. To exclude that the presumably lower affinity between Cks1 and Sic1T173A is simply due to lower Cks1 levels in sic1T173A strains, we compared the Cks1 expression levels in WT and sic1T173A cells. Cks1 levels were equally high in exponentially growing cells and rapidly dropped with similar kinetics in rapamycin-treated cells of both strains (Figure 2d). Directed control co-immunoprecipitation (co-IP) analyses, in which we pulled down HA3-tagged Cks1 and assessed the levels of associated Sic1 or Sic1T173A, also confirmed our finding that Cks1 associates with Sic1, but not with Sic1T173A, in rapamycin-treated cells (Figure 2e). These results indicate that the CDK phospho-adaptor subunit Cks1 specifically docks to pThr173 within Sic1. However, the interpretation of these experiments may be subjected to the following caveats, namely (i) that the lysates from rapamycin-treated sic1T173A cells contained much lower amounts of the instable Sic1T173A than the ones of WT cells containing the stable Sic1 protein, and (ii) that our co-IP experiments may, to some extent, probe the association of Cks1 with the N-terminal Cks1-docking sites in Sic1/Sic1T173A. To address both issues at once, we carried out additional Cks1 co-IP experiments using myc13-tagged fragments of Sic1 (Sic1150-285), which lacked the N-terminal phosphodegron and Cks1-docking sites that drive Sic1 degradation. As expected, these Sic1 fragments, whether they contained or not the Thr173-to-Ala mutation, were equally stable under all conditions tested. The respective co-IP experiments with Cks1 confirmed the ones in which full-length variants of Sic1 were expressed (Figure 2e and f). Together with the fact that the pThr173 and its neighboring residues (Pro-Gly-Thr173-Pro) perfectly match the Cks1-binding consensus motif (F/I/L/P/V/W/Y)XTP [10], our results therefore assign pThr173 a role as bona fide Cks1-docking site.

The C-terminal RXL motif and the pThr173 residue coordinately bind Clb5-CDK-Cks1

Our co-IP studies presented in Figure 2e and f indicated that Cks1 also associates with Sic1, in a Thr173-dependent manner, in exponentially growing cells. This is not surprising, as low levels of Sic1-pThr173 molecules can also be detected in these cells [17] (see also below). Importantly, however, rapamycin treatment strongly increased the relative amount of Sic1 that was precipitated with Cks1 (Figure 2e and f), suggesting that Cks1 molecules, if not degraded, get rapidly trapped by Sic1-pThr173 in rapamycin-treated cells. This mechanism likely contributes to the swift clearance of free Cks1 molecules and may potentially contribute to a more robust inhibition of Clb5-CDK-Cks1 complexes due to their insertion between the Clb5-tethering RXLCDKI domain and the pThr173 Cks1-docking site in Sic1. Several additional experiments support this model. For instance, mutation of the RXLCDKI motif in Sic1 strongly weakened the association between Clb5 and this Sic1 mutant protein. This resulted in the release of Clb5-CDK from Sic1 inhibition, as Clb5-CDK immune complexes isolated from rapamycin-treated Sic1R262A/L264A-expressing cells (unlike the ones from WT cells) exhibited largely unrestrained Clb5-CDK activity in vitro (Figure 3a). In addition, even though Clb5 pulled down Sic1 and Sic1T173A with equal efficiency from extracts of rapamycin-treated cells (see also Figure 2b), only Clb5-CDK-Sic1T173A immune complexes exhibited measurable levels of unrestrained Clb5-CDK activity (Figure 3a). These data show that the capacity of Sic1 to dock Cks1 on pThr173 grants more stringent inhibition of Clb5-CDK in vivo. This also explains why recombinant Sic1 and Sic1T173A were found to be equally effective CDKIs when assayed in vitro in the absence of Cks1 (Figure 2c). Lastly, as expected if both Sic1R262A/L264A and Sic1T173A were defective in separate mechanisms to adequately inhibit of Clb5-CDK-Cks1 complexes in vivo (and hence were prone themselves to continued flagging for degradation by the latter complexes), both Sic1 variants were unstable in an additive manner when assayed in rapamycin-treated cells (Figure 3b). Thus the robust inhibition of Clb5-CDK-Cks1 complexes in vivo requires the coordinate binding of Clb5 to RXLCDKI and Cks1 to pThr173 within Sic1 (Figure 3c).

