Requirement of ATM for Rapid p53 Phosphorylation at Ser46 without Ser/Thr-Gln Sequences

Abstract

p53 phosphorylation at Ser46 following DNA damage is important for preferential transactivation of proapoptotic genes. Here, we report that ataxia-telangiectasia mutated (ATM) kinase is responsible for Ser46 phosphorylation of p53 during early-phase response to DNA damage. To elucidate the direct phosphorylation of p53 at Ser46 by ATM, an ATM mutant (ATM-AS) sensitive to ATP analogues was engineered. In vitro kinase assays revealed that p53 was phosphorylated at Ser46 by ATM-AS, even when ATP analogues were used as phosphate donors, although this phosphorylation site is not in an SQ motif, a consensus ATM site. Furthermore, Ser46 phosphorylation by ATM was dependent on the N- and C-terminal domains of p53, unlike Ser15 phosphorylation. Immunofluorescence analyses showed that Ser46-phosphorylated p53 was observed as foci in response to DNA damage and colocalized with γ-H2AX or Ser1981-phosphorylated ATM. These results suggest that ATM phosphorylates a noncanonical serine residue on p53 by mechanisms different from those for the phosphorylation of Ser15.

The tumor suppressor protein p53 activates the transcription of numerous target genes involved in cell cycle arrest, apoptosis, and DNA repair (5, 15, 35). Upon various cellular stresses, p53 is phosphorylated and acetylated at multiple sites to activate downstream target genes (13, 31, 36).

Phosphorylation of p53 at Ser15 leads to the dissociation of MDM2, an E3 ubiquitin ligase, from p53 to prevent MDM2-dependent p53 degradation (36). We have previously shown that Ser46 on p53 is phosphorylated following DNA damage and that this phosphorylation contributes to the expression of p53-regulated apoptosis-inducing protein 1 (p53AIP1) (33). Ser46 phosphorylation also contributes to the preferential transactivation of other proapoptotic genes, such as Noxa and PUMA, to prevent tumor formation (18, 27). Although p38 mitogen-activated protein (MAP) kinase, protein kinase C δ (PKCδ), homeodomain-interacting protein kinase 2 (HIPK2), and dual-specificity tyrosine phosphorylation-regulated kinase 2 (DYRK2) have been reported to phosphorylate p53 at Ser46 in response to UV or adriamycin (ADR), a radiomimetic DNA-damaging agent, these enzymes are controversial candidates for direct kinases for Ser46 phosphorylation occurring in early phase (within 1 h) in response to ionizing radiation (IR) (6, 11, 16, 41, 49).

Ataxia-telangiectasia mutated (ATM) is a member of the phosphatidylinositol 3-phosphate kinase (PI3-K) family and is crucial for the initiation of signaling pathways following exposure to IR. Functional defects of the gene encoding ATM cause the human genetic disorder ataxia-telangiectasia (A-T). The major hallmarks of A-T are neurodegeneration, immunodeficiency, genomic instability, and cancer predisposition (26). Following exposure to IR, ATM phosphorylates Ser/Thr-Gln (S/T-Q) sequences on numerous proteins participating in DNA damage responses (29). Among these proteins, p53 phosphorylation at Ser15 is a well-known target of ATM (3, 7, 21).

Here, we found that ATM directly phosphorylates p53 at Ser46 as well as Ser15 and that ATM is required for acute DNA damage response to induce Ser46 phosphorylation. Unlike Ser15 phosphorylation, the Ser46 phosphorylation by ATM requires both proline-rich and C-terminal domains of p53. Furthermore, Ser46-phosphorylated p53 is partially colocalized with IR-activated ATM that is known to localize at DNA double-strand break (DSB) sites. Interestingly, Ser46 phosphorylation by IR-activated ATM is induced within 1 h and ATM is required for early-phase response to DNA damage.

MATERIALS AND METHODS

RNA interference (RNAi) experiments and RT-PCR.

For the expression of short hairpin RNA (shRNA), oligonucleotides containing sequences homologous to ATM (5′-GATCCCCAAGCTATCAGAGAAGCTAATAAATTCAAGAGATTTATTAGCTTCTCTGATAGCTTTTTTTGGAAA-3′ and 5′-AGCTTTTCCAAAAAAAGCTATCAGAGAAGCTAATAAATCTCTTGAATTTATTAGCTTCTCTGATAGCTTGGG-3′) or to HIPK2 (5′-GATCCCCGAAAGTACATTTTCAACTGTTCAAGAGACAGTTGAAAATGTACTTTCTTTTTGGAAA-3′ and 5′-AGCTTTTCCAAAAAGAAAGTACATTTTCAACTGTCTCTTGAACAGTTGAAAATGTACTTTCGGG-3′) (10) were synthesized and the duplex oligonucleotide DNA was inserted into the pSUPER.retro vector (Oligoengine) to generate pSR-ATM and pSR-HIPK2, respectively. These plasmids were digested to obtain DNA fragments containing the H1 promoter and DNA coding for shRNA, and these DNA fragments were inserted into pLenti6.2/V5-DEST (Invitrogen) to generate pL-shATM and pL-shHIPK2, respectively. Lentiviruses were produced in accordance with the manufacturer's instructions (Invitrogen) and used to infect MCF7 or U2OS cells. To generate stable cell lines, infected cells were selected with blasticidin (Invitrogen). The sequences of primers for reverse transcription-PCR (RT-PCR) were as follows: 5′-GGCCTCACATGTGCAAGTTTTC-3′ and 5′-TTGGTAGGTATCAAGGAGGCTC-3′ for HIPK2 and 5′-TCCACAGTCTTCTGGGTGGCAGTGA-3′ and 5′-GGGGAGCCAAAAGGGTCATCATCTC-3′ for glyceraldehyde-3-phosphate dehydrogenase (GAPDH). For experiments with short interfering RNA (siRNA), Allstars negative-control siRNA and Hs_ATM_5_HP validated siRNA were purchased from Qiagen. Sequences of siRNAs and primers for RT-PCR for HIPK2 and DYRK2 are described by Hofmann et al. and Taira et al., respectively (17, 41). For p53 knockdown, siRNA described previously was used (12). Each 100 pmol of siRNA was transfected with HiPerfect transfection reagent (Qiagen) and RNAiMax transfection reagent (Invitrogen) into MCF7 or U2OS cells (2 × 10⁵ cells). At 48 h after transfection, cells were used for assays.

