DNA Damage-Induced Acetylation of Lysine 3016 of ATM Activates ATM Kinase Activity

The ATM protein kinase is essential for cells to repair andsurvive genotoxic events. The activation of ATM's kinase activityinvolves acetylation of ATM by the Tip60 histone acetyltransferase.In this study, systematic mutagenesis of lysine residues wasused to identify regulatory ATM acetylation sites. The resultsidentify a single acetylation site at lysine 3016, which islocated in the highly conserved C-terminal FATC domain adjacentto the kinase domain. Antibodies specific for acetyl-lysine3016 demonstrate rapid (within 5 min) in vivo acetylation ofATM following exposure to bleomycin. Furthermore, lysine 3016of ATM is a substrate in vitro for the Tip60 histone acetyltransferase.Mutation of lysine 3016 does not affect unstimulated ATM kinaseactivity but does abolish upregulation of ATM's kinase activityby DNA damage, inhibits the conversion of inactive ATM dimersto active ATM monomers, and prevents the ATM-dependent phosphorylationof the p53 and chk2 proteins. These results are consistent witha model in which acetylation of lysine 3016 in the FATC domainof ATM activates the kinase activity of ATM. The acetylationof ATM on lysine 3016 by Tip60 is therefore a key step linkingthe detection of DNA damage and the activation of ATM kinaseactivity.

INTRODUCTION

Top

INTRODUCTION

Ataxia telangiectasia (A-T) is an inherited disease characterizedby immune deficiencies, neurodegeneration, susceptibility tocancer, and sensitivity to ionizing radiation (28). The A-Tgene product, the ATM protein kinase, is activated in responseto DNA double-strand breaks (DSBs) (24, 35) and phosphorylatesmultiple DNA damage response proteins, including Nbs1, p53,chk2, and SMC1 (reviewed in reference 24). The phosphorylationof these proteins by ATM is essential for correct activationof cell cycle checkpoints and for the initiation of DNA repair.Consequently, cells lacking functional ATM protein exhibit defectsin DNA repair and a loss of cell cycle checkpoints (24, 28,35), resulting in increased sensitivity to ionizing radiation.

A central question in studying the ATM protein has been to determinethe mechanism by which ATM's kinase activity is activated byDNA damage. Activation of ATM's kinase activity is associatedwith increased autophosphorylation of ATM at multiple sites(23), including serine 1981 (2). This autophosphorylation ofATM is proposed to initiate a dimer-monomer transition and therelease of active ATM monomers (2), although the exact contributionof ATM autophosphorylation to ATM activation is still underdebate (31). Several additional molecular events, includingthe Mre11-Rad50-Nbs1 (MRN) DNA binding complex (8, 11, 14, 41)and changes in chromatin structure (2, 4), also contribute toactivation of ATM's kinase activity. Mutations in the individualprotein components of the MRN complex can reduce or abolishthe activation of ATM's kinase activity by DNA damage (8, 11,14, 41). The recent characterization of a conserved C-terminaldomain in Nbs1 which is essential for recruitment of ATM tothe MRN complex and for the efficient activation of ATM's kinaseactivity (14, 42) further supports a key role for MRN in ATMactivation. Finally, biochemical studies have shown that MRNcan activate ATM's kinase activity in vitro (25, 26), suggestingthat inactive ATM is recruited to DSBs by MRN prior to MRN-dependentactivation of ATM's kinase activity. However, other studiesindicate that disruption of chromatin structure in the absenceof DNA damage can activate ATM's kinase activity (2) and thatkinase-inactive ATM is not recruited to enzymatically inducedDSBs (4), suggesting that ATM is activated in the nucleoplasmby DNA damage-induced changes in chromatin structure. In thiscase, the MRN complex may function as an adaptor complex whichmaintains ATM in an active form at the DSB (2, 4, 22). However,the mechanism by which interactions between ATM and MRN or structuralchanges in chromatin activate ATM's kinase activity is not known.

Previously, we demonstrated that the Tip60 histone acetyltransferase(HAT) is essential for the activation of ATM's kinase activity(19, 37). The HAT activity of Tip60 is increased by DNA damage,and the loss of Tip60 blocks the autophosphorylation and subsequentactivation of ATM's kinase activity (37). However, althoughactivation of Tip60's HAT activity is associated with an increasein the overall acetylation level of ATM, neither the acetylationsite nor the functional significance of ATM acetylation is known.We now demonstrate that Tip60 directly acetylates ATM on lysine3016 and that this acetylation is required for activation ofATM's kinase activity. The DNA damage-dependent acetylationof ATM on lysine 3016 by Tip60 is therefore the critical stepwhich links the detection of DSBs to the activation of ATM kinaseactivity.

