Regulated gene expression makes an important contribution to cell cycle control mechanisms. In fission yeast, a group of genes is coordinately expressed during a late stage of the cell cycle (M phase and cytokinesis) that is controlled by common cis-acting promoter motifs named pombe cell cycle boxes (PCBs), which are bound by a trans-acting transcription factor complex, PCB binding factor (PBF). PBF contains at least three transcription factors, a MADS box protein Mbx1p and two forkhead transcription factors, Sep1p and Fkh2p. Here we show that the fission yeast Cdc14p-like phosphatase Clp1p (Flp1p) controls M–G1 specific gene expression through PBF. Clp1p binds in vivo both to Mbx1p, a MADS box-like transcription factor, and to the promoters of genes transcribed at this cell cycle time. Because Clp1p dephosphorylates Mbx1p in vitro, and is required for Mbx1p cell cycle-specific dephosphorylation in vivo, our observations suggest that Clp1p controls cell cycle-specific gene expression through binding to and dephosphorylating Mbx1p.
Gene expression at particular times during the mitotic cell division cycle is widespread amongst eukaryotes and makes an important contribution to this fundamental process (Spellman et al., 1998; Rustici et al., 2004). A major area of interest concerns understanding how this regulated gene expression is integrated into general cell cycle controls.
In fission yeast, at least four waves of gene expression have been described, and in each case the transcription factor complex that regulates their specific expression has been identified (McInerny, 2004; Bähler, 2005). One wave occurs at the M–G1 interval, with a number of genes transcribed at this cell cycle time that encode proteins required for cytokinesis and cell division (Anderson et al., 2002). The transcription factor complex that regulates this wave of gene expression has been identified and named PBF (pombe cell cycle box binding factor). PBF consists of at least three components: Mbx1p, a MADS box-like transcription factor, and two forkhead transcription factors, Sep1p and Fkh2p (Buck et al., 2004; Bulmer et al., 2004). It is thought that primary regulation of gene expression occurs through the forkhead transcription factors, with Sep1p activating and Fkh2p repressing gene expression. However, PBF is also regulated directly by the polo-like kinase Plo1p, which binds to and phosphorylates Mbx1p (Papadopoulou et al., 2008). Because both fkh2+ and plo1+ are themselves transcribed at M–G1 under PBF control, this suggests that feedback loops might operate to both turn on and turn off cell cycle gene expression at this cell cycle time. Related mechanisms have been identified in budding yeast and humans, showing that they are conserved and so play a significant part in cell cycle controls in all eukaryotic organisms (Darieva et al., 2006; Fu et al., 2008).
Cdc14p proteins form a class of phosphatases that have important and diverse roles in controlling cell cycle progression, through regulating a number of crucial target substrate proteins, to couple nuclear and cytoplasmic divisions (Simanis, 2003; Stegmeier and Amon, 2004; Vázquez-Novelle et al., 2005; Clemente-Blanco et al., 2006; Clifford et al., 2008a). In budding yeast, one function of Cdc14p is to control cell cycle-specific gene expression through regulating the nuclear localisation of transcription factors required for this process. These transcription factors include Swi5p, Ace2p and Swi6p, whose dephosphorylation permits their nuclear entry, and consequent activation of transcription of target genes at M–G1 and G1–S, respectively (Visintin et al., 1998; Geymonat et al., 2004).
In this paper we identify a new function for the fission yeast Cdc14p-like phosphatase, Clp1p (Flp1p). We show that Clp1p controls cell cycle-specific gene expression at the M–G1 interval through a novel mechanism, binding to and dephosphorylating Mbx1p on the promoters of genes transcribed at this cell cycle time. Thus it appears that this Cdc14p-like phosphatase has the ability to interact with gene promoters to directly regulate gene expression, a new property for this type of protein.
Clp1p binds to and dephosphorylates Mbx1p
To identify proteins that bind to Clp1p, we purified tandem affinity purification (TAP) complexes from mitotically arrested cells expressing TAP-tagged Clp1p (Gould et al., 2004). Two-dimensional liquid chromatography tandem mass spectrometry was used to characterise the protein composition of these complexes. One protein identified was Mbx1p, with 50 peptides recovered providing 29% sequence coverage. We confirmed that this interaction occurs both in vivo and in vitro by co-immunoprecipitation, pull-down and yeast two-hybrid experiments. Mbx1p–13myc from fission yeast extracts was detected in a pull-down assay with purified MBP–Clp1p (Fig. 1A). The interaction was direct because GST–Mbx1p purified from bacterial extracts pulled-down purified MBP–Clp1p (Fig. 1B). The interaction between Clp1p and Mbp1p was also shown using yeast two-hybrid (Fig. 2C, data not shown). Furthermore, the interaction was cell cycle-specific because Clp1p–3HA was only detected binding in vivo to Mbx1p–13myc during late M phase after release from an nda3 M phase arrest (Fig. 1C).