Immunoblotting analysis.

To prepare whole-cell lysates, cells were collected and stored at −80°C. The cells were thawed in chilled IP150 buffer (50 mM HEPES [pH 7.0], 150 mM NaCl, 1 mM EDTA, 2.5 mM EGTA, 10% glycerol, 1 mM dithiothreitol [DTT], 1 mM Na₃VO₄, and 5 mM NaF) containing 0.1% NP-40 and a protease inhibitor cocktail consisting of 10 μg/ml pepstatin A, 10 μg/ml antipain, 10 μg/ml chymostatin, 10 μg/ml leupeptin, 10 μg/ml E-64, and 10 μg/ml phenylmethylsulfonyl fluoride (PMSF). The lysates were centrifuged for 15 min in a microcentrifuge at 4°C, and the supernatants were collected and boiled in sodium dodecyl sulfate (SDS) sample buffer. The samples were separated on SDS-PAGE gels and blotted onto Immobilon-P transfer membrane (Millipore). The membranes were blocked with blocking solution containing 5% nonfat dry milk in phosphate-buffered saline (PBS) containing 0.1% Tween 20 (PBS-T) for 1 h at room temperature and then incubated with primary antibodies diluted in Can Get Signal Solution 1 (Toyobo) overnight at 4°C. After three washes with PBS-T, the membranes were incubated for 1 h with secondary antibodies in PBS-T containing 1% nonfat dry milk at room temperature. Immunoblots were visualized by chemiluminescence (Western Lightning; Perkin-Elmer). Anti-ATM antibody (2C1) was purchased from GeneTex, Inc.; anti-p53 antibody (9282), anti-phospho-Ser15-of-p53 antibody (9284), and anti-phospho-Thr68-of-Chk2 antibody (2661) were from Cell Signaling Technology; anti-phospho-Ser1981-of-ATM antibody (10H11.E12) was from Rockland Inc.; anti-glutathione S-transferase (anti-GST) antibody (B14) was from Santa Cruz Biotechnology; and anti-FLAG (M2) antibody was from Sigma. Antibodies to phospho-Ser392 of p53 were described previously (28, 33, 40). Anti-phospho-Ser46-of-p53 mouse monoclonal antibody was generated with a synthetic phosphorylated peptide (Lab of Monoclonal Antibody Co., Inc.). Signals from immunoblots were quantified by Multi-gauge v.3.0 (Fujifilm).

Plasmids and purification of recombinant proteins.

An expression construct of FLAG-tagged wild-type ATM was a gift from M. Kastan. To construct expression vectors for the mutant ATM, parts of the FLAG-tagged wild-type ATM were mutated using a QuikChange site-directed mutagenesis kit (Stratagene). Cells transiently transfected with constructs were thawed in IP150 buffer supplemented with 0.3% Nonidet P-40, followed by centrifugation. ATM was purified using anti-FLAG M2 agarose (Sigma) and then eluted into IP150 buffer with 3× FLAG peptides (Sigma) after two washes with IP150 buffer with 0.3% Nonidet P-40, two washes with IP500 buffer containing 500 mM NaCl with 0.3% Nonidet P-40, and two washes with IP150 buffer without detergent. For kinase assays using ATM-AS, a series of ATMs, including ATM-WT and ATM-KD, were prepared in IP150 or IP500 supplemented with 0.03% Nonidet P-40. GST fusion proteins were expressed in Escherichia coli BL21 cells from pGEX-4T-1 or pGEX-6P-1 vectors (GE Healthcare).

In vitro kinase assay.

ATP was purchased from Cell Signaling Technology, and all ATP analogues were from Biolog. In vitro kinase assays were performed in kinase buffer (9802) (25 mM Tris-HCl [pH 7.5], 5 mM β-glycerophosphate, 2 mM dithiothreitol, 0.1 mM Na₃VO₄, and 10 mM MgCl₂) (Cell Signaling Technology) supplemented with 10 mM MnCl₂ and 0.1 mM ATP or ATP analogues, at 30°C for 30 or 60 min. Reactions were stopped by the addition of SDS sample buffer followed by boiling. Reaction products were subjected to immunoblotting.

Confocal immunofluorescent microscopy.

Cells were fixed with 3.7% formaldehyde in PBS for 10 min at room temperature, permeabilized with 0.5% Triton X-100 in PBS for 10 min at room temperature, and then blocked with PBS containing 3% bovine serum albumin (BSA) and 0.1% goat serum for 1 h at room temperature. Cells were incubated with primary antibodies overnight at 4°C and then with secondary antibodies (Alexa Fluor 488 goat anti-rabbit IgG or Alexa Fluor 594 goat anti-mouse IgG; Invitrogen) for 1 h at room temperature. Vectashield with DAPI (4′,6-diamidino-2-phenylindole; Vector Laboratories Inc.) was used as the mounting medium. The primary antibodies used for immunostaining were anti-p53 (Ab-6; Calbiochem), anti-phospho-Ser1981 of ATM (10H11.E12; Rockland Inc.), anti-phospho-Ser15 of p53 (40), anti-phospho-Ser46 of p53 (33), anti-phospho-Thr68 of Chk2 (2661; Cell Signaling Technology), anti-phospho-Ser957 of SMC1 (NB100-205; Novus Biologicals), anti-γ-H2AX (JBW103; Upstate Biotechnology Inc.), and anti-PML antibody (PG-M3; Santa Cruz Biotechnology). All primary antibodies were diluted in Can Get Signal A (Toyobo). These immunostained objects were observed under a confocal immunofluorescent microscope using a Zeiss LSM5 Exciter system equipped with Zen software (Carl Zeiss Inc.). For preextraction experiments, cells were extracted with 0.2% Triton X-100 for 3 min at room temperature prior to 3.7% formaldehyde fixation, followed by the same procedures described above. Fluorescence resonance energy transfer (FRET) signals were obtained as described previously (23). In brief, cells were exposed to 488-nm light to excite Alexa Fluor 488 (donor fluorescence) and the emission of Alexa Fluor 568 (acceptor molecules) was scanned to detect FRET.

