Stoichiometric Phosphorylation of Human p53 at Ser315Stimulates p53-dependent Transcription*

p53 protein activity as a transcription factor can be activated in vivo by antibodies that target its C-terminal negative regulatory domain suggesting that cellular enzymes that target this domain may play a role in stimulating p53-dependent gene expression. A phospho-specific monoclonal antibody to the C-terminal Ser315phospho-epitope was used to determine whether phosphorylation of endogenous p53 at Ser315 can be detected in vivo, whether steady-state Ser315 phosphorylation increases or decreases in an irradiated cell, and whether this phosphorylation event activates or inhibits p53 in vivo. A native phospho-specific IgG binding assay was developed for quantitating the extent of p53 phosphorylation at Ser315 where one, two, three, or four phosphates/tetramer could be defined after in vitro phosphorylation by cyclin-dependent protein kinases. Using this assay, near-stoichiometric Ser315 phosphorylation of endogenous p53 protein was detected in vivo after UV irradiation of MCF7 and A375 cells, coinciding with elevated p53-dependent transcription. Transfection of the p53 gene with an alanine mutation at the Ser315 site into Saos-2 cells gave rise to a form of p53 protein with a substantially reduced specific activity as a transcription factor. The treatment of cells with the cyclin-dependent protein kinase inhibitor Roscovitine promoted a reduction in the specific activity of endogenous p53 or ectopically expressed p53. These results indicate that the majority of p53 protein has been phosphorylated at Ser315 after irradiation damage and identify a cyclin-dependent kinase pathway that plays a role in stimulating p53 function.

The cells of higher eukaryotes are subjected to multiple forms of cellular stress, and the biochemical processes that ensure such damage is sensed and repaired are critical for the prevention of diseases such as cancer. Activation of p53-induced growth arrest or apoptosis is a key end point to the signal transduction pathways induced by many distinct forms of cellular damaging agents, including ionizing and nonionizing radiation (1,2), anti-metabolites that inhibit ribonucleotide biosynthesis (3,4), heat shock (5), hypoxia (6), and low extracellular pH (7). The biochemical activity of p53 most tightly linked to its biological function involves its ability to bind to DNA sequence-specifically (8) and to function as a transcription factor (9). p53 is composed of at least four functional domains that regulate its activity as a transcription factor: (i) an N-terminal trans-activation domain that is required for interaction with components of the transcriptional machinery, including p300 (10,11); (ii) the central conserved core DNAbinding domain containing most of the inactivating mutations found in human tumors (12); (iii) a tetramerization domain (13); and (iv) a C-terminal negative regulatory domain whose phosphorylation, acetylation, or SUMOylation correlates with activation of the latent sequence-specific DNA binding function of p53 (14 -19). Each of these domains on p53 contain multiple sites for modification by both covalent and noncovalent interactions, and it now seems likely that it is the combined action of many enzymes that coordinately modulate p53-dependent gene expression in response to cellular stress. Given that 2-fold changes in the gene dosage of p53 can have dramatic affects on tumor incidence in vivo (20,21), it seems evident that these post-translational events modulating the specific activity of p53 will play an important role in regulating its tumor suppressor function.
The N-terminal regulatory domain of p53 contains a highly conserved 15-amino acid domain that directs the binding of p53 to the positive effector p300 or the inhibitor MDM2, the balance of which modulates p53-dependent tumor suppression (22). The regulation of the p53-p300 interaction is modulated in part by phosphorylation within the BOX-I domain, as p300/CBP binding to p53 in the N terminus and subsequent acetylation of the C-terminal domain of p53 can be stimulated by Ser 15 phosphorylation within the p300 docking site (23,24). Consistent with these data, mutation of full-length p53 protein at multiple sites including Ser 15 can reduce its specific activity as a transcription factor in vivo (25,26). Enhanced phosphorylation of endogenous p53 protein at Ser 15 following DNA damage, quiescence, or senescence (27) can occur through the action of an ATM/ATR/DNA-PK kinase-dependent pathway (28 -30). These studies identified one key signal transduction cascade that could stimulate p53-dependent transcription via modification of the N-terminal domain of p53.
Two other sites of post-translational modification are now known to be clustered within this BOX-I regulatory domain, with a similar paradigm being supported: phosphorylation of p53 within the N-terminal BOXI domain at Thr 18 or Ser 20 can destabilize the p53-MDM2 complex or stabilize p53-p300 pro-* 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. tein interactions (31)(32)(33). 1 The phosphorylation sites at Thr 18 or Ser 20 exhibit distinct types of regulation, depending upon the context or damaging agent. Cycling human cells with a wild-type p53 pathway constitutively modify p53 at the Ser 20 site (35,36), and the most likely role for Ser 20 phosphorylation under these conditions is to produce a transcriptionally competent pool of p53 protein that binds to p300. 1 These data are consistent with the requirement for an active pool of p53 in a cycling cell to produce p21 WAF1 protein (37) and provide a mechanism for producing MDM2 protein in a cycling cell to maintain the negative feedback degradation loop (38). Oxidant stresses can result in hypo-phosphorylation at Ser 20 (35,36), which will presumably destabilize the p300-p53 complex and place an excessive oxidant burden on the p53 pathway. The ionizing radiation-induced form of p53 protein is phosphorylated at Ser 20 by a Chk2-dependent pathway (39 -41), ensuring that the p300-p53 complex is stabilized and active in an irradiated cell for the induction of p53-dependent gene expression. The Thr 18 phosphorylation can potently inhibit MDM2 binding or stabilize p300 binding, 1 and the Thr 18 site is phosphorylated in human breast cancers (31), induced during senescence (27) or transiently following ionizing radiation (33).