MATERIALS AND METHODS

Top

MATERIALS AND METHODS

Cells. 293T, HeLa (American Type Culture Collection, VA), and GM5849A-T cells (Coriel Institute, NJ) were cultured according tothe suppliers' recommendations. Transfection of cells (usingFugene-6 [Roche, IN]), establishment of stable cell lines byG418 selection, and clonogenic cell survival assays were doneas previously described (15, 37). Small interfering RNAs (siRNAs)T3 (GGAAGCUGCUGAUCGAGUUUU) and T4 (GACGUAAGAACAAGAGUUAUU) andgreen fluorescent protein (GFP; Dharmacon, CO) were transfectedinto cells by using Lipofectamine 2000 (Invitrogen, CA) as previouslydescribed (37).

Mutagenesis. Two overlapping fragments at the C terminus of ATM were subclonedfrom pcDNA3.1-ATM into pBluescript, creating pBS-SPBA (containinga 3,582-base-pair internal SpeI-BamHI fragment) and pBS-XBXH(containing an approximately 1,800-base-pair XbaI-XhoI fragmentcarrying the entire 3' end of the cDNA, with the XhoI site locatedin pcDNA3.1). Mutagenesis was carried out using QuikChange IIXL site-directed mutagenesis kits (Stratagene, CA) and the primerslisted in Table S1 in the supplemental material. Following sequencing,mutations were released from pBS-SPBA and pBS-XBXH by digestionwith SpeI-BlpI and BlpI-XhoI, respectively. The correspondingfragments were then excised from pcDNA3.1-ATM, and the mutatedversions were ligated into the plasmid to create full-lengthATM constructs.

Immunoprecipitation, Western blot analysis, HAT assay, and kinase assay. General protocols for lysis buffers, immunoprecipitation, andWestern blot analysis were done as described previously (19,37). ATM immunoprecipitation was routinely performed using ATMantibody PC116 (EMD Biosciences, CA). Cells were lysed in ATMlysis buffer (20 mM HEPES, pH 7.4, 150 mM NaCl, 0.2% Tween 20,1.5 mM MgCl₂, 1 mM EGTA, 2 mM dithiothreitol, 50 mM NaF, 500µM NaVO₄, 1 mM phenylmethylsulfonyl fluoride, 0.1 µg/mlaprotinin, 0.1 µg/ml leupeptin) and cleared by centrifugation.For kinase assays, ATM was immunoprecipitated and washed sequentiallyin ATM lysis buffer followed by ATM kinase buffer (10 mM HEPES,pH 7.4; 10 mM MgCl₂; 50 mM NaCl; 10 mM MnCl₂). Immunoprecipitateswere incubated in kinase buffer (50 µl) supplemented with50 µM ATP, p53 peptide (2 µg; EPPLSQEAFADLWKK),and 10 µCi [ ${gamma}$ -³²P]ATP for 15 min at 30°C (37). ForHAT assays, ATM or Tip60 was immunoprecipitated and washed sequentiallyin ATM lysis buffer followed by HAT assay buffer (50 mM Tris,pH 8; 10% glycerol; 0.1 mM EDTA; 1 mM dithiothreitol). Immunoprecipitateswere then incubated in HAT assay buffer (60 µl) supplementedwith acetyl-coenzyme A (100 µM) and either biotinylatedhistone H4 peptide, biotinylated atm^wt peptide (ERVLMRLQEKLKGVEEGT),or atm^KR peptide (ERVLMRLQEALKGVEEGT) (0.5 µg) for 30min at 30°C. An aliquot of the reaction mix was immobilizedon a streptavidin plate, and acetylation was detected usinga HAT enzyme-linked immunosorbent assay according to the manufacturer'sinstructions (Upstate Biotechnology, NY). For both kinase andHAT assays, each datum point represents the average for threeseparate immunoprecipitations, with each derived from approximately3 x 10⁷ cells and containing equivalent amounts of protein.Antibodies used were as follows: ATM antibodies PC116 (EMD Biosciences,CA), 5C2, 2C1 (both from Genetex, TX), and pS1981 ATM (RocklandImmunochemicals, PA) and antibodies to ß-actin, chk2,hemagglutinin (HA), myc, phospho-chk2 (all from Cell Signaling,MA), Tip60, acetylated lysine (AcLys; Upstate Biotech, NY),p53, and pS15-p53 (EMD Biosciences, CA). A rabbit polyclonalantibody (AcK6) was raised against peptide CRVLMRLQE[KAc]LKGand affinity purified using immobilized acetylated peptide followedby absorption against immobilized nonacetylated peptide. Theresulting antiserum was used at a 1:50 dilution for Westernblot analysis.

Immunofluorescence. Cells were seeded on LabTek II chamber slides (Nunc, NY). Cellswere fixed as previously described (15) and stained with primaryantibody ATM 5C2 (Genetex, TX) followed by immunoglobulin G(IgG)-fluorescein isothiocyanate (Santa Cruz, CA). Slides weremounted with Fluoromount-G (Southern Biotech, AL) and visualizedwith a Nikon Eclipse TE 2000 microscope.

