Reactive Oxygen Species-induced Phosphorylation of p53 on Serine 20 Is Mediated in Part by Polo-like Kinase-3*

.

Reactive oxygen species (ROS), 1 ubiquitously present, are very reactive and cause damage to biological molecules, including DNA. ROS are potentially mutagenic and may be involved in activation of protooncogene and inactivation of tumor suppressor genes (1,2). Thus, ROS are suspected to represent important human carcinogens (3,4). Oxidative signals, either external or internal, are thought to be detected by sensor molecules and mediated by cellular signal transduction systems, which eventually results in cell cycle arrest, senescence, or apoptosis in normal diploid fibroblast cells. ATM has been proposed to be a sensor of oxidative damage of cellular macromolecules such as DNA (5). The tumor suppressor protein p53 appears to be a major effector of the genotoxic stress-signaling pathway that is mediated by ATM (6). In fibroblast cells, p53 protein level is increased upon H 2 O 2 treatment, and the level of p53 is correlated with replicative senescence and apoptosis (7). In the p66shc Ϫ/Ϫ cells, p53 activation and its target gene p21 expression are impaired in response to oxidative stress (8). However, a p53-independent pathway that mediates H 2 O 2induced G 2 /M growth arrest has also been reported (9).
Members of the Polo family of protein kinases, conserved through evolution, have been characterized in yeast (10), Caenorhabditis elegans (11), Drosophila melanogaster (12), Xenopus laevis (13), mouse (14,15), and human (16,17). The founding member of this family, Polo, was originally identified in the fruit fly and was shown to be a serine-threonine kinase required for mitosis (12). Mammalian cells contain at least three proteins (Plk1, Plk2, and Plk3) that exhibit marked sequence homology to Polo (14,15,18,19). As cells progress through the cell cycle, Plk proteins undergo substantial changes in abundance, kinase activity, or subcellular localization. In human cells, the amounts of Plk1 protein and its kinase activity peak at mitosis (18). During mitosis, Plk1 transiently associates with mitotic structures such as the spindle apparatus, kinetochores, and centrosomes (20). Recent studies have shown that Plk1 contributes to a variety of mitotic (or meiotic) events, including activation of cyclin B-Cdc2, breakdown of the nuclear membrane, centrosome maturation, and formation of the bipolar spindle at the onset of mitosis (21)(22)(23). Plk1 also controls the exit of cells from mitosis by regulation of the anaphasepromoting complex (24). Plk3 shows little resemblance to Plk1 with regard to function in mammalian cell cycle regulation. Thus, the abundance of Plk3 remains relatively constant during the cell cycle, and its kinase activity peaks during late S and G 2 phases (25). Furthermore, Plk3 phosphorylates Cdc25C on serine 216, resulting in inhibition of the activity of this protein (25), whereas phosphorylation of Cdc25C by Plx1, a Xenopus Plk1 ortholog, results in activation of this protein (13).
Polo family kinases also participate in the response to DNA damage (26 -28). For example, Cdc5, a budding yeast ortholog of Drosophila polo, promotes adaptation to cell cycle arrest at the DNA damage checkpoint (29). The electrophoretic mobility of Cdc5 in denaturing gels is affected by prior subjection of cells to DNA damage, and this modification is dependent on Mec1, Rad53 (a yeast Chk1 homolog), and Rad9 (26). In addition, a functionally defective Cdc5 mutant protein suppresses a Rad53 checkpoint defect, whereas overexpression of Cdc5 overrides checkpoint-induced cell cycle arrest (27), suggesting that Cdc5 acts downstream of Rad53. Moreover, DNA damage appears to interfere with the activation of Plk1 in mammalian cells, resulting in down-regulation of the kinase activity of this protein.
In contrast, expression of dominant negative mutants of Plk1 * This work was supported in part by Public Health Service Award CA47729 (to W. D.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
overrides the induction of G 2 arrest by DNA damage (28).
We have been studying the biological role of polo-like kinase-3 (Plk3, previously named Prk) during normal and abnormal cell growth (17,25,30,31). Here we report that ROS induces activation of Plk3 as well as p53, which is correlated with p53 phosphorylation on multiple serine sites. Activation of both Plk3 and p53 is ATM-dependent. In addition, we have obtained experimental evidence strongly suggesting that Plk3 mediates ROS-induced serine 20 phosphorylation of p53.