RESULTS

ATM is required for p53 phosphorylation at Ser46 following exposure to IR but not to UV.

It was reported that Ser46 phosphorylation of p53 is abrogated following exposure to IR in A-T lymphoblasts (38). To identify the kinase that phosphorylates p53 at Ser46 in response to IR, we first assessed whether ATM is required for this phosphorylation in other cell lines. Human mammary carcinoma MCF7 cells were infected with lentiviruses encoding a short hairpin RNA (shRNA) against ATM to generate stable cell lines devoid of ATM expression (pL-shATM) (Fig. 1A). The phosphorylation at Ser46 as well as Ser15 (3, 7, 21) was delayed and attenuated in ATM-depleted cells (Fig. 1B). Similar results were obtained using human osteosarcoma U2OS cells depleted of ATM (Fig. 1C and D). On the other hand, downregulation of ATM expression had no effect on phosphorylation at Ser392 of p53 in either MCF7 or U2OS cells (Fig. 1B and D). In contrast, depletion of ATM had little effect on phosphorylation at Ser46 or Ser15 after exposure to UV (Fig. 1E and F). We also performed immunoblotting to assess Ser46 phosphorylation at various doses of IR. Ser46 phosphorylation was obviously detected at a high dose of IR compared to Ser15 phosphorylation, consistent with previous reports (33) (Fig. 2). These findings suggest that ATM is responsible for the phosphorylation of p53 at Ser46 after exposure to IR but not to UV.

FIG. 1.

ATM is required for the phosphorylation of p53 at Ser46 in response to IR but not UV. (A and C) Expression level of ATM. MCF7 cells (A) or U2OS cells (C) were infected with control lentiviruses (pL-vec) or lentiviruses encoding shRNA to ATM (pL-shATM), and cell lysates from the selected stable cells were subjected to immunoblotting. Coomassie brilliant blue (CBB) staining is also shown as a loading control. (B and D) Time kinetics of accumulation and phosphorylation of p53 in MCF7 cells and its derivatives after exposure to IR of 10 Gy (B) or in U2OS cells and its derivatives after exposure to IR of 20 Gy (D). Cell lysates from the above stable cells were analyzed by immunoblotting with antibodies against p53 (α-p53), phospho-Ser15 (α-p-Ser15), phospho-Ser46 (α-p-Ser46), and phospho-Ser392 (α-p-Ser392) of p53. CBB staining shows the amount of total protein applied to each lane. (E and F) Time kinetics of accumulation and phosphorylation of p53 in MCF7 cells and its derivatives after exposure to UV of 20 J/m² (E) or in derivatives of U2OS cells after exposure to UV of 30 J/m² (F). Cell lysates from the above stable cells were subjected to immunoblotting as described for panel B. The expression level of ATM is shown in panels A and C. The immunoblots in Fig. 1B and D to F were scanned to compare the amount of phosphorylation at Ser46 or Ser392 in ATM-depleted cells (pL-shATM) and in MCF7 or U2OS cells. The amount of phosphorylated p53 at each residue was calculated by dividing the value of phosphorylated p53 by that of total p53 at that time point.

FIG. 2.

Dose- and time-dependent accumulation of p53 and Ser46 phosphorylation. (A) Time kinetics of accumulation and phosphorylation of p53 in MCF7 cells after exposure to IR of the indicated doses. Cell lysates were subjected to immunoblotting as described for Fig. 1B. (B and C) Time kinetics of accumulation and phosphorylation of p53 in MCF7 cells and its derivatives after exposure to IR of 2 Gy (B) or 4 Gy (C). Cell lysates from the above stable cells were subjected to immunoblotting as described for Fig. 1B. The expression level of ATM is shown in Fig. 1A.

ATM phosphorylates p53 at Ser46 in vitro.

To examine whether ATM associates with kinases that phosphorylate p53 at Ser46, FLAG-tagged ATM transiently expressed in 293T cells was purified by immunoprecipitation with anti-FLAG antibody and used for in vitro kinase assays using full-length p53 fused to GST (GST-p53) as a substrate (Fig. 3A). When wild-type ATM (ATM-WT) was mixed with GST-p53, ATM-WT phosphorylated p53 at Ser46 as well as at Ser15, although neither a kinase-dead ATM mutant (ATM-KD) nor eluates from cells transfected with an empty vector (control) produced the same phosphorylation patterns (Fig. 3A). On the other hand, p53 phosphorylation at Ser392 by ATM was not detected, as shown in Fig. 3A. Given that ATM purified from 293T cells contains the kinase activity to phosphorylate p53 at Ser46, similar experiments were carried out using different cell lines, including human lung carcinoma H1299 and MCF7 cells. Recombinant ATM purified from these cells also phosphorylated p53 at Ser46 as well as Ser15 (Fig. 3B and C), suggesting that the Ser46 kinase activity in the immunoprecipitates was not a cell-line-specific phenomenon. Moreover, an in vitro kinase assay with GST-p53 carrying an alanine substitution at Ser15 (S15A) or Ser46 (S46A) showed that each phosphorylation event is mutually independent (Fig. 3D). The effect of ATM kinase inhibitors, wortmannin and KU-55933, on Ser46 kinase activity was assessed. p53 phosphorylation at Ser46 was blocked by both ATM kinase inhibitors as efficiently as that at Ser15 (Fig. 3E and F), suggesting that Ser46 kinase activity in the immunoprecipitate was dependent on ATM, consistent with the result that ablation of ATM caused impaired IR-induced Ser46 phosphorylation (Fig. 1B and D).