Figure 3
figure 3

The C-terminal RXL-motif and pThr173 coordinately bind Clb5-CDK-Cks1. (a) Clb5-CDK immune complexes from rapamycin-treated sic1T173A and sic1R262A/L264A cells, but not the ones from respective wild-type (WT) cells, exhibit cyclin-dependent kinase (CDK) activity. Plasmid-expressed Clb5-HA3 was IPed from extracts of rapamycin-treated (1 h) cells that expressed genomically tagged Sic1-myc13-, Sic1T173A-myc13- or Sic1R262A/L264A-myc13. Cell lysates (Input) and anti-HA immunoprecipitates (IP: anti-HA) were analyzed by immunoblotting with anti-HA and anti-myc antibodies. Clb5-CDK immune complexes were probed for CDK activity as in Figure 2c. (b) Combined loss of Cln1/2 and Clb5 suppresses the instability of Sic1T173A in rapamycin-treated cells. The levels of Sic1 variants were determined in exponentially growing (RAP, 0 h) and rapamycin-treated (1–4 h) strains with the indicated genotypes using polyclonal anti-Sic1 antibodies. Adh1 levels served as loading controls. (c) Model for the target of rapamycin complex 1 (TORC1)-/nutrient-regulated switch of Sic1 from a target to an inhibitor of Clb5-CDK-Cks1. In cells growing on rich medium, Cks1-CDK-Clb5 is found to dock onto Sic1 via Cks1 binding to pThr5 and pThr33 (only the latter binding is shown) and via Clb5 binding to RXL motifs (only RXL2 of four RXL motifs is shown). This allows phosphorylation of the neighboring phosphodegron clusters at Thr45/48 and Ser69/76/80 [7, 9, 10] and subsequent degradation of Sic1. Upon TORC1 inactivation or nutrient starvation, a distinct docking configuration of Cks1-CDK-Clb5 onto Sic1 happens, involving binding of Cks1 to Thr173 and of Clb5 to a C-terminal RXL motif within the CDKI domain. This precludes further phosphorylation of Sic1 by Cks1-CDK-Clb5 and favors its stabilization. For further details, see text. (d) Cks1-Clb5 complexes are released from CDKI-binding in rapamycin-treated sic1T173A cells. Genomically tagged Cks1-HA3 was IPed from extracts of rapamycin-treated (30 min) WT (SIC1) or sic1T173A cells that co-expressed genomically tagged Clb5-myc13. Cell lysates (Input) and anti-HA immunoprecipitates (IP: anti-HA) were analyzed by immunoblotting with anti-myc, anti-HA, and anti-Sic1 antibodies. (e) Rapamycin-treated sic1T173A cells exhibit higher levels of Cks1-Cln2 complexes. Genomically tagged Cln2-myc13 was IPed from extracts of rapamycin-treated (30 min) WT (SIC1) or sic1T173A cells that co-expressed genomically-tagged Cks1-HA3. Cell lysates (Input) and anti-HA immunoprecipitates (IP: anti-HA) were analyzed by immunoblotting with anti-myc and anti-HA antibodies.

Our results imply that the sequestration of Cks1 within Sic1-associated Clb5-CDK-Cks1 complexes is part of a program that prevents Cks1 from directing free Cln/Clb-CDKs toward substrates (including Sic1 itself) when TORC1 is low. Our co-IP analyses using extracts from rapamycin-treated cells endorsed this hypothesis, as Cks1 interacted well with Clb5 and Sic1 in WT cells, while Cks1 only bound Clb5 but not Sic1T173A in sic1T173A cells (Figure 3d). Under the same conditions, Cks1 also increasingly bound Cln2 in sic1T173A cells when compared with WT cells (Figure 3e). Finally, our genetic experiments revealed that only the combined loss of Clb5 and Cln1/2, but not their individual loss, fully suppressed the instability of the Sic1T173A allele in rapamycin-treated cells. The B-type cyclin-CDKI Sic1 therefore has a hitherto undescribed role in limiting the activity of Cln-CDKs by sequestering their regulatory adaptor Cks1 upon TORC1 inactivation.