RESULTS

Top

RESULTS

Previously, we demonstrated that the Tip60 HAT plays a key rolein the activation of ATM (19, 37) and that ATM is acetylatedin cells exposed to agents which induce DNA DSBs. Here we reporton studies designed to identify acetylation sites on the ATMprotein. ATM acetylation can be monitored using a pan-specificacetyl-lysine antibody, previously characterized by us (37),which specifically detects the Tip60-dependent acetylation ofATM. The specificity of the pan-specific acetyl-lysine antibodyis demonstrated in Fig. 1A and B. As shown in Fig. 1A, a previouslyvalidated siRNA was used to deplete cellular Tip60 protein levels(19, 37). As shown in Fig. 1B, cells exposed to the radiomimeticagent bleomycin displayed increased ATM acetylation as wellas autophosphorylation of serine 1981, which can be used tomonitor the intrinsic kinase activity of ATM (2). Depletionof Tip60 by siRNA reduced both the acetylation and autophosphorylationof ATM (Fig. 1B) and blocked the ability of ATM to phosphorylatedownstream effector proteins (37). The pan-specific acetyl-lysineantibody can therefore detect Tip60-dependent acetylation ofATM. This antibody was then used in a focused, site-directedlysine mutagenesis study designed to identify ATM acetylationsites. Lysines were selected for mutagenesis based on threecriteria. First, regulatory acetylations are predicted to occurin the conserved FAT/kinase domain/FATC region at the C terminusof ATM. Second, regulatory lysines should be conserved amongATMs from higher eukaryotes. Third, HATs frequently acetylatemultiple lysines located within a few amino acids of each other.Based on these criteria, 17 regions within the C terminus (aminoacids 1698 to 3056) containing closely spaced, conserved lysineresidues were identified (see Table S1 in the supplemental material).Seventeen full-length ATM cDNA constructs, each containing twoclosely spaced lysine-to-alanine mutations, were then constructedand expressed in ATM-negative A-T fibroblasts. Each constructwas then screened for DNA damage-induced acetylation by treatingthe cells with the radiomimetic agent bleomycin. Although mutationof several lysines in the FAT and kinase domains destabilizedthe ATM protein (see Table S2 in the supplemental material),only the ATM mutant ATMK17, containing the double mutation K3016/K3018A,lacked inducible acetylation (Fig. 1C). Lysines 3016 and 3018were then mutated individually to arginine to determine whichof these lysine residues was acetylated. Figure 1D demonstratesthat ATM^K3016R lacked both acetylation and autophosphorylationon serine 1981, whereas ATM^K3018R was acetylated and autophosphorylatedto the same extent as wild-type ATM. These results imply thatlysine 3016 of ATM is acetylated when cells are exposed to DNAdamage. Because mutations within the C termini of ATM and relatedproteins can destabilize ATM (see Table S2 in the supplementalmaterial) (3, 29), lysine 3016 was mutated to alanine (to abolishthe positive charge), to threonine or glutamic acid (to altercharge), or to the acetyl-lysine mimic glutamine. The mutantswere expressed at similar levels to that of wild-type ATM protein(Fig. 1E). However, none of the mutants, including that containingthe acetyl-lysine mimic glutamine, exhibited increased ATM acetylationor autophosphorylation in response to DNA damage. These resultsindicate that lysine 3016 is required for activation of ATM'skinase activity but does not influence the overall stabilityof the ATM protein. Furthermore, the acetyl-lysine mimic glutaminedoes not substitute for acetyl-lysine in ATM.

View larger version (56K):
[in this window]
[in a new window]

FIG. 1. Lysine 3016 of ATM is acetylated after DNA damage. (A) 293T cells were transfected with siRNA targeting GFP or Tip60. Extracts were immunoprecipitated with either IgG (lane 1), Tip60 antibody (lanes 2 and 3, upper panel), or ATM antibody (lanes 2 and 3, lower panel). ATM and Tip60 were detected by Western blotting (WB). (B) 293T cells were transiently transfected with siRNA targeting either GFP (siGFP) or Tip60 (siTip60). Forty-eight hours later, cells were exposed to solvent (–) or bleomycin (5 µM for 30 min). Cell extracts were immunoprecipitated with IgG (lane 1) or anti-ATM antibody PC116 (lanes 2 to 5), and ATM protein levels (ATM), ATM acetylation (AcLys), and autophosphorylation (pS1981) of ATM were measured by Western blot analysis. (C) ATM-deficient GM5849 A-T cells transfected with vector, ATM, or ATM mutant K2 (K1772/1773A), K6 (K2204/2207A), or K17 (K3016/3018A) were exposed to bleomycin (+; 5 µM) for 30 min. ATM was immunoprecipitated with ATM antibody PC116, and ATM and ATM acetylation (AcLys) were measured by Western blotting. (D and E) GM5849 A-T cells transfected with vector, ATM, or ATMs with the indicated point mutations were exposed to bleomycin (+; 5 µM) for 30 min. ATM was immunoprecipitated, and Western blot analysis was used to monitor ATM levels, ATM acetylation (AcLys), and ATM autophosphorylation on serine 1981 (pS1981). (F) Location and sequence comparison of ATM acetylation sites.

