Regulation of Cdc25A Half-life in Interphase by Cyclin-dependent Kinase 2 Activity*

Cdc25A regulates cell cycle progression, has oncogenic and anti-apoptotic activity, and is over-expressed in many human tumors. Phosphorylation by Chk1 and Cds1/Chk2 down-regulates Cdc25A levels in response to genotoxic stresses. Nevertheless, it remains unclear whether Chk1 and Cds1/Chk2 are uniquely responsible for regulating Cdc25A stability during interphase or if other kinase activities contribute. Here we report that treatment of HeLa cells with the cyclin-dependent kinase inhibitor roscovitine caused a concentration- and time-dependent increase in Cdc25A protein levels. Transfection with dominant-negative Cdk mutants dem-onstrated that only a Cdk2 mutant increased Cdc25A protein levels; Cdk1 and Cdk3 mutants had no effect. The increased Cdc25A protein levels were the result of an increase in the half-life of the protein; no increase in Cdc25A mRNA levels was observed. These results dem-onstrate Cdk2 kinase activity contributes to the labile nature of Cdc25A during interphase and redefine the nature of the Cdc25A-Cdk2 autoamplification feedback loop. The dual-specificity phosphatases catalyze cell cycle progression by dephosphorylating and activating the cyclin-de-pendent kinases (Cdk) 1 ; the three human cdc25 homologs, Cdc25A, Cdc25B, Cdc25C, different phases of the cell cycle. Cdc25C in mitosis catalyzes mitotic Cdk1/cyclin and MCF-7 human mammary adenocarcinoma cells (American Tissue Culture Collection, Manassas, VA) were maintained in Dulbecco’s min-imum essential medium containing 10% fetal bovine serum (HyClone, Logan, UT) and 1% penicillin-streptomycin (Invitrogen) in a humidified atmosphere of 5% CO 2 at 37 °C. Western Blotting— Cells were harvested and lysed in a HEPES lysis (30 m M HEPES, 1% Triton X-100, 10% 5 m M MgCl 2 , m M NaF, m M EGTA, M NaCl, M 4 , (cid:2) soybean trypsin inhibitor, 10 (cid:2) g/ml leupeptin, 10 (cid:2) g/ml aprotinin, 100 (cid:2) g/ml 4-(2-aminoethyl)benzenesulfonyl fluoride, 6.4 mg/ml Sigma phosphatase substrate), incubated on ice for 30 min, and centrifuged at 13,000 (cid:2) g to clear the lysates. Protein content was determined by the Bradford method. Total cell lysates (30–50 (cid:2) g protein) were resolved by SDS-PAGE and transferred to nitrocellulose membranes (Schleicher & Schuell). Membranes were incubated in blocking solution and probed with primary antibodies overnight. Positive antibody reactions were visualized using peroxidase-conjugated secondary antibodies and an enhanced chemiluminescence detection system (Renaissance, Perkin-Elmer Life Sciences) according to the manufacturer’s instructions. For quantitation of protein expression levels, x-ray films were scanned on an Amersham Biosciences personal SI densitometer and analyzed using the ImageQuant software package (version Amersham Biosciences). Transfections— HeLa

The Cdc25 dual-specificity phosphatases catalyze cell cycle progression by dephosphorylating and activating the cyclin-dependent kinases (Cdk) 1 ; the three human cdc25 homologs, Cdc25A, Cdc25B, and Cdc25C, regulate different phases of the cell cycle. Most investigators believe that Cdc25C functions primarily in mitosis and catalyzes mitotic progression by activating Cdk1/cyclin B, as Cdc25C is the human Cdc25 isoform most homologous to yeast and Xenopus Cdc25 (1)(2)(3)(4). Cdc25B was originally characterized as functionally redundant to Cdc25C because of its ability to activate Cdk1/cyclin B (5). More recently, Cdc25B has been implicated as the activator of Cdk1/cyclin B at the onset of mitosis and of Cdk2/cyclin A in late G 2 and as the target of a Chk1-and Cds1/Chk2-independent G 2 /M checkpoint (6 -8). Because the primary Cdk substrate for Cdc25A seems to be Cdk2/cyclin E, Cdc25A was relegated to promoting the G 1 /S cell cycle transition and S phase progression (9,10). Cdc25A protein levels and activity, however, remain present after S phase and seem to increase as cells enter mitosis (9,11). It has been recently reported that Cdk1/cyclin B-mediated phosphorylation of Cdc25A increases its stability in mitotic cell populations, further supporting a role for Cdc25A in mitosis, although the functional significance of elevated Cdc25A activity throughout G 2 and mitosis remains unclear (12).