Flanking the tetramerization domain of p53 in the extreme C terminus is a negative regulatory domain whose post-translational modification also plays an important role in modulating the specific activity of p53 in vivo. One function for this regulatory C-terminal domain is to maintain p53 protein in a latent state for specific DNA binding. Deletion of this domain or stoichiometric phosphorylation at Ser 392 activates the latent specific DNA binding function of p53 in vitro by an allosteric mechanism (14). Increased steady-state phosphorylation at Ser 392 of endogenous human p53 protein in cells occurs following UV-C and X irradiation damage (42), but whether CK2, PKR or other enzymes catalyze this reaction is not yet known (43)(44)(45). In addition, microinjection or intracellular synthesis of an antibody (pAb421) that binds to the C-terminal negative regulatory domain near the phosphorylation sites can activate p53-dependent gene expression in vivo (46 -48), suggesting that C-terminal modification can be a rate-limiting step in stimulating p53 function in vivo. Together, these data provide the basis for the paradigm that a signaling pathway modulating the specific activity of p53 after cellular damage targets the C-terminal negative regulatory domain by stimulating the latent specific DNA binding function of the p53. Other enzymes that are now known to target sites within the C-terminal domains of p53 and stimulate its specific DNA binding function have also been suggested to play a coordinated role in the damage response, including increases in steady-state acetylation at Lys 320 and/or Lys 382 by p300/CBP (16) and increases in SUMOylation at Lys 386 (17,18).
One site of phosphorylation in the C terminus for which a role in the stress-induced activation of p53 has not been defined is the cyclin-dependent kinase site at Ser 315 . Conventional mapping techniques have shown that the highly conserved Ser 315 site is phosphorylated in cells using conventional [ 32 P]orthophosphate labeling methods and is a good in vitro substrate for the G 2 and S phase cyclin-dependent kinases (49,50). Phosphorylation of p53 at this site has been shown to significantly enhance the sequence-specific DNA binding activity of p53 protein in vitro (50,51), possibly by cooperating with other modifications within the extreme C-terminal negative regulatory domain (52) or by alternate N-terminal modifications (53). Conversely, biophysical studies have shown that phosphorylation at the Ser 315 site reverses the stabilizing and activating effects of Ser 392 phosphorylation on tetramer formation (54). Elevated phosphorylation at the cdk/cdc phosphorylation site induced by okadaic acid treatment of cells increased the specific activity of p53 as a DNA-binding protein from the mdm2 or p21 waf1 promoters but reduced activity from the cyclin G promoter (55). Mutation of the cyclin-dependent kinase phosphorylation site of rat p53, Ser 313 , to alanine also had differential affects on reducing or increasing the specific activity of ectopically expressed rat p53 from distinct promoters (55). Together, these data indicate that the Ser 315 site phosphorylation can have an inhibitory or a stimulatory role in modulating p53-dependent transcription, depending upon the context.
Given the central role played by the cyclin-dependent kinases in the regulation of the eukaryotic cell cycle and the evidence indicating that Ser 315 modification of p53 can be inactivating or activating, we set out to further investigate the role played by phosphorylation of endogenous p53 protein at Ser 315 in cells with a wild-type p53 pathway. Standard radiolabeling techniques used to monitor phosphorylation in cells actually perturbs the base-line phosphorylation-status of p53 and induces a p53-dependent growth arrest and will therefore alter cyclin-dependent kinase activity (35). To circumvent this problem, we have utilized a noninvasive approach to dissect signaling to Ser 315 and to begin to address the role of cyclindependent kinases in the regulation of p53 function. A phospho-specific monoclonal antibody to Ser 315 has been generated and characterized to demonstrate that phosphorylation of endogenous human p53 in MCF7 and A375 cells at Ser 315 is near-stoichiometric after cellular irradiation. Mutation of the Ser 315 phosphorylation site on p53 to alanine reduces the specific activity of p53 as a transcription factor in vivo, and the use of the cyclin-dependent protein kinase inhibitor Roscovitine reduces the specific activity of wild-type p53 in vivo. These data identify a regulatory site in which the phosphorylation stoichiometry of endogenous p53 protein is established in vivo and link enhanced Ser 315 phosphorylation by a cyclin-dependent protein kinase pathway to an increase in the specific activity of p53 as a DNA-binding protein and transcription factor in vivo.

EXPERIMENTAL PROCEDURES
Reagents, Enzymes, and Proteins-Anti-p53 antibodies DO-1, DO-12, BP.10, pAb421, CM5, and CM1 have been described previously (43,56). Anti-sep70 is a monoclonal raised to the chaperone SEP70. 2 FPS392 is a monoclonal antibody specific to p53 phosphorylated at Ser 392 and was described previously (42). For the generation of FPS315, mice were immunized with a keyhole limpet hemocyanin-conjugated phosphopeptide NNTSSS PO4 PQPKKKPLDG corresponding to amino acids 310 -325 on human p53 (synthesized by Dr. G. Bloomberg, University of Bristol). Monoclonal antibodies were generated according to established procedures (57) and characterized as described in Fig. 1 and by Ref. 58. Expression of full-length human p53 in Escherichia coli or in Sf9 insect cells and fractionation by heparin-Sepharose chromatography or by phosphocellulose cation exchange chromatography was performed as described previously (59). Recombinant human cyclin A, cdk2, cyclin B, and cdc2 were expressed in Sf9 insect cells as individual subunits, and the holoenzyme complex was reconstituted in the kinase reactions as indicated previously (58).
Peptide Library Phage Display to Define the FPS315 Epitope-Enzyme-linked immunosorbent assay wells were coated with purified FPS315 and used to select bacteriophage from libraries containing random peptides inserted within the phage III coat protein according to the manufacturer's protocol (New England Biolabs). Two separate libraries were screened: PhD12 contains random linear 12mer peptides, whereas C7C contains 7-mer cyclic peptides, the structure of which is constrained by a disulfide bridge between two cysteine residues at either end of the insert peptide. After two cycles of selection and amplification, individual phage clones were screened for FPS315 binding by enzyme-linked immunosorbent assay, and the inserts of strongly positive clones were sequenced and depicted in Table I.
