The eukaryotic cell cycle comprises an ordered series of events, orchestrated by the activity of cyclin-dependent kinases (Cdks), leading from chromosome replication during S phase to their segregation in mitosis. The unidirectionality of cell-cycle transitions is fundamental for the successful completion of this cycle. It is thought that irrevocable proteolytic degradation of key cell-cycle regulators makes cell-cycle transitions irreversible, thereby enforcing directionality1,2,3. Here we have experimentally examined the contribution of cyclin proteolysis to the irreversibility of mitotic exit, the transition from high mitotic Cdk activity back to low activity in G1. We show that forced cyclin destruction in mitotic budding yeast cells efficiently drives mitotic exit events. However, these remain reversible after termination of cyclin proteolysis, with recovery of the mitotic state and cyclin levels. Mitotic exit becomes irreversible only after longer periods of cyclin degradation, owing to activation of a double-negative feedback loop involving the Cdk inhibitor Sic1 (refs 4, 5). Quantitative modelling suggests that feedback is required to maintain low Cdk activity and to prevent cyclin resynthesis. Our findings demonstrate that the unidirectionality of mitotic exit is not the consequence of proteolysis but of systems-level feedback required to maintain the cell cycle in a new stable state.
After completion of chromosome segregation during mitosis, the activity of the key cell-cycle kinase Cdk is downregulated to promote mitotic exit and the return of cells to G1. This involves ubiquitin-mediated degradation of mitotic cyclins under control of the anaphase promoting complex (APC), a multisubunit ubiquitin ligase. Mitotic cyclins are initially targeted for degradation by the APC in association with its activating subunit Cdc20 (APCCdc20). Later, declining Cdk levels and activation of the Cdk-counteracting phosphatase Cdc14 allow a second APC activator, Cdh1, to associate with the APC (APCCdh1)6,7,8. Cyclin proteolysis, a thermodynamically irreversible reaction, is thought to be responsible for the irreversibility of mitotic exit1,2,3. However, de novo protein synthesis can counteract degradation and constitutes a similar thermodynamically irreversible process, driven by ATP hydrolysis. In a cellular setting, therefore, protein levels are defined by reversible changes to the rates of two individually irreversible reactions: protein synthesis and degradation. These considerations have led to the hypothesis that it is not proteolysis itself, but systems-level feedback that affects synthesis and degradation rates, making cell-cycle transitions irreversible9.
To test this hypothesis, we investigated the contribution of cyclin proteolysis to the irreversibility of budding yeast mitotic exit (Fig. 1a). We arrested budding yeast cells in mitosis with high levels of mitotic cyclins by depleting Cdc20 under control of the MET3 promoter. In these cells, we induced Cdh1 expression from the galactose-inducible GALL promoter (an attenuated version of the GAL1 promoter) for 30 min6. We expressed a Cdh1 variant, Cdh1(m11), that activates the APC even in the presence of high Cdk activity owing to mutation of 11 Cdk phosphorylation sites. This led to efficient degradation of the major budding yeast mitotic cyclin Clb2 (Fig. 1b). The mitotic Polo-like kinase, another APCCdh1 target10, was also efficiently degraded, whereas levels of the S phase cyclin Clb5, a preferential substrate for APCCdc20 (ref. 11), remained largely unaffected (Supplementary Fig. 1). Clb2 destruction was accompanied by dephosphorylation of known mitotic Cdk substrates, seen by their change in electrophoretic mobility (Fig. 1b). These included three proteins whose dephosphorylation contributes to spindle elongation and chromosome segregation, Sli15, Ase1 and Ask1 (refs 12–14). Their dephosphorylation depended on the activity of the mitotic exit phosphatase Cdc14 (Supplementary Fig. 2). Mitotic spindles that were present in the metaphase-arrested cells disassembled as Clb2 levels declined, accompanied by outgrowth of pronounced astral microtubules (Fig. 1c and Supplementary Fig. 3), reminiscent of spindle breakdown at the end of mitosis. APCCdh1(m11)-mediated destruction of the spindle stabilizing factor Ase1 (Fig. 1b), in addition to its dephosphorylation, may contribute to this phenotype15.