An essential cellular alteration for malignant transformation is deregulation of cell cycle control proteins (13). Indeed, over-expression of Cdc25A has been reported to transform normal mouse embryonic fibroblasts in cooperation with an oncogenic isoform of Ras (Ha-Ras G12V ) or in an Rb Ϫ/Ϫ background (14) and Cdc25A over-expression has been documented in numerous human cancers (15). The oncogenic activity of Cdc25A can be attributed to its impingement on several signaling pathways regulating cell cycle checkpoints, growth factor, and hormonal mitogenesis, apoptosis, and senescence (15,16).
Because the oncogenic potential of Cdc25A is potentially dependent on both its abundance and its catalytic activity, the mechanisms that regulate Cdc25A activity and expression level are of considerable interest. Transcriptional regulation of the Cdc25A promoter has been attributed to both E2F and c-Myc transcription factors and seems to be cell cycle-dependent, with increases in Cdc25A mRNA occurring predominantly prior to S phase, consistent with its essential role in S phase induction (9,10,(17)(18)(19). Post-translational modification of Cdc25A has both positive and negative effects on its activity and protein levels. Cdc25A protein levels are tightly regulated by proteasomemediated degradation pathways that may involve multiple ubiquitin ligases (11, 20 -23). Cdc25A is phosphorylated by Cdk2/cyclin E in a positive feedback loop, which increases the activity of both proteins sufficiently to cross the threshold required for the G 1 /S transition (24,25). Cdc25A-activating phosphorylations have also been attributed to Raf-1 and Pim-1 kinases (26,27). Although the phosphorylation of Cdc25A by Cdk1/cyclin B has not been reported to increase its phosphatase activity per se, it does lead to increased protein stability in mitotic cell populations (12). On the other hand, Cdc25A protein stability is negatively regulated in a cell cycle checkpointdependent manner by phosphorylation at serine 123, enabling poly-ubiquitination and subsequent degradation (11,21,22). However, the mechanisms regulating Cdc25A protein stability in the absence of genetic insults remain unclear.
Because of the highly labile nature of Cdc25A protein, we hypothesized that a candidate for interphase regulation of Cdc25A protein levels in the absence of genetic insults would be a proximal downstream effector. By analogy, the stability of Cdc25B is decreased following phosphorylation by one of its proximal downstream effectors, Cdk1/cyclin A (28). To explore the possibility that Cdk-mediated phosphorylation of Cdc25A contributes to its inherent instability in interphase, we treated HeLa cells with the Cdk inhibitors roscovitine or olomucine and found that inhibition of Cdk activity resulted in a concen-tration-and time-dependent increase in Cdc25A protein levels. Because of the selectivity profile for roscovitine and olomucine, we employed dominant-negative Cdk mutants to determine which Cdk activity contributed to altered Cdc25A protein levels. Cdc25A protein levels were uniquely increased by a dominant-negative Cdk2 mutant; no increase was seen with either dominant-negative Cdk1 or Cdk3 mutants. The increased Cdc25A protein levels appeared to be the result of an increase in the half-life of the protein and no increase in Cdc25A mRNA levels was observed. These results support the hypothesis that Cdk2 kinase activity contributes to the labile nature of Cdc25A during interphase and describe how Cdc25A protein levels can be maintained under strict control until increased protein levels are necessary as cells approach mitosis.