Mbx1p undergoes cell cycle-dependent phosphorylation, which is directly regulated by the polo-like kinase, Plo1p (Papadopoulou et al., 2008). We mapped the amino acid residues in Mbx1p that are modified by phosphorylation using tandem mass spectrometry and identified four serine residues (Fig. 2A; supplementary material Fig. S1); a further serine was identified by database analysis. Of these, the amino acids around two (46 and 106) resembled the consensus sites for polo-kinases, whereas another two (183 and 372) resembled those for cyclin-dependent kinase 1 (Cdc2p) (Nakajima et al., 2003). We examined the significance of the phosphorylation of four of these serine residues by mutating them in pairs either to aspartic acid or alanine, mimicking permanently phosphorylated or unphosphorylatable forms. These phosphorylation sites are relevant to Mbx1p function because a combination of serine to alanine changes at residues 46 and 106 reduced the in vivo interactions of Mbx1 with both Clp1p and Plo1p; similarly, serine to alanine changes at residues 183 and 372 displayed reduced in vivo interaction with Plo1p, but not Clp1p (Fig. 2C). Perhaps most strikingly, serine to aspartic acid changes at residues 46 and 106 resulted in a defective cell morphology and cell separation phenotypes, with cell clumping reminiscent of mbx1Δ (Fig. 2D) (Buck et al., 2004). These mutations also allowed these cells to mate with both h+ and h− strains, as observed with mbx1Δ (Papadopoulou et al., 2008).
We also explored the ability of Clp1p to dephosphorylate Mbx1p. In separate in vitro kinase reactions, Mbx1p was exposed to Clp1p after pre-phosphorylation by either Cdk1 or Plo1p (Fig. 2B, data not shown). Under such conditions, Clp1p efficiently dephosphorylated Mbx1p phosphorylated by Cdk1, but not by Plo1p. Consistent with this function, we observed that a phosphatase-dead mutant (with cysteine 286 changed to serine) (Wolfe et al., 2006) affected Mbx1p cell cycle-specific phosphorylation in vivo: the presence of lower mobility, hyperphosphorylated forms of Mbx1p were detected in the mutant clp1-C286S compared with wild type (Fig. 1C).
Clp1p binds to PCB promoter DNAs in vivo
The observation that Clp1p binds to Mbx1p suggested that this phosphatase might regulate M–G1 gene expression by binding to promoter DNAs. We tested this using chromatin immunoprecipitation (ChIP) analysis of Clp1p, using a tagged version of the protein under the control of its native promoter to detect whether it binds in vivo to promoters of genes containing PCB motifs (to which PBF binds) expressed at M–G1 (Papadopoulou et al., 2008). Clp1p specifically bound to the promoters of PCB-regulated genes (Fig. 3A). Furthermore, binding was cell cycle-specific: minimum Clp1p binding occurred when target gene expression was maximum and vice versa (Fig. 3B), implying a repressive function for Clp1p in controlling gene expression. The ability of Clp1p to bind to PCB promoters is dependent on Mbx1p, because ChIP analysis in cells lacking Mbx1p revealed that Clp1p cannot bind to PCB promoters (Fig. 3C). Similarly, Plo1p binding to PCB promoters is also dependent on Mbx1p (Fig. 3C).
Clp1p is required for M–G1 specific gene expression in fission yeast
The observation that Clp1p binds to PCB promoters through Mbx1p suggests that it has a role in controlling M–G1 specific gene transcription. We therefore examined M–G1 gene expression in clp1+ deletion mutant (clp1Δ) cells, synchronised either by centrifugal elutriation (Fig. 4A) or by cdc25-22 transient block and release (Fig. 4B). cdc15+ and spo12+ mRNAs were present through more of the cell cycle, compared with the distinct peaks at M–G1 observed in wild-type cells. Consistent with this observation, Cdc15p protein levels are elevated in clp1Δ cells (Clifford et al., 2008b). Such increased gene expression outside the M–G1 interval in cells lacking Clp1p supports the idea that it has a repressive role in controlling gene transcription at these cell cycle times. This phenotype is exacerbated by combining clp1Δ with mid1Δ (Argarwal et al., 2010), suggesting that the Clp1p and Mid1p proteins have overlapping repressive roles in controlling M–G1 gene expression in fission yeast.