The Gwl Rim15 integrates signals from both TORC1 and Cln-CDK to control Sic1

To substantiate our model in which TORC1 regulates the binding between Cks1 and its docking site at Thr173 in Sic1 via Mpk1 and the Gwl pathway (Rim15-Igo1/2-PP2ACdc55), we next quantified the relative levels of Sic1 that were precipitated with Cks1 in exponentially growing and rapamycin-treated WT, mpk1Δ and cdc55Δ cells. As expected, inhibition of TORC1 strongly stimulated the relative affinity between Sic1 and Cks1 (see also Figure 2e and f), and this depended on the presence of Mpk1 and the Thr173 residue in Sic1, while loss of Cdc55 strengthened the respective interaction (Figure 4a). Thus, by inhibiting Mpk1 and the Gwl pathway, TORC1 antagonizes Sic-Thr173 phosphorylation, and, as a consequence, the sequestration of Cks1 by Sic1 in exponentially growing cells. Nevertheless, exponentially growing cells also contain low levels of Sic1-pThr173 (see above) and the mutation of Thr173 in Sic1 per se, whether combined or not with cdc55Δ, measurably weakened the affinity between Sic1 and Cks1 in exponentially growing cells (Figure 4a). We considered it therefore possible that an additional mechanism exists, which fine-tunes the Sic1-Thr173 phosphorylation levels in proliferating cells. To analyze whether this presumed mechanism operates at different stages of cycling cells, we determined the levels of both Sic1 and Sic1-pThr173 in synchronized cells that were released from an α-factor arrest. Surprisingly, the phosphorylation of Sic1-Thr173 was clearly cell cycle regulated. It correlated closely with the relative amount of Sic1, being high in G1 cells and low in cells that have entered S-phase, as judged by the sequential waves of Cln2 and Clb5 in these cells (Figure 4b). Notably, in this context, both phosphorylation analyses of the Clb5-CDK target Sld2 and standard flow cytometric experiments previously revealed that, when released from an α-factor arrest, Sic1T173A-expressing cells enter S phase significantly earlier than WT cells (Figure 5b and e in Adrover et al. [30]). Combined with our finding that the Cln-/Clb5/6-CDK activity (assayed by monitoring the relative phosphorylation of Thr5 in Sic1/Sic1T173A) raises more rapidly and subsequently oscillates at a significantly higher level in sic1T173A than in WT cells when released from an α-factor arrest (Supplementary Figure S4), these data indicate that the phosphorylation of Thr173 in Sic1 is also physiologically relevant in proliferating cells.