Western Blotting-Cells were harvested and lysed in a HEPES lysis buffer (30 mM HEPES, 1% Triton X-100, 10% glycerol, 5 mM MgCl 2 , 25 mM NaF, 1 mM EGTA, pH 8, 10 mM NaCl, 2 mM Na 3 VO 4 , 10 g/ml soybean trypsin inhibitor, 10 g/ml leupeptin, 10 g/ml aprotinin, 100 g/ml 4-(2-aminoethyl)benzenesulfonyl fluoride, 6.4 mg/ml Sigma 104 phosphatase substrate), incubated on ice for 30 min, and centrifuged at 13,000 ϫ g to clear the lysates. Protein content was determined by the Bradford method. Total cell lysates (30 -50 g protein) were resolved by SDS-PAGE and transferred to nitrocellulose membranes (Schleicher & Schuell). Membranes were incubated in blocking solution and probed with primary antibodies overnight. Positive antibody reactions were visualized using peroxidase-conjugated secondary antibodies and an enhanced chemiluminescence detection system (Renaissance, Perkin-Elmer Life Sciences) according to the manufacturer's instructions. For quantitation of protein expression levels, x-ray films were scanned on an Amersham Biosciences personal SI densitometer and analyzed using the ImageQuant software package (version 4.1, Amersham Biosciences).
Transfections-HeLa cells were transfected with plasmids encoding dominant-negative mutants of Cdk1, Cdk2, and Cdk3 using Lipo-fectAMINE PLUS™ in serum-containing medium according to the manufacturer's instructions. Media containing the DNA-lipid complexes was removed after 3 h and replaced with complete growth medium, and cells were harvested after 48 h. Protein lysates were prepared and analyzed by SDS-PAGE and Western blot analysis as described above.
RNA Isolation and Northern Blotting-Total RNA was isolated from HeLa cells using an RNeasy Kit (Qiagen, Valencia, CA). RNA concentrations were determined spectrophotometrically using a DU640 spectrophotometer (Beckman Instruments, Fullerton, CA). Northern blotting was performed using NorthernMax™ system (Ambion, Austin, TX) according to the manufacturer's instructions. Briefly, 5 g of total RNA was separated on 1% denaturing agarose gel containing 2.2 M formaldehyde, transferred to Nytran® SuPerCharge membrane (Schleicher & Schuell), UV cross-linked, and processed for detection of mRNA. A 711-base pair DIG-labeled antisense single-stranded DNA probe was generated by asymmetric PCR amplification using a PCR DIG Probe Synthesis Kit. Briefly, the template for probe synthesis was a 711-bp PCR product at the 3Ј end of human Cdc25A cDNA; this template was generated by conventional PCR methodology using the following primers: 5Ј-AAGAGGAGGAAGAGCATGTC-3Ј (Primer A) and 5Ј TCAGA-GCTTCTTCAGACGACЈ3Ј (Primer B). The DIG-labeled probe was generated by asymmetric PCR from this template using primer B. Overnight hybridization of the probe to the immobilized RNA was carried out in ULTRAhyb™ Ultrasensitive Hybridization Buffer (Ambion), and the membrane was processed using DIG Wash and Block Buffer Set. The hybridized probe-anti-DIG-AP complex was visualized on x-ray film (Eastman Kodak Co.) after incubation of the membrane with CDP Star. Relative intensities of the hybridization signals were quantified as described above for Western blotting.

RESULTS
Exposure of asynchronous HeLa cells to the Cdk inhibitor roscovitine (10 M) resulted in a marked increase in Cdc25A protein levels at 24 h (Fig. 1). In accordance with previously published results (28), roscovitine treatment of HeLa cells resulted in an increase in Cdc25B levels, presumably because of inhibition of Cdk1/cyclin A-mediated targeting of Cdc25B for proteasomal-mediated degradation (Fig. 1). Roscovitine treatment had no effect on Cdc25C protein levels, the activity of which is regulated predominantly by cytoplasmic sequestration and inactivation (Fig. 1) (30 -32). Cdc25A protein levels increased in a concentration-and time-dependent manner (Figs. 2 and 3), suggesting that this increase was due to the specificity of roscovitine as a Cdk inhibitor. Similar results were obtained with olomucine, a Cdk inhibitor with a similar selectivity profile but with reduced potency (Fig. 2C and data not shown). Because Cdc25A protein levels were elevated rapidly, namely within 1 h of roscovitine treatment (Fig. 3), it seems unlikely that the increased Cdc25A protein levels were simply due to cell cycle perturbation.