We also examined M–G1 gene expression and Clp1p binding in clp1-3A cells, a gain-of-function mutant that is not inhibited in phosphatase activity as effectively as the wild type (Wolfe et al., 2006). The clp1-3A mutation resulted in delayed PCB-regulated gene expression, with cdc15+ mRNA levels peaking after cdc22+, an MBF-regulated gene that is expressed at G1–S (Fig. 5A). In wild-type cells, cdc15+ mRNA levels peak before those of cdc22+ mRNA (Fig. 5A) (Anderson et al., 2002). Furthermore, ChIP analysis revealed that Clp1p-3A was unable to contact PCB promoters in vivo (Fig. 5B).
clp1+ shows genetic interactions with plo1+ and genes encoding components of PBF
To explore further the role of Clp1p in cell cycle regulation, we searched for synthetic phenotypes of mutations in clp1 (clp1Δ, gain-of-function mutant clp1-3A, and phosphatase-dead mutant clp1-D257A) with mutations in plo1, fkh2, sep1 or mbx1. A summary of the results of these various crosses is shown in Fig. 6A. Strikingly, a clp1Δ fkh2Δ double deletion mutant was found to be synthetically lethal. Because Fkh2p is thought to negatively regulate M–G1 gene expression, with deletion of fkh2+ causing overexpression of target genes (Buck et al., 2004; Bulmer et al., 2004; Rustici et al., 2004), the effect of also deleting clp1+, another repressor, might exacerbate de-regulated expression of target genes to lethal levels. The three mutants in clp1 also showed synthetic phenotypes with plo1, reducing the restrictive temperature of the temperature-sensitive mutant plo1-ts35 from >35°C to 30°C. The phenotype of plo1-ts35 clp1-D257A cells was particularly severe, with the cells exhibiting malformed shape and slow growth. Again, these synthetic phenotypes support the idea that clp1+ and plo1+ interact to control cellular events. Consistent with this suggestion, Clp1p cannot bind to PCB promoters in vivo in plo1-ts35 cells (data not shown).
We also examined the effect of deletions of mbx1+ and clp1+ on the lethal phenotype observed when overexpressing plo1+ (Ohkura et al., 1995). Deletions of either suppressed the lethal phenotype caused by overexpressing plo1+ (Fig. 6B), although the cells did not return to a wild-type phenotype (data not shown), again consistent with the encoded proteins functioning together in vivo.
Clp1p regulation of cell cycle-specific gene expression
The observations described here provide the first evidence for a new function for the fission yeast Cdc14p-like phosphatase Clp1p in controlling cell cycle-specific gene expression at M–G1. We reveal that Clp1p binds Mbx1p to directly regulate gene expression. Cdc14p in budding yeast has been shown to control cell cycle-specific gene expression, but in this organism the mechanism is more indirect, with the phosphatase controlling the nuclear location of transcriptions factors required for this process (Visintin et al., 1998; Geymonat et al., 2004). Dephosphorylation of Swi5p, Ace2p or Swi6p allows their entry into the nucleus to then bind to promoters and stimulate gene expression. Thus, in these cases Cdc14p has a positive, activating role in allowing gene expression to occur at the end of the cell cycle.
A repressive role for Clp1p in controlling gene expression
In fission yeast we propose a novel mechanism by which Clp1p controls cell cycle-specific gene expression. Here Clp1p more directly effects transcription by binding to a transcription factor and dephosphorylating it while it is bound to gene promoters (Fig. 7).
Another striking difference between Cdc14p and Clp1p is that the evidence points to Clp1p having a negative, repressive role in controlling cell cycle-specific gene expression. Removal of Clp1p from cells through chromosomal deletion of clp1+ (clp1Δ) does not prevent M–G1 gene expression, as would be expected if Clp1p had a positive, activating role, but instead results in gene expression not restricted to M–G1. This phenotype suggests that Clp1p normally has a function to prevent gene expression outside the M–G1 interval. This conclusion is supported by other observations with Clp1p. ChIP analysis through the cell cycle reveals that this phosphatase only binds to promoters when genes are not being transcribed, and is present when their expression is inhibited. Furthermore, combining clp1Δ with mid1Δ results in higher overexpression (Argarwal et al., 2010), suggesting that the two have overlapping repressive roles in controlling M–G1 gene expression.
Genetic experiments also support the hypothesis that Clp1p has a negative role in gene expression. A repressive function has been ascribed to the Fkh2p transcription factor, with fkh2Δ cells showing related phenotypes to clp1Δ (Buck et al., 2004; Bulmer et al., 2004; Papadopoulou et al., 2008). It is notable that the double deletion mutant fkh2Δ clp1Δ is synthetically lethal, which could be explained by the combinatorial removal of both repressive proteins to allow such high levels of overexpression of target genes that this is lethal to cells.