Human Cell Lines, Culture Conditions, and Cell Lysis-A375 malignant melanoma cells have been described previously (42,43). A375 and MCF-7 cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum (Life Technologies, Inc.). For UV-C irradiation cells were first washed with Hank's balanced salt solution (Life Technologies, Inc.) and then irradiated in the absence of medium using a model 2400 Stratalinker (Stratagene), before being refed with fresh medium. Roscovitine was obtained from Calbiochem/Novabiochem and stored at Ϫ20°C as 50 mM aliquots in Me 2 SO. High salt nuclear extract of human cell lines were performed as described (43,60) and used for the radiolabeled DNA binding assay, immunoprecipitation, and kinase activity assays. Sf9 cells expressing p53 protein were lysed by a whole cell lysis method (43). For immunoblotting, cells were scraped off the dishes on ice-cold phosphate-buffered saline, pelleted by centrifugation, and snap frozen. Frozen pellets were lysed for 15 min at 4°C in denaturing urea buffer (6.4 M urea, 0.1 M dithiothreitol, 0.05% Triton X-100, 25 mM NaCl, and 20 mM HEPES, pH 7.6) or Nonidet P-40 nondenaturing lysis buffer (43), where indicated, and lysates were clarified by centrifugation at 13,000 ϫ g for 10 min. Protein concentration was determined by the method of Bradford (Bio-Rad), and aliquots were stored at Ϫ70°C until required. Transient transfections were performed as indicated (61). Plasmids used included pCMVwtp53 (obtained by cloning wtp53 cDNA into the BamHI site of pcDNA3.1(ϩ)), pCMVala315p53, and pCMVala341p53 (obtained by in situ directed mutagenesis using the pCMVwtp53 as a template and the following primers containing the desired mutation: ala315 (F): 5Ј-aac aac acc agc tcc gct ccc cag cca aag-3Ј; ala315 (R): 5Ј-ctt tgg ctg ggg agc gga gct ggt gtt gtt-3Ј; ala341 (F): 5Ј-gag cgc ttc gag atg gcc cga gag ctg aat-3Ј; ala341 (R): 5Ј-att cag ctc tcg gag cgg ctc gaa gcg ctc-3Ј). pCMV␤-gal was a gift from Dr. Alain Puisieux (Centre Leon Berard, Lyon, France). Each plasmid (1 g) was used in transfection assays unless differently stated. A375 cells or Saos-2 cells grown as indicated above were incubated with concentrations of DNA-LipofectAMINE complexes according to manufacturer's protocols and incubated for the times indicated in the figure legends. Following cell manipulation, luciferase and ␤-galactosidase activity were analyzed and normalized as indicated, or the p53 protein levels were determined by independent lysis in a Nonidet P-40 nondenaturing buffer as indicated previously (43).
Immunochemical Methods-Standard immunoblotting was performed using established techniques, as described previously (43). Immunoprecipitation of recombinant p53 protein was performed using RIPA buffer (150 mM NaCl, 10 mM Tris-HCl, pH 7.2, 1% sodium deoxycholate, 1% Triton X-100, and 0.1% SDS) containing 1 mM benzamidine, 50 mM NaF, 5 mM dithiothreitol. Protein G-Sepharose beads (Amersham Pharmacia Biotech) were preblocked in 3% BSA, 3 and then added to 200 l of buffer containing 1% BSA, with 1 g of antibody, and 100 ng of p53 (Heparin fraction II; Ref. 59). Reactions were incubated for 18 h at 4°C, and then beads were washed extensively before being boiled in Laemmli buffer and analyzed by Western blotting. For immunoprecipitation of p53 from cell lysates DO-1 was first cross-linked to protein G-Sepharose beads using dimethylpimylimidate (Sigma) (57), and noncross-linked antibody was then eluted with 0.1 M glycine, pH 2.5. Immunoprecipitation was allowed to proceed for 3 h at 4°C. Luminographic quantitation of p53 protein levels using enzyme-linked immunosorbent assay was performed as indicated (58).
Sequence-specific DNA Binding Assays-Electrophoretic mobility shift analysis was performed differently depending on whether recombinant protein or cell line nuclear extracts were being assayed. Recombinant proteins were assayed at 4°C in 10 l of a buffer containing 15% glycerol, 25 mM HEPES, pH 7.6, 10 mM KCl, 0.02% Triton X-100, 1 mg/ml BSA, 5 mM dithiothreitol, 1 mM MgCl 2 , 1 mM benzamidine, 100 ng of sonicated herring sperm DNA, and 1 ng of radiolabeled PG or p21 consensus oligonucleotide (50,62). For the assay of nuclear lysates 5 g (4 l) of protein in nuclear lysis buffer B (20% glycerol, 20 mM HEPES, pH 7.6, 1.5 mM MgCl 2 , 0.1% IGEPAL CA-630, 5 mM dithiothreitol, 1 mM benzamidine) was added to a 30 l of a buffer containing 20 l of cytoplasmic lysis buffer A (as buffer B, but with 10 mM KCl, 1 mg/ml BSA, and 100 ng of sonicated herring sperm DNA). Reactions were allowed to assemble for 10 min at 22°C before the addition of 1 ng of radiolabeled p53CON oligonucletide (GGACATGCCCGGGCATGTCC) and the indicated monoclonal antibody. Reaction products were processed using native polyacrylamide gel electrophoresis as indicated previously (62). The native gel agarose band shift assay was performed as indicated (56) with the following modifications. Briefly, latent p53 protein (460 ng) was incubated in a buffer containing 5 mM Tris-HCl, pH 7.5, 0.5 mM EDTA, 50 mM KCl, 0.01% Triton X-100, 5 mM MgCl 2 , and 1 mM ATP without or with the indicated kinases (PKC or cyclin A-cdk2) or monoclonal antibody. BP.10 was used in place of pAb421 as the C-terminal activating antibody with the minimal epitope defined as being Gln 375 -Ser 376 -Thr 377 -Ser 378 (56). Following an incubation at 30°C for 60 min, 1 g of pPGM-1 plasmid DNA restricted with PvuII (to give a large vector fragment and a smaller 474-base pair DNA fragment containing the p53 consensus site PG, which was originally cloned into the HindIII site of pBluescript; Ref. 62) was added to the reaction and the products were electrophoresed on a 0.8% agarose gel in 1ϫ TBE (89 mM Tris borate, 2 mM EDTA) at 10 V/cm for 200 min at 4°C. After electrophoresis, the gels were stained with ethidium bromide (1 g/ml) for 30 min, rinsed in water, and photographed.
Kinase Reactions-Phosphorylation of E. coli-expressed human p53 was performed at 30°C in kinase buffer (10% glycerol, 25 mM HEPES, pH 7.6, 0.05 M KCl, 0.5 mg/ml BSA, 1 mM NaF, 1 mM benzamidine, 5 mM MgCl 2 , 1 mM dithiothreitol, 250 M to 1 mM ATP). For radioactive kinase assays [␥-32 P]ATP was also included in the reactions. Reaction products were then resolved by SDS-polyacrylamide gel electrophoresis, and gels were either exposed directly to film in the case of radioactive assays or immunoblotted with phospho-specific antibodies as indicated (42,58).

RESULTS
The Development of an Immunochemical Assay to Quantitate in Vitro Ser 315 Phosphorylation of p53-Phospho-specific antibody reagents have recently proved to be valuable noninvasive reagents for the study of signal transduction pathways that target p53 in vivo, especially because conventional 32 P labeling methods commonly used to identify phosphorylation sites induce a p53-dependent growth arrest and alter phosphorylation of p53 protein in normal human diploid fibroblasts (35). Given that cyclin-dependent kinase activity will change after such cell damage, the development of a noninvasive probe to this known cyclin-dependent protein kinase site on p53 is important in beginning to define an accurate role for this pathway in modulating p53 function.
The monoclonal antibody described in this study specific for phospho-Ser 315 (named FPS315) was raised against the keyhole limpet hemocyanin-conjugated phospho-peptide NNT-SSS PO4 PQPKKKPLDG corresponding to amino acids 310 -325 on human p53, and its use in a Cdk2 and Cdc2 kinase assay has been reported (58). Purification of the monoclonal antibody from hybridoma tissue culture supernatants using protein A-Sepharose was required to obtain pure IgG devoid of contaminating phosphatases found in serum supernatant, and this highly purified IgG was used in the in vitro assays described below. Phage peptide display libraries were first used to define the essential residues within the FPS315 epitope (Table I). This assay was developed because unpublished data from our lab have shown that the specificity of some phospho-antibodies generated to N-terminal phosphorylation sites on p53 can be toward phospho-amino acids rather than phospho-epitopes. Given the widespread growing use of phospho-specific monoclonal and polyclonal antibodies, a rigorous characterization is therefore required to ensure the integrity of the phospho-specific IgG. Positive peptide phage clones selected by two rounds of biopanning that bound to the monoclonal antibody FPS315 all contained an invariant PQP motif corresponding to Pro 316 -Gln 317 -Pro 318 on human p53; however, differences were observed in amino acids flanking the PQP motif depending upon the combinatorial library utilized. The linear 12-mer peptide library gave rise to a predominant set of clones with a stabilizing glutamate (negatively charged phosphate mimetic) at the position of Ser 315 and two other clones with a glycine or serine at the position of Ser 315 (Table I). The cyclic 7-mer peptide library gave rise to clones without the expected stabilizing aspartate at the position of Ser 315 but yielded a stabilizing Lysine at position Lys 320 that was not observed using the linear combinatorial library. Although the differences in the affinity between each of these peptide epitope clones are not known, these experiments demonstrate that an important recognition feature of the monoclonal antibody FPS315 involves its specificity for amino acids flanking the Ser 315 phosphorylation site of p53 and that this antibody is therefore not just detecting a phospho-serine moiety.
Denaturing immunoblots of human recombinant p53 protein demonstrate that FPS315 only recognizes p53 phosphorylated at Ser 315 (Fig. 1A, top panel, lane 2) but does not recognize unphosphorylated p53 (Fig. 1A, top panel, lane 1) or Ser 392 phosphorylated p53 (Fig. 1A, top panel, lane 3). A duplicate immunoblot was probed with antibody DO12 to demonstrate that the p53 protein levels are equal in all three lanes (Fig. 1A,  bottom panel). FPS315 also displayed a striking specificity for the native phospho-Ser 315 p53 protein because immunoprecipitation by this antibody is dependent on the prior phosphorylation of p53 at Ser 315 (Fig. 1B, top panel, lane 2), whereas FPS315 does not immunoprecipitate unphosphorylated p53 (Fig. 1B, top panel, lane 1) or Ser 392 phosphorylated p53 (Fig.  1B, top panel, lane 3). As controls for this immunoprecipitation, DO-12 antibody could bind equally well to all three isoforms of p53 (Fig. 1B, middle panel), whereas a control monoclonal antibody (named sep70) shows that Ser 315 phosphorylated p53 is not binding nonspecifically to the solid phase (Fig. 1B, bottom  panel).
The most informative immunochemical assay we developed for studying Ser 315 phosphorylation of p53 was the electrophoretic mobility shift DNA binding assay. The central feature of this assay first takes advantage of the fact that a bivalent IgG molecule can bind in cyclic manner to a multivalent antigen (63,64), producing a stable complex containing one or two IgG molecules per p53 tetramer (59). Previous biochemical studies have also indicated that the number of antibodies bound to one tetrameric p53-DNA complex can be quantitated by counting the integral number of stable, intermediate IgG-p53-DNA complexes separated by native gel electrophoresis (59). In theory, the extent of Ser 315 phosphorylation on p53 tetramers could similarly be quantitated by determining whether zero (0% phosphorylation), one (50% phosphorylation), or two (100% phosphorylation) bivalent FPS315 IgG molecules are bound to and supershift the tetrameric p53-DNA complex.