MATERIALS AND METHODS
Cell Culture-Various cell lines, including ATM-deficient cell line (ATCC number CRL-1702), were purchased from ATCC. CRL-1702 has been characterized as ATM Ϫ/Ϫ (32). GM00637 cell line (human fibroblast) was originally from the Coriell Institute for Medical Research. HeLa, A549, GM00637, DU145, LNCap, and PC-3 cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum and antibiotics (100 g/ml penicillin and 50 g/ml streptomycin sulfate) with 5% CO 2 . DAMI, HEL, and HL-60 cells were cultured in RPMI 1640 medium, and Daudi cells were culture in McCoy's medium supplemented with fetal bovine serum and antibiotics as above.
Immunoblotting-Cells treated with H 2 O 2 (200 M unless otherwise specified) or adriamycin (100 M) were collected and lysed (25). In some experiments, caffeine (2 M) was supplemented to the cultured cells for 30 min prior to the treatment with H 2 O 2 or adriamycin. Equal amounts (40 g) of protein lysates from the treated cells were analyzed by SDS-polyacrylamide gel electrophoresis followed by immunoblotting with antibodies (New England Biolabs) to phosphorylated p53 (specifically phosphorylated on serine 9, serine 15, or serine 20), p21, or Bax. The same blots were also stripped and reprobed with antibodies to regular p53 (Santa Cruz Biotechnology). Signals were detected with horseradish peroxidase-conjugated goat secondary antibodies (Sigma) and enhanced chemiluminescence reagents (Amersham Pharmacia Biotech).
Protein Kinase Assays-Immunocomplex kinase assays were performed essentially as described previously (25). In brief, A549 cells were exposed to H 2 O 2 (200 M) for various times, lysed, and subjected to immunoprecipitation with antibodies to Plk3. The resulting precipitates were resuspended in a kinase buffer (10 mM Hepes-NaOH (pH 7.4), 10 M MnCl 2 , 5 mM MgCl 2 ), and the kinase reaction was initiated by the addition of [␥-32 P]ATP (2 Ci) (Amersham Pharmacia Biotech) and ␣-casein (Sigma). After incubation for 30 min at 37°C, the reaction mixtures were analyzed by SDS-polyacrylamide gel electrophoresis and autoradiography. Recombinant His 6 -Plk3, produced and purified as described previously (25,30), was assayed for kinase activity as a positive control. In some kinase assays, GST-p53 was incubated with His 6 -Plk3 or His 6 -Plk3 K52R in the kinase buffer supplemented with "cold" ATP. After reaction, Plk3-phosphorylated GST-p53 samples, as well as nonphosphorylated GST-Plk3, were blotted for serine 20 phosphorylation.
Coimmunoprecipitation Analysis-GM00637 cell lysates were incubated for 30 min at room temperature in a total volume of 500 l with of 20 l of protein A/G-agarose bead slurry (Santa Cruz Biotechnology). After removal of the beads, the supernatant was supplemented with either rabbit polyclonal (PharMingen) or mouse monoclonal antibodies to Plk3, or with control immunoglobulins followed by incubation with constant agitation for overnight at 4°C. Protein A/G-agarose beads (20 l) were then added to each immunoprecipitation mixture, and the incubation was continued for 1 h at room temperature. Immunoprecipitates were collected by centrifugation, washed three times with the cell lysis buffer, and subjected to immunoblot analysis with monoclonal antibodies to serine 20-phosphorylated p53.
Transient Transfection-GM00637 cells were transfected, using the LipofectAMINE method (Life Technologies, Inc.), with constructs expressing Plk3 or Plk3 K52R25 or with the vector pCR592. One day after transfection, cells were treated with or without H 2 O 2 for 30 min. Cell lysates were prepared and blotted for Plk3, p53, or serine 20-phosphorylated p53.

RESULTS
Although recent studies have shown that phosphorylation of p53 plays an important role in stabilization and activation of this tumor suppressor protein in cells exposed to ionizing radiation (IR) or UV (6, 33), the mechanism by which ROS-induced p53 accumulation/activation remains unclear (34). To determine whether ROS activates p53 through phosphorylation, GM00637D cells were exposed to H 2 O 2 for various times, and p53 phosphorylation status was analyzed by immunoblotting using phospho-specific antibodies. Fig. 1A shows that upon H 2 O 2 treatment, p53 was rapidly phosphorylated on serine 20 and serine 15 in GM00637D cells. Serine 9 phosphorylation was also induced with a slow kinetics (Fig. 1A, lane 5). However, serine 392 phosphorylation was not detected (data not shown). These results indicate that p53 accumulation upon oxidative stress as reported by von Harsdorf and Dietz (35) is at least partly due to phosphorylation of p53 on serine 15 and serine 20, because these two residues are located within the domain of the protein that interacts with HDM2 (human ortholog of murine double minute-2 protein, MDM2), resulting in stabilization of the normally short-lived p53 protein in response to the stress (36).