Figure 4
figure 4

The Gwl Rim15 integrates target of rapamycin complex 1 (TORC1) and Cln-CDK signals to control the G1–S transition. (a) The Gwl pathway controls the association between Cks1 and Sic1. Genomically HA3-tagged Cks1 was IPed from extracts of exponentially growing cells (genotypes indicated) that were treated (RAP, +), or not (RAP, −), for 2 h with rapamycin. Bars denote the relative amount of Sic1 that co-precipitated with Cks1 (see also Figure 2e for details). The respective ratios were all normalized to the value in exponentially growing wild-type (WT) cells (n=3; ±s.d.). (b) Rim15 activity is cell cycle regulated. For synchronization, exponentially growing WT cells expressing genomically tagged Sic1-myc13 were treated for 2 h with α-factor (5 μg ml−1). Following α-factor release, samples were collected at the indicated time points. Sic1-myc13-pThr173, Sic1-myc13, Igo1-pSer64, Igo1, Cln2 and Clb5 levels were detected by immunoblot analysis using anti-pThr173, anti-myc, anti-Igo1-pSer64, anti-Igo1, anti-Cln2 and anti-Clb5 antibodies, respectively. Notably, we have used our anti-Sic1-pThr173 and Igo1-pSer64 antibodies in previous studies in which we have also shown that both antibodies recognize the respective phospho-residues in Sic1 (pThr173) and Igo1 (Ser64) with high specificity [17, 31]. The asterisk denotes an unspecific band. (c) Cdc28 antagonizes Igo1-Ser64 phosphorylation. Exponentially growing (0 min) cdc28as and rim15Δ cdc28as cells were treated for the times indicated with 1NM-PP1 (0.5 μM) to inactivate Cdc28as. The levels of both Igo1 and the phosphorylation of the Rim15-target residue Ser64 in Igo1 were determined by immunoblot analysis using anti-Igo1 and phospho-specific anti-Igo1-pSer64 antibodies. To assay TORC1 activity, the levels of both Sch9 and the phosphorylation of the TORC1-target residue Thr737 in Sch9 were determined by immunoblot analysis using anti Sch9 and phospho-specific anti-Sch9-pThr737 antibodies. (d) rim15Δ cdc28as cells expressing plasmid-encoded kinase-inactive GFP-Rim15KD and the genomically tagged nuclear marker Hhf2-tDimer were grown exponentially. They were left untreated (EXP), or treated for 1 h with 1NM-PP1, rapamycin (RAP) or 1NM-PP1 combined with rapamycin. The subcellular distribution of green fluorescent protein (GFP)-Rim15KD and Hhf2-tDimer was analyzed by fluorescence microscopy. The numbers indicate the percentage of cells (n>100) with clear nuclear accumulation of the GFP-fusion protein. (e) Exponentially growing rim15Δ cells expressing plasmid-encoded GFP-Rim15KD were synchronized by α-factor treatment as in (b) and analyzed by fluorescence microscopy at the times indicated following α-factor release. (f) Cln2-CDK but not Clb5-CDK complexes interact with Rim15 in exponentially growing cells. Plasmid-encoded GST-Rim15 was pulled down from extracts of untreated (0 h) and rapamycin-treated (RAP; 1 h) rim15Δ cells that co-expressed Clb5-HA3, Cln2-HA3 or Lst7-HA3. Cells carrying an empty vector were used as control. Cell lysates (Input) and GST pulldown fractions were analyzed by immunoblotting with anti-HA and anti-GST antibodies. (g) Cln-CDK phosphorylates Rim15 in vitro. Cln2-HA3-CDK and Clb5-HA3-CDK immunoprecipitated from exponentially growing yeast cells were tested for their capacity to phosphorylate in vitro a bacterially purified GST-Rim15 fragment encompassing amino acids 944–1149 of Rim15 (Rim15KI, for kinase insert), GST alone or the GST-Sic11-100 fragment (see Figure 2c). Sypro-Ruby staining of the gel (left panel) and the corresponding autoradiograph [32P] are shown. GST-Sic11-100 served as positive control for both Clb5-/Cln2-CDK activities. Asterisks denote respective autophosphorylation bands. (h) Inactivation of Cdc28as correlates with hypophosphorylation of Rim15-pThr1075 in vivo. Cells with the indicated genotypes were grown exponentially and then treated (+), or not (−), for 1 h with 1NM-PP1. The levels of Rim15-myc13 and the phosphorylation of the residue Thr1075 in Rim15 were determined via immunoblot analysis using anti-myc and phospho-specific anti-Rim15-pThr1075 antibodies, respectively. Cells expressing Rim15T1075A-myc13 were used as control.

Interestingly, the rapid disappearance of phosphorylated Sic1-pThr173 species (when cells were released from the α-factor arrest) temporally coincided with a significant decrease in the level of the active version of the PP2ACdc55-inhibitory endosulfine Igo1 (that is, Igo1-pSer64) [20]. Because the latter represents a direct readout for the activity of the Gwl Rim15 in vivo [31], this indicated that Rim15 may be inhibited by CDK at the onset of the G1–S transition phase and be released from this inhibition when CDK is low (that is, in late mitosis and G1). Consistent with this idea, inactivation of Cdc28as with 1NM-PP1 stimulated the activity of Rim15, which was not due to indirect effects on TORC1, because the phosphorylation of Thr737 in Sch9, a proxy for TORC1 activity, remained largely constant throughout the experiment (Figure 4c). Notably, the activation of Rim15 following CDK inhibition was paralleled by an increased appearance of Rim15KD-GFP in the nuclei of exponentially growing cells (Figure 4d), which also fits well with the model that Rim15 activation requires its transfer into the nucleus [24]. Finally, in line with the implicit assumption that CDK inhibits Rim15 by preventing its accumulation in the nucleus, the dynamics of Rim15 inactivation correlated well with its relocation from the nucleus to the cytoplasm when cells were released from an α-factor arrest (Figure 4b and e).