FIG. 3.

ATM phosphorylates p53 fused to GST (GST-p53) at Ser46 in vitro. (A) Kinase activity of ATM. HEK 293T cells were transiently transfected with an empty vector (control) or an expression construct of FLAG-ATM wild-type (ATM-WT) or kinase-dead ATM (ATM-KD). The kinase activity of ATM purified with an anti-FLAG antibody was analyzed by an in vitro kinase assay using GST-p53 as a substrate. Phosphorylated p53 was detected by immunoblotting with the indicated antibodies. ATM in the reaction mixture was also detected by anti-ATM antibody. Cell lysates from MCF7 cells treated with 3 μM ADR are used for positive controls (p.c.) of antibodies. (B and C) Kinase activity of ATM in lysates from different cell lines. H1299 (B) or MCF7 (C) cells were transiently transfected with an empty vector (control) or an expression construct of FLAG-ATM-WT. The kinase activity of ATM purified with an anti-FLAG antibody was analyzed by an in vitro kinase assay using GST-p53 as a substrate. Phosphorylated p53 was detected by immunoblotting as described for panel A. (D) GST-p53 carrying an alanine substitution at Ser15 (S15A) or at Ser46 (S46A) was subjected to the in vitro kinase assay as in panel A. (E) Effect of an ATM kinase inhibitor on the Ser46 kinase activity of ATM. The in vitro kinase assay was performed as in panel A with different concentrations of wortmannin. DMSO, dimethyl sulfoxide. (F) Effect of KU-55933 on the Ser46 kinase activity of ATM. The in vitro kinase assay was performed as in panel A with or without 10 μM KU-55933. α, anti.

Construction of ATM sensitive to ATP analogues.

It is believed that the S/T-Q motif is important for the phosphorylation of substrates by ATM (22, 34), and ATM phosphorylates a typical SQ motif in p53 at Ser15 (3, 7, 21). On the other hand, the Ser46 phosphorylation site is not in the SQ motif and therefore is a noncanonical ATM site (data not shown). To assess whether ATM kinase directly phosphorylates Ser46, a combined chemical and genetic approach was adopted. This strategy involved the alteration of a conserved bulky residue in the ATP-binding pocket of kinase to small amino acids. The resulting mutant with an enlarged ATP-binding pocket was able to bind ATP analogues that wild-type kinase cannot accept. This method has allowed the search for bona fide substrates of various kinases, including the Src family kinase (39), stress-activated protein kinase/Jun N-terminal kinase (SAPK/JNK) (14), and cyclin-dependent kinase 1 (cdk 1) (43). The X-ray crystallographic structure of ATP-bound PI3-Kγ belonging to the PI3-K family showed that the Tyr867 residue on PI3-Kγ contacts the N⁶ position of the adenine ring of ATP (45). The sequence alignment of ATP-binding pockets among members of the PI3-K family showed that this tyrosine residue is conserved and corresponds to Tyr2755 in ATM (data not shown). To engineer ATM sensitive to ATP analogues (ATM-AS), a Tyr2755 residue in ATM was replaced with alanine. An ATM-AS molecule was expressed and could be immunoprecipitated with anti-FLAG antibody similarly to ATM-WT. Importantly, Tyr2755 is deep in the ATP-binding pocket and far from the substrate binding site in order not to alter the substrate specificity of the kinase. To assess whether ATM directly phosphorylates p53 at Ser46, in vitro kinase assays with ATM-AS were performed (Fig. 4A). Natural ATP or various N⁶-substituted ATP analogues were used as phosphate donors, and ATM kinase activity was determined using a phospho-Ser15-specific antibody. ATM-WT phosphorylated p53 at Ser15 in the presence of natural ATP; however, when bulky ATP analogues were used, ATM-WT was not able to transfer a phosphate group to Ser15 (Fig. 4A), suggesting that the ATP analogues used did not associate with ATM-WT. On the other hand, ATM-AS exhibited markedly improved specificity for N⁶-substituted ATP analogues (Fig. 4A). When N⁶-1-methylbutylated-ATP (1-MeBu) was used as a phosphate donor, the kinase activity of ATM-WT was minimal and ATM-AS phosphorylated p53 at Ser15 more efficiently than ATM-WT (Fig. 4A, row IV).

FIG. 4.

ATM directly phosphorylates Ser46 of p53 using a mechanism different from that for Ser15. (A) Recombinant ATM was purified from 293T cells expressing FLAG-tagged ATM using anti-FLAG antibody and detected by immunoblotting with anti-ATM antibody. Kinase activity of ATM-WT or an ATM mutant sensitive to ATP analogues (ATM-AS and ATM-Y2755A) was measured by in vitro kinase assays using GST-p53. In this assay, natural ATP or various N⁶-substituted ATP analogues (row I, ATP; row II, N⁶-benzyl ATP [N⁶-Bn-ATP]; row III, N⁶-phenylethyl ATP [N⁶-PhEt-ATP]; row IV, N⁶-1-methylbutyl ATP [N⁶-1-MeBu-ATP]; row V, N⁶-2-methylbutyl ATP [N⁶-2-MeBu-ATP]) were used as phosphate donors, and the kinase activity was determined using α-p-Ser15 of p53 antibody. (B) In vitro kinase assays of ATM-WT, ATM-AS, and ATM-KD using GST-p53 as the substrate. Immunoprecipitates from 293T cells transfected with empty vector (control) were also subjected to the same assay. ATP or N⁶-1-MeBu-ATP (1-MeBu) was used as a phosphate donor for the kinase reaction. “−” indicates a sample without a phosphate donor in the reaction mixture. α, anti.

Kinase domain of ATM directly phosphorylates Ser46 of p53.

