ATM Mediates Phosphorylation at Multiple p53 Sites, Including Ser46, in Response to Ionizing Radiation*

The p53 tumor suppressor protein preserves genome integrity by regulating growth arrest and apoptosis in response to DNA damage. In response to ionizing radiation (IR), ATM, the gene product mutated in ataxia telangiectasia, stabilizes and activates p53 through phosphorylation of Ser15 and (indirectly) Ser20. Here we show that phosphorylation of p53 on Ser46, a residue important for p53 apoptotic activity, as well as on Ser9, in response to IR also is dependent on the ATM protein kinase. IR-induced phosphorylation at Ser46 was inhibited by wortmannin, a phosphatidylinositol 3-kinase inhibitor, but not PD169316, a p38 MAPK inhibitor. p53 C-terminal acetylation at Lys320 and Lys382, which may stabilize p53 and activate sequence-specific DNA binding, required Ser15phosphorylation by ATM and was enhanced by phosphorylation at nearby residues including Ser6, Ser9, and Thr18. These observations, together with the proposed role of Ser46 phosphorylation in mediating apoptosis, suggest that ATM is involved in the initiation of p53-dependent apoptosis after IR in human lymphoblastoid cells.

In response to DNA damage, the p53 tumor suppressor protein is phosphorylated on each of the seven serines and one threonine the in the first 50 amino acids of its N-terminal transactivation domain as well as at several sites in its carboxyl (C)-terminal tetramerization/regulatory domain (1,2). As a transcription factor, p53 induces or represses several genes that regulate cell cycle arrest, DNA repair or apoptosis, including p21 WAF1 , MDM2, GADD45, p53R2, and p53AIP1. Recent studies suggest that specific p53 phosphorylation events are important for the activation or repression of specific promoters (3)(4)(5)(6). Optimal induction and activation of p53 after exposure to IR requires phosphorylation by the ATM protein kinase (1,2,7). ATM is thought to directly phosphorylate Ser 15 in vivo (8,9) and also is required for phosphorylation of Ser 20 through activation of the Chk2 protein kinase, which phosphorylates Ser 20 in vitro (10 -12). However, the potential role for ATM in regulating p53 modifications at other sites has not previously been explored.

EXPERIMENTAL PROCEDURES
Cell Cultures and Inhibitors-Epstein-Barr virus immortalized normal (GM02254) and A-T 1 (GM01526) human lymphoblast cultures were obtained from the Human Genetic Mutant Cell Repository (Camden, NJ). H1299 (ATCC CRL-5803), a human lung carcinoma cell line that is null for both TP53 alleles, and A549 (ATCC CCL-185), a human lung carcinoma cell line that expresses wild-type p53, were obtained from the American Type Culture Collection (Manassas, VA). All cells were grown in Dulbecco's modified minimal essential medium (Invitrogen) supplemented with 15% (lymphoblasts) or 10% (H1299) fetal bovine serum, 100 nM glutamine and penicillin/streptomycin in a humidified atmosphere with 5% CO 2 . Wortmannin (Sigma) was prepared as a 10 mM stock and PD169316 (Calbiochem, Inc.) as a 1 mM stock in Me 2 SO; both were stored at Ϫ20°C and diluted into the cell media immediately before use.
Induction of DNA Damage, Immunoprecipitation, and Western Immunoblot Analysis-Asynchronously growing cultures in 75-cm 2 flasks were irradiated using a Shepherd Mark I 137 Cs irradiator at a dose rate of 3.2 Gy/min. To detect p53 acetylation, the deacetylase inhibitor trichostatin A (Wako, Osaka, Japan) was added at a final concentration of 5 M 4 h before harvesting. Cultures were harvested at the indicated times after treatment, washed twice with ice-cold phosphate-buffered saline, and lysed in ice-cold lysis buffer (50 mM Tris-HCl at pH 7.5, 5 mM EDTA, 150 mM NaCl, 1% Triton X-100, 50 mM NaF, 10 mM sodium pyrophosphate, 25 mM ␤-glycerolphosphate, 1 mM sodium orthovanadate, 1 mM sodium molybdate, 10 g/ml aprotinin, 10 g/ml leupeptin, 5 g/ml pepstatin, 0.5 mM phenylmethylsulfonyl fluoride). Immunoprecipitation and Western blot analyses were performed as described (13,14). Anti-p53 monoclonal antibody DO-1 was purchased from Santa Cruz Biotechnology Inc.; anti-p53 polyclonal antibody Ab-7 was from Calbiochem, Inc. In all Western blot analyses, uniform protein loading was confirmed by Coomassie Brilliant Blue staining of the SDS-polyacrylamide gels after transfer to the polyvinylidene difluoride membranes.