Lysine 3016 is located within the conserved FATC domain at theC terminus of ATM and is conserved between ATMs from highereukaryotes (Fig. 1F). The ATM acetylation site has sequencehomology to previously identified Tip60 acetylation sites inthe androgen receptor (ARKLK) (16) and the myc oncogene (SEKLA)(30). Overall, the data in Fig. 1 indicate that ATM is induciblyacetylated on lysine 3016 in response to DNA damage. However,we cannot exclude the possibility that there are additionalacetylation sites within the ATM protein which are not detectedby the pan-specific acetyl-lysine antibody used here. The ATM^K3016Aand ATM^K3016R mutants were chosen for further analysis and yieldedessentially identical results.

Previously, we demonstrated that the acetylation of ATM (detectedusing the pan-specific acetyl-lysine antibody used for Fig.1) was abolished when Tip60 was inactivated using either siRNA(Fig. 1B) (37) or dominant-negative Tip60 proteins (37). Figure1 demonstrates that this Tip60-dependent acetylation of ATMoccurs on lysine 3016. To determine if Tip60 could directlyacetylate lysine 3016 of ATM, a peptide corresponding to aminoacids 3007 to 3024 of ATM (atm^wt) and a control peptide in whichlysine 3016 was replaced with arginine (atm^KR) were synthesized.HA-Tip60 and HA-Tip60 with point mutations in the HAT domain(Tip60^HD) were immunopurified from HeLa cells as described previously(37, 38). As shown in Fig. 2A, Tip60 acetylated both the histoneH4 and atm^wt peptides, and the ability of Tip60 to acetylatethese peptides was increased following exposure to bleomycin(Fig. 2A). The higher levels of acetylation of the H4 peptidereflect the presence of three Tip60 acetylation sites in thehistone H4 peptide (21), in contrast to one in the atm^wt peptide.When lysine 3016 was replaced with arginine, the atm^KR peptidewas minimally acetylated by Tip60 and displayed no detectableincrease in HAT activity in response to bleomycin. Furthermore,when a Tip60 protein with inactivating mutations in the HATdomain (Tip60^HD) was used (18), minimal acetylation of eitherthe H4, atm^wt, or atm^KR peptide was seen (Fig. 2A). These resultsdemonstrate that Tip60 can directly acetylate a peptide containinglysine 3016 of ATM in vitro. In a further series of experiments,immunopurification of the ATM-Tip60 complex from cells alsodemonstrated that Tip60 acetylated lysine 3016 in vitro (seeFig. S1 in the supplemental material). Figure 2A therefore demonstratesthat Tip60 can directly acetylate peptides containing lysine3016 of ATM in vitro.

View larger version (45K):
[in this window]
[in a new window]

FIG. 2. Lysine 3016 of ATM is acetylated in vivo. (A) Tip60 was immunopurified from HeLa cells expressing either HA-Tip60 (Tip60) or HAT-inactive HA-Tip60 (Tip60^HD). Cell extracts were incubated with either HA antibody or IgG and washed as described in Materials and Methods to immunopurify Tip60. The ability of immunopurified Tip60 to acetylate peptides derived from either ATM or the N terminus of histone H4 was determined. Peptides used were as follows: amino acids 3007 to 3024 of ATM (atm^wt; ERVLMRLQEKLKGVEEGT); amino acids 3007 to 3024 of ATM, with lysine 3016 replaced with arginine (atm^KR; ERVLMRLQERLKGVEEGT); or amino acids 1 to 21 of histone H4 (SGRGKGGKGLGKGGAKRHRKV). All peptides contained a C-terminal biotin label. Cells were treated for 30 min with bleomycin (+; 5 µM), where indicated. Results are means ± standard deviations (SD) (n = 3). (B) GM5849 A-T cells transfected with vector, ATM, or ATM^K3016A were exposed to bleomycin (+; 5 µM) for the indicated times (minutes). ATM was immunoprecipitated, and Western blot analysis (WB) was used to monitor ATM levels, ATM autophosphorylation on serine 1981 (pS1981), and ATM acetylation, using the K3016Ac-specific antiserum AcK6. (C) 293T cells were irradiated (2 Gy) and allowed to recover for the indicated times (minutes). ATM was immunoprecipitated, and Western blot analysis was used to monitor ATM levels, ATM-pS1981, and ATM acetylation, using the K3016Ac-specific antiserum AcK6.