To explore whether ATM itself directly phosphorylates Ser46 on p53, in vitro kinase assays using ATM-AS and N⁶-1-MeBu-ATP were performed. If ATM directly phosphorylates Ser46 on p53, ATM-AS would phosphorylate Ser46 as well as Ser15 with N⁶-1-MeBu-ATP. If ATM activates a separate Ser46 kinase that is contained in the ATM immune complex, ATM-AS would phosphorylate Ser15 but not Ser46 with N⁶-1-MeBu-ATP, as unmodified kinases cannot use ATP analogues as a phosphate donor. The in vitro kinase assay revealed that the phosphorylation pattern of Ser46 was quite similar to that of Ser15 (Fig. 4B). ATM-AS phosphorylated both serine residues more efficiently than normal ATP even when N⁶-1-MeBu-ATP was used (Fig. 4B, lanes 8 and 9). In contrast, ATM-WT phosphorylated both Ser15 and Ser46 of p53 only in the presence of normal ATP (Fig. 4B, lanes 5 and 6). These phosphorylations were not due to contamination of ATP associated with the immunoprecipitates because the immunoprecipitate complex did not phosphorylate p53 in the absence of phosphate donors (Fig. 4B, lanes 1, 4, 7, and 10). These findings suggest that ATM itself directly phosphorylates Ser46 without any effector kinase.

ATM requires N- and C-terminal domains of p53 for Ser46 phosphorylation.

To investigate how ATM phosphorylates a noncanonical target such as Ser46 on p53, the efficiency of Ser46 phosphorylation was analyzed using various deletion mutants of p53. The consensus sequences required for phosphorylation by ATM were identified using a short peptide library (22, 34). However, there is a possibility that ATM might require the whole structure of substrates for sufficient phosphorylation. Therefore, the phosphorylation efficiency between full-length p53 and a short peptide containing Ser15 or Ser46 was compared to evaluate whether Ser46 phosphorylation is a conformation-dependent event. ATM was able to phosphorylate Ser15 on short peptides as efficiently as full-length p53 (Fig. 5A, left). In contrast, no Ser46 phosphorylation was detected on a short peptide (Fig. 5A, right), indicating that the whole structure of p53 is required for Ser46 phosphorylation. To exclude the possibility that Ser46 was not exposed because a small region of p53 was fused to a larger GST protein, in vitro kinase assays using chemically synthesized p53 peptide were performed (data not shown). Dot blot analysis showed that ATM did not phosphorylate short peptides containing Ser46 without GST (data not shown). Furthermore, antibodies specifically recognized chemically synthesized phosphorylated forms of corresponding short peptides (data not shown). Taken together, these data suggest that the entire p53 structure is required for Ser46 phosphorylation.

FIG. 5.

ATM requires the whole structure of p53 for Ser46 phosphorylation. (A) Comparison experiments of phosphorylation by ATM between p53 full-length and short peptides containing Ser15 (left panel) or Ser46 (right panel) from p53. Full-length p53 and short peptides were produced as GST fusion proteins and used for in vitro kinase assay. To compare the phosphorylation efficiencies, full-length p53 and short peptides were mixed and the mixtures were subjected to immunoblotting with indicated antibodies. Specificity of α-p-Ser46 antibody was examined using peptides phosphorylated or not on Ser46 (data not shown). (B) Schematic representation of deletion mutants of p53. Various deletion mutants of p53 were used for the experiments shown in panels C and D. ΔTAD1 corresponds to the deletion of amino acids 1 to 39 of p53, ΔPro corresponds to the deletion of amino acids 64 to 92, and ΔC corresponds to the deletion of amino acids 292 to 393. (C and D) Phosphorylation of p53 deletion mutants in vitro (C) or in vivo (D). (C) In vitro kinase assays using deletion mutants of p53 fused to GST as the substrates. Immunoblotting with anti-GST antibody shows the amount of each substrate in the reaction mixture. (D) Deletion mutants of p53 were transiently expressed in H1299 cells, and at 24 h after transfection, cells were exposed to 0 Gy (−) or 10 Gy (+) of IR. The phosphorylation of p53 at 1 h after IR was detected by immunoblotting. α, anti.

To map other regions that may be required for Ser46 phosphorylation, various deletion mutants of p53 were used (Fig. 5B). Deletion of the proline-rich (ΔPro) or C-terminal (ΔC) domains of p53 caused decreased Ser46 phosphorylation in vitro (Fig. 5C). In contrast, deletion of the N-terminal TAD1 domain had little or no effect on Ser46 phosphorylation.

To confirm the effect of the deletion of p53 on Ser46 phosphorylation in vivo, the deletion mutants of p53 used for in vitro kinase assays were expressed transiently in H1299 cells and the transfected cells were exposed to IR. The proline-rich and C-terminal domains were required for IR-dependent Ser46 phosphorylation (Fig. 5D), which was consistent with in vitro data shown above. These findings imply that the whole structure of p53 is required for Ser46 phosphorylation by ATM, although ATM is directed to Ser15 by amino acid sequences surrounding it.

ATM phosphorylates Ser46 of p53 at the sites of DSBs.