After 50 min, when Clb2 levels became almost undetectable, we turned off APCCdh1(m11) by inactivating a temperature-sensitive APC core subunit encoded by the cdc16-123 allele16. Notably, in response to APCcdc16-123 inactivation, Clb2 levels recovered and Sli15, Ase1 and Ask1 reappeared in their mitotic hyperphosphorylated forms. Mitotic spindles formed again, suggesting that cells had re-entered a mitotic state. Spindles appeared morphologically intact, but were longer after cyclin reaccumulation (3.9 ± 0.8 μm; mean ± s.d.) compared to metaphase spindles at the beginning of the experiment (2.1 ± 0.6 μm). A probable reason for this lies in compromised sister chromatid cohesion after some, albeit inefficient, inactivation of the anaphase inhibitor securin by APCCdh1(m11) (Supplementary Fig. 4)17. Fluorescence-activated cell sorting (FACS) analysis of DNA content confirmed that cells maintained a 2C DNA content throughout the experiment (Fig. 1d). This demonstrates that cyclin destruction promotes mitotic exit events, but it is not sufficient to render them irreversible. Cyclin resynthesis can reverse mitotic exit. Note that reversibility of mitotic exit under these conditions did not depend on the persistence of the S phase cyclin Clb5 (Supplementary Fig. 1).
We next addressed what makes mitotic exit irreversible, if not cyclin destruction. When we repeated the experiment, but continued Cdh1(m11) induction and inactivated APCcdc16-123 only after 60 min, Clb2 did not reaccumulate and over half of the cells subsequently completed cytokinesis and entered G1 (Fig. 2a). This suggests that after longer periods of Clb2 destruction mitotic exit becomes irreversible. Western blotting showed that around the time when Clb2 loss turned irreversible, the Cdk inhibitor Sic1 accumulated4,5. We therefore asked whether Sic1 accumulation was responsible for the irreversibility of mitotic exit. When we repeated the experiment using a sic1Δ strain, Clb2 reappeared after APCcdc16-123 inactivation, and only a minority of cells proceeded to cytokinesis (Fig. 2a). In the absence of Sic1, mitotic exit remained reversible even after prolonged Clb2 destruction for 90 min (Supplementary Fig. 5). Sic1 is part of a double-negative feedback loop in which Cdk downregulation allows the Cdc14 phosphatase to dephosphorylate Sic1, as well as its transcription factor Swi5, increasing the expression and stability of the Cdk inhibitor5,9. Because Clb2 positively regulates its own synthesis18,19, Clb2 inhibition by Sic1 may be required to prevent Clb2 resynthesis.
These findings indicate that mitotic exit remained reversible for up to 50 min because the Sic1 feedback loop had not yet been sufficiently activated. Consistent with this possibility, dephosphorylation-dependent translocation of Swi5 from the cytoplasm to the nucleus, indicative of Swi5 activation during mitotic exit5,20, was inefficient under these conditions (Fig. 2b). Similarly, the levels of Sic1 that became detectable remained below the levels observed in cells undergoing mitotic exit after release from the metaphase block by Cdc20 reinduction (Fig. 2c). If irreversibility of mitotic exit is due to activation of the Sic1-dependent feedback loop, irreversibility should be advanced by increasing Cdc14 phosphatase activity, or by directly enhancing Sic1 levels. As predicted, ectopic expression of either Cdc14 or a version of Sic1 that is stable in the presence of high Cdk activity owing to mutation of three Cdk phosphorylation sites21, Sic1(m3), made Clb2 destruction irreversible under conditions that otherwise permit Clb2 resynthesis (Fig. 3a). This confirms that activation of the Sic1-dependent feedback loop limits the irreversibility of mitotic exit.
It has been shown that mammalian mitotic exit can be driven in the absence of cyclin proteolysis by chemical inhibition of Cdk activity. After short, but not longer periods, of Cdk inhibition mitotic exit remained reversible3. We repeated these experiments in budding yeast cells carrying the ATP analogue (1NM-PP1)-sensitive Cdk allele cdc28-as1 (ref. 22). As in mammalian cells, we observed reversible mitotic exit events, exemplified by dephosphorylation of the Cdk substrate Orc6 (ref. 23), after transient Cdk inhibition for 10 min. After 50 min of inhibitor treatment, Orc6 dephosphorylation turned irreversible. Irreversibility again correlated with, and depended on, the accumulation of Sic1 (Fig. 3b). This suggests that feedback-loop activation is responsible, and that cyclin destruction is not required, for irreversible mitotic exit. In the presence of cyclin destruction, a shorter period of Cdk inhibition was sufficient to render mammalian mitotic exit irreversible3. Although this was taken to demonstrate a requirement for cyclin destruction, we suggest that activation of a feedback loop correlated with cyclin destruction that made mitotic exit irreversible.