Plasmids and cDNA Constructs-Wild-type and the L22Q/W23S mu-* 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. tant p53 sequences were subcloned from the p53 expression vectors pC53-SN3 (16) or pCB6ϩp53-22/23 (17) into the pCAGGS expression plasmid behind the CAG (cytomegalovirus enhancer-chicken ␤-actin hybrid) promoter (18). The serine codons at amino acid positions 6, 9, 15, 20, 33, 37, and 46, and the threonine codon at 18, were changed to alanine by site-directed mutagenesis. The entire p53 sequence in each vector was confirmed by DNA sequencing.
Transient Transfections-The day prior to transfection, 10 6 H1299 cells were seeded in each 10-cm tissue culture plate. On the following day, the cultures were transfected with the expression vectors for wildtype p53 or p53 mutants using LipofectAMINE PLUS Reagent (Invitrogen) as recommended by the manufacturer. Cells were exposed to 8 Gy IR 18 h after transfection, and the cultures were harvested for immunoprecipitation and Western blot analysis 2 h after IR.

Identification of ATM-dependent p53
Phosphorylation-To determine whether other p53 posttranslational modifications depend on ATM, we prepared a panel of polyclonal antibodies that, respectively, recognize p53 modified at each of 12 sites at which it is phosphorylated or acetylated (13)(14)(15)19). These antibodies were then used to examine the time course of phosphorylation at each site after exposure of normal (GM02254) and A-T (GM01526) human lymphoblasts to 8 Gy IR (Fig. 1). Acetylation at two C-terminal sites, Lys 320 and Lys 382 , also was examined. p53 accumulated rapidly and reached a maximum at 4 h after irradiation in normal lymphoblasts, as detected with the p53-specific monoclonal antibody DO-1. As expected (19), delayed p53 accumulation was seen in A-T lymphoblasts. p53 increased slowly and was less that half that in normal cells before 2 h after IR; p53 then increased to ϳ80% of that in normal lymphoblasts by 4 h after irradiation. Although low levels of constitutive phosphorylation were observed at Ser 6 , Ser 33 , Ser 315 , and Ser 392 , phosphorylation increased rapidly at each of these sites and at Ser 9 , Ser 15 , Ser 20 , and Ser 46 , after exposure of normal lymphoblasts to IR. Phosphorylation at Thr 18 may be cell line-dependent, and no IR-induced increase in phosphorylation was observed at Thr 18 in either normal or A-T lymphoblasts. Likewise, phosphorylation at Ser 37 is thought to occur primarily through activation of ATR after exposure to e.g. UV light (20), and phosphorylation of Ser 37 was not observed in either the normal or A-T lymphoblasts. A very similar pattern of IR-induced p53 phosphorylation was observed in A549 lung and MCF7 breast carcinoma cell lines (data not shown), indicating that these modifications are not specific to lymphoblasts. In normal lymphoblasts, increased phosphorylation at Ser 6 , Ser 9 , Ser 15 , Ser 20 , Ser 33 , Ser 46 , Ser 315 , and Ser 392 was observed within 15 min after IR. In contrast, in A-T lymphoblasts, phosphorylation at Ser 15 was significantly delayed (until ϳ2 h after IR) and reduced compared with normal lymphoblasts, consistent with a previous report (19). Phosphorylation at Ser 20 also was largely abrogated in A-T cells as reported previously (21), although some Ser 20 phosphorylation was observed by 2 h after IR.