The immunoblot and immunoprecipitation ( Fig. 1) showed that unphosphorylated p53 protein expressed in E. coli is not bound by FPS315. Consistent with this, unphosphorylated p53 protein is not supershifted by FPS315 in a DNA binding assay ( Fig. 2A, lane 1). However, a time course of phosphorylation performed under conditions where phosphorylation is substoichiometric began to give rise to a supershift of the p53-DNA complex and proceeds steadily from a slower migrating species obtained when one FPS315 IgG molecule is bound unstably to one singly phosphorylated tetramer ( Fig. 2A, lane 1 versus  lanes 3 and 4) to the more slowly migrating complex obtained with one FPS315 IgG molecule bound stably to one doubly phosphorylated tetramer (i.e. 50% phosphorylation of the tetramer; Fig. 2A, lane 1 versus lanes 7-9). Further phosphorylation results in the production of three phosphates/tetramer and a corresponding increase in the intensity of a slower migrating complex (Fig. 2A, lane 9 versus lanes 1-8). Changing the kinase conditions in vitro can also give rise to altering extents of phosphorylation. Stoichiometric phosphorylation of p53 tetramers (approaching 100%) can be observed as defined by nearly complete supershifting of the p53-DNA complexes (Fig.  2B, lane 6 versus lanes 1-5). In addition, where ATP is limiting, ϳ50% of the tetramers can be phosphorylated twice as defined by the presence of equal amounts of unshifted p53-DNA complexes and a partially shifted p53-DNA complexes containing one FPS315-IgG per p53-DNA complex (Fig. 2B, lane 4 versus  lane 3). As a control, the complete reaction containing high levels of ATP and kinase but in the presence of the cdk/cdc inhibitor Roscovitine (250 M) do not yield an FPS315-shifted p53-DNA complex (Fig. 2B, lane 2 versus lane 1). These data indicate that the extent of Ser 315 phosphorylation can be determined by quantitating the number of FPS315 IgG molecules bound to native p53 tetramers and is used as an assay to address the extent of Ser 315 site phosphorylation on endogenous p53 protein in vivo.
The p53 Protein Induced by UV Irradiation Is Stoichiometrically Phosphorylated at Ser 315 -Although FPS315 is highly specific for its phospho-epitope, the affinity is relatively low, and the IgG cannot be used to detect changes in endogenous p53 protein phosphorylation by direct immunoblotting of lysates (data not shown). As a result, p53 protein was first immunoprecipitated from lysates isolated from cycling and irradiated cells to concentrate the p53 protein for immunoblotting. Using an immunoprecipitation assay, increases in Ser 315 phosphorylation can be detected 5 h after DNA damage (Fig.  3A, bottom panel, lane 3 versus lanes 1 and 2) under conditions where p53 protein is induced (Fig. 3A, top panel, lanes 2 and 3  versus lane 1). However, these immunoblotting data do not address the stoichiometry in vivo, an important milestone in determining the significance of an enzyme pathway interacting with p53 in cells. For example, if Ser 315 phosphorylation of p53 was very low (i.e. less than 1%), then the significance of this pathway would assume less importance. However, if phosphorylation was very high (i.e. greater than 50%) and the majority of the p53 tetramers activated by radiation interacted at some stage with this Ser 315 site kinase, then it would define a pathway likely to be important in affecting p53 function.
Using the native gel electrophoretic IgG binding assay, we elected to use two distinct cell types expressing p53 protein for examining the Ser 315 phosphorylation stoichiometry of p53 synthesized in vivo. Recombinant human p53 protein expressed in Sf9 insect cells results in multiple isoelectric forms representing differential phosphorylation states of p53 (31) that can be partially separated by heparin-Sepharose chromatography into latent and activated isoforms (65), and insect cell-expressed p53 has been reported to phosphorylate p53 at the CDC2/CDK2 phosphorylation site (51). As such, we first used the native gel electrophoretic IgG binding assay to determine whether Ser 315 phosphorylation of human p53 occurring in vivo could be quantitated in this cell system. The activated isoform of p53 protein purified on heparin-Sepharose from Sf9 cells (Fig. 3B, lane 1) is phosphorylated at the pAb421 epitope (65) and is not shifted by this phosphorylation-sensitive monoclonal antibody (Fig. 3B, lane 2). FPS315 supershifted the majority of activated isoform of p53 protein expressed in Sf9 cells to produce the species containing two FPS315-IgG per p53-DNA complex (Fig. 3B, lanes 3 and 4), indicating that the majority of the p53 has four phosphates per tetramer. A small proportion of the activated p53 contained two phosphates per tetramer as defined by the faster migrating species containing one FPS315-IgG per p53-DNA complex (Fig. 3B, lanes 3 and 4). Together, these data demonstrate that in vitro phosphorylation of bacterially expressed p53 protein (Fig. 2) or in vivo phosphorylation of p53 at Ser 315 in insect cells (Fig. 3B) can be quantified using the native gel electrophoretic IgG binding assay and provided the foundation to determine whether the stoichiometry of Ser 315 phosphorylation could be quantified on endogenous p53 protein in human cells.
Exposure of the human cancer cell line MCF7 to UV irradiation caused a clear increase in the levels of endogenous wildtype p53-DNA complexes (Fig. 3C, lane 3 versus lane 1), and the majority of this complex was further shifted when FPS315 was added to the binding reaction (Fig. 3C, lane 4 versus lanes  3 and 2). These data indicate that the phosphorylation at Ser 315 induced by radiation in this cell line occurs on greater than 95% of the endogenous p53 tetramers and that the majority of wild-type p53 protein activated by UV irradiation interacts with the Ser 315 kinase pathway in vivo. A similar level of Ser 315 phosphorylation was observed in irradiated A375 cells (see below; Fig. 6, lane 6 versus lane 5). A standard immunoblotting method was also employed to determine whether the transiently transfected p53 gene produces a protein that is phosphorylated at Ser 315 following irradiation, because this technique produces more p53 protein than is normally present endogenously, and this allows the detection of p53 phosphorylation using a direct immunoblotting assay. The transfection of a gene encoding wild-type p53 into the p53-null cell line Saos-2 resulted in a protein that exhibits a basal level of Ser 315 phosphorylation (Fig. 3D, lane 1), and after UV irradiation the transfected p53 protein displays a transient increase in Ser 315 phosphorylation 4, 8, and 12 h post-UV (Fig. 3D, lanes 3-5  versus lane 1), providing additional evidence that phosphorylation of p53 occurs at Ser 315 following irradiation in vivo.
Phosphorylation of p53 at Ser 315 Activates the Latent Sequence-specific DNA Binding Activity of p53 in Vitro-Before proceeding to investigate whether Ser 315 phosphorylation of endogenous p53 protein is regulated in vivo, we first wished to further define the effects of phosphorylation of p53 at Ser 315 on the in vitro sequence-specific DNA binding function of p53, because it is not clear whether modification of this site alone is activating or destabilizing to the DNA binding function of p53 (50,54,55). When unphosphorylated p53 protein purified from a bacterial expression system is added to DNA binding assays, the protein is inactive for DNA binding (Fig. 4B, lane 2) unless the activating antibody BP.10 is added (Fig. 4B, lane 3). The latent activity of unphosphorylated p53 could be activated by enzymes that target the C-terminal regulatory domain including PKC (Fig. 4B, lane 6 versus lanes 4 and 5) and cyclin A-cdk2 (Fig. 4B, lane 9 versus lanes 7 and 8). The form of p53 protein  5 and 6). After an incubation at 30°C for 30 min, the reactions were stopped by adding an equal volume of DNA binding buffer containing radiolabeled consensus site DNA and 250 M of the cdk/cdc inhibitor Roscovitine at 4°C. The extent of phosphorylation was then determined using the DNA binding assay with the activating antibody pAb421 (lanes 1-6), and the phospho-specific antibody FPS315 was added to the indicated reactions (lanes 2, 4, and 6). The arrows mark the mobility of pAb421-p53-DNA complexes bound to one FPS315 IgG molecule per p53 tetramer (2 phosphates per p53 tetramer) or to two FPS315 IgG molecules per p53 tetramer (4 phosphates per p53 tetramer). activated by cyclin A-cdk2 (Fig. 4A, lane 4 versus lane 2) could be supershifted by FPS315 (Fig. 4A, lane 5 versus lanes 4 and  3), further indicating that near-stoichiometric phosphorylation has occurred on p53 in vitro by the action of cyclin A-cdk2. Together, these data indicate that Ser 315 phosphorylation can activate the sequence-specific DNA binding function of p53, consistent with a previous report that this phosphorylation can be stimulatory rather than inhibiting (50).

Mutation of the Ser 315 Phosphorylation Site to Alanine Reduces the Specific Activity of p53 as a Transcription Factor-
Given that most of the p53 protein interacts with a Ser 315 kinase pathway after irradiation (Fig. 3), these data suggest that this phosphorylation may be stimulatory because it coincides with p53 activation in a damaged cell. As such, the transcription activity of p53 was quantitated in the p53-null cell line Saos-2 after cotransfection with a p53-responsive reporter gene and either wild-type p53 or the phosphorylation mutant p53-Ala 315 gene. Although transient cotransfection of the wild-type p53 gene gave rise to a relatively high level of reporter activity from vectors containing the p53 consensus binding site from the p21 WAF1 or bax promoters (Fig. 5, A and  B), transient cotransfection of the p53-Ala 315 gene gave rise to lower activity from these promoters (Fig. 5, A and B). As a negative control, the Ala 341 substitution mutant, which produces a monomeric protein of low activity, gave rise to less activity than the wild-type p53 and the Ala 315 substitution mutant (Fig. 5, A and B). Endogenous p21 WAF1 protein was also induced to a lower extent in Saos-2 cells after transfection of the p53-Ala 315 gene, in comparison with the wild-type p53 gene (Fig. 5C), further indicating that an intact Ser 315 phosphoryl-ation site is required for the highest level of p53 activity.
Roscovitine Reduces the Specific Activity of p53 Protein in Vivo-The data in Figs. 4 and 5 suggest that phosphorylation of p53 increases its DNA binding and transcription activity in vivo by a Ser 315 kinase-dependent pathway. Because cyclin-dependent protein kinases are the only known enzymes that can target this consensus cdk/cdc phosphorylation site (50), it is predicted that inhibition of endogenous cyclin-dependent kinase activity through the use of the cdk/cdc inhibitor Roscovitine would reduce the specific activity of p53 in vivo. We therefore utilized Roscovitine to determine whether endogenous p53 protein activity and transiently transfected p53 protein activity would be reduced after inhibition of cdk/cdc function in vivo. The majority of wild-type p53 tetramers in A375 cells have been phosphorylated at least twice after UV irradiation, as defined by the extent of supershift of the p53-DNA complex by the monoclonal antibody FPS315 (Fig. 6, lane 6 versus lanes 5  and 3). The inclusion of the cdk/cdc inhibitor Roscovitine at the time of irradiation prevented much of the supershift by FPS315 (Fig. 6, lane 9 versus lane 8), indicating that this cdk/cdc inhibitor can prevent Ser 315 phosphorylation of the p53 protein after DNA damage. It should be noted, however, that Roscovitine also produced a striking increase in p53 protein levels in the irradiated cells (Fig. 6, lane 7 versus lanes 4 and 1; and see below Fig. 7B), despite the fact that the p53 protein induced was hypo-phosphorylated at Ser 315 .
Endogenous p53 activity in A375 cells from a p53-responsive reporter was also tested in the absence or presence of Roscovitine to determine whether the specific activity of p53 is reduced after cdk/cdc inhibition. When A375 cells are transiently trans-  1 and 2) and p53-DNA complexes bound to one or two FPS315 IgG molecules (lanes 3 and 4) as indicated in the panel by the arrow pointing to two phosphates per tetramer (2 P-tetramer) or four phosphates per tetramer (4 P-tetramer). C, quantitating Ser 315 phosphorylation of p53 in irradiated MCF7 cells. MCF7 cells were either unirradiated (lanes 1 and 2) or exposed to 20 J/m Ϫ2 UV (lanes 3 and 4). 5 h after irradiation, p53 protein was assayed from lysates for its ability to bind a radiolabeled consensus oligonucleotide as indicated under "Experimental Procedures" with the addition of the indicated antibodies pAb421 or FPS315. The arrows mark the migration of either: p53-DNA complexes without FPS315 (lane 3) and p53-DNA complexes bound to one or two FPS315 IgG molecules (lane 4) as indicated in the panel by the arrow pointing to two phosphates per tetramer (2 P-tetramer) or four phosphates per tetramer (4 P-tetramer). Similar results quantitating Ser 315 phosphorylation of p53 in irradiated A375 cells were obtained as in described in the legend to Fig. 6. D, direct blotting can be used to detect increases in Ser 315 phosphorylation of transfected p53 in irradiated Saos-2 cells. The p53 gene was transfected into the p53-null cell line Saos-2, and 24 h after transfection, the cells were untreated (lane 1) or exposed to 20 J/m Ϫ2 UV followed by lysis 0, 4, 8, 12, 16, or 32 h post-irradiation (lanes 2-7, respectively). After direct immunoblotting of lysates, the levels of phosphorylated p53 were analyzed by incubation with FPS315 and processed as indicated under "Experimental Procedures." p53 protein phosphorylated by cyclin A-cdk2 is included as a phosphorylated control (lane 8) as described previously (58) and is marked by the arrow.