Integration into general cell cycle control mechanisms
An important area of study in cell cycle-regulated transcription concerns understanding how each wave of gene expression is integrated into the wider context of the cell cycle. In budding yeast, it has been proposed that each wave of gene expression is controlled by the previous wave, thereby making a ‘transcription cycle’ within a cycle (Simon et al., 2001; Tyers, 2004). However, in fission yeast such continuous linked waves of gene expression are not complete, and thus some of the waves must be linked to the cell cycle by other, independent control mechanisms (McInerny, 2004; Bähler, 2005). Here, we show that Clp1p directly controls M–G1 gene expression. As Clp1p is an established cell cycle control protein that regulates many important events during M phase and cytokinesis, these experiments offer a direct way in which gene expression at this cell cycle time is linked to the processes of division and cell duplication.
Conservation of controls
The observations presented in this paper show that fission yeast Clp1p has a direct role in controlling cell cycle-specific gene expression, a new function for this phosphatase. A related function for Cdc14p has been identified in budding yeast in controlling at least two waves of cell cycle gene expression, although the mechanistic details are different (Visintin et al., 1998; Geymonat et al., 2004). Budding yeast Cdc14p has also been shown to reduce transcription by RNA polymerase I during anaphase (Clemente-Blanco et al., 2009). Thus, it appears that the regulation of gene expression during the cell cycle by Cdc14p phosphatases might be a widespread and general property of this class of proteins in yeasts. It will be fascinating to see if this property extends to higher eukaryotes, including humans.
Materials and Methods
Yeast strains and growth conditions
The general molecular procedures of Sambrook et al. (Sambrook et al., 1989) were performed, with the standard methodology and media used for the manipulation of Schizosaccharomyces pombe as described in Moreno et al. (Moreno et al., 1991). The strains used in this study are shown in supplementary material Table S1. mbx1 phospho-mutants were cloned into plasmid pJK148 containing wild-type mbx1+ promoter and terminator sequences, and integrated in single copy at the leu1-32 locus into an mbx1Δ strain. Cells were routinely grown in complete yeast extract (YE) medium at 25°C or 30°C. Synchronous cultures of cdc25-22 cells were prepared by transient temperature shifts to 36°C for 4 hours. Septation indices of such cultures were counted microscopically and plotted to indicate synchrony. Overexpression of plo1+ was from pREP1 (Maundrell, 1993). Cells were grown in Eagle's modified medium (EMM) with 5 mg ml−1 thiamine (nmt1 promoter ‘off’) to early exponential stage, washed three times in thiamine-free EMM and then grown for 15 hours in EMM without thiamine (nmt1 promoter ‘on’).
DNA constructs and RNA manipulations
GST-mbx1+, MBP-clp1+ and CDK1 constructs have been described (Clifford et al., 2008b; Papadopoulou et al., 2008). The plo1+, mbx1+ and clp1+ two-hybrid constructs have been described (Clifford et al., 2008b; Papadopoulou et al., 2008). The mbx1+ plasmid has been described (Buck et al., 2004). Site-directed mutagenesis was performed using the pALTER vector (Promega). S. pombe total RNA was prepared and analysed by northern blotting (Buck et al., 2004). Levels of individual transcripts were quantified by normalisation against levels of rRNA in each sample.
Immunoprecipitations, pull-downs and western blot analysis
Preparation of whole cell extracts, immunoprecipitations, pull-downs and western blot analysis were performed as described (Clifford et al., 2008b; Papadopoulou et al., 2008). Purification of Clp1p–TAP and subsequent identification of Clp1p and proteins by mass spectrometry was performed as described (Gould et al., 2004; Roberts-Galbraith et al., 2009). Mass spectrometry was used to identify the Mbx1p phosphorylation sites in Clp1p–TAP eluates, and in GST–Mbx1p phosphorylated in vitro by purified His–Plo1p (Papadopoulou et al., 2008).
Kinase and phosphatase assays
Bacterially produced recombinant GST–Mbx1p was purified and left on glutathione beads. Kinase assays were conducted as described (Wolfe et al., 2006) with the following modifications: Cdk kinase assay buffer consisted of 50 mM Tris-HCl, pH 7.4, 10 mM MgCl2, 2 mM DTT; Plo1p kinase assay buffer was composed of 10 mM Tris-HCl, pH 7.4, 10 mM MgCl2, 150 mM NaCl, 2 mM DTT. Clp1p phosphatase assays were as described (Clifford et al., 2008b).
↵* Present address: Department of Biology, University of Western Ontario, London, Ontario N6A-5B7, Canada
Supplementary material available online at http://jcs.biologists.org/lookup/suppl/doi:10.1242/jcs.073056/-/DC1
- Accepted September 6, 2010.
- © 2010.