Unexpectedly, phosphorylation at Ser 46 also was defective in A-T lymphoblasts after exposure to IR, and phosphorylation at Ser 9 was reduced and delayed (Fig. 1, A and B). These data suggest that Ser 9 and Ser 46 also may be phosphorylated by an ATM-activated protein kinase (or possibly ATM itself); alternatively, in response to IR, these phosphorylations may depend upon prior phosphorylation of Ser 15 or Ser 20 . In response to UV irradiation, Ser 46 can be phosphorylated by the p38 MAPK (22), and recently Ser 46 was reported to be phosphorylated by homeodomain-interacting protein kinase 2 (HIPK2), which also is activated by exposure of cells to UV light (23,24). The kinase that phosphorylates Ser 46 after IR has not been identified.
To further establish the importance of ATM in mediating phosphorylation of Ser 46 , we examined the effect of wortmannin, an inhibitor of ATM (25), and PD169316, a p38 MAPK-specific inhibitor (26), on Ser 46 phosphorylation in A549 cells in response to IR. Fig. 2 shows that increasing concentrations of wortmannin inhibited IR-induced phosphorylation at both Ser 15 and Ser 46 , consistent with dependence on ATM, while increasing concentrations of PD169316 had no effect on phosphorylation at either site. Similar results were obtained in BT normal lymphoblasts (data not shown). Phosphorylation at each of the sites shown in Fig. 1 after exposure to IR also was examined in two other pairs of normal and A-T lymphoblasts, C3ABR and AT24RM, and BT (normal) and L3 (A-T), with FIG. 1. Phosphorylation of p53 on multiple N-terminal serines in response to ionizing radiation requires ATM. A, normal human lymphoblasts (GM02254) and A-T lymphoblasts (GM01526) were treated with 8 Gy IR and harvested at the indicated times. Extracts were immunoprecipitated using anti-p53 antibodies and analyzed for p53 and phosphorylation by Western immunoblot analysis; phosphorylation site-specific antibodies or monoclonal antibody to p53 (DO-1) are indicated (right). Extracts of SHEP human neuroblastoma cells, treated with adriamycin at 0.2 g/ml for 8 h, served as a positive control (PC) (35). Phosphorylation at Ser 9 , Ser 15 , Ser 20 , and Ser 46 was attenuated in the A-T lymphoblasts. B, the Western blot in A was scanned to compare the amount of phosphorylation at individual residues in A-T cells (GM01526) with the same residue in normal lymphoblasts (GM02254). The amount of phosphorylation at each residue was first normalized by dividing its value by the value for the amount of p53 at that time point. The normalized ratio was then multiplied by 100. Shown are data for 5 residues: Ser 6 , Ser 15 , Ser 20 , Ser 46 , and Ser 392 . The results show that phosphorylation at Ser 15 , Ser 20 , and Ser 46 (and Ser 9 , data not shown) is reduced and delayed in A-T cells compared with normal cells. The amount of p53 in A-T cells was less than half of that in normal cells up to 0.5 h after exposure and then increased to ϳ80% of the level in normal cells at 4 h. similar results; in neither A-T cell line did IR induce phosphorylation of Ser 46 , while robust phosphorylation at this site was observed in normal lymphoblasts (data not shown). That the p53 in A-T cells is capable of being phosphorylated in response to DNA damage was shown by examining the response in GM02254 and GM01526 cells exposed to UV light. In this case, no significant difference in p53 phosphorylation between the normal and A-T lymphoblasts was observed (data not shown), indicating that phosphorylation at Ser 9 , Ser 15 , Ser 20 , and Ser 46 was unlikely to be masked in A-T cells. These findings fit previous observations that ATM acts specifically in the cellular responses to IR (7). Thus, the N-terminal p53 phosphorylation sites can be grouped into two categories with respect to dependence on ATM for rapid phosphorylation in response to IR. One group, consisting of Ser 6 , Ser 33 , Ser 315 , and Ser 392 , is independent of ATM and constitutively phosphorylated at low levels; the other group, consisting of Ser 9 , Ser 15 , Ser 20 , and Ser 46 , is ATMdependent for a rapid response to IR-induced damage, presumably DNA double strand breaks.