Before examining the in vivo acetylation of ATM, immunofluorescencestaining was used to confirm that ATM^K3016R was located in thenucleus of the cell (see Fig. S2 in the supplemental material).To monitor the in vivo acetylation of lysine 3016 of ATM, asequence-specific antibody was raised against acetylated lysine3016. As shown in Fig. 2B, this antibody (AcK6) detected rapid(within 5 min) DNA damage-dependent acetylation of ATM. No significantacetylation or autophosphorylation of ATM was detected in eitherATM-deficient A-T cells or A-T cells expressing ATM^K3016A forup to 2 h following exposure to bleomycin, demonstrating thatthe AcK6 antibody specifically recognizes ATM acetylated onlysine 3016. To examine the acetylation of endogenous ATM, 293Tcells were exposed to low doses of ionizing radiation. Figure2C demonstrates the rapid acetylation of lysine 3016 of ATM(detected with the sequence-specific AcK6 antibody) and correspondingautophosphorylation of serine 1981 following exposure to ionizingradiation. Lysine 3016 of ATM is therefore rapidly acetylatedin vivo in response to DNA damage, and this acetylation occurswith the same rapid kinetics as those for the previously describedautophosphorylation of ATM on serine 1981 (2).

Previously, we demonstrated that ATM and Tip60 interact andthat this interaction requires the FATC domain at the C terminusof ATM (19). Because lysine 3016 is located within the FATCdomain of ATM (Fig. 1F), we determined if the mutation of lysine3016 affected either ATM-Tip60 interactions or activation ofTip60's HAT activity by DNA damage. Immunoprecipitation studiesdemonstrate that mutation of lysine 3016 does not alter theinteraction between ATM and Tip60 (see Fig. S3A in the supplementalmaterial). As shown in Fig. 3A, ATM was immunoprecipitated fromATM-deficient A-T cells complemented with either vector, ATM,or ATM^K3016A, and the associated Tip60-dependent HAT activitywas measured. In the absence of ATM, minimal Tip60 HAT activitywas measured, and this was not increased by bleomycin (Fig.3A, vector). However, the HAT activities of Tip60 associatedwith both ATM and ATM^K3016A were increased to similar levelsin response to bleomycin treatment (Fig. 3A). Mutation of lysine3016 therefore does not affect either the activation of Tip60'sHAT activity or the interaction between ATM and Tip60.

View larger version (44K):
[in this window]
[in a new window]

FIG. 3. Mutation of lysine 3016 of ATM blocks activation of ATM's kinase activity. (A) GM5849 A-T cells transfected with vector, ATM, or ATM^K3016A were left untreated (–) or exposed to bleomycin (+; 5 µM) for 30 min, and ATM was immunoprecipitated. The HAT activity of the Tip60 associated with ATM was measured using histone H4 peptide as the substrate. Results are means ± SD (n = 3). (B) GM5849 A-T cells transfected with vector, ATM, ATM^K3016R, or ATM^K3018R were exposed to bleomycin (+; 5 µM) for 20 min. p53, phospho-p53 (pS15-p53), chk2, phospho-chk2 (pT68-chk2), and ß-actin were detected by Western blotting (WB) of whole-cell extracts. (C) GM5849 A-T cells transfected with vector, ATM, ATMkd, or ATM^K3016R were left untreated (–) or exposed to bleomycin (+; 5 µM) for 30 min, and ATM was immunoprecipitated. The intrinsic kinase activity of ATM was then measured using a p53 peptide (containing serine 15) as the substrate. Results are means ± SD (n = 3). (D) GM5849 A-T cells expressing ATM or ATM^K3016R were preincubated with OA (0.5 µM) or dimethyl sulfoxide (–) for 20 min and then irradiated (IR; 2 Gy), and cell extracts were prepared 40 min later. ATM was immunoprecipitated, and the levels of ATM, acetylated ATM (AcLys), and ATM autophosphorylated on serine 1981 (pS1981) were determined. In addition, whole-cell extracts were probed by Western blot analysis to determine the levels of p53 and p53 phosphorylated on serine 15 (pS15-p53).

Next, we examined if lysine 3016 was required to activate ATM'skinase activity by monitoring the ability of ATM^K3016R to phosphorylatep53 and chk2 in vivo. A potential problem in analyzing the ATM^K3016Rmutant is that although the introduced arginine maintains theoverall charge, it may introduce structural changes which affectATM function independently of the loss of the acetylation site.To address this point, the activity of ATM^K3016R was comparedwith that of ATM^K3018R, in which the adjacent lysine at aminoacid 3018 was mutated to arginine (described in Fig. 1D). A-Tcells expressing each construct were treated with bleomycin,and the ability of ATM, ATM^K3016R, and ATM^K3018R to phosphorylatep53 and chk2 was determined. Both wild-type ATM and ATM^K3018Rinduced the ATM-dependent phosphorylation of p53 and chk2 inresponse to DNA damage (Fig. 3B), whereas ATM^K3016R failed toelicit phosphorylation of either p53 or chk2, consistent withthe failure of ATM^K3016R to exhibit either autophosphorylationor acetylation (Fig. 1). Similar results were seen with ATM^K3016A(see Fig. S3B in the supplemental material). Mutation of lysine3016, but not of the adjacent residue lysine 3018, thereforeblocks acetylation of ATM and the subsequent phosphorylationof p53 and chk2 by ATM. Furthermore, the ability of ATM^K3018Rto function like wild-type ATM indicates that this region ofATM can tolerate lysine-to-arginine substitutions without affectingfunction, suggesting that the loss of activity of ATM^K3016Ris due to a loss of acetylation rather than to structural effects.