Phosphorylation and activation of p53 upon challenge with genotoxic stress such as IR and UV often results in cell cycle arrest (6). In fact, the trans-activation by p53 of genes such as those encoding p21 and Bax proteins is thought to be responsible at least in part for cell cycle arrest and apoptosis, respectively, in cells subjected to genotoxic stress (37). To determine whether H 2 O 2 -induced p53 phosphorylation is correlated with its functional activation, we measured expression of its target genes p21 and Bax. Fig. 1B  DNA damage caused by IR activates p53 through phosphorylation on multiple residues, and this activation is ATMdependent (33). To determine whether ROS-induced p53 phosphorylation was also ATM-dependent, we treated GM00637 cells with caffeine, an ATM/ATR inhibitor, prior to exposure of the cells to H 2 O 2 or IR-mimetic drug adriamycin. Fig. 2A shows that caffeine (CFN) partially blocked H 2 O 2 -stimulated phosphorylation of p53 on serine 15 and serine 20 (lanes 2 and 5), whereas it completely inhibited adriamycin (ADR)-induced phosphorylation of p53 on all three residues (lanes 3 and 6). These observations suggest that p53 activation by ROS is at least in part dependent on ATM and/or ATR. To further confirm that ATM was important in mediating p53 phosphorylation by H 2 O 2 , we exposed ATM-deficient CRL-1702 cells to H 2 O 2 . Fig. 2B shows that in the ATM-deficient cells after H 2 O 2 treatment enhancement in phosphorylation of p53 on neither serine 20 nor serine 15 was observed, indicating that ROS-induced phosphorylation and activation of p53 is ATM-dependent.
Our laboratory has been studying human Plk3, which is involved in regulating cell cycle progression (17,25,30). As an initial step to identify protein kinase(s) responsible for phosphorylation of p53 induced by ROS, we examined the possibility of Plk3 activation by H 2 O 2 , because Plk3 phosphorylates the same residue of Cdc25C (serine 216) as that targeted by Chk1 and Chk2. Chk1 and Chk2 are also reported to phosphorylate p53 on serine 20 (38,39). A549 cells, expressing good levels of Plk3, were treated with ROS for various times. Plk3 immunoprecipitated from the treated cells was analyzed for its kinase activity using casein as substrate as described previously (25). Fig. 3A shows that compared with the control (lane 1) Plk3 kinase activity was rapidly activated in A549 cells (lane 2) and maintained for at least 1 h. To determine whether Plk3 activation was ATMdependent, CRL-1702 cells treated with H 2 O 2 were collected, and Plk3 immunoprecipitates were assayed for Plk3 kinase activity. Fig. 3B shows that whereas recombinant Plk3 phosphorylated casein effectively (lane 4), no difference in Plk3 kinase activity was detected between untreated control (lane 1) and H 2 O 2treated CRL-1702 cells (lane 2), suggesting that Plk3 activation also requires ATM. In addition, Plk3 activation was caffeinesensitive because pretreatment of A549 cells with caffeine completely blocked activation of Plk3 by H 2 O 2 (data not shown).
To determine the possibility that Plk3 was involved in mediating H 2 O 2 -induced p53 phosphorylation, we screened a dozen cell lines for Plk3 expression. We observed (Fig. 4A) that Daudi (B lymphoblastic leukemic cells with wild-type p53 (40)) did not express detectable levels of Plk3, whereas other tested cell lines expressed various levels of this protein. Further analysis with polymerase chain reaction confirmed that no Plk3 expression was detectable in Daudi cells (data not shown). To determine whether the absence of Plk3 expression affected p53 phosphorylation, we analyzed p53 phosphorylation on both serine 20 and serine 15 residues in Daudi cells exposed to H 2 O 2 . Fig. 4B shows that p53 phosphorylation on serine 15 is rapidly induced and maintained for at least 2 h in Daudi cells (lanes 2-5). In contrast, no serine 20 phosphorylation was observed. Chk2 is reported to phosphorylate p53 on serine 20 (39). Reprobing the same blot with antibody to Chk2 revealed that

FIG. 2. H 2 O 2 -induced p53 phosphorylation is ATM-dependent.
A, GM00637 cells pretreated with caffeine (CFN) were exposed to H 2 O 2 or adriamycin (ADR) for 30 min. Equal amounts of protein lysates from the treated cells were analyzed for p53 phosphorylation using phospho-specific antibodies to serine 15 or serine 20. The same cell lysates were also blotted for all forms of p53. Daudi cells expressed abundant Chk2 (Fig. 4B). These observations suggest that Plk3 is involved in regulating serine 20 phosphorylation of p53.
We next asked whether Plk3 directly phosphorylated p53. In vitro kinase assays showed (Fig. 5A) that recombinant histidine-tagged Plk3 (His 6 -Plk3) phosphorylated GST-p53 (lane 2), as well as casein (lane 1), but not GST alone (lane 5), indicating that Plk3 targets the p53 moiety of GST-p53. A kinase-defective mutant of Plk3, His 6 -Plk3 K52R , in which lysine 52 was replaced with arginine, did not significantly phosphorylate GST-p53 (Fig. 5A, lane 3). To further examine whether the serine 20 residue of p53 was a phosphorylation target of Plk3, we incubated GST-p53 with His 6 -Plk3 or His6-Plk3 K52R in the kinase buffer supplemented with ATP. In vitro phosphorylated GST-p53 samples, as well as nonphosphorylated GST-p53, were blotted for serine 20 phosphorylation. Fig. 5B shows that purified GST-p53 was not recognized by the antibody to serine 20-phosphorylated p53 (lane 4). However, when phosphorylated in vitro by His 6 -Plk3, but not by His 6 -Plk3 K52R , GST-p53 exhibited a strong phosphoserine 20 epitope (lane 5). Given that Plk3 kinase activity and serine 20 phosphorylation of p53 are induced by H 2 O 2 , these observations strongly suggest that serine 20 is an in vivo target of Plk3 during H 2 O 2 -induced stress response.
To explore the physical interaction between p53 and Plk3, we immunoprecipitated Plk3 from cells treated with or without H 2 O 2 , and Plk3 immunoprecipitates were then blotted for the presence of serine 20-phosphorylated p53. Fig. 5C shows that neither the control IgGs appreciably precipitated serine 20phosphorylated p53 from the H 2 O 2 -treated cells (lane 1) nor Plk3 antibody brought down the phospho-p53 from the untreated control cells (lane 2) However, Plk3 antibody precipitated p53 that was phosphorylated on serine 20 from cells treated with H 2 O 2 .
To further demonstrate that Plk3 regulated serine 20 phosphorylation of p53 in vivo, GM00637 cells were transfected with constructs expressing either Plk3 or Plk3 K52R . One day after transfection, both Plk3 proteins were expressed (Fig. 6A, GST and ␣-casein were used as negative and positive controls, respectively. After kinase reaction, samples were fractionated on SDS-polyacrylamide gel electrophoresis followed by autoradiography. B, GST-p53 was phosphorylated in vitro by His 6 -Plk3 and His 6 -Plk3 K52R in the kinase buffer supplemented with "cold" ATP. The reaction samples and protein lysates from H 2 O 2 -treated GM00637 cells (lane 1) were then blotted with the antibody to phosphoserine 20 of p53. Partial degradation of GST-p53 was observed (lane 5). C, equal amounts of protein lysates from GM00637 cells treated with or without H 2 O 2 were immunoprecipitated with the antibody to Plk3 or control IgGs. Immunoprecipitates were then blotted for serine 20-phosphorylated p53. GM00637 cell lysates were used as a positive control. lanes 3 and 4) at a level higher than the endogenous one (the band with a slower mobility). The fast mobility of both transfected Plk3 proteins was due to a short truncation at the amino terminus. Further analysis of the transfected cells showed (Fig.  6B) that no significant enhancement in serine 20 phosphorylation was detected when cells were transfected with either Plk3 (lane 2) or Plk3 K52R (lane 3) compared with cells transfected with vector alone (lane 1). However, when Plk3-transfected cells were exposed to a low concentration of H 2 O 2 , a significant increase in serine 20 phosphorylation was detected (lanes 2 and 5). In contrast, no such enhancement in serine 20 phsophorylation was detected in cells transfected with Plk3 K52R (lanes 3 and 6). These observations suggest that Plk3 needs to be activated by ROS before it can fully phosphorylate its physiological substrates. DISCUSSION The mechanism by which mammalian cells transmit signals in response to oxidative damage remains unclear. Here we report that ROS phosphorylates and activates p53 tumor suppressor protein. Consequences of p53 activation are either cell cycle arrest or apoptosis. We have observed that p53 activation in response to H 2 O 2 treatment results in significant increase in expression of p21, but not of Bax (Fig. 1B), which is consistent with our observation that the concentration of H 2 O 2 used in our experiments did not cause significant apoptosis of GM00637 cells (data not shown). However, we cannot exclude the possibility of Bax activation by post-translational mechanisms. Interestingly, it has been proposed that p53 may cause cell death by directly stimulating mitochondria to produce an excess amount of toxic ROS in some cells (33). Thus, a feedback loop between p53 and ROS may exist, which is presumably to amplify the stress signal, resulting in accelerated programmed cell death when damage caused by a genotoxic stress is beyond repair.