fected with a p53-responsive reporter, the basal level of p53 activity is reduced from 4 to 8 h post drug-addition (Fig. 7A) and from 12 to 24 h, and the total levels of p53 activity began to return to base-line levels (Fig. 7A), presumably because of Roscovitine-metabolism and inactivation. Normalizing p53 protein levels in the absence or presence of Roscovitine (Fig. 7, B and C) to the total activity (Fig. 7A) gave rise to the changes in p53 specific activity summarized in Fig. 7D. The latter data indicate that the specific activity of p53 is dramatically reduced by the cdk/cdc inhibitor and provide independent evidence complimenting data in Fig. 5 for an activating role for a cdk/cdc family member in the p53 response. Similar results were observed by transfecting the p53 gene in Saos-2 cells and examining the changes in the activity of p53 in the absence or presence of Roscovitine. p53-dependent gene expression in Saos-2 cells cotransfected with a p53-dependent luciferase reporter is reduced 4 or 8 h after Roscovitine treatment (Fig. 8A) under conditions where transfected p53 protein levels were unaffected by the drug (Fig. 8B). Thus, the specific activity of both endogenous and ectopically expressed p53 protein can be reduced by inhibiting the cdk/cdc pathway with the kinase inhibitor Roscovitine. DISCUSSION A variety of biochemical, cellular, and genetic approaches have been developed to indicate that post-translational modification of p53 protein might be central to the control of its transactivation function. In particular, the artificial manipulation and activation of p53 in vivo by using an antibody that mimics kinases by targeting the C-terminal negative regulatory domain of p53 (14,46,48) or an antibody that mimics N-terminal p53-BOXI domain kinases by disrupting MDM2 binding (66) suggests that the function and stability of endogenous p53 protein in cells is dependent upon post-translational modification. The major issues addressed in this report are whether Ser 315 phosphorylation predominates in vivo in cycling or damaged cells, given the conflicting data on the effects of Ser 315 phosphorylation on p53 activity. Our data demonstrate that phosphorylation of p53 at Ser 315 primes and stimulates its latent DNA binding function in vitro and that endogenous p53 protein is phosphorylated near-stoichiometrically at Ser 315 in response to irradiation predominantly through the activity of cdk/cdc-dependent pathway. Given that the A375 cell line we use to dissect signaling to p53 also triggers Ser 392 phosphorylation after cell irradiation (42,43), these data indicate that two distinct C-terminal kinase pathways modify p53 in response to DNA damage, consistent with a requirement for enhanced C-terminal modification of p53 as a component of the radiation response. However, because the antibody to the Ser 392 phosphorylation site cannot be used to address stoichi-  1-3) or exposed to 20 J/m Ϫ2 UV (lanes 4 -9) and refed with medium alone (lanes 1-6) or exposed to 20 J/m Ϫ2 UV followed by 20 M Roscovitine (lanes 7-9). 5 h after irradiation, p53 protein was assayed from lysates for its ability to bind a radiolabeled consensus oligonucleotide as indicated under "Experimental Procedures" with the addition of the indicated antibodies pAb421 or FPS315. The arrows mark the migration of either: p53-DNA complexes without FPS315 (lanes 5 and 8) and p53-DNA complexes bound to one or two FPS315 IgG molecules (lanes 6 and 9) as indicated in the panel by the arrow pointing to two phosphates per tetramer (2 P-tetramer) or four phosphates per tetramer (4 P-tetramer).
ometry, 4 as is the case with the antibody to the Ser 315 phosphorylation site (Figs. 2 and 3), a direct comparison of the efficiency of these two kinase pathways is not yet possible.
Studying the role of p53 phosphorylation in vivo in controlling p53 function is complicated given that the act of 32 P labeling, which is commonly used to map phosphorylation sites, can actually induce a p53-dependent growth arrest and alter immunoreactivity including Ser 20 dephosphorylation in normal human fibroblasts (35,36). Thus, the development of noninvasive monoclonal antibodies provide valuable tools for studying steady-state signaling to p53, especially for the cyclin-dependent kinase site pathway, which can be perturbed by cell damage. However, such phospho-specific reagents require careful in vitro characterization prior to use, because it cannot be assumed that an immune response to a hapten produces a monospecific IgG. For example, monoclonal antibodies to Nterminal BOX-I phosphorylation sites have given an insight into the type of immune response possible in a polyclonal IgG prep to a phospho-epitope. Although phospho-antibodies were obtained that were apparently specific for the phospho-Ser 20 or phospho-Thr 18 epitopes, there are some that display a prefer-ence only for phospho-amino acid or for epitope containing a stabilizing phosphate at any of the nearby phospho-acceptor sites Ser 15 , Thr 18 , and Ser 20 (31). Thus, in this report we suggest that the definition of specificity for a phospho-specific IgG requires the use of phage peptide display to define the IgG protein-protein contact site and the use of an in vitro kinase assay to determine whether the IgG is specific for denatured and/or native full-length protein.