Regulation of Cdc25A protein levels by DNA damage checkpoints has been reported previously to be a p53-independent event; Cdc25A levels are also known to be affected by the high risk human papillomavirus (HPV) E7 oncoprotein (19,22,33). To test whether the increase in Cdc25A protein levels due to Cdk inhibition was dependent on p53 or HPV status (HeLa cells are HPV-positive), we treated MCF-7 human mammary adenocarcinoma cells, wild type for p53 and HPV-negative, with increasing concentrations of roscovitine for 24 h. As seen in HeLa cells, Cdc25A protein levels in MCF-7 cells increased in a concentration-dependent manner following roscovitine treatment, indicating that increases in Cdc25A levels resulting from Cdk inhibition were independent of p53 activity or HPV status (Fig. 2B).
Because roscovitine is a broad spectrum Cdk inhibitor, we next determined which Cdk was involved in regulating Cdc25A protein levels. HeLa cells were transfected with dominantnegative mutants of Cdk1, Cdk2, and Cdk3, as these are the predominant roscovitine-sensitive Cdks in HeLa cells. Only genetic inhibition of Cdk2 kinase activity resulted in increased Cdc25A protein levels; genetic inhibition of a Cdk1 or Cdk3 had no effect on Cdc25A levels (Fig. 4). We observed no significant alteration in the HeLa cell cycle profile 48 h after transfection with the dominant-negative Cdk2 mutant, consistent with the recent report by Tetsu and McCormick (34). Thus, the increase in Cdc25A protein levels after ectopic expression of the dominant-negative mutant of Cdk2 was not secondary to cell cycle arrest. These results indicate that Cdk2 kinase activity plays an important role in regulating Cdc25A protein levels in asynchronous cells.
Because Cdc25A expression can be regulated at both the transcriptional and post-translational levels, we next investigated the mechanism responsible for increases in Cdc25A levels following inhibition of Cdk2 activity. In response to genetic insults or inhibition of DNA synthesis, Cdc25A is phosphoryl- ated and targeted for rapid ubiquitin-mediated proteolytic degradation by the checkpoint kinases Chk1 and Cds1/Chk2 (11,21,22). In addition, transcription from the Cdc25A promoter can be activated by E2F and c-Myc transcription factors (17)(18)(19). Because Cdc25A has been reported to be a highly labile protein in interphase and an increase in the half-life of a labile protein would result in a significant accumulation of that protein, we explored whether inhibition of Cdk2 kinase activity altered the half-life of Cdc25A in asynchronous cells. Following a 24-h treatment with roscovitine or vehicle control, HeLa cells were treated with 10 g/ml cycloheximide for 0 -60 min, and Cdc25A levels were examined by Western blotting. The basal half-life of Cdc25A was 6.26 Ϯ 0.78 min, which is in agreement with previous reports (11,20). Roscovitine-mediated inhibition of Cdk2 kinase activity doubled the half-life of Cdc25A (Fig. 5), which readily could account for the observed time-dependent increases in Cdc25A protein levels. To confirm that the increased Cdc25A protein levels were not affected by a transcriptionally mediated mechanism, Cdc25A mRNA levels were ex-amined by Northern blotting. Roscovitine treatment of HeLa cells did not significantly increase Cdc25A mRNA levels, confirming that Cdk2 kinase activity affects Cdc25A protein levels by a post-transcriptional mechanism (Fig. 6). These results were independently confirmed by reverse transcriptase-PCR (data not shown).

