The DNA Damage Response Kinases DNA-dependent Protein Kinase (DNA-PK) and Ataxia Telangiectasia Mutated (ATM) Are Stimulated by Bulky Adduct-containing DNA

Abstract

A variety of environmental, carcinogenic, and chemotherapeutic agents form bulky lesions on DNA that activate DNA damage checkpoint signaling pathways in human cells. To identify the mechanisms by which bulky DNA adducts induce damage signaling, we developed an in vitro assay using mammalian cell nuclear extract and plasmid DNA containing bulky adducts formed by N-acetoxy-2-acetylaminofluorene or benzo(a)pyrene diol epoxide. Using this cell-free system together with a variety of pharmacological, genetic, and biochemical approaches, we identified the DNA damage response kinases DNA-dependent protein kinase (DNA-PK) and ataxia telangiectasia mutated (ATM) as bulky DNA damage-stimulated kinases that phosphorylate physiologically important residues on the checkpoint proteins p53, Chk1, and RPA. Consistent with these results, purified DNA-PK and ATM were directly stimulated by bulky adduct-containing DNA and preferentially associated with damaged DNA in vitro. Because the DNA damage response kinase ATM and Rad3-related (ATR) is also stimulated by bulky DNA adducts, we conclude that a common biochemical mechanism exists for activation of DNA-PK, ATM, and ATR by bulky adduct-containing DNA.

Introduction

Environmental carcinogens, such as UV light and benzo(a)pyrene, form bulky base adducts on DNA that interfere with many aspects of DNA metabolism, including transcription and DNA replication. To minimize the detrimental effects of DNA damage on genome stability and cell viability, signal transduction pathways termed DNA damage checkpoints delay or prevent cell cycle progression to allow time for DNA repair (1, 2). This DNA damage response (DDR)² is composed of a cascade of protein phosphorylation events that ultimately determines cell fate. Defects in the repair or response to bulky DNA adducts are associated with a number of human disorders, including neurodegeneration and cancer (1, 2).

Initiation of the DDR is governed by protein kinases and accessory factors that directly sense the DNA damage and then respond by phosphorylating and activating additional protein factors that control cell cycle progression, DNA repair, and apoptosis (1, 2). In humans, DDR signaling is initiated by the phosphoinositide 3-kinase-related protein kinases (PIKKs) ATR, ATM, and DNA-PK (1, 2).

Although many proteins become phosphorylated in cells exposed to genotoxic stressors (3), the DDR transducer and effector proteins p53, Chk1 (checkpoint kinase 1), and replication protein A (RPA) play well recognized roles in controlling DNA repair, cell cycle progression, and cell fate (2, 4–7). Upon DNA damage, all three proteins become phosphorylated on specific amino acid residues that regulate their DNA damage checkpoint functions. The phosphorylation status of these proteins can therefore be used as a common biochemical read-out for activation of the DDR.

Precisely how bulky DNA adducts are sensed by the cell to elicit DNA damage checkpoint responses is not completely understood. Evidence from many model systems, including yeast, Xenopus egg extracts, and mammalian cells, indicates that the stalling of RNA and DNA polymerases at bulky DNA base adducts during transcription and DNA replication serves as a potent inducer of the DNA damage checkpoint response (1, 2, 8, 9). Similarly, the repriming of DNA synthesis downstream of bulky DNA adducts, which leaves postreplication gaps throughout the genome, may also serve as a signal for checkpoint activation (10, 11). Furthermore, the actual removal of bulky DNA adducts by the process of nucleotide excision repair leaves short 24–32-single-stranded DNA (ssDNA) gaps that may also activate this response (12–17). Based on these data, current models propose that the generation of ssDNA, a common intermediate at sites of repair and transcription and replication stalling, serves as the trigger for DNA damage checkpoint activation. According to this model, this ssDNA rapidly becomes bound by the major eukaryotic ssDNA-binding protein RPA. Through a direct interaction of the ATR-interacting protein ATRIP (ATR-interacting protein) with the 70-kDa RPA1 subunit of RPA, ATR can be recruited to sites of DNA damage to initiate signaling responses to bulky DNA damage (18–20).

However, several other lines of evidence suggest that alternative mechanisms exist for initiating DNA damage checkpoint responses to bulky lesions, including through the direct recognition of bulky DNA adducts by either ATR alone or in combination with its kinase-activating protein partner TopBP1 (topoisomerase II-binding protein) (21–24). Similarly, direct protein-protein interactions of ATR with various nucleotide excision and mismatch repair factors may also control the DDR to different forms of DNA base damage (25–29). It therefore remains unclear whether additional pathways may control the cellular response to bulky DNA adducts.

To gain new insights into the mechanisms of DNA damage checkpoint activation by bulky DNA lesions, we designed a cell-free assay using mammalian cell nuclear extract, damaged DNA, and checkpoint target protein substrates. Importantly, this in vitro system induces robust checkpoint responses, including the phosphorylation of Chk1, p53, and RPA on residues essential for checkpoint responses in vivo. Furthermore, these responses occur independent of DNA repair, transcription, replication, and chromatin modification. Using this assay, we have discovered that bulky DNA adducts directly activate the protein kinases DNA-PK and ATM to phosphorylate multiple checkpoint target proteins.

EXPERIMENTAL PROCEDURES

Cell Lines

The Chinese hamster ovary (CHO) cell lines AA8 and V3 (gift from Dale Ramsden, University of North Carolina) were grown in α-minimum Eagle's medium supplemented with 10% (v/v) fetal bovine serum (FBS) and penicillin/streptomycin. M059K (ATCC CRL-2365) and M059J (ATC CRL-2366) cells were maintained in DMEM/F-12 containing 10% FBS and penicillin/streptomycin. The SV40-immortalized XP-A fibroblast cell line XP20S (30) was grown in DMEM plus 10% FBS and penicillin/streptomycin. Nuclear extracts were prepared from either adherent cells (150-mm plates) or suspension cells, essentially as described previously (31). Extracts were dialyzed overnight against Buffer D (25 mm HEPES-KOH, pH 7.9, 100 mm KCl, 10% (v/v) glycerol, 0.2 mm EDTA, 1 mm DTT). Nuclear extract concentrations ranged from 3 to 8 mg/ml using the Bio-Rad protein assay with bovine serum albumin (BSA) as a standard.

Protein Purification

Kinase-dead, hexahistidine-tagged Chk1 was expressed in Sf21 cells and purified using nickel-NTA-agarose, as described previously (21). RPA was expressed in BL21 (DE3) cells and purified as described previously (32, 33). GST-tagged, full-length p53 (Addgene plasmid 10852) (34) was expressed in BL21 (DE3) cells and purified with glutathione-Sepharose 4B (Amersham Biosciences) resin according to the manufacturer's recommendations. Purified DNA-PK was purchased from Promega (catalog no. V581A). FLAG-tagged ATM kinase was expressed in HEK293 cells and purified using anti-FLAG-agarose (Sigma), as described previously (35).

Chemicals and Antibodies

The DNA-PK inhibitor NU7026 and ATM inhibitor KU-55933 were purchased from Sigma and dissolved in DMSO. N-acetoxy-2-acