CHK1 phosphorylates CDC25B during the cell cycle in the absence of DNA damage

CDC25 dual specificity phosphatases play a crucial role in controlling eukaryotic cell-cycle progression under both normal conditions and after checkpoint activation, through the dephosphorylation and activation of cyclin-dependent kinases (CDKs) (Morgan, 1995). In human cells, the CDC25 family includes three members, CDC25A, CDC25B and CDC25C (Galaktionov and Beach, 1991; Nagata et al., 1991; Sadhu et al., 1990). Although all three CDC25 proteins have been shown to function during mitosis, CDC25B is thought to act as an initiator of early mitotic events, because its overexpression in interphase cells induces premature mitosis (Baldin et al., 1997b; Gabrielli et al., 1996; Karlsson et al., 1999; Lammer et al., 1998; Nilsson and Hoffmann, 2000). However, despite recent progress, the exact role of CDC25B remains unclear. The expression of CDC25B is cell-cycle regulated and its activity is detectable from S phase to mitosis (Cazales et al., 2005). Moreover, CDC25B has been shown to activate CDK1–cyclin-B complexes at the centrosome and to regulate centrosomal microtubule nucleation at the G2-M transition (Gabrielli et al., 1996; Lammer et al., 1998; Lindqvist et al., 2005). It has been reported that the centrosomal and cytoplasmic activation of CDK1–cyclin-B precedes its translocation to the nucleus, where the accumulation of active complexes irreversibly drives the cell into mitosis (De Souza et al., 2000; Jackman et al., 2003). Therefore, at the G2-M boundary the centrosome appears to functionally integrate the pathways contributing to the commitment of the cell to enter mitosis. According to this concept, we have recently shown that the aurora-A kinase phosphorylates CDC25B in vivo at the centrosomes and this phosphorylation appears to participate in regulating entry into mitosis (Cazales et al., 2005; Dutertre et al., 2004). Recent work has demonstrated that the human checkpoint kinase CHK1 localizes to interphase centrosomes during normal cell cycle and, thereby, prevents the activation of centrosome-associated CDK1 and early mitotic events, such as centrosome separation and microtubule nucleation (Kramer et al., 2004). However the mechanism by which centrosome-associated CHK1 prevents CDK1 activation has not been established. CDC25 phosphatases are targets for the kinase CHK1 in response to checkpoint activation (Donzelli and Draetta, 2003). Indeed, upon DNA damage CHK1 phosphorylates CDC25C on S216, resulting in its cytoplasmic sequestration and G2-M checkpoint activation (Matsuoka et al., 1998; Peng et al., 1997; Sanchez et al., 1997), and also phosphorylates CDC25A on multiple sites, triggering its proteasome-mediated degradation as well as S and G2 checkpoint activation (Goloudina et al., 2003; Hassepass et al., 2003; Molinari et al., 2000; Sorensen et al., 2003; Xiao et al., 2003; Zhao et al., 2002). However, more recent findings have shown that CHK1 is also a major regulator of CDC25A activity throughout an unperturbed cell cycle (Chen et al., 2003; Lam and Rosen, 2004; Sorensen et al., 2003; Uto et al., 2004; Zhao et al., 2002). Although CHK1 has been shown to phosphorylate CDC25B in vitro (Kramer et al., 2004; Mils et al., 2000; Sanchez et al., 1997), similar regulation of CDC25B during an unperturbed cell cycle has not yet been documented. In the present study, we show that CDC25B is phosphorylated at the centrosome by CHK1 during unperturbed cell division.

To gain insight into a potential role for the checkpoint kinase CHK1 in the regulation of CDC25B, mass spectrometry analyses of in vitro CHK1-phosphorylated recombinant CDC25B protein were performed. This study led to the identification of a number of phosphorylated residues (supplementary material Fig. S1). Among these was serine residue 563 (S563), which had been previously identified as being phosphorylated by CHK1 (Chen et al., 2003; Uto et al., 2004), and serine residues 229 or 230 (S229 or S230, respectively), although mass spectrometry was insufficient to distinguish between these two residues. Similar analyses of CDC25B protein purified from exponentially growing U2OS cells also revealed phosphorylation of S563 and S229 or S230, indicating that phosphorylation of these residues occurs in vivo (data not shown). Since, as shown in Fig. 1A, S230 lies within an RPSS motif similar to the consensus CHK1 phosphorylation motif (RxxS) (O'Neill et al., 2002), and corresponds to S178 of human CDC25A that is phosphorylated by CHK1 in vivo (Chen et al., 2003; Sorensen et al., 2003), we postulated that S230 (and not S229) is the residue phoshorylated by CHK1 in vitro.

Affinity-purified polyclonal antibodies were raised against CDC25B phosphorylated at S230 and S563 (anti-CDC25B-S230-P and anti-CDC25B-S563-P, respectively). These antibodies recognized S230-P and S563-P CHK1 but not unphosphorylated recombinant CDC25B proteins, and the immunodetection could be competitively inhibited by the immunogenic phosphorylated peptides (Fig. 1B), confirming S230 and S563 as CHK1 phosphorylation sites. To further validate the specificity of these CDC25B phosphorylation-specific antibodies, HeLa cells were transiently transfected with plasmids allowing the expression of enhanced yellow fluorescent protein (EYFP)-tagged wild-type CDC25B or phosphorylation-site mutant proteins S230A and S563A. As shown in Fig. 1C, the phosphorylation-specific antibodies detected phosphorylation on S230 and S563 in cells expressing the wild type CDC25B protein but not in cells expressing the mutant proteins.

In vivo phosphorylation of CDC25B on S230 was next examined in western blotting experiments. Since we have not been able to detect the S230-phosphorylated forms of CDC25B at the endogenous protein level (the level of endogenous CDC25B is quite low in this cell line and the anti-CDC25B-S230-P antibody is probably not sensitive enough in a western blot), we used a U2OS cell line that conditionally expressed hemagglutinin (HA)-tagged CDC25B (HA-CDC25B) protein (Theis-Febvre et al., 2003) and that was synchronized by a double thymidine block and released. As shown in Fig. 2A, upon progression from S phase to G2- and M-phases, CDC25B electrophoretic mobility was progressively modified, reflecting expected changes in its phosphorylation status (Baldin et al., 1997a). Immunoblotting with the anti-S230-P antibody revealed the progressive accumulation of CDC25B phosphorylated on S230 as cells progressed through S phase and G2- and M-phases. This indicates that CDC25B is phosphorylated in vivo on S230, and that this phosphorylation is regulated during cell-cycle progression. We next investigated the subcellular localization of endogenous CDC25B phosphorylated at S230 in HeLa cells by immunofluorescence staining (Fig. 2B,C). S563 phosphorylation could not be examined, because the antibody was inefficient in immunofluorescence analyses. To accurately determine the timing of S230 phosphorylation in S phase, HeLa cells were transfected with a plasmid expressing DNA ligase I fused to the DsRed1 fluorescent protein [which has previously been used as a marker to determine the position of cells within S-phase (Easwaran et al., 2005)] and subjected to immunofluorescence staining with anti-S230-P and γ-tubulin antibodies. As shown in Fig. 2B, the S230-phosphorylated form of CDC25B was detected at low levels at the centrosome and in the cytoplasm from early to late S phase. This phosphorylated form was also detected later in interphase cells as well as in mitotic cells, where the signal intensifies as CDC25B accumulates (Cazales et al., 2005) (Fig. 2C). When mitotic cells reached late prometaphase the staining pattern became much more diffuse (Fig. 2C), but the centrosomal staining was not abolished, as confirmed by confocal microscopy (data not shown). As similarly reported for CHK1 (Kramer et al., 2004), this localization was not modified after a 2-hour nocodazole treatment, indicating that CDC25B specifically interacts with centrosomal components (data not shown). Moreover, this centrosome localisation was observed in several cell types, including U2OS (supplementary material Fig. S2) and Hep2 cells (data not shown) and was independent of the fixation method used. The staining pattern observed with the anti-S230-P antibody was abolished by competition with the recombinant CDC25B protein previously phosphorylated in vitro by CHK1 kinase, but not by the unphosphorylated recombinant CDC25B protein, as shown in U2OS cells expressing HA-tagged CDC25B (Fig. 2D). Identical results were also obtained in HeLa cells (data not shown). Moreover, in HeLa cells that had been treated with small interfering RNAs (siRNAs) against CDC25B, downregulation of endogenous CDC25B resulted in a strong reduction in the percentage of cells displaying centrosomal CDC25B signal – as detected by the anti-S230-P antibody – further confirming that this signal was specific for the phosphorylated form of CDC25B (Fig. 2E).

To determine whether S230 is an in vivo substrate for the CHK1 kinase, S230 phosphorylation of CDC25B was examined by western blot in U2OS cells conditionally expressing HA-tagged CDC25B protein, following incubation with the CHK1 inhibitor UCN-01 (Graves et al., 2000). The results in Fig. 3A revealed that S230 phosphorylation of CDC25B was inhibited in cells exposed to UCN-01 for 1 hour, 6 hours after the release from thymidine block (when S230 phosphorylation of CDC25B becomes detectable by western blot, Fig. 2A). Similarly, S563 phosphorylation was completely abolished. As a control, we also looked at the phosphorylation of S353, a residue that we have previously shown to be targeted by other kinases in vivo (Baldin et al., 2003; Dutertre et al., 2004), and found that it was not modified. The inhibition of S230 phosphorylation of CDC25B observed after UCN01 treatment was also correlated with a strong reduction of Y15 phosphorylation of CDK1 (Fig. 3A) and with the induction of premature mitosis (data not shown), as has already been described (Niida et al., 2005). To further validate the results from the chemical inhibition experiment, we knocked down the expression of CHK1 kinase by siRNA. Following transfection with CHK1-directed siRNA oligonucleotides, expression of CHK1 was abolished and, under our conditions, phosphorylation of CDC25B on S230 and also on S563 were completely inhibited (Fig. 3B). By contrast, transfection of cells with scrambled siRNA had no effect on CHK1 expression or CDC25B phosphorylation (Fig. 3B). We next examined S230 phosphorylation by immunofluorescence in these CHK1-siRNA-transfected cells. As shown in Fig. 3C, CHK1 siRNA abolished the phosphorylation of CDC25B on S230 in transfected cells. Identical results were also observed in CHK1-siRNA-transfected HeLa cells (data not shown). Taken together, these results indicate that, CDC25B is phosphorylated in vivo on S230 by the CHK1 kinase and this phosphorylation event occurs at the centrosome during the cell cycle, in the absence of DNA damage.

Overexpressed CDC25B causes S-phase and G2-phase cells to prematurely enter mitosis (Karlsson et al., 1999). Therefore, we next tested whether, when overexpressed, the S230A mutant induces premature mitosis. HeLa cells were transiently transfected with plasmids expressing the EYFP-tagged wild-type CDC25B (EYFP-CDC25B) or the S230A mutant (EYFP-CDC25B S230A) and the percentage of EYFP-positive cell displaying a mitotic phenotype was counted. Our results indicated that the S230A mutation consistently increases the mitosis-inducing activity of CDC25B (Fig. 4A). This data suggests that phosphorylation of S230 inhibits the activity of CDC25B.

CDC25B is phosphorylated by several protein kinases including CK2, PKB/Akt, CDK/cyclin, p38, MAPKAP kinase 2, pEg3 and aurora A. These phosphorylation events have been shown to regulate CDC25B activity, stability and/or subcellular localization (Baldin et al., 1997a; Baldin et al., 2003; Bulavin et al., 2002; Davezac et al., 2002; Lindqvist et al., 2004; Manke et al., 2005; Mirey et al., 2005; Sanchez et al., 1997; Theis-Febvre et al., 2003).

Here, we demonstrate that CHK1 phosphorylates CDC25B on S230 in vivo and that this phosphorylated form of CDC25B localizes to the centrosomes from S phase until mitosis during normal cell-cycle progression. Since identical results were obtained in immunofluorescence studies using U2OS cells expressing HA-tagged CDC25B as well as HeLa cells, we can exclude the possibility that overexpression of CDC25B contributes to CHK1 activation. This observation correlates with the report showing the association of the CHK1 kinase with interphase centrosomes in several cell types (Kramer et al., 2004). It has recently been reported that CHK1 depletion leads to premature activation of CDK1–cyclin-B and mitotic catastrophy (Niida et al., 2005), indicating that CHK1 plays a key role in the control of the timely regulated progression of the cell into mitosis. Indeed centrosomal CHK1 has been shown to inhibit centrosomal CDK1 activity, which in turn activates centrosomal mitotic events (Kramer et al., 2004). As a CHK1 substrate, CDC25B is likely to be an essential player in this pathway. We propose a model (Fig. 4B) in which we postulate that, in an unperturbed cell cycle, CHK1 is involved in negatively regulating the activity of CDC25B through phosphorylation of several sites, including S230, thereby preventing premature activation of CDK1–cyclin-B at the centrosomes. In agreement with this hypothesis we find that mutation of S230 to the non-phosphorylatable alanine residue increases the mitosis-inducing activity of CDC25B. Moreover, it has also been reported that CDC25B with the S230A mutation was able to induce endogenous CDK1–cyclin-B1 activity more efficiently than wild-type CDC25B (Giles et al., 2003). When cells reach the G2-M transition, CHK1 has been reported to leave the centrosomes (Kramer et al., 2004), allowing CDC25B to become proficient to activate CDK1–cyclin-B1. Since CHK1-dependent phosphorylation of CDC25B on S230 can still be detected in the cell during mitosis, it is probable that additional sites are phosphorylated by other (mitosis-inducing) kinases. These additional phosphorylation events would allow the activation of CDC25B, first at the centrosomes before nuclear-envelope disassembly and then, more diffusely, throughout the cell, to counterbalance the negative effect of CHK1. According to this model, we have recently shown that, aurora-A kinase phosphorylates CDC25B in vivo at the centrosomes and this phosphorylation participates in the positive regulation of entry into mitosis (Cazales et al., 2005; Dutertre et al., 2004). Moreover, our data (Fig. 2A) show that, S230 phosphorylation is abundant in the slower electrophoretic mobility form of CDC25B that accumulates when cells progressively reach mitosis and probably corresponds to an active form of CDC25B that is phosphorylated on multiple sites by other kinases. However, the mechanism by which the resulting combination of phosphorylation events by multiple kinases modulates the activity of the CDC25B protein remains unclear.

Another question that remains to be elucidated is how CDC25B phosphorylation by CHK1 negatively regulates its activity. S230 has been described to be one of the three 14-3-3-protein-binding sites identified for CDC25B (together with S151 and S323) (Giles et al., 2003; Uchida et al., 2004) and it has been reported that 14-3-3 binding inactivates CDC25B by blocking the access of its substrate CDK/cyclin to the catalytic site (Giles et al., 2003). The proposed model is that S151 and S230, located in the N-terminal domain, both cooperate in binding dimeric 14-3-3 protein molecules anchored to the S323 site to close the structure and decrease accessibility to the catalytic site (Giles et al., 2003). Thus, a single mutation in one of the two N-terminal binding sites (e.g. S230) changes the equilibrium towards the more open and more active conformation. Interestingly, S151 and S323 were also identified in our mass spectrometry analyses (supplementary material Fig. S1) as in vitro CHK1-phosphorylated sites. Moreover, a subset of cellular 14-3-3 proteins have been found to be associated with centrosomes (Andersen et al., 2003; Pietromonaco et al., 1996), suggesting that, at the centrosomes, CDC25B phosphorylated on S230 by CHK1 may promote 14-3-3 binding. We have previously shown that interactions between isoforms of 14-3-3 protein and CDC25B do not increase upon phosphorylation of CDC25B by CHK1 (Mils et al., 2000). However, in that study we also report that binding of CDC25B to the σ isoform of 14-4-3, the preferred subtype for S230 binding (Uchida et al., 2004) was not assessed. Further studies are therefore required to determine whether the phosphorylation of CDC25B on S230 by CHK1 at the centrosomes inhibits its activity by promoting the binding of 14-3-3 σ.

In conclusion, centrosomal localization of both CHK1 (Kramer et al., 2004) and CDC25B (Dutertre et al., 2004) (and this study) support the idea that major regulatory pathways are integrated at the level of the centrosome to control entry into mitosis after successful completion of S phase and G2 phase (Jackman et al., 2003). Whether the CHK1-CDC25B pathway revealed in this study is also involved in the DNA-damage-induced inhibition of centrosomal CDK1 activity remains to be elucidated.

HeLa cells and a U2OS derivative cells conditionally expressing HA-tagged CDC25B3 (Theis-Febvre et al., 2003) were grown as previously described (Davezac et al., 2000). U2OS cells conditionally expressing HA-tagged CDC25B3 were synchronized by a double thymidine block as previously described (Dutertre et al., 2004). Expression of HA-tagged CDC25B3 was induced for at least 16 hours before harvesting. Cell synchronization was monitored by flow cytometry after propidium iodide staining. Cells were treated for 1 hour with 100 nM UCN-01 (a generous gift of B. Gass, NCI, Rockville, MD). HeLa cells were transfected with pEYFPC1 (control vector), pEYFPC1-CDC25B wild type, pEYFPC1-CDC25B-S230A or pEYFPC1-CDC25B-S563A using Lipofectin Reagent (Invitrogen) following manufacturer's instructions.

Cells were lysed as previously described (Dutertre et al., 2004). Cell lysates were electrophoresed on a 7.5% SDS-PAGE gel and analysed by western blotting. Overexpressed CDC25B3 was detected using anti-HA antibody (clone 12CA5, Roche). CDC25B rabbit polyclonal antibody (C-20) was purchased from Santa-Cruz Biotechnology. The rabbit polyclonal anti-CDC25B S230-P was described previously (Dutertre et al., 2004). The rabbit polyclonal antibodies against phosphorylated S230 and phosphorylated S563 were raised against peptides EAFAQRPSpSAPDLMC and LKTRpSWAGERSRC, respectively and affinity-purified as previously described (Dutertre et al., 2004) (Eurogentec SA). These antibodies were always used in the presence of non-phosphorylated peptide (20 μM). CHK1 (G-4) monoclonal antibody was obtained from Santa Cruz. Phosphorylated-CHK1 (at residue S345) polyclonal antibody was from Upstate. Anti-cdc2 and anti-cdc2 phosphorylated at residue Y15 were purchased from Cell Signaling Technology. Anti-actin (C4) monoclonal antibody was from Chemicon.

Recombinant HA-CDC25B protein (1 μg) was incubated in 20 μl of kinase buffer (50 mM Tris-Hcl pH 7.5, 10 mM MgCl2, 1 mM DTT and 100 μM ATP) with purified recombinant CHK1 kinase (100 ng, Upstate) at 30°C for 30 minutes. The reaction was stopped by the addition of Laemmli sample buffer then separated by SDS-PAGE and analysed by western blotting.

Cells were seeded onto glass coverslips then fixed and permeabilized 24 hours later as described (Davezac et al., 2000). Cells were subjected to a 1-minute pre-extraction with Triton X-100, followed by fixation in acetone-methanol (1:1 v/v) at –20°C for 7 minutes. Cells were further incubated at 37°C for 1 hour with CDC25B-S230-P and γ-tubulin antibodies. DNA was visualized using DAPI. Mouse anti-γ-tubulin (T6557) was from Sigma. For immunostaining, the antibody against S230-P CDC25B was used at a dilution of 1:500 in HeLa cells and 1:1000 in U2OS cells expressing HA-CDC25B3, in the presence of non-phosphorylated peptide (1000× molar excess). For the competition experiment, unphosphorylated or in vitro CHK1-phosphorylated recombinant HA-CDC25B protein (500 ng) was added during incubation with the primary antibody. Secondary antibodies labelled with Alexa Fluor-488 or Alexa Fluor-594 (Molecular Probes) were used at the dilution of 1:800 for 1 hour at room temperature. Images were acquired using a DM5000 microscope (Leica) fitted with a Roper COOL Snap ES CCD camera, and subsequently processed using the MetaMorph and Photoshop software packages.

For siRNA-mediated ablation of CDC25B, HeLa cells were transfected by Oligofectamine reagent (Invitrogen) twice for a total of 72 hours with Cdc25B-siRNA directed against the sequence 5′-UCCUCCCUGUCGUCUGAAU-3′ as described (Kramer et al., 2004; Lindqvist et al., 2004). Unrelated Drosophila sequence was used as control. For siRNA-mediated ablation of CHK1, siRNA duplexes directed against the sequence 5′-GAAGCAGUCGCAGUGAAGA-3′ as described (Ahn et al., 2003). were electroporated (Nucleofector Amaxa) in U20S cells following the manufacturer's instructions, and cells were examined by immunostaining or western blotting 48 hours post-transfection. Scrambled siRNA oligonucleotides, purchased from Dharmakon, were used as control.

We thank Jean Pierre Bouché, Martine Cazales and Odile Mondesert for their participation in this work. We are grateful to Cristina Cardoso for the gift of the DsRed1-DNA ligase expression plasmid, Clare McGowan for Chk1 plasmids and the NCI for the gift of UCN-01. This work was supported by grants from the CNRS, l'Université Paul Sabatier, EDF, Genopôle Toulouse Midi-Pyrénées (to B.M.) and la Ligue National Contre le Cancer to BD (Equipe labellisée). E.S. was a recipient of a post-doctoral fellowship from the Fonds de la recherche en santé du Québec. R.B. is a recipient of a post-doctoral fellowship from the French Ministry of Research.