After exposure to IR, activated ATM is recruited to DSBs and phosphorylates various substrates, including Chk2, SMC1, and H2AX (4, 24, 47). To investigate where ATM phosphorylates Ser46 on p53, antibodies specific for detection of Ser15- and Ser46-phosphorylated p53 were used for immunofluorescence analyses. The specificity of these antibodies was confirmed by immunofluorescence with p53 mutants bearing a mutation at Ser15 or at Ser46 and by immunoblotting with samples from cells with or without DNA damage (data not shown). Following exposure of cells to IR, p53 accumulated in the nuclei (Fig. 6A and B) and Ser15-phosphorylated p53 (Fig. 6A and C) exhibited a diffuse nuclear distribution. On the other hand, confocal immunofluorescent microscopic analyses with an antibody against phospho-Ser46 of p53 showed that Ser46-phosphorylated p53 was observed as foci (Fig. 6B) that partially colocalized with γ-H2AX (Fig. 6D) and Ser1981-phosphorylated ATM (Fig. 6E) but not the PML body (Fig. 6F), which has been known to form another nuclear focus. The foci of fluorescence signals from phospho-Ser46 of p53 in immunofluorescence were confirmed to be p53 itself by RNAi-directed ablation of p53 expression (Fig. 6A and B). Dose dependency of IR and time kinetic experiments showed that foci of phospho-Ser46 gradually increased in a dose-dependent manner and were observed even in cells irradiated at a lower dose (2 Gy, 30 min) (Fig. 7), at which Ser46 phosphorylation was not detected by immunoblot assays. The reason why phospho-Ser46 was not detected in immunoblots may be explained by Ser46-phosphorylated p53 in cells being diluted to perform immunoblot assays, whereas phospho-Ser46 concentrates as foci, subsequently observed by immunofluorescence. We also conducted fluorescence resonance energy transfer (FRET) experiments to show colocalization of phospho-Ser46 with DSB sites. As shown in Fig. 6F, a FRET signal was detected in immunofluorescence with anti-phospho-Ser46 and anti-γ-H2AX antibodies; whereas FRET from immunofluorescence with anti-phospho-Ser46 and anti-PML antibodies was not observed, confirming that Ser46 phosphorylation indeed colocalized with DSB sites. Furthermore, we assessed whether DSB foci (γ-H2AX) colocalizes with Chk2 and SMC1. Immunofluorescent analyses showed that phosphoforms of Chk2 and SMC1 were partially colocalized with DSB foci (Fig. 8), as seen in phospho-Ser46-p53/γ-H2AX (Fig. 6D). We also assessed the effect of permeabilization of cells on the phospho-Ser46 focus formation. When cells were treated with a detergent (0.2% Triton X-100) prior to fixation with formaldehyde, most p53 molecules, including focus-associated p53, were washed out (Fig. 6E). These observations are similar to previous findings that focus-associated Chk2 immediately dissociates from chromatin after activation by ATM-mediated phosphorylation (4). These data demonstrate that Ser46 phosphorylation of p53 by IR-activated ATM occurs at the sites of DSBs, and it seems that some ATM substrates such as p53 and Chk2 are caught and immediately released on ATM-associated foci.

FIG. 6.

Ser46, but not Ser15, of p53 is colocalized with activated ATM at DSB sites. (A and B) At 48 h after transfection with either control or p53 siRNA, MCF7 cells were exposed to 0 Gy (−) or 10 Gy (+) of IR and subjected to confocal immunofluorescent analysis at 30 min after irradiation. (A) Immunofluorescence with anti-phospho-Ser15-of-p53 (green) and anti-p53 (red) antibodies. (B) Immunofluorescence with anti-phospho-Ser46-of-p53 (green) and anti-p53 (red) antibodies. (C) Immunofluorescence with anti-phospho-Ser15-of-p53 (green) and anti-γ-H2AX (red) antibodies. (D) Immunofluorescence with anti-phospho-Ser46-of-p53 (green) and anti-γ-H2AX (red) antibodies. (E) Release of Ser46-phosphorylated p53 from chromatin by preextraction of a detergent prior to formaldehyde fixation. MCF7 cells were exposed to 0 Gy (−) or 10 Gy (+) of IR and subjected to confocal immunofluorescent analysis 30 min after irradiation. To assess whether activated ATM (Ser1981-phosphorylated ATM), Ser46-phosphorylated p53, or γ-H2AX is tightly associated with DSB, preextraction (PE) before fixation was performed. (F) Immunofluorescence with anti-phospho-Ser46-of-p53 (green) and anti-γ-H2AX or anti-PML (red) antibodies. Nuclei were stained with DAPI (blue). FRET signals occurring because of colocalization between Ser46-phosphorylated p53 and γ-H2AX or PML are shown (bottom panels).

FIG. 7.

Dose- and time-dependent accumulation of phospho-Ser46 foci. (A) MCF7 cells were exposed to 0 Gy (−) or 10 Gy (+) of IR and subjected to confocal immunofluorescent analysis at 1 or 2 h after irradiation. Immunofluorescence with anti-phospho-Ser46-of-p53 (green) and anti-γ-H2AX (red) antibodies. DAPI (blue) is also shown. (B) MCF7 cells were exposed to the indicated doses of IR and subjected to confocal immunofluorescent analysis at 30 min after irradiation. Immunofluorescence with anti-phospho-Ser46-of-p53 (green) and anti-γ-H2AX (red) antibodies. DAPI (blue) is also shown. α, anti.

FIG. 8.

Release of other ATM substrates from chromatin by preextraction of a detergent prior to formaldehyde fixation. MCF7 cells were exposed to 0 Gy (−) or 10 Gy (+) of IR and subjected to confocal immunofluorescent analysis 30 min after irradiation. To assess whether activated ATM (Ser1981-phosphorylated ATM) or protein phosphorylated by ATM (Thr68-phosphorylated Chk2, Ser957-phosphorylated SMC1, or γ-H2AX) is tightly associated with DSB, preextraction (PE) before fixation was performed. α, anti.

Ser46 is preferentially phosphorylated by ATM in the early inductive phase of response to DNA damage.