The nucleocytoplasmic distribution of Rim15 is intimately controlled by the phosphorylation status of two residues (Ser1061 and Thr1075) within adjacent classical 14-3-3-binding domains that mediate cooperative association with dimeric 14-3-3 proteins in the cytoplasm [32]. Both residues are targeted by known protein kinases, namely, the TORC1 effector Sch9 (Rim15-Ser1061) and the phosphate-responsive cyclin-CDK Pho80-Pho85 module (Rim15-Thr1075) [21, 23, 24]. Because both CDKs, Pho85 and Cdc28, are known to redundantly phosphorylate the same CDK consensus sites in some of their effector proteins [33, 34], we speculated that Cdc28 could, in a most simple model, modulate the subcellular localization of Rim15 by regulating the phosphorylation of the previously identified Rim15-Thr1075 residue together with Pho85. Supporting this hypothesis, we found Rim15 to specifically interact with Cln2, but not with Clb5 or the control protein Lst7, when probed in exponentially growing cells (Figure 4f). Moreover, Cln2-Cdc28, but not Clb5-Cdc28, was able to phosphorylate Rim15 in vitro (Figure 4g), and inactivation of Cdc28as with 1NM-PP1, similar to the loss of Pho85 [24], significantly reduced the Rim15-pThr1075 levels in vivo (Figure 4h). Therefore, Cln-CDK antagonizes Rim15, in parallel to or possibly downstream of Pho80–85 [35], by phosphorylation of its Thr1075 residue. Together with the TORC1/Sch9-mediated phosphorylation of Ser1061 in Rim15, this ensures cooperative cytoplasmic retention of inactive Rim15 by 14-3-3 proteins, which is also validated by the observation that Cln-CDK inactivation and rapamycin treatment triggered nuclear accumulation of Rim15KD in an additive manner (Figure 4d). Taken together, our current study advocates a model in which the Gwl Rim15 coordinates the G1–S transition in response to cell cycle and nutritional cues by integrating the signals that emanate from Cln-CDK and TORC1-Sch9, respectively.

Analogies between the roles of budding yeast and metazoan Gwl pathways

The budding yeast and metazoan Gwl pathways are organized in a strikingly similar way. Indeed, as shown here and in previous studies [36], they are both subjected to a direct control by CDKs, which likely serves to sharpen the onset of the Gwl-controlled cell cycle transition(s). Curiously, however, the main role of the Gwl pathway appears to be different in budding yeast and metazoans: yeast Gwl preferably antagonizes G1–S [17, 37, 38], while metazoan (and likely also fission yeast) Gwl primarily promotes the G2–M cell cycle transition [17, 3941]. The Gwl pathway may therefore have been rewired during the evolution of higher eukaryotes to control G2–M instead of G1–S. Alternatively, the observed divergent roles of the Gwl pathways in budding yeast and metazoans could have evolved from an ancestral version of this pathway that functioned in both G1–S and G2–M, which raises the possibility that higher eukaryotes may have retained some aspects of the Gwl-mediated G1–S control mechanism identified in this study. In this context, it is compelling that mammalian cells express the cyclin-CDK2 inhibitor p27Kip1, which is structurally similar to Sic1 and also functions in G0/1–S cell cycle transition control [42]. Moreover, similar to Sic1, p27Kip1 is flagged for degradation by the combined action of cyclin-CDK, Cks1 and the SCFSkp2 ubiquitin ligase complex, while it is stably maintained in G0 cells only when phosphorylated at specific amino-acid residues [43]. Considering the importance of p27Kip1 and Cks1 in suppressing and promoting cancer progression, respectively [43, 44], these analogies warrant future studies that address a potentially conserved role of the Gwl pathway in G1–S transition control.