Dependence of C-terminal Acetylation on N-terminal Phosphorylation-Previously, we and others suggested that acetylation of human p53 after DNA damage may be mediated through phosphorylation at N-terminal sites (4,15,27). To address this issue in vivo, we first examined the time course of p53 acetylation at Lys 320 and Lys 382 after exposure to 8 Gy IR in normal and A-T lymphoblasts (Fig. 3A). In normal lymphoblasts, acetylation at Lys 320 was observed within 1 h after IR and at Lys 382 by 2 h; both sites were well acetylated at 4 h after IR. In contrast, in A-T lymphoblasts, acetylation at both sites was significantly delayed and reduced (by 20 -40%) at 4 h after IR compared with normal lymphoblasts (Fig. 3B). No differences in the acetylation of these sites was observed between normal and A-T lymphoblasts after UV radiation (data not shown).
To determine which N-terminal phosphorylation site(s) are important for acetylation, we examined mutant p53s in which individual N-terminal phosphorylation sites were changed to Ala by transient transfection of p53-negative H1299 cells with the indicated p53 expression vectors (Fig. 3C). p53 protein levels from each transfection were comparable as shown by staining with the polyclonal, p53-specific antibody Ab-7, and wild-type p53 was acetylated at both sites with or without exposure to IR. Acetylation without IR was not unexpected, since the transfection procedure elicits a strong "stress-like" response in human cells, as reported previously (14,22). In contrast to wild-type p53, acetylation of Lys 382 in the p53 S15A mutant, as well as the L22Q/W23S double mutant, was virtually abrogated, while acetylation of Lys 320 was reduced significantly. Acetylation at Lys 382 also was significantly reduced by mutations that changed Ser 6 , Ser 9 , or Thr 18 to Ala. In contrast, changing Ser 20 , Ser 33 , Ser 37 , or Ser 46 to Ala had little, if any, effect on acetylation at Lys 382 . Except for the Ser 15 to Ala change, the effects of phosphorylation site mutants on the acetylation of Lys 320 were less clear, in part, due to weakness of the Ac-Lys 320 signal. DISCUSSION Previous studies showed that phosphorylation of p53 at Ser 15 and Ser 20 in response to IR is mediated by the ATM protein kinase and that these modifications are important for stabilizing and activating p53 as a transcription factor. We show here, for the first time, that phosphorylation of Ser 9 and Ser 46 also are dependent on the ATM kinase ( Figs. 1 and 4). Phosphorylation of Ser 46 was shown to be important for the induction of apoptosis in response to damage caused by exposure of epithelial-derived cell lines to UV light (6,22), and two protein kinases capable of phosphorylating Ser 46 , p38 MAPK (22) and HIPK2 (23,24), both of which are activated after exposure of cells to UV light, have been described. In contrast to UV light, activation of p53 after exposure of epithelial cells to IR primarily induces cell cycle arrest in G 1 . However, human lymphoid and neuronal cells are much more sensitive to p53-dependent, IR-induced apoptosis, and both p53 and ATM are required for its induction (28,29). We suggest that in lymphoid cells, activation of ATM in response to DNA double strand breaks mediates activation of an unidentified protein kinase that phosphorylates p53 at Ser 46 , possibly through recruitment of p53DINP1, the product of a recently identified p53-inducible  3. Efficient acetylation of p53 on C-terminal lysine residues in response to ionizing radiation requires phosphorylation at multiple N-terminal residues. A, normal lymphoblasts (GM02254) and A-T lymphoblasts (GM01526) were treated with 8 Gy IR and analyzed as described in the legend to Fig. 1 except that immunoblot analysis was with antibodies specific for acetylation at Lys 320 (PAbLys(Ac)320) or Lys 382 (PAbLys(Ac)382) as indicated. A549 human lung carcinoma cells, harvested 