Although several protein kinases that are capable of phosphorylating Ser46 have been identified, HIPK2 and DYRK2 are the most prominent kinases responsible for Ser46 phosphorylation (6, 9, 33, 41, 49). According to previous reports, HIPK2 phosphorylates Ser46 following exposure to UV (16), but it is controversial whether this kinase actually responds to double-strand breaks because Taira et al. have reported that HIPK2 is a specific kinase serving in a UV-mediated pathway (9, 16, 41). Another Ser46 kinase, DYRK2, phosphorylates following treatment with adriamycin (ADR), a radiomimetic DNA-damaging reagent (41). Although Ser46 phosphorylation observed in this report was obtained at late-phase response to DNA damage, a kinase responsible for Ser46 phosphorylation occurring at early-phase response to IR remains unidentified. Therefore, we next investigated whether ATM is required for Ser46 phosphorylation occurring at early-phase response to IR. In ATM-deficient AT2KY cells, Ser46 was not phosphorylated at early-phase response to DNA damage but became phosphorylated at late phase (Fig. 9A). To define the relation between ATM and other Ser46 kinases, Ser46 phosphorylation was assessed over 24 h under conditions of more than 80% reduction of ATM, DYRK2, and HIPK2 expression (Fig. 9B to E). Ser46 phosphorylation rapidly occurred after exposure to IR in HIPK2-depleted cells as well as in control cells, and it peaked at 2 h after IR treatment (Fig. 9C). In contrast, depletion of ATM causes decreased Ser46 phosphorylation occurring at early-phase response. However, Ser46 phosphorylation was detected after 24 h at a steady-state level and a significant effect on Ser46 phosphorylation by depletion of HIPK2 was not detected by immunoblotting with anti-Ser46 antibody. Thus, although HIPK2 was reported to be responsible for Ser46 phosphorylation observed at 24 h after IR in MCF7 cells (9), HIPK2 knockdown did not alter Ser46 phosphorylation (Fig. 9C). Damage response following treatment with ADR was also assessed using siRNAs (Fig. 9D and E) because DYRK2 was reported to phosphorylate Ser46 at 24 h after exposure to ADR in human osteosarcoma U2OS cells (41). Treatment of U2OS cells with ADR induced rapid p53 phosphorylation at Ser46, and it continued over 24 h in control cells (Fig. 9E). In cells transfected with siRNA for ATM (siATM), Ser46 phosphorylation induced by ADR was significantly attenuated in early phase (Fig. 9E), but it gradually recovered. In contrast, in cells devoid of DYRK2, Ser46 phosphorylation was not affected during early response times (i.e., 6- and 12-h time points), but levels decreased by 24 h as expected (Fig. 9E). Again, HIPK2 knockdown using siRNA did not alter Ser46 phosphorylation and had only a marginal effect on Ser46 phosphorylation (Fig. 9E). On the other hand, upon UV treatment, Ser46 phosphorylation is severely affected by HIPK2 knockdown (Fig. 9F), suggesting that HIPK2 seems to be a kinase directed by UV irradiation, at least under our conditions. These observations suggest that ATM is selectively responsible for Ser46 phosphorylation in early-inductive-phase response to DSBs.

FIG. 9.

ATM preferentially phosphorylates p53 at Ser46 in early-phase response to DSBs. (A) Time kinetics of accumulation and phosphorylation of p53 in human fibroblast TIG7 cells or AT2KY cells, human fibroblast cells derived from an A-T patient, treated with 3 μM ADR for indicated periods. Cell lysates were assayed as described for Fig. 1B. Stars indicate bands specific to antibodies. (B and C) ATM is required for Ser46 phosphorylation of p53 in early-phase response to IR. (B) Expression of ATM or HIPK2 in MCF7 cells infected with control lentiviruses (pL-control) or lentiviruses encoding shRNA to HIPK2 (pL-shHIPK2) or ATM (pL-shATM). Expression of ATM was examined by immunoblotting. Expression of HIPK2 was examined by RT-PCR. Coomassie brilliant blue (CBB) staining and RT-PCR of GAPDH are shown for loading controls. (C) Time kinetics of accumulation and phosphorylation of p53 after cellular exposure to IR of 10 Gy. Cell lysates were assayed as described for Fig. 1B. (D to F) ATM is required for Ser46 phosphorylation of p53 in early-phase response to ADR. (D) Expression of ATM, DYRK2, or HIPK2 in U2OS cells transfected with either control siRNA (siControl), siRNA to ATM (siATM), DYRK2 (siDYRK2), or HIPK2 (siHIPK2). The expression level of ATM was determined by immunoblotting, and expression levels of DYRK2 and HIPK2 were determined by RT-PCR. As loading controls, CBB staining and RT-PCR of GAPDH are also indicated. (E and F) Accumulation of p53 and Ser46 phosphorylation occurred in early-phase response to ADR or UV irradiation. U2OS cells transfected with indicated siRNAs were treated with 0.5 μM ADR (E) or 30 J/m² UV (F) for indicated periods, and cell lysates were assayed as described in Fig. 1B. The amount of Ser46-phosphorylated p53 was calculated by dividing the value of phosphorylated p53 by that of total p53 at that time point.

DISCUSSION

Here, we found that Ser46 on p53 is directly phosphorylated by ATM in response to IR. Although ATM preferentially phosphorylates S/T-Q sequences (22, 34), the current experiments revealed that Ser46 is phosphorylated by ATM in a conformation-dependent manner, although this site is an S-P sequence (data not shown). Analyses with deletion mutants of p53 revealed that the proline-rich and C-terminal regions of p53, but not the TAD1 domain, are required for the Ser46 phosphorylation by ATM (Fig. 5C and D). Since the proline-rich domain, but not the TAD1 domain, is reported to be indispensable for p53-mediated apoptosis (44, 46), the current finding that the proline-rich domain is required for Ser46 phosphorylation may reflect the selectivity of p53-induced apoptosis. Again, ATM also requires the C-terminal region of p53 to phosphorylate Ser46 (Fig. 5C and D), although DYRK2 can phosphorylate Ser46 on GST-p53(1-92) (41). There are an increasing number of reports that ATM phosphorylates non-S/T-Q sequences such as Ser1893 (S-E) on ATM (25) and Ser1189 (S-P) or Ser1452 (S-G) on Brca1 (8), although in none of these cases has it been demonstrated that ATM directly phosphorylates these sites. ATM may phosphorylate these non-S/T-Q sequences by recognizing the whole or partial structure of these proteins.

In this study, an ATP analogue-accepting ATM mutant (ATM-AS) system was constructed by alanine substitution for Tyr2755 on ATM to confirm that ATM directly phosphorylates Ser46 on p53 (Fig. 4B). Since Tyr2755 on ATM is conserved in the PI3-K family (data not shown), this substitution could be applied to search for direct targets of not only ATM but also other PI3-Ks.