The inability of ATM^K3016R to phosphorylate p53 or chk2 suggestsa defect in the kinase activity of ATM^K3016R. The kinase activityof ATM was measured by immunoprecipitating ATM and then incubatingit with a substrate peptide containing the phosphorylation siteof the p53 protein. Immunoprecipitation with IgG or from cellsexpressing a kinase-inactive ATM (ATMkd) yielded minimal levelsof ATM kinase activity (Fig. 3C). ATM had significant basallevels of ATM kinase activity, which were increased threefoldfollowing exposure to bleomycin. In contrast, whereas ATM^K3016Rhad the same levels of basal kinase activity as ATM, the kinaseactivity of ATM^K3016R was not increased by bleomycin exposure(Fig. 3C). This implies that the function of acetylation isto activate ATM's kinase activity. To further explore the functionalproperties of ATM^K3016R, we took advantage of previous workdemonstrating that ATM and protein phosphatase 2A exist as acomplex in the cell (17). Inhibition of protein phosphatase2A by okadaic acid (OA) leads to rapid accumulation of ATM autophosphorylatedon serine 1981, a process requiring ATM's kinase activity. However,this form of ATM does not exhibit elevated kinase activity andis functionally inactive (17). Therefore, we examined if OAtreatment of ATM^K3016R could induce autophosphorylation of ATMon serine 1981. Figure 3D demonstrates that OA induced the autophosphorylationof both ATM and ATM^K3016R on serine 1981. However, only wild-typeATM showed inducible acetylation and autophosphorylation inresponse to ionizing radiation or combined DNA damage and OAtreatment. Mutation of lysine 3016 of ATM therefore does notaffect the ability of ATM to undergo autophosphorylation onserine 1981, indicating that substitution of arginine for lysineat amino acid 3016 does not affect the intrinsic kinase activityof ATM. Autophosphorylation of ATM is postulated to be a keystep in the production of active ATM (2, 23). This raises thepossibility that sequential exposure of ATM^K3016R-expressingcells to OA (to induce autophosphorylation of serine 1981) followedby ionizing radiation may bypass the need for acetylation, leadingto the generation of an active ATM molecule able to phosphorylatekey target proteins, including p53. A-T cells expressing wild-typeATM did not exhibit p53 phosphorylation following OA treatmentbut displayed robust p53 phosphorylation after either ionizingradiation or ionizing radiation plus OA (Fig. 3D), as previouslyshown (17). However, OA treatment of ATM^K3016R cells, eitheralone or in combination with ionizing radiation, did not generatean active ATM molecule able to phosphorylate p53 in vivo (Fig.3D). Figure 3D therefore clearly shows that ATM^K3016R retainsthe ability to undergo autophosphorylation on serine 1981 inresponse to OA exposure but does not undergo autophosphorylationin response to DNA damage signals.

Next, we determined if the failure to activate the kinase activityof ATM^K3016R was reflected in a loss of functional activity.First, the ability of ATM^K3016R to complement the increasedradiosensitivity of A-T cells was examined. As shown in Fig.4A, the expression of ATM or ATM^K3018R complemented the radiosensitivityof A-T cells, whereas expression of ATM^K3016R had no effect,consistent with the lack of activation of the kinase activityof ATM^K3016R and its inability to phosphorylate key target proteins.Autophosphorylation of ATM is proposed to initiate a dimer-monomertransition, releasing active ATM monomers (2, 23). To determinethe contribution of acetylation to the dimer-monomer transition,the ability of ATM, ATMkd, and ATM^K3016A to undergo dimer-monomerconversion was examined. A-T cells were stably transfected withequimolar ratios of HA-ATM and myc-ATM expression vectors, andthe dimer-monomer transition was monitored by examining dissociationof the HA-ATM/myc-ATM dimer by coimmunoprecipitation techniques.Figure 4B demonstrates the coexpression of HA-ATM and myc-ATMin A-T cells. In some experiments, a nonspecific band of 340kDa was detected for the ATM-negative A-T cells by the HA andmyc antibodies (Fig. 4B and C); however, this band was not detectedwith ATM antibodies (Fig. 1 and 2). Following exposure to bleomycin,wild-type ATM, but not ATMkd or ATM^K3016A, was autophosphorylatedon serine 1981. Similarly, wild-type ATM, but not ATM^K3016A,was acetylated. Interestingly, ATMkd, which does not possesskinase activity and does not autophosphorylate (40), was alsoacetylated. Immunoprecipitation with HA antibody brought downboth HA-ATM and myc-ATM in all unstimulated cells (Fig. 4C).Following exposure to bleomycin, wild-type HA-ATM no longerinteracted with myc-ATM, whereas both ATMkd and ATM^K3016A remainedin the dimeric form. Importantly, since ATMkd was still acetylatedon lysine 3016, this indicates that acetylation on its own isnot sufficient for initiating the dimer-monomer transition ofthe ATM protein, suggesting that additional events, such asautophosphorylation of ATM, may also be required (2, 23). TheAcK6 anti-acetyl K3016 antiserum was not competent for immunofluorescence,precluding its use for monitoring recruitment of the acetylatedATM to DSBs.