A second issue to address concerning post-translational modification of p53 is not only whether it changes in response to cellular stimuli but the extent to which sites are modified in vivo. For example, the original 32 P mapping data demonstrated the Ser 315 is an in vivo phosphorylation site (49), and recent mass spectrometric data have shown that the Ser 315 site is modified in an ionizing irradiated cell (67). However, both methods cannot address reliably the stoichiometry of modification in vivo, which is the advantage of the phospho-specific monoclonal antibody that can detect integral phosphates on the endogenous p53 tetramer in a noninvasive manner. The only previous report addressing the extent of modification of p53 was the phosphorylation at Ser 20 in cycling cells, which was defined to be at least 70% of the total p53 population (36). The other phospho-specific monoclonal antibodies developed to p53 protein at Ser 15 (36) and Ser 392 (42) cannot be used to define the stoichiometry of p53 modification in vivo, because these antibodies are only specific for phospho-denatured p53 protein, and the native IgG binding assay cannot be used to address phosphorylation stoichiometry (data not shown). Thus, the unique feature of FPS315 monoclonal antibody has been that is displays specificity for the native phospho-tetramer, thus allowing a quantitation of the extent of phosphorylation at Ser 315 . The development of the native electrophoretic ligand binding assay was an important milestone in beginning to address the issue of stoichiometry of phosphorylation of endog- Qualitative changes in p53 protein levels from detergent lysed cells were analyzed by standard immunoblotting procedures using the antibody DO-1 (top panel, Immunoblot) and quantitative changes in p53 protein levels were analyzed using a two-site capture enzyme-linked immunosorbent assay by luminographic methods (bottom panel, Luminography). C, chemiluminescent quantitation of p53. The raw quantitative data derived from luminographic analysis of p53 protein levels in Fig. 8B (bottom panel) were quantitated in an ascent fluoroscan luminometer, and the relative levels of p53 protein (relative light units, RLU) are plotted as a function of time after addition of Roscovitine. D, the specific activity of p53. The total levels of luciferase activity quantitated in A were normalized to total p53 protein levels in C, and the data are plotted as p53-dependent activity (in relative light units, RLU) as a function of time after addition of Roscovitine.
FIG. 8. The cdk inhibitor Roscovitine reduces the specific activity of p53 in transiently transfected Saos-2 cells. A, p53-dependent Luciferase activity. The p53-null cell line Saos-2 was transiently cotransfected with the indicated p53-expression constructs (wild-type p53 (black bars) and the Ala 315 mutant p53 (white bars)) and the p21 WAF1 -luciferase promoter. 24 h after transfection, the cells were treated with a Me 2 SO control or Roscovitine (40 M) and lysed for analysis of luciferase activity at the indicated times post-drug addition (4 and 8 h). The activity is plotted as wild-type p53 and the Ala 315 mutant p53 activity (in relative light units, RLU) as a function of time after addition of Roscovitine. B, p53 protein levels in transfected Saos-2 cells after treatment with Roscovitine. Saos-2 cells were transiently cotransfected with the indicated p53-expression constructs (wild-type p53) and the p21 WAF1 -luciferase promoter. 24 h after transfection, the cells were treated with a Me 2 SO control or Roscovitine (40 M) and lysed for analysis of p53 protein levels at the indicated times post-drug addition (4 and 8 h) using standard immunoblotting procedures with the antibody DO-1. enous p53 at one of its highly conserved phosphorylation sites.
The enhanced phosphorylation of p53 at Ser 315 after irradiation is consistent with models suggesting that C-terminal modifications play a role in stimulating the specific DNA binding function of p53. Additional evidence suggests that modification of the C-terminal domain of p53 plays a role in controlling the transcriptional activity of p53 in cells. A Cdk2 homologue has been implicated in stimulating the specific activity of recombinant human p53 as a transcription factor in yeast (68). In addition, multiple basic-to-hydrophobic amino acid mutations within the PKC motif (69) can dramatically increase the specific activity of human p53 as a transcription factor in cells, presumably by neutralizing negative regulatory domain interaction with its binding site in the central domain. Finally, selective mutation of murine p53 at the CK2 site can regulate its growth suppressor function possibly via transrepression (70,71) and can increase its transcriptional activity in contact-inhibited murine fibroblasts (72). These results suggest that the structural integrity of p53 surrounding the CDC2, PKC, and the CK2 site phosphorylation sites can modulate the specific activity of p53 in cells. Consistent with this hypothesis, one rate-limiting step in activating the function of p53 as a transcription factor has been determined; the form of p53 latent for sequence-specific DNA binding (65) can be activated using the monoclonal antibody pAb421 as a sequence-specific transcription factor using in vitro transcription systems (73) and by microinjection into cells containing p53-responsive reporter genes (14,46). These results highlight the importance of discovering signaling pathways and enzymes implicated in site-specific modification of the C terminus of p53.
It remains unclear which enzymes might be targeting p53 at Ser 315 in vivo. Given that p53 can be modified in vitro at Ser 315 by cyclin B-and cyclin A-dependent kinases (50) and that cyclin A-cdc2 is an key cell cycle regulator (74 -77), determining whether cyclin B-cdc2, cyclin A-cdk2, or cyclin A-cdc2 is the major Ser 315 kinase in vivo would begin to address the pathway implicated in stoichiometric modification at Ser 315 . Our preliminary data indicate that although cyclin A-cdk2 is the major Ser 315 kinase, we cannot rule out a contributory role by cyclin B-cdc2, cyclin A-cdc2, or other enzymes that may target the SP motif. Precedents for distinct kinases targeting the same phosphorylation site on p53 have been reported for ATM or ATR at Ser 15 (28 -30) and CK2 or PKR at Ser 392 (44). In addition, the major Ser 315 kinase activity is down-regulated by irradiation (data not shown), thus classifying the enzyme as similar to certain cyclin-dependent kinases that are down-regulated by cell damage. Although it is generally held that cdk2 activity is down-regulated after DNA damage to ensure the induction of a G 1 /S arrest, positive roles for cdk2 have been identified in the radiation response. Cyclin A-cdk2 can induce the formation of apoptotic bodies in Xenopus egg extracts, and elevated levels of cyclin A-cdk2 correlate with radiation-induced apoptosis in developing Xenopus embryos before the midblastula transition (78). In addition, in irradiated thymocytes that undergo apoptosis in a p53-dependent manner, an induction of Cdk2 accelerates apoptosis, whereas its inhibition blocks radiation-induced cell death (79). However, the mechanism of enhanced steady-state Ser 315 phosphorylation is not known and may involve a coordinated change in the Ser 315 kinase as well as the antagonizing human CDC14 phosphatase that targets the Ser 315 site (34,80). Thus, our data highlighting the stoichiometric phosphorylation of p53 at the cyclin-dependent kinase site in vivo and the correlation between elevated Ser 315 phosphorylation with relatively high levels of p53-dependent transcription, suggests the existence of a radiation-responsive kinase/phosphatase axis that links p53 function to DNA-damage dependent checkpoint control.