To theoretically investigate the contribution of feedback to the irreversibility of budding yeast mitotic exit, we used a mathematical model24 to describe the cell-cycle control network that operates during mitotic exit (Fig. 4a, a detailed description is found in Supplementary Fig. 6). In our experiments, Cdk downregulation begins by Cdh1(m11)-mediated Clb2 degradation, or 1NM-PP1 addition, which causes Sic1 accumulation because of the double-negative feedback loops. If Clb2 proteolysis is terminated by inactivating APCcdc16-123, or if 1NM-PP1 is removed, before Sic1 reaches a threshold, Sic1 accumulation becomes transient and Cdk activity will recover (Fig. 4b, e). In the absence of Sic1, mitotic exit will therefore always remain reversible (Fig. 4d, g). Clb2 destruction becomes irreversible only if Sic1 levels have reached a threshold that maintains Cdk activity low enough to prevent Clb2 resynthesis (Fig. 4c). Mitotic exit after chemical Cdk inhibition turns irreversible when Sic1 levels are sufficient to maintain Cdk inhibition independently of 1NM-PP1 (Fig. 4f). Note that during normal mitotic exit, Cdk downregulation is initiated by APCCdc20-mediated Clb2 destruction, and APCCdh1 activation by the decreasing Cdk/Cdc14 ratio forms an additional double-negative feedback loop that maintains low Cdk activity redundantly with Sic1. In mammalian cells, the antagonistic relationship between Cdk and Cdh1, and between Cdk and its inhibitory tyrosine phosphorylation25,26, create double-negative feedback loops of Cdk inactivation that probably contribute to irreversibility of mitotic exit.
Protein destruction is a commonly used mechanism controlling key cell-cycle transitions. The rise in Cdk activity at the G1/S transition is accompanied by the degradation of Cdk inhibitors. The irreversibility of this transition, however, probably stems from positive feedback during Cdk activation27. Likewise, even though the G2/M transition in the vertebrate cell cycle involves proteolysis of the Cdk inhibitory kinase Wee1, this transition shows the characteristics of a bistable switch, driven by feedback between Cdk, Wee1 and the Cdk-activating phosphatase Cdc25 (refs 28, 29). Although not irreversible in a cellular context, protein degradation, and in particular protein resynthesis, occur at a slower timescale than the addition or removal of posttranslational modifications. Proteolysis thereby introduces an element of distinct dynamic nature into the cell-cycle control network, the consequences of which merit further investigation.
A list of strains used in this study is found in Supplementary Table 1. Epitope tagging of endogenous genes and gene deletions were performed by gene targeting using PCR products. Integrative expression vectors for Cdh1(m11) and Cdc14 under control of the GALL and GAL1 promoters, respectively, were as described6,13; the YIplac211GAL1-SIC1(m3)-HA vector was a gift from E. Schwob.
Cells were grown at 25 °C in synthetic complete (SC) medium lacking methionine, with 2% raffinose as carbon source, and arrested in metaphase by depletion of Cdc20 under control of the MET3 promoter by addition of 2 mM methionine for 5 h. Expression of Cdh1(m11), Sic1(m3) or Cdc14 was induced by addition of 2% galactose. Induction was terminated by addition of 2% glucose, and APCcdc16-123 was inactivated by shifting the culture to a waterbath at 37 °C. Alternatively, metaphase-arrested cells were released into synchronous mitotic progression by Cdc20 reinduction after filtration and resuspension in methionine-free medium. Protein extracts were prepared using the TCA method30. Antibodies used for western blotting and immunostaining were: anti-HA 12CA5, anti-myc 9E10, anti-Pk SV5-Pk1 (Serotec), anti-Orc6 SB49 (a gift from B. Stillman), anti-Clb2 (y-180), anti-Clb5 (yN-19), and anti-Sic1 antisera (FL-284, all Santa Cruz Biotechnology), anti-Tub1 YOL1/34 (Serotec), anti-actin N350 (Amersham), anti-Tub4 serum (a gift from J. Kilmartin), and an antiserum raised against recombinant Cdc14 purified after overexpression in Escherichia coli.
We thank C. Bouchoux, J. Kilmartin, E. Schwob and W. Zachariae for antibodies and constructs, and members of our laboratory for discussion and critical reading of the manuscript. This work was supported by a European Commission Marie Curie Individual Fellowship to S.L.-A., and the BBSRC and EC FP7 (O.K.).
This file contains Supplementary Figures 1-6 with Legends, Supplementary Data, Supplementary References and Supplementary Table 1.
About this article
Bulletin of Mathematical Biology (2015)