Materials and Methods

Strains, plasmids and growth conditions

Strains and plasmids are listed in Supplementary Tables S1 and S2, respectively. Notably, Sic1 alleles exhibit several polymorphisms, one of which includes an additional Glu residue at position 129 of the protein in some strain backgrounds, including the JK9-3D used here (see Saccharomyces Genome Database). To remain consistent with the previously published literature, we refer to Sic1-Thr173 throughout our study although the respective residue corresponds to Sic1-Thr174 in JK9-3D. Saccharomyces cerevisiae yeast cells were pregrown overnight at 30 °C in standard rich medium (YPD) with 2% glucose or in synthethic defined medium (0.17% yeast nitrogen base, 0.5% ammonium sulfate, 0.2% appropriate dropout mix (USBiological, Salem, MA, USA) and 2% glucose) for maintenance of plasmids. Prior to the experiments, cells were diluted to an OD600 of 0.15 and grown until they reached an OD600 of 0.6. Rapamycin (LC Laboratories, Woburn, MA, USA), 1NM-PP1 (Cayman Chemicals, Ann Arbor, MI, USA), and α-factor (KareBay Biochem, Monmouth Junction, NJ, USA) were used at final concentrations of 200 ng ml−1, 0.5 μM, and 5 μg ml−1, respectively.

Immunoblot analyses

Protein extraction was performed by mild alkali treatment of cells followed by boiling in standard electrophoresis buffer [45]. Sodium dodecyl sulfate-polyacrylamide gel electrophoresis and immunoblot analyses were performed according to standard protocols. Phosphorylation of Sic1 variants was assessed by loading whole-cell extracts on 9% sodium dodecyl sulfate-polyacrylamide gel electrophoresis gels containing 50 μM Phos-tag (Wako, Neuss, Germany). The sources and dilutions of antibodies used are listed in Supplementary Table S3.


For co-IP analyses, lysates were prepared by disruption of frozen cells in lysis buffer (50 mM TRIS (pH 7.5), 1 mM EDTA, 150 mM NaCl, 0.5% NP40 and 1× protease and phosphatase inhibitor cocktails (Roche, Rotkreuz, Switzerland)) with glass beads (0.5 mm diameter) using a Precellys cell disruptor (Bertin Technologies, Montigny-le-Bretonneux, France) and subsequent clarification by centrifugation (5 min at 14 000 r.p.m.; 4 °C). HA- and myc-tagged proteins were immunoprecipitated with an anti-HA and anti-myc magnetic matrix (Pierce, Lausanne, Switzerland), respectively. GST-tagged proteins were pulled down with glutathione-sepharose beads (GE Healthcare, Otelfingen, Switzerland).

In vitro kinase assays

Protein kinases were immunoprecipitated as above. For CDK assays, Cln2-HA3 or Clb5-HA3 immunocomplexes were incubated at 30 °C for 20 min (or for the times indicated) in kinase buffer (125 mM TRIS pH 7.5, 50 mM MgCl2, 2.5 mM dithiothreitol and either 10 mM ATP for non-radioactive kinase assays or 20 μM ATP with 10 μCi of γ-32P ATP (Hartmann Analytic, Braunschweig, Germany) for radioactive kinase assays). The reactions were stopped with sample buffer and boiled for 5 min. For the CDK inhibition assays, the immunoprecipitated kinase complexes were incubated for 1 h on ice with GST-Sic1, GST-Sic1T173A or GST-Sic1R262A L264A (that were subjected to in vitro phosphorylation by Mpk1 before use) and then assayed in CDK assays using GST-Sic11-100 as substrate. The levels of Thr5 in Sic1 were detected with phosphospecific anti-Sic1-pThr5 antibodies.

Fluorescence microscopy

Images were captured with an inverted Spinning Disk Confocal Microscope (VisiScope CSU-W1, Puchheim, Germany) that was equipped with an Evolve 512 (Photometrics, Tucson, AZ, USA) EM-CCD camera and a 100× 1.3 NA oil immersion Nikon CFI series objective (Egg, Switzerland).