8 h after exposure to 25 J/m 2 UV-C, served as a positive control (PC) (15). B, the ratio of acetylation at Lys 320 and Lys 382 in A-T to that in normal lymphoblasts (ϫ100) was calculated after normalization for p53 amounts as described for phosphorylation in the legend to Fig. 1. The data show that acetylation in A-T cells was dramatically delayed before approaching levels in normal cells by 4 h after irradiation. C, H1299 cells (TP53 Ϫ/Ϫ ) were transiently transfected with vectors that expressed wild-type or p53 mutants with single amino acid substitutions (serine/threonine to alanine) at the indicated sites and were exposed or not to 8 Gy IR. Extracts were prepared 18 h after transfection and 2 h after irradiation for immunoprecipitation and Western blot analysis as described (see "Experimental Procedures"). Ab-7 is a p53-specific polyclonal antibody. gene that facilitates p53 phosphorylation at Ser 46 in response to IR (30). This protein kinase is unlikely to be p38 MAPK, which phosphorylates Ser 46 in response to UV, since PD169316, a p38 kinase-specific inhibitor, failed to block phosphorylation of Ser 46 (Fig. 2).
We also found that ATM is required for phosphorylation of p53 at Ser 9 . We previously reported that Ser 6 and Ser 9 became strongly phosphorylated in response to both IR-and UV-induced DNA damage and suggested that Ser 9 may be phosphorylated by CK1 in response to phosphorylation of Ser 6 (13). In vitro CK1 phosphorylates serines and threonines two residues distal to a phosphorylated serine or threonine. The data presented here suggest that in response to IR, phosphorylation of Ser 9 may be independent of phosphorylation at Ser 6 and dependent upon activation of an unknown protein kinase by ATM. Thus, as for Ser 20 , the kinase that phosphorylates Ser 9 in response to IR is likely to be activated by ATM. We cannot, however, rule out the possibility that recognition of Ser 9 by its kinase requires either Ser 15 or its phosphorylation.
The functional consequences of phosphorylation at Ser 6 and Ser 9 are unknown. Changing either serine to alanine had little effect on the ability of chimeric Gal4-p53(1-42) to activate transcription of a reporter in transient transfection assays (4), consistent with a report that these and other N-terminal phosphorylations are not essential for p53 activity (31). However, coupled with our previous results (15) and those of others (4,27) showing that phosphorylation of Ser 15 recruits CBP/p300 to p53, our current data (Fig. 3C) supports a cascade model in which phosphorylation at several N-terminal sites, of which phosphorylation of Ser 15 appears to be most important, promotes acetylation at C-terminal sites (Fig. 4). Our data further suggest that the N-terminal region important for regulating p53 interactions with HATs includes the first 18 residues of human p53, since abrogation of phosphorylation by changing Ser 20 , Ser 33 , Ser 37 , and Ser 46 to Ala did not affect C-terminal acetylation. While the double mutant L22Q/W23S also blocked C-terminal acetylation as did the equivalent mutations in mouse p53 (32), these changes most likely profoundly affect the structure of the amphipathic helical region (amino acids [17][18][19][20][21][22][23][24][25][26][27][28] that is likely to be important for interactions with both HATs and MDM2 (33). Thus, phosphorylation of Ser 6 and Ser 9 , along with phosphorylation of Thr 18 , may serve to amplify a primary signal initiated by the ATM-mediated phosphorylation of Ser 15 that is important for recruiting HATs to p53. We note, however, that changing Ser 18 of mouse p53, the equivalent of human Ser 15 , to Ala did not block the acetylation of mouse p53 at Lys 317 and Lys 379 , equivalent to human Lys 320 and Lys 382 , respectively, in ES cells (34). Thus, regulation of p53 acetylation through phosphorylation may not be equivalent in mice and humans.