DISCUSSION
Cdc25A biology is undergoing a paradigm shift, drifting away from its narrow role as critical regulator of the G 1 /S transition to a more broad responsibility in the cell cycle with an essential function in mitosis. Specifically, it is now known that Cdc25A levels are at their highest during late G 2 /M and that degradation of Cdc25A is necessary for the G 2 /M checkpoint in response to DNA damage (11,12). The original models describing the regulation of Cdc25A are being refined to include recent data thoroughly detailing protein stability as one of its key regulatory mechanisms (11, 12, 20 -23). The relationship between Cdc25A and Cdk2 was originally that of an autoamplification feedback loop in which Cdk2 contributed to the activation of Cdc25A and Cdc25A contributed to the activation of Cdk2 to amplify the activities of both proteins to a high enough level to enable progression through the G 1 /S transition (9). Here we report that Cdk2 kinase activity contributes to the labile nature of Cdc25A in interphase, and this kinase activity may in fact be the same Cdk2 kinase activity originally reported to activate Cdc25A. Our results contribute to the understanding of this Cdc25A-Cdk feedback loop and support a mathematical model suggesting that hyperphosphorylation of Cdc25A by Cdk2 may contribute to its degradation (9,35). By directly linking Cdc25A stability to the activity of its substrates, physiologic levels of Cdc25A can be maintained in a tight feedback loop to prevent catastrophic deregulated Cdc25A protein levels or activity. This relationship between increased activity and decreased protein stability has been described for another protein phosphatase, PTEN (phosphatase and tensin homolog). PTEN phosphorylation maintains the protein in a stabilized state with decreased phosphatase activity; upon loss of phosphorylation in the C-terminal PTEN tail, catalytic activity is increased and protein stability is decreased (36). Although our data do not specifically detail the nature of the Cdk2-cyclin complex that contributes to the inherent instability of Cdc25A in interphase or the detailed molecular mechanism involved, there are several possible testable hypotheses. It has been reported recently that Cdc25A can associate with elements of the SCF ubiquitin ligases and may be a target of the APC cdh1 and SCF ubiquitin ligases (20). However, it remains uncertain how Cdc25A might be targeted to these ubiquitin ligases. Other cell cycle regulatory proteins, notably p27 and cyclin E, are targeted by the SCF ubiquitin ligase for proteolytic degradation by a phosphorylation-dependent mechanism, whereas conversely, phosphorylation may not be necessary for p21 and cyclin D degradation mediated by SCF ligases (37,38). It remains unclear whether phosphorylation of Cdc25A is a necessary event preceding ubiquitin ligase association, as Cdc25A is phosphorylated prior to its degradation in response to genetic insults and is rescued from proteolytic degradation in mitosis by Cdk1/cyclin B-mediated phosphorylation (12). However, our results support a role for Cdk2-mediated phosphorylation of either Cdc25A itself or a specific effector protein(s) necessary for the rapid degradation of Cdc25A. Although our results cannot rule out the involvement of Chk1 and Cds1/Chk2 as downstream effectors of the Cdk2 kinase activity responsible for Cdc25A degradation, their role may be unique to proteasomal targeting of Cdc25A following cellular stress and may not play a role in regulating Cdc25A levels in the absence of cellular stress. A similar model seems to regulate Cdc25C, which is inactivated in the G 2 /M checkpoint by checkpoint kinase(s)-dependent phosphorylation, 14-3-3 association and cytoplasmic sequestration. Cdc25C, however, is maintained inactive and sequestered in the cytoplasm during interphase in the absence of cellular stress by the Cdc25Cassociated protein kinase (C-TAK1) (39).
Based on our results and the above mentioned studies, we propose the following model. Cdc25A levels are up-regulated by transcription in late G 1 to a level that enables Cdk2/cyclin E activation to promote transition from G 1 into S phase. Then Cdc25A levels are carefully controlled via Cdk2/cyclin E and Cdk2/cyclin A kinase activity through S phase and into late G 2 phase. Once Cdk1/cyclin B is activated, Cdc25A protein levels are released from this strict regulatory loop and are permitted to increase as the cells approach the G 2 /M transition and reach their maximal levels, which are required for mitosis (11,12).
As deregulated cell cycle progression is one of the hallmarks of cancer, regulation of cell cycle proteins has taken a prominent position in efforts to design new therapeutic approaches, including inhibitors of Cdks (13). While rational drug discovery efforts are laudable, our results provide a cautionary note. They suggest that inhibitors of the catalytic activity of Cdk2 may have the unexpected consequence of elevating the expression of the proto-oncogene Cdc25A.