View larger version (41K):
[in this window]
[in a new window]

FIG. 4. Lysine 3016 of ATM is required for ATM to regulate radiosensitivity and dimer-monomer transition. (A) GM5849 A-T cells stably transfected with vector (•), ATM ( ${circ}$ ), ATM^K3016R ( ${triangleup}$ ), or ATM^K3018R ( ${blacksquare}$ ) were irradiated, and cell survival was measured using a clonogenic cell survival assay. Results are means ± SD (n = 3). (B) A-T cells were stably cotransfected with HA- and myc-tagged ATM constructs, such that both HA-ATM and myc-ATM were expressed in the same cell. Cell lines coexpressing HA- and myc-tagged ATM, HA- and myc-tagged ATMkd, or HA- and myc-tagged ATM^K3016A were exposed to bleomycin (5 µM for 30 min). Levels of HA-ATM, myc-ATM, pS1981, and acetylation of lysine K3016 (AcK6) were measured. (C) Cells were treated as described above and then immunoprecipitated with HA antibody. Interaction between HA-ATM and myc-ATM was then assessed by Western blot (WB) analysis to detect HA-ATM and coprecipitating myc-ATM.

DISCUSSION

Top

DISCUSSION

Previously, we established that Tip60 was essential for activationof ATM's kinase activity (19, 37, 38) and that DNA damage increasedthe overall level of ATM acetylation in a Tip60-dependent manner.The results reported here have identified lysine 3016 of ATMas the amino acid acetylated by Tip60 (19, 37) and provide evidencethat this acetylation of ATM is required for the activationof ATM's kinase activity. The acetylation of lysine 3016 ismaximal within 5 min and is temporally indistinguishable fromthe time course of autophosphorylation of serine 1981 of ATM(Fig. 2) (2, 37). However, a kinase-inactive ATM protein wasstill acetylated on lysine 3016, demonstrating that ATM acetylationis independent of ATM's kinase activity. This implies that ATMacetylation precedes ATM autophosphorylation and defines acetylationof lysine 3016 of ATM as the earliest posttranslational modificationof ATM identified to date.

The results indicate that mutation of lysine 3016 blocks theactivation of ATM's kinase activity and subsequent autophosphorylationand dimer-monomer transition. Furthermore, ATM^K3016R does notphosphorylate downstream target proteins or complement the increasedradiosensitivity of A-T cells. These results are consistentwith an essential role for acetylation in the activation ofATM kinase activity. A potentially confounding problem is thatmutation of lysine 3016, within the FATC domain of ATM, mayhave a negative impact on the correct function of the kinasedomain independent of the loss of acetylation. However, we carriedout several controls which suggest that this is not the case,as follows: (i) ATM^K3016R retains normal basal kinase activityand lacks only inducible activation; (ii) unlike many mutationsin the kinase domain identified in A-T patients, substitutinglysine 3016 for arginine does not destabilize ATM or alter itsintrinsic kinase activity; (iii) introducing a lysine-to-argininemutation at an adjacent lysine (amino acid 3018) does not affectATM function, indicating that the region of ATM containing lysine3016 can tolerate small changes in protein architecture; and(iv) ATM^K3016R can still undergo autophosphorylation on serine1981 following the addition of OA, indicating that mutationof lysine 3016 to arginine does not disrupt the critical intermolecularprotein-protein interactions within the ATM dimer which arerequired for the autophosphorylation of ATM by the kinase domain(2). Taken together, we interpret these results to support amodel in which the failure of ATM^K3016R to increase its kinaseactivity in response to DNA damage is due to a loss of acetylationof lysine 3016 rather than to nonspecific effects related tothe lysine-to-arginine substitution. Thus, the primary functionof acetylation of lysine 3016 is to upregulate ATM's kinaseactivity.

