Pro-apoptotic Role of Cdc25A

Cdc25A is a dual specificity protein phosphatase that activates cyclin/cyclin-dependent protein kinase (Cdk) complexes by removing inhibitory phosphates from conserved threonine and tyrosine in Cdks. To address how Cdc25A promotes apoptosis, Jurkat cells were treated with staurosporine, an apoptosis inducer. Upon staurosporine treatment, a Cdc25A C-terminal 37-kDa fragment, designated C37, was generated by caspase cleavage at Asp-223. Thr-507 in C37 became dephosphorylated, which prevented 14-3-3 binding, as shown previously. C37 exhibited higher phosphatase activity than full-length Cdc25A. C37 with alanine substitution for Thr-507 (C37/T507A) that imitated the cleavage product during staurosporine treatment interacted with Cdc2, Cdk2, cyclin A, and cyclin B1 and markedly activated cyclin B1/Cdc2. The dephosphorylation of Thr-507 might expose the Cdc2/Cdk2-docking site in C37. C37/T507A also induced apoptosis in Jurkat and K562 cells, resulting from activating cyclin B1/Cdc2 but not Cdk2. Thus, this study reveals that Cdc25A is a pro-apoptotic protein that amplifies staurosporine-induced apoptosis through the activation of cyclin B1/Cdc2 by its C-terminal domain.

Cdc25A with oncogenic potential works together with either H-Ras G12V or deprivation of RB1 in tumorigenic transformation of primary mouse embryonic fibroblasts (9). Hemizygous loss of Cdc25A in mice suppresses mammary tumorigenesis induced by H-Ras 12V or Her2 (10), indicating that Cdc25A is a rate determiner in the mammary transformation. Furthermore, the overexpression of Cdc25A is correlated with poor clinical outcomes in a variety of human cancers.
The stability of Cdc25A is regulated by the ubiquitin-proteasome pathway. During mitotic exit and the early G 1 phase, Cdc25A is degraded through Cdh1-directed anaphase-promoting complex, which also requires a KEN box motif between residues 141 and 143 (11). Subsequently, the phosphorylation of Ser-76 and Thr-80 by the Plk-3/GSK-3␤ pathway promotes Cdc25A ubiquitination and turnover during the middle and late G 1 phase (12). During the S phase, SCF ␤-TRCP regulates Cdc25A degradation in an unperturbed cell division cycle and in response to DNA damage, requiring phosphorylation of serines 76, 79, 82, and 85 (13)(14)(15). The Cdc25A destruction induced by ionizing and UV irradiation requires the phosphorylation of Ser-123 mediated by the ATM/Chk2 or ATR/Chk1 pathway (7,16,17) and the phosphorylation of Ser-76, respectively (18,19). Rapid Cdc25A degradation in response to DNA damage is a part of the DNA damage checkpoint mechanism that protects genomic integrity by arresting the cell cycle progression and allowing cells to repair damaged DNA.
Although Cdc25A plays a clear role in both cell cycle regulation and cell cycle checkpoint control, it is unknown how Cdc25A promotes apoptosis. It has been reported that the ectopic expression of Cdc25A induces apoptosis in 3T3 LI cells depleted of growth factors (20), but this mechanism remains unclear. Cdc25A activates Cdc2 or Cdk2 by removing inhibitory phosphates from Cdks. Cdc2 and Cdk2 are activated in a caspase-dependent manner during Fas-or staurosporine-induced apoptosis (21). Moreover, the activation of Cdc2 is required for apoptosis induced by fragmentin-2 or staurosporine (26). The activation mechanism for Cdc2 or Cdk2 during apoptosis remains unclear. After examining the amino acid sequence of Cdc25A, we discovered several possible caspase consensus motifs. Based on this information, we hypothesize that Cdc25A is a substrate for caspase, and the Cdc25A fragment generated by caspase cleavage activates Cdc2 and/or Cdk2, resulting in apoptosis. Thus, we used staurosporine, a typical inducer for apoptosis, to elucidate the possible role of Cdc25A in promoting apoptosis in Jurkat cells. This study reports a novel molecular mechanism in which Cdc25A, a pro-apoptotic protein, promotes apoptosis. Upon staurosporine treatment, a Cdc25A C-terminal 37-kDa fragment, designated C37, was generated by caspase cleavage at Asp-223, and its Thr-507 became dephosphorylated. C37 with alanine substitution for Thr-507 (C37/T507A) interacted with Cdc2, Cdk2, cyclin A, and cyclin B1 and significantly activated cyclin B1 and Cdc2associated kinase activities. The dephosphorylation of Thr-507 might expose the Cdc2/Cdk2-docking site in C37. In the absence of staurosporine, C37/T507A induced apoptosis, resulting from activating cyclin B1/Cdc2 but not Cdk2. In summary, staurosporine causes caspase to cleave Cdc25A and promotes the dephosphorylation of Thr-507. As a result, the cyclin B1-Cdc2 complex is activated but not through the activation of Cdk2, which leads to apoptosis. Therefore, this study presents a novel pathway of staurosporine induced-apoptosis.
Plasmid pcDNA3-CD20 was used to construct serial plasmids expressing fusion proteins with a presumed caspase consensus motif from Cdc25A between N-terminally tagged His 6 and CD20. CD20 cDNA was amplified by PCR using pCD20 as a template and two primers (supplemental Table S2) and subsequently subcloned to pcDNA3 through XhoI. pcDNA3-His-CD20 was prepared by subcloning a CD20 BamHI/XbaI fragment from pcDNA3-CD20 to pcDNA3-His. CD20 in pcDNA3-His-CD20 is not in the reading frame until the adaptor (supplemental Table S3) encoding a presumed caspase consensus motif containing eight amino acids from Cdc25A was subcloned to pcDNA3-His-CD20 through BamHI and NotI.
All of the mutation sites were mutated by site-directed mutagenesis using the QuikChange XL site-directed mutagenesis kit (Stratagene) and primers listed in supplemental Table S4. Each mutation site was verified by sequencing.
Transfection and Small Interfering RNA Knockdown Experiments-To transfect Jurkat or K562 cells, 2 ϫ 10 5 cells were first suspended in 10 l of buffer R containing different amounts of plasmid DNA. Jurkat cells were then electroporated with MicroPorator MP-100 (Digital Bio) at a pulse voltage of 1240, pulse width of 30, and pulse number of 1. Alternatively, K562 cells were electroporated at a pulse voltage of 1000, pulse width of 50, and pulse number of 1. The cells were then seeded to a well containing 500 l of prewarmed RPMI complete medium. Depending on the experiments, cells were harvested at different time periods after electroporation. HeLa or 293T cells were transfected with Lipofectamine 2000 following the steps recommended by the manufacturer (Invitrogen).
Knockdown of Cdc25A, Cdc2, Cdk2, or cyclin B1 in Jurkat cells was achieved through delivery of small interfering RNA (siRNA) (supplemental Table S5) by electroporation. 2 ϫ 10 5 Jurkat cells were suspended in 10 l of buffer R containing either 8.4 M Cdc25A, Cdc2, or Cdk2 siRNA or 7.4 M cyclin B1 siRNA. Depending on the experiments, cells were harvested at different time periods after electroporation. Green fluorescent protein and luciferase siRNAs (supplemental Table S5) were used as controls. Small interfering RNAs were synthesized by Ambion.
To detect sub-G 1 cells, cells were fixed with 0.5% paraformaldehyde in phosphate-buffered saline (pH 8.0) at room temperature for 10 min, washed with phosphate-buffered saline, permeabilized with ice-cold 70% ethanol, stained by 30 g/ml propidium iodide, and analyzed with a FACSCalibur cytometer.
To examine phosphorylation status of Thr-507 in Cdc25A, Jurkat cells were electroporated with pcDNA3-3ϫFLAG Cdc25A/224 -524, treated with vehicle or 300 nM staurosporine at a 37°C incubator for 2 h, and lysed in MCLB buffer with the described supplements. Each cell lysate was incubated overnight with M2 antibody-agarose beads at 4°C, and each bead pellet was washed with MCLB buffer six times. Proteins immobilized in the beads were resolved by SDS-PAGE and analyzed by immunoblotting using antibodies as indicated in the figures.
To determine whether there is a Cdc2/Cdk2-docking site in Cdc25A/224 -524/T507A, 293T cells were transfected with either pcDNA3-3ϫFLAG Cdc25A/224 -524/T507A or Cdc25A/224 -524/R446L/R450L/Y455A/T507A, suspended in buffer A (2) with the supplements, and lysed by passage through a 27-gauge needle 30 times. Each cell lysate was precleared by centrifugation at 100,000 ϫ g for 1 h followed by incubation with 10 l of protein A-agarose beads at 4°C for 2 h. Each precleared cell lysate was incubated with 10 l of M2 antibodyagarose beads at 4°C for 2 h. The resulting immunoprecipitates were washed with buffer A six times or buffer A three times plus buffer A with 30 mM NaCl three times (for immunoprecipitation using Cdc2 antibody) and analyzed by SDS-PAGE and immunoblotting using antibodies as indicated in figures.
Expression and Purification of Bacterial GST Fusion Proteins and Phosphatase Activity Assay-DH5␣ cells were transformed with various pGEX2TЈ constructs encoding GST-tagged Cdc25A or its mutants. GST-tagged Cdc25A and mutants were purified following the method described previously (2). Proteins immobilized in glutathione beads (GE Healthcare) were eluted with 40 mM glutathione reduced form in 50 mM Tris-HCl (pH 8.0) and dialyzed against 20 mM Tris-HCl (pH 8.0) containing 1 mM dithiothreitol. Purified proteins were concentrated with a YM-30 Centricon (Armicorn) and stored at Ϫ80°C.
The phosphatase activity of 5 or 10 g of GST-tagged Cdc25A or its mutants was assayed using 3-O-methylfluorescein phosphate (Sigma) as substrates. The detailed steps of this process have been described previously (2). The absorbance values at 477 nm were normalized with the number of moles of GST-tagged Cdc25A or its mutants. The absorbance value at 477 nm of 59 pmol of GST-Cdc25A was set at 1.
Kinase Assay-293T cells were transfected with empty vector (pcDNA3-3ϫFLAG) or vector expressing 3ϫFLAG-tagged Cdc25A/224 -524/T507A, and lysed with MCLB with the supplements 24 h after transfection. Cell lysates were precleared by centrifugation at 100,000 ϫ g for 1 h and then incubated with protein A-agarose beads at 4°C for 2 h. Each cell lysate containing 150 g of protein was immunoprecipitated with 2 g of cyclin A or Cdk2 antibody immobilized on 10 l of protein A-agarose beads or 5 l of anti-cyclin B1-agarose beads (Santa Cruz Biotechnology). Alternatively, 300 g of protein was immunoprecipitated overnight with 10 l of anti-Cdc2-agarose beads (Santa Cruz Biotechnology) at 4°C. Each bead pellet was washed with MCLB buffer four times and then subjected to kinase assay using a Cdk1/Cdc2 kinase assay kit (Upstate). The kinase reaction incubation time was 10 min for cyclin A, cyclin B1, or Cdk2 immunoprecipitates, and 30 min for Cdc2 immunoprecipitates.

Cdc25A Involvement in Staurosporine-induced Apoptosis Is through Its 37-kDa C-terminal Fragment Generated by Caspase
Cleavage-To understand how Cdc25A promotes apoptosis, Jurkat cells were treated with staurosporine or anti-Fas antibody, functioning as a ligand for Fas, to induce apoptosis. At 2 h after staurosporine treatment, a 37-kDa Cdc25A C-terminal fragment was generated (Fig. 1A). This Cdc25A C-terminal fragment also appeared during 500 or 750 ng/ml anti-Fas antibody treatment (Fig. 1B). Poly(ADP-ribose) polymerase, a known caspase substrate, was proteolyzed concomitantly with the generation of the Cdc25A C-terminal fragment (Fig. 1, A  and B). These results suggest that caspase mediates the generation of the Cdc25A C-terminal fragment.
Cdc25A is an activator for Cdc2 and Cdk2. Cdc2 and Cdk2 were activated in a caspase-dependent manner during staurosporine-induced apoptosis, as reported by Zhou et al. (21). Tyr-15 in Cdc2 and Cdk2 was gradually dephosphorylated 1 h after staurosporine treatment, as shown in immunoblot analysis ( Fig. 1C) and phospho-specific flow cytometric analysis (Fig.  1D). The extent of Cdc2 and Cdk2 dephosphorylation was consistent with an increase in caspase 3 activity and apoptotic cells during staurosporine treatment ( Fig. 1, C-E). These results agree with the report of Zhou et al. (21) and show that staurosporine treatment induces apoptosis in Jurkat cells, activates caspases and Cdc2/Cdk2, and promotes the generation of a 37-kDa Cdc25A C-terminal fragment.
To determine whether the Cdc25A C-terminal fragment is generated by caspase cleavage and plays a role in staurosporineinduced apoptosis, Jurkat cells were pretreated with vehicle or caspase inhibitor I in the presence or absence of staurosporine.
Caspase inhibitor I pretreatment blocked the generation of the Cdc25A C-terminal fragment ( Fig.  2A) and inhibited staurosporine-induced apoptosis (Fig. 2B). Furthermore, the knockdown of endogenous Cdc25A using siRNA against Cdc25A suppressed staurosporineinduced apoptosis, as indicated by immunoblotting (Fig. 2C, left panel) and a decrease in annexin V-positive cells after staurosporine treatment (Fig. 2C, right panel). These results suggest that Cdc25A is a proapoptotic protein and is involved in staurosporine-induced apoptosis through the Cdc25A C-terminal fragment generated by caspase cleavage.
In Cdc25A, Aspartate 223 Is the Caspase Cleavage Site to Generate the 37-kDa C-terminal Fragment, and There Are Eight Other Caspase Cleavage Sites N-terminal of Aspartate 223-C-terminal His-tagged Cdc25A was similarly cleaved as endogenous Cdc25A to generate the Cdc25A C-terminal fragment during staurosporine treatment (Fig. 3A). The Cdc25A C-terminal fragments accumulated abundantly at 1 h after staurosporine treatment (Fig. 3A, lane 3) and decreased thereafter (lane 4). The accumulation and decline of the ectopic Cdc25A C-terminal fragment levels were faster than that of endogenous Cdc25A (Fig. 1A). This phenomenon may be due to a slight activation of caspase by the ectopic expression of Cdc25A and/or electroporation.
To identify the caspase cleavage site to generate the Cdc25A Cterminal fragment, site-directed mutagenesis was performed. Mutation at Asp-223 to alanine blocked the generation of the Cdc25A C-terminal fragment after staurosporine treatment (Fig. 3B, lane 5) as compared with the C-terminal Histagged Cdc25A (lane 3). Moreover, the C-terminal His-tagged Cdc25A/224 -524 migrated in the gel at the same position as that of the Cdc25A C-terminal fragment generated after staurosporine treatment (Fig. 3B, lanes 3 and 7). The migration of the C-terminal His-tagged Cdc25A/224 -524 without staurosporine treatment was slightly slower than that with treatment ( Fig. 3B, lanes 6 and 7), which was due to the protein kinase inhibition effect of staurosporine (22). The levels of the C-ter-FIGURE 1. Cdc25A is cleaved to generate a 37-kDa C-terminal fragment, and tyrosine 15 in Cdc2/Cdk2 is dephosphorylated during staurosporine treatment. 1 ϫ 10 6 /ml Jurkat cells were treated with 300 nM staurosporine (A) or 500 or 750 ng/ml anti-Fas antibody (Ab) (B) for various time periods, as indicated. Cell lysate from each indicated time point was analyzed by SDS-PAGE and immunoblotting with antibody against poly-(ADP-ribose) polymerase (PARP), Cdc25A (against an epitope at the C terminus), or ␣-tubulin. The protein level of ␣-tubulin was used as a loading control. C, 1 ϫ 10 6 /ml Jurkat cells were treated with 300 nM staurosporine for various times, as indicated. Immunoblot analysis was performed using phospho-Tyr-15-Cdc2 (against phosphorylated tyrosine 15 in Cdc2 and Cdk2), Cdc2, Cdk2, cyclin A, cyclin B1, cleaved caspase 3 (only against activated caspase 3), or ␣-tubulin antibody. D, Jurkat cells were treated with staurosporine for various time periods as in C, immunostained with phospho-Tyr-15-Cdc2 antibody and anti-rabbit IgG antibody conjugated with Alexa Fluor-546, and analyzed by flow cytometry. The percentage of cells with dephospho-Tyr-15-Cdc2/ Cdk2 after staurosporine treatment was indicated by subtraction from the overlap before and after staurosporine treatment. Data are displayed as means Ϯ S.D., n ϭ 3. E, Jurkat cells were treated with staurosporine for various time periods as in C and subjected to annexin V staining as a marker of early apoptosis and flow cytometric analysis. Data are displayed as means Ϯ S.D., n ϭ 3.
minal His-tagged Cdc25A/224 -524 with and without staurosporine treatment were very similar (Fig. 3B, lanes 6 and 7), whereas the level of the N-terminal His-tagged Cdc25A/1-223 after staurosporine treatment decreased dramatically (lanes 8 and 9). Furthermore, the Cdc25A C-terminal fragment generated during Fas-dependent apoptosis was the same as that generated during staurosporine-induced apoptosis (Fig. 3C). These results show that Asp-223 is the cleavage site to generate the Cdc25A C-terminal domain, and there are other possible caspase cleavage site(s) in Cdc25A/1-223.
To further identify the caspase cleavage site(s) in Cdc25A/1-223, serial constructs were made. These constructs encoded fusion proteins with a presumed caspase consensus motif consisting of eight amino acids from Cdc25A between the N-terminal His 6 tag and C-terminal CD20 (Fig. 3D, top). The mobility of these fusion proteins in gel with or without staurosporine treatment was examined using CD20 antibody. In addition to using the Asp-223 caspase cleavage motif as a positive control, 16 possible caspase cleavage sites in the first 223 amino acids of Cdc25A were tested (Fig. 3D). Results show that Asp-223 is the most frequent cleavage site, followed by aspartates 128, 130, and 215, while aspartates 81, 87, 136, 190, and 216 are less cleaved. Furthermore, compared with wild type Cdc25A, the substitution of aspartates 81, 87, 128, 130, 136, 190, 215, 216, and 223 with alanines inhibited the Cdc25A cleavage induced by staurosporine treatment (Fig. 3E). Thus, in addition to aspartate 223, there are eight caspase cleavage sites in Cdc25A/1-223.
Phosphatase Activity of Cdc25A/224 -524 Is 3-Fold Higher than That of Full-length Cdc25A, and Threonine 507 Is Dephosphorylated after Staurosporine Treatment, Which Cannot Recruit 14-3-3 Binding-The Cdc25A C-terminal fragment generated by caspase cleavage at Asp-223 contains a phosphatase catalytic domain from residues 336 to 497. To compare the phosphatase activity of wild type Cdc25A with its mutants, bacterially produced GST-tagged Cdc25A, Cdc25A/C431S (phosphatase dead mutant), Cdc25A/1-223,and Cdc25A/224 -524 were purified by pulldown using glutathione beads. These proteins were purified to 95% homogeneous, as indicated by Coomassie Blue staining (Fig. 4A). Phosphatase activity was assayed using 3-O-methylfluorescence phosphate, an artificial substrate. The phosphatase activity of GST-Cdc25A/224 -524 was 3-fold higher than that of GST-Cdc25A, whereas that of GST-tagged Cdc25A/C431S and Cdc25A/1-223 was nondetectable (Fig.  4B). These results suggest that the conformation of the active site in Cdc25A/224 -524 was quite unlike that in the full-length Cdc25A, resulting in an increase of Cdc25A/224 -524 phosphatase activity.
In an unperturbed cell division cycle, Thr-507 is phosphorylated by Chk1 at the G 1 /S boundary until the G 2 phase. The phosphorylation of Thr-507 recruits 14-3-3 binding, which shields the cyclin B1-docking site. In the G 2 phase, Thr-507 is dephosphorylated, and Cdc25A can then access and activate cyclin B1/Cdc2, promoting the G 2 /M transition (2). Because staurosporine inhibits Chk1 activity (22), the phosphorylation status of Thr-507 was examined using phospho-Cdc25A-Thr-507 antibody. Thr-507 was indeed dephosphorylated at 2 h after staurosporine treatment (Fig. 4C), and in turn 14-3-3 binding could not be recruited.
In Addition to Amplification of Staurosporine-induced Apoptosis, the Ectopic Expression of Cdc25A/224 -525/T507A Induces the Activation of Cdc2/Cdk2 and Caspase 3 and Apoptosis-The phosphatase activity of Cdc25A/224 -524 was higher than that of full-length Cdc25A, and staurosporine treatment induced dephosphorylation of Thr-507. To determine the function of Cdc25A/224 -524 with the dephosphorylated Thr-507 generated during staurosporine treatment, Jurkat cells were electroporated with the expression plasmid with two promoters to drive any of the His-tagged Cdc25A/1-223, Cdc25A/224 -524, and Cdc25A/224 -524/T507A and copepod green fluorescence protein (copGFP) expression. At 10 h after electroporation, the protein levels of these Cdc25A truncated mutants and copGFP were determined by immunoblotting (Fig. 5A, top panel), and some cells were treated with staurosporine for another 12 h. The percentage of apoptotic cells in the cells expressing copGFP before (black columns) or after (gray columns) staurosporine treatment was represented by the sub-G 1 cell population (Fig. 5A, bottom panel). Compared with the expression of copGFP alone, the ectopic expression of either His-tagged Cdc25A/224 -524 or Cdc25A/224 -524/ T507A amplified staurosporine-induced apoptosis but not Histagged Cdc25A/1-223. The ectopic expression of either Histagged Cdc25A/224 -524 or Cdc25A/224 -524/T507A for 24 h also promoted the dephosphorylation of Tyr-15 in Cdc2/Cdk2 ( Fig. 5B) and apoptosis, as indicated by the activation of caspase-3 (Fig. 5C, top panel) or an increase in terminal deoxynucleotidyltransferase dUTP nick end labeling-positive cells (Fig. 5C, bottom panel). Moreover, compared with Histagged Cdc25A/224 -524, the ectopic expression of His-tagged Cdc25A/224 -524/T507A had a greater effect on the amplifica-tion of staurosporine-induced apoptosis, activation of Cdc2/ Cdk2 and caspase 3, and induction of apoptosis. copGFP was cleaved by an unknown protease or caspase during the ectopic expression of Cdc25A/224 -524 with or without T507A mutation (Fig. 5, A, top panel, and B, top panel). However, the electroporation efficiency did not decrease, indicating that the FIGURE 3. Identification of caspase cleavage sites in Cdc25A. A, Jurkat cells were electroporated with vectors expressing C-terminal His-tagged Cdc25A, incubated at 37°C for 6 h, and treated with 300 nM staurosporine for various times as indicated. The cell lysate made from nonelectroporated Jurkat cells was as a control lysate. Immunoblotting was performed using antibodies as indicated. B, Jurkat cells were co-electroporated with vectors expressing CD20 and either C-terminal His-tagged Cdc25A, Cdc25A/D223A, or Cdc25A/224 -524 or N-terminal His-tagged Cdc25A/1-223, incubated at 37°C for 6 h, and treated with 300 nM staurosporine or vehicle for 1.5 h. Immunoblotting was performed using antibodies as indicated. The protein level of CD20 is as an electroporation efficiency control. C, Jurkat cells were electroporated with vectors expressing either C-terminal His-tagged Cdc25A, Cdc25A/D223A, or Cdc25A/224 -524, incubated at 37°C for 6 h, and treated with or without 500 ng/ml anti-Fas antibody for 1 h. Immunoblotting was performed using antibodies as indicated. D, to identify caspase cleavage sites in Cdc25A, the adaptor encoding GG-XXXXXDXX-GL, with BamHI at 5Ј end and NotI at 3Ј end, was subcloned between His 6 and CD20 in pcDNA3-His-CD20 (top). XXXXXDXX is a presumed caspase consensus motif in Cdc25A. If the D in XXXXXDXX is a caspase cleavage site, staurosporine treatment speeds up the movement of the fusion protein containing CD20 in gel compared with that without treatment. HeLa cells were transiently transfected with plasmids expressing the fusion proteins containing presumed caspase consensus motifs as indicated, incubated at 37°C for 24 h, and treated with 1 M staurosporine or vehicle for 6 h. Immunoblotting was performed with CD20 antibody. E, Jurkat cells were co-electroporated with vectors expressing CD20 and either N-terminal His-tagged Cdc25A or Cdc25A/D81A, -D87A, -D128A, -D130A, -D136A, -D190A, -D215A, -D216A, and D223A (Cdc25A/D81, 87, 128, 130, 136, 190, 215, 216, and 223A), incubated at 37°C for 6 h, and treated with 300 nM staurosporine or vehicle for 1.5 h. Immunoblot analysis was performed using antibodies as indicated.
cleaved copGFP still had fluorescent activity. Interestingly, the level of cleaved copGFP is correlated with the apoptotic effect induced by the ectopic expression of Cdc25A/224 -524 with or without T507A (Fig. 5, B, top panel, and C).
We identified nine caspase cleavage sites in Cdc25A. Amino acids around Asp-81 and 87 are involved in the regulation of Cdc25A protein stability. Substituting these two aspartates with alanines dramatically increased Cdc25A protein levels (data not shown). Moreover, Asp-81 and 87 are minor caspase cleavage sites. Thus, we used the caspase-resistant mutant that contained mutations at aspartates 128, 130, 136, 190, 215, 216, and 223 to alanines to further confirm that the generation of Cdc25A/224 -524 with dephosphorylated Thr-507 during staurosporine treatment amplified staurosporine-induced apoptosis. The overexpression of wild type Cdc25A also induced apoptosis in Jurkat cells (data not shown), so in Fig. 5D, the ectopic Cdc25A and the caspase-resistant mutant were expressed for 1 h, and the dosage of staurosporine was decreased to 150 nM. One hour after electroporation, the protein levels of His-tagged Cdc25A and the caspase-resistant mutant and copGFP were assessed by immunoblotting (Fig. 5D, top panel), and some cells were treated with 150 nM staurosporine for another 1.5 h. The percentage of early apoptotic cells in the cells expressing copGFP before (black columns) or after (gray columns) staurosporine treatment was represented by the annexin V-positive cell population (Fig.  5D, bottom panel). Compared with His-tagged wild type Cdc25A, the ectopic expression of the caspase-resistant mutant amplified staurosporine-induced apoptosis to as less extent, even though the protein level of the caspase-resistant mutant was much higher than that of wild type. These results further confirm that Cdc25A is a pro-apoptotic protein that amplifies staurosporine-induced apoptosis through its C-terminal domain.
The ectopic expression of Cdc25A/224 -524/T507A induced apoptosis in Jurkat cells. To determine whether Cdc25A/224 -524/T507A also induced apoptosis in the other cell types, Histagged Cdc25A/224 -524/T507A was expressed in K562 cells for 12 h. The protein levels of His-tagged Cdc25A/224 -524/ T507A and copGFP were examined by immunoblotting (Fig.  5E, top panel), and the percentage of apoptotic cells in the cells expressing copGFP was determined by annexin V staining (Fig. 5E, bottom panel). Results show that the ectopic expression of Cdc25A/224 -524/T507A also induced apoptosis in K562 cells.   210 in Cdk2 (23,24). Cdc25A also contains three consensus amino acids in Cdc25B as follows: Arg-446 and 450 and Tyr-455. Furthermore, Cdc2 also contains two consensus amino acids in Cdk2 at Asp-207 and 211. To elucidate how Cdc25A/ 224 -524/T507A activates Cdc2/Cdk2, Arg-446 and 450, two presumed Cdc2/Cdk2-binding sites, were mutated to leucine. At 24 h post-electroporation, the protein levels of His-tagged Cdc25A/224-524/T507A and Cdc25A/224-524/R446L/R450L/ T507A and copGFP were detected by immunoblotting (Fig. 6A,  top panel). The activation of Cdc2/Cdk2 and the percentage of apoptotic cells in the cells expressing copGFP were determined by phospho-specific flow cytometric analysis and the sub-G 1 cell population (Fig. 6, A, bottom panel, and  B). The substitution of Arg-446 and 450 to leucine significantly suppressed the activation of Cdc2/ Cdk2 and apoptosis induced by the ectopic expression of His-tagged Cdc25A/224 -524/T507A. Because the substitution of leucine for Arg-446 and 450 did not completely abolish Cdc25A/224 -524/T507A activity for Cdc2/Cdk2, Tyr-455, the other presumed Cdc2/Cdk2binding site, was mutated to alanine. The mutation at both Arg-446 and 450 to leucine and Tyr-455 to alanine dramatically inhibited the pro-apoptotic function of Cdc25A after staurosporine treatment for 12 h (Fig. 6C, bottom panel). However, this was not due to the protein level of His-tagged Cdc25A/ R446L/R450L/Y455A being lower than that of wild type (Fig. 6C, top  panel). These results suggest that Arg-446 and 450 and Tyr-455 are involved in the activation of Cdc2/ Cdk2 and the induction of apoptosis.
Substitution of Leucine for Arginine 446 and 450 and Alanine for Tyrosine 455 Attenuates the Interaction between Cdc25A/224 -524/ T507A and Cdc2, Cdk2, Cyclin A, or Cyclin B1, and the Ectopic Expression of Cdc25A/224 -524/T507A Significantly Increases Cyclin B1 and Cdc2-associated Kinase Activities-To confirm that Arg-446 and 450 and Tyr-455 in Cdc25A mediate the docking of Cdc2/Cdk2, we performed co-immunoprecipitation experiments. 293T cells were transiently transfected with plasmids expressing either 3ϫFLAGtagged Cdc25A/224 -524/T507A or Cdc25A/224 -524/R446L/R450L/Y455A/T507A. The cell lysates were subjected to immunoprecipitation using M2 antibody against 3ϫFLAG epitope, and immunoprecipitates were analyzed by immunoblotting using M2 antibody or antibody against Cdc2, Cdk2, cyclin B1, or cyclin A (Fig. 7, A-C). These results show that the amounts of Cdc2, Cdk2, cyclin B1, and cyclin A co-immunoprecipitated with Cdc25A/224 -524/ T507A are greater than those co-immunoprecipitated with Cdc25A/224 -524/R446L/R450L/Y455A/T507A, confirming that Arg-446 and 450 and Tyr-455 mediate Cdc2 and Cdk2 binding. However, the substitution of leucine for argi-FIGURE 6. Substitutions of leucine for both arginine 446 and 450 and plus alanine for tyrosine 455 inhibit apoptosis induced by Cdc25A/224 -524/T507A and the amplification of apoptosis promoted by Cdc25A after staurosporine treatment, respectively. A, Jurkat cells were electroporated with plasmids expressing copGFP alone or co-expressing copGFP and either His-tagged Cdc25A/224 -524/T507A or Cdc25A/224 -524/ R446L/R450L/T507A. At 24 h post-electroporation, SDS-PAGE analysis was followed by immunoblotting using antibodies as indicated (top), and the activity of Cdc2/Cdk2 in copGFP-positive cells was detected by phospho-flow cytometry after immunostaining with phospho-Tyr-15-Cdc2 antibody and anti-rabbit IgG antibody conjugated with Alexa Fluor 647 (bottom). B, at 24 h post-electroporation, some cells were also subjected to propidium iodide staining and flow cytometry. The percentage of apoptotic cells in copGFP-positive cells was represented by the sub-G 1 cell population. C, Jurkat cells were electroporated with plasmids expressing copGFP alone or co-expressing copGFP and either His-tagged Cdc25A or Cdc25A/R446L/R450L/Y455A. At 8 h post-electroporation, immunoblotting using the indicated antibodies was analyzed (top), and a portion of cells was treated with 300 nM staurosporine for another 12 h. The percentage of apoptotic cells in copGFP-positive cells before (black bars) or after (gray bars) staurosporine treatment was examined by propidium iodide staining followed by flow cytometry and represented by the sub-G 1 cells population (bottom). Column data are displayed as mean Ϯ S.D., n ϭ 3.

Apoptosis Induced by the Ectopic Expression of
by immunoblotting using the indicated antibodies and annexin V staining, respectively. These results show that knockdowns of Cdc2 and cyclin B1 (Fig. 8, A and B), but not Cdk2 and cyclin A (data not shown), dramatically inhibit the apoptosis induced by the ectopic expression of His-tagged Cdc25A/224 -524/ T507A. Thus, the apoptosis induced by Cdc25A/224 -524 with dephosphorylated Thr-507 generated by staurosporine treatment is due to the activation of cyclin B1/Cdc2. This study shows that Cdc25A is involved in staurosporineinduced apoptosis through the knockdown of Cdc25A (Fig.  2C). Cdc25A/224 -524/T507A interacts with and activates Cdc2 and Cdk2 (Fig. 7, A, B, and D), whereas the apoptosis induced by the ectopic expression of Cdc25A/224 -524/T507A is due to the activation of cyclin B1/Cdc2 (Fig. 8, A and B). To further confirm that the function of Cdc25A/224 -524 with dephosphorylated Thr-507 in staurosporine induced apoptosis, we performed knockdown experiments using siRNA against Cdc2 or Cdk2. Knockdowns of Cdc2 and Cdk2 in Jurkat cells were carried out for 36 h, and then some cells were treated with staurosporine for another 4 h. The protein levels before staurosporine treatment and percentage of apoptotic cells before or after staurosporine treatment were detected by immunoblotting using the indicated antibodies and annexin V staining, respectively (Fig. 8C). These results indicate that knockdown of Cdc2, but not Cdk2, attenuates staurosporine-induced apoptosis. Thus, Cdc25A is a pro-apoptotic protein that amplifies staurosporine-induced apoptosis by activating cyclin B1/Cdc2 through its C-terminal fragment.

DISCUSSION
This study proposes a novel function of Cdc25A, a proapoptotic protein, in staurosporine-induced apoptosis (Fig. 9). Staurosporine treatment activates caspase and induces apopto-  . Model for the pro-apoptotic role of Cdc25A. Staurosporine treatment activates caspases and induces apoptosis in Jurkat cells. Cdc25A is cleaved by caspase(s) at aspartate 223 to generate an active C-terminal fragment, designated C37. Staurosporine, a broad spectrum protein kinase inhibitor, also promotes dephosphorylation of Thr-507 in C37. Therefore, 14-3-3 binding cannot be recruited, as shown by Chen et al. (2), and the Cdc2/Cdk2docking site is exposed. C37 with dephosphorylation of Thr-507 activates cyclin B1/Cdc2 and then caspases and eventually induces apoptosis. Cdc25A is a pro-apoptotic protein, cleaved by caspase(s) to acquire an apoptotic function. This study reveals that the activation of cyclin B1/Cdc2 by C37 with dephosphorylated Thr-507 is a new staurosporine-induced apoptotic pathway. sis in Jurkat cells (Fig. 1, C and E). Cdc25A is a caspase substrate. Staurosporine induces the cleavage of Cdc25A, which can be inhibited by caspase inhibitor I ( Fig. 2A). Caspase mainly cleaves Cdc25A at Asp-223, and it slightly cleaves Cdc25A at several residues in the N-terminal 223 amino acids during staurosporine treatment (Fig. 3, B and D). The Cdc25A C-terminal fragment containing the last 301 residues, C37, has a higher phosphatase activity than full-length Cdc25A (Fig. 4B). Staurosporine, a broad spectrum protein kinase inhibitor, also induces the dephosphorylation of Thr-507 (Fig. 4C). C37 with or without the substitution of alanine for Thr-507 amplifies staurosporine-induced apoptosis, promotes the activation of Cdc2/Cdk2 and caspase 3, and induces apoptosis (Fig. 5, A-C). C37 with the substitution of alanine for threonine 507 (C37/T507A), mimicking the Cdc25A cleavage product generated during staurosporine treatment, has a greater effect than C37 (Fig. 5, A-C). Arg-446 and 450 and Tyr-455 in C37/T507A mediate Cdc2/ Cdk2 activation and binding. It is shown by that mutation at both Arg-446 and 450 to leucine or both of these mutation sites plus Tyr-455 to alanine weakens the activation of Cdc2/Cdk2 and apoptosis and also reduces the interaction with Cdc2, Cdk2, cyclin B1, or cyclin A (Fig. 6, A and B, and Fig. 7, A-C). The ectopic expression of C37/T507A dramatically activates cyclin B1-and Cdc2-associated kinase activities but only slightly activates cyclin A-and Cdk2-associated kinase activities (Fig. 7D). Moreover, the apoptosis induced by the ectopic expression of C37/T507A is due to the activation of cyclin B1/Cdc2, as indicated by our Cdc2 or cyclin B1 siRNA knockdown experiment (Fig. 8, A and B). Because the ectopic expression of either C37/T507A or Cdc25A amplifies staurosporineinduced apoptosis, and knockdown of Cdc2 attenuates staurosporine-induced apoptosis (Fig. 5A, Fig. 6C, and Fig. 8C), Cdc25A is a pro-apoptotic protein that amplifies staurosporine-induced apoptosis through the activation of cyclin B1/Cdc2 by C37 with dephosphorylated Thr-507. The phosphorylation of Thr-507 is required for recruiting 14-3-3 binding (2). C37/T507A has a greater effect on the activation of Cdc2/Cdk2 and the induction of apoptosis than C37. This suggests that the dephosphorylation of Thr-507 exposes the Cdc2/ Cdk2-docking site in C37. The ectopic expression of C37 activates Cdc2/Cdk2 and caspase 3 and promotes apoptosis (Fig. 5, B and C), indicating that Cdc25A also functions as a pro-apoptotic protein in the Fas-dependent apoptotic pathway.
A recent study showed that Cdc25A was cleaved by caspase 3 at Asp-223 to generate an active C-terminal fragment that activated Cdk2 and induced apoptosis (25). However, our present findings show that the same C-terminal fragment generated during staurosporine treatment, but with dephosphorylated Thr-507, induces apoptosis through the activation of cyclin B1/Cdc2 but not Cdk2. Furthermore, the ectopic expression of C37/T507A markedly activates Cdc2-associated kinase activity but only slightly activates Cdk2-associated kinase activity. This means that Cdc2 and Cdk2 can be activated to a different extent by the ectopic expression of C37/T507A. Although premature activation of Cdc2 is required for apoptosis induced by a lymphocyte granule protease or staurosporine (26), the mechanism is not clear. Our results provide strong evidence explaining why the premature activation of Cdc2 is involved in staurosporine-induced apoptosis and even lymphocyte granule protease-induced apoptosis. The mechanisms for staurosporine-induced apoptosis are quite complicated. In addition to caspases, staurosporine-induced apoptosis can proceed through cathepsin D (27). Staurosporine also triggers DNA double strand breaks and inhibits transcription during early treatment (28). Our results show that C37 with the dephosphorylation at Thr-507 generated during early staurosporine treatment is involved in the early stages of apoptosis. Previous research shows that Cdc2 induces neuronal apoptosis in the absence of electrical activity or growth factors through the phosphorylation of Bad at Ser-128 (29), whereas the phosphorylation of Bad at Ser-128 does not promote apoptosis in non-neural cells (30). The identification of the downstream substrate of cyclin B1/Cdc2 will further elucidate the mechanism of apoptosis induced by C37/T507A.
Cdc25A activates different types of cyclin-Cdk complexes by directly associating with different cyclins. Interestingly, our study shows that there is a Cdc2/Cdk2-docking site at C37. Thus, C37 can activate Cdc2 and Cdk2. The property of C37 is quite different from that of full-length Cdc25A. Using small artificial molecules as substrates, the phosphatase activity of C37 is 3-fold higher than that of full-length Cdc25A, suggesting the active site conformation of C37 is unlike that of Cdc25A, and the catalytic activity of C37 is much higher than that of full-length Cdc25A. Compared with C37/T507A, the substitution of leucine for both Arg-446 and 450 and alanine for Tyr-455 only decreases the interaction between C37/T507A and either Cdc2, Cdck2, cyclin A, or cyclin B1. This suggests that other residues are involved in Cdc2/Cdk2 binding. We only showed that Cdc2, Cdk2, cyclin A, or cyclin B1 were co-immunoprecipitated with C37/T507A. It is unknown if C37 interacts with other cyclins and Cdks.
The ubiquitin-mediated proteasome pathway regulates Cdc25A protein stability in unperturbed and perturbed cell division cycles. The protein stability of C37 is likely more stable than that of Cdc25A, because almost all of the residues and the motifs required for the ubiquitin-mediated proteasome pathways are located at N-terminal 223 residues. The phosphorylation status of Thr-507 may regulate the interaction of Cdc2 and Cdk2 with C37. Thr-507 is phosphorylated by Chk1 from the G 1 /S boundary until the G 2 phase (2). The phosphorylation of Thr-507 may recruit 14-3-3 binding and shield the Cdc2/Cdk2docking site, which prevents C37 from interacting with and activating Cdc2 and Cdk2 from the G 1 /S boundary until the G 2 phase. The amino acid sequence around residue 223 in Cdc25A is VDLLD 223 , which is more homologous to the consensus motif preferred by caspase 2, 3, or 7 cleavage (31). A previous study showed that Cdc25A was cleaved by recombinant caspase 3 at aspartate 223 in an in vitro cleavage assay (25). Procaspases 2, 3, 6, 7, and 9 are overexpressed in many cancer cell lines (32,33). Procaspase 3 is also overexpressed in colon cancer tissues, compared with the adjacent non-tumor tissues (33). When the procaspase protein level reaches a threshold, most procaspases undergo rapid autolytic activation (34). Therefore, the results of this study imply that cancer cells produce C37 through caspase cleavage, and that C37, which is more active and stable than Cdc25A, activates Cdc2 and Cdk2 in cancer cells to affect cancer cell proliferation.