ATM attenuated Ser46 phosphorylation following exposure to IR or ADR but not to UV (Fig. 1). This is consistent with previous reports that ATM is activated by DSBs (19). Immunofluorescence analyses showed that Ser46-phosphorylated p53 colocalized with γ-H2AX and IR-activated ATM at foci of DSBs (Fig. 6D and E); in contrast, Ser15-phosphorylated p53 was diffusely distributed in the nuclei (Fig. 6A and C). Several groups showed that Ser15 on p53 may be phosphorylated by ATM prior to the recruitment of IR-activated ATM to DSBs, because phosphorylation of Ser15 on p53 does not require the recruitment of ATM to DSBs and this occurs in the initial step of ATM activation (2, 20, 24). Regarding the subnuclear localization of phospho-Ser15 of p53, there is a different report that Ser15-phosphorylated p53 does not form foci at DNA damage sites (4), in agreement with our data reported here. In contrast, it has also been reported that p53 phosphorylated at Ser15, but not bulk p53, formed foci that colocalized with γ-H2AX (1). Why does such a discrepancy happen? One possible explanation is that it may be dependent on cell lines used. The focus formation of phospho-Ser15 of p53 may be observed in normal diploid fibroblasts but not cancerous cell lines such as U2OS and MCF7 cells used in our assay. In addition, it remains enigmatic how damaged cells decide their fate: to repair and live or to die. The current observation that Ser46 is phosphorylated by ATM at DSBs may explain the previous report that phosphorylation of Ser15 on p53 is much more sensitive to cell damage and occurs rapidly compared to phosphorylation at Ser46, which is important for induction of apoptosis (33).

Time course experiments revealed that depletion of ATM selectively interfered with Ser46 phosphorylation at early phase (Fig. 9C and E). In response to IR, Ser15 is sequentially phosphorylated by ATM at an earlier inductive phase and by an A-T and Rad3-related (ATR) kinase at a later steady-state phase (Fig. 10) (42). Our data also suggest that Ser46 is sequentially phosphorylated by ATM and other kinases at different time kinetics to maintain the level of Ser46 phosphorylation (Fig. 10), and presumably apoptotic signals triggered by strong DNA damage would be transduced into damaged cells so that they can efficiently die. It has been reported that DYRK2 translocates into the nucleus to phosphorylate Ser46 in response to ADR (41) and that HIPK2 is stabilized by IR or treatment with ADR to phosphorylate Ser46 (37, 48). Moreover, it has recently been reported that Siah-1 binds to HIPK2 to degrade it and that the phosphorylation of Siah-1 at Ser19 by ATM causes the disruption of Siah-1-HIPK2 interaction to stabilize HIPK2 (48). However, these kinases are unlikely to be responsible for Ser46 phosphorylation in early phase because the depletion of ATM selectively interfered with Ser46 phosphorylation at early phase and that of DYRK2 did at late phase in response to treatment with ADR in our present study (Fig. 9E). Following exposure to IR, depletion of ATM significantly abrogated rapid Ser46 phosphorylation at early phase, but we observed only a marginal effect of HIPK2 on Ser46 phosphorylation under conditions in which HIPK2 knockdown causes impaired Ser46 phosphorylation directed by UV irradiation (Fig. 9C, E, and F). Moreover, it has been reported that HIPK2 is phosphorylated in response to UV but not gamma irradiation (16) and in most reports that HIPK2 is a Ser46 kinase and activated by UV irradiation (11, 16, 41). Our data presented here support the latter observation (11, 16, 41) that HIPK2 is a Ser46 kinase, specifically serving in a UV-mediated pathway. Regarding ATM dependency of DYRK2, DYRK2 accumulates in nuclei more than 8 h following ADR treatment (41), but Ser46 of p53 is phosphorylated within 6 h after DNA double-strand breaks under our conditions. Such a discrepancy may be due to a higher sensitivity of anti-Ser46 antibody used in our study. Therefore, probably, DYRK2 seems to be dependent on ATM and may be required for Ser46 phosphorylation occurring at late-phase response of DNA double-strand breaks.

FIG. 10.

A proposed model for contribution of kinases to p53 phosphorylation. In response to DNA damage, Ser15 is reported to be sequentially phosphorylated by ATM at an earlier inductive phase and followed by ATR at a later steady-state phase. Ser46 is also sequentially phosphorylated by ATM and other kinases at different time kinetics to maintain the level of Ser46 phosphorylation.

In our present study, IR-activated ATM is partially colocalized with Ser46-phosphorylated p53 in response to DNA damage, suggesting that p53 phosphorylation at Ser46 is required to trigger early apoptotic signals to damaged cells. The reason why Ser46-phosphorylated p53 is recruited to DSB sites is still unclear. It has been reported that Chk2 is phosphorylated by ATM near DSB sites and that activated Chk2 releases from DSB sites to function during G₁ and G₂ checkpoints (30, 32). Similarly, activated ATM phosphorylates p53 at Ser46 near DSB sites and Ser46-phosphorylated p53 may release from DSB sites to promote transcription of proapoptotic genes. Under various stresses, including UV and IR, DYRK2 and HIPK2 may work together with ATM to phosphorylate p53 at Ser46, leading to apoptosis to ensure that severely damaged cells are completely killed, though further investigations are needed to elucidate the molecular mechanisms of ATM-triggered p53-mediated apoptosis.

In conclusion, our present study strongly supports the idea that ATM directly phosphorylates p53 at Ser46 and is required for Ser46 phosphorylation occurring in early-phase DNA damage response. The direct link of ATM to Ser46 phosphorylation of p53 provides new insights into ATM-mediated p53-dependent apoptosis, though it cannot be ruled out that ATM may require adaptor proteins for Ser46 phosphorylation near DSB sites to facilitate its phosphorylation. Here, we show that ATM is likely to directly phosphorylate Ser46 of p53, but the kinase activity for Ser46 may be dependent on unidentified posttranslational modifications of ATM in response to severe DNA damage. In addition, it is quite difficult to determine bona fide in vivo kinases responsible for Ser46, so we cannot exclude the possibility that ATM is indirectly involved in the phosphorylation of Ser 46. Therefore, further studies are required to clarify whether ATM is a bona fide in vivo kinase responsible for Ser46 phosphorylation.