Insights into how acetylation may regulate the intrinsic kinaseactivity of ATM are provided by recent structural studies onATM and the structurally related DNA-PKcs protein. PIKK proteinfamily members, including ATM and DNA-PKcs, contain a conservedFAT domain/kinase domain/FATC domain structure (6). Low-resolutionstructural modeling of the DNA-PKcs and ATM proteins indicatesthat the FAT and FATC domains protrude from either side of thekinase domain (27, 33, 36). The FAT domain is weakly conservedbetween family members, whereas the kinase and FATC domainsare highly conserved (1). FATC domains are highly flexible ${alpha}$ -helical structures (10, 33) and are required for the kinaseactivities of the mTor (39), DNA-PKcs (3, 32), and ATM (19)proteins. Studies indicate that the interaction between DNA-PKand DNA induces conformational changes in the FATC and FAT domains(33, 36), and these conformational changes are predicted toinfluence the catalytic activity of DNA-PK and ATM (33). Acetylationof lysine 3016 within the flexible FATC domain of ATM may thereforeinfluence ATM kinase activity. This is further supported byreports that a tumor-derived mutation near the kinase domainof the ATM-like p110 ${alpha}$ phosphatidylinositol 3-kinase subunit resultsin constitutive kinase activity (20, 34). Sequence alignmentof ATM and p110 ${alpha}$ indicates that one of these p110 ${alpha}$ mutations (atamino acid 1054) occurs in the same relative position as lysine3016 of ATM. This implies that the C-terminal regions of p110 ${alpha}$ and ATM may be essential for regulating kinase activity. Basedon the results presented here and the results of structuralstudies (10, 33, 36), we propose that acetylation of lysine3016, which is located at the junction between the FATC andkinase domains, alters the conformation of the FATC domain.This altered conformation of the FATC domain could allow substrateproteins access to the kinase domain as well as positively regulatingthe intrinsic kinase activity of the kinase domain. Furtherstructural studies are required to address this point.

A key question is to address the mechanistic links between ATMacetylation and autophosphorylation. The autophosphorylationof serine 1981 of ATM was originally proposed to be essentialfor ATM activation and to play a key role in initiating thedimer-monomer transition and the release of active ATM monomers(2). However, recent biochemical studies have shown that invitro activation of ATM kinase activity can be achieved in theabsence of significant serine 1981 autophosphorylation (12,25). In addition, a mouse model expressing a mutation in themouse equivalent of serine 1981 of ATM (serine 1987) displaysnormal ATM function. These observations indicate that autophosphorylationof serine 1981 is dispensable for ATM function under some conditions(31). Recently, several additional ATM autophosphorylation siteswere identified (23). Mutation of these sites impaired the abilityof ATM to phosphorylate downstream targets and to correct theradiosensitive phenotype of A-T cells (23). Therefore, althoughsome studies demonstrate that autophosphorylation of serine1981 is dispensable for ATM activity (12, 25, 31, 42), the identificationof additional ATM autophosphorylation sites raises the possibilitythat, similar to the case for DNA-PKcs (5, 9), multiple ATMautophosphorylation sites contribute to ATM activation. Theresults presented here indicate that the main function of ATMacetylation is to activate ATM's kinase activity. In this model,acetylation of ATM precedes and is required for the activationof ATM's kinase activity. Once activated, ATM kinase activityinduces autophosphorylation of ATM at multiple sites, includingserine 1981 (2, 23), and subsequent conversion of ATM dimersto monomers. Thus, while ATM autophosphorylation and subsequentdimer-monomer transitions may be important for achieving fullATM activation, the results presented in this report indicatethat acetylation of ATM, rather than autophosphorylation, isthe key step in upregulating ATM's kinase activity.

The results reported here and in previous works from us (19,37) and other groups (13) demonstrate that Tip60 plays a keyrole in the acetylation and activation of ATM and indicate acentral role for Tip60 in the early events leading to activationof ATM's kinase activity. However, other significant contributorsto ATM activation include both the MRN complex (8, 11, 22, 25,41) and changes in chromatin structure (2). A wide range ofexperimental approaches have established that the MRN complexplays a key role in several aspects of ATM activation, includingthe recruitment of ATM to DSBs (14, 42) and modulation of theactivation of ATM's kinase activity by DSBs (8, 22, 25, 26,41). This raises the possibility that MRN may regulate ATM activationby influencing the HAT activity of Tip60, either through directinteractions between MRN and Tip60 or through positioning ofthe ATM-Tip60 complex at DSBs. Tip60 also contains a chromodomain,a protein domain with the potential to interact with methylatedlysine residues on histones (7). Changes in chromatin structurefollowing ionizing radiation (either localized to the DSB orgenerated over large chromatin domains) may expose methylatedhistones as potential binding sites for Tip60's chromodomain.Such interactions between Tip60 and altered chromatin structuresadjacent to DSBs may be essential for upregulating Tip60's HATactivity and for the acetylation and activation of the ATM protein.Tip60 may therefore provide the key step which links previousobservations on the relative contributions of the MRN complex,chromatin structural changes, and ATM acetylation with the activationof the ATM protein kinase by DNA damage.

ACKNOWLEDGMENTS

This work was supported by grants to B.D.P. from the NationalCancer Institute (CA93602 and CA64585). Y.S. and K.R. were supportedby National Institutes of Health training grant T32 CA09078,and Y.X. was supported by NIAID center grant U19A1067751.

We thank Susan Lees-Miller and Sheng-Chung Lee for helpful discussions.

FOOTNOTES

* Corresponding author. Mailing address: Dana-Farber Cancer Institute, JF516, 44 Binney St., Boston, MA 02115. Phone: (617) 632-4946. Fax: (617) 632-4599. E-mail: brendan_price{at}dfci.harvard.edu

Published ahead of print on 8 October 2007.

Supplemental material for this article may be found at http://mcb.asm.org/.

REFERENCES

Top