Recent advances indicate that reversible phosphorylation plays an important role in the DNA damage checkpoint activation. In fact, p53 is rapidly phosphorylated upon exposure of cells to IR or UV (6). Our current studies demonstrated that oxidative stress activates p53 also through phosphorylation on multiple residues. The kinetics of ROS-induced phosphorylation of p53 on various serine residues is apparently different (Fig. 1A), suggesting the involvement of several protein kinases. It is also likely that phosphorylation of certain residues may facilitate the subsequent phosphorylation of other residues. Consistent with the latter scenario, phopshorylation of threonine 18 by casein kinase II requires prior phosphorylation of serine 15 by ATM upon DNA damage (41).
Our current studies indicate that Plk3 is directly involved in H 2 O 2 -induced phosphorylation of p53 on the serine 20 residue. First, induction of both p53 phosphorylation and Plk3 kinase activity by H 2 O 2 is ATM-dependent (Figs. 2B and 3B). Second, H 2 O 2 does not induce serine 20 phosphorylation of p53 in Daudi cells that express Chk2 but no detectable levels of Plk3 (Fig. 4). Third, Plk3, but not Plk3 K52R , directly phosphorylates GST-p53 (but not GST alone) in vitro, and Plk3-phosphorylated GST-p53 contains a strong serine 20 epitope (Fig. 5). Fourth, Plk3 interacts with serine 20-phosporylated p53 when cells are exposed to H 2 O 2 (Fig. 5C). Fifth, ectopic expression of Plk3, but not the kinase-defective mutant Plk3 K52R , results in significantly enhanced phosphorylation of p53 on serine 20 after H 2 O 2 treatment.
Our studies, together with previous observations (30), suggest that Plk3 may act in parallel with Chk1 and Chk2, downstream of ATM or ATR. Plk3 may preferentially transduce signals generated by a specific genotoxic stress such as H 2 O 2 , just as Chk1 and Chk2 are differentially activated by UV radiation and IR, respectively (6). The observation that serine 20 phosphorylation of p53 was not induced by H 2 O 2 in Daudi cells that express abundant Chk2 but no detectable Plk3 supports this notion. On the other hand, given that Cdc5 acts downstream of Rad53 in yeast (26), it is also possible that Plk3 may lie downstream of Chk2 (and/or Chk1). Plk3 may integrate the signals from ATM-Chk2 and ATR-Chk1 and induce cell cycle arrest or apoptosis by phosphorylating either Cdc25C on serine 216 or p53 on serine 20. Consistent with the latter scenario, Plk3 is activated by IR-mimetic drug adriamycin and UV radiation (data not shown) in addition to H 2 O 2 .

FIG. 6. Plk3 regulates serine 20 phosphorylation of p53 in vivo.
A, GM00637 cells were transfected with constructs expressing Plk3 or Plk3 K52R or with the vector alone. One day after transfection, cells were lysed, and equal amounts of proteins from the transfected cells were blotted with the antibody to Plk3. B, GM00637 cells transfected with various constructs as indicated were treated with or without H 2 O 2 (20 M) for 30 min. Equal amounts of proteins from various treatments were blotted with antibodies to serine 20-phosphorylated p53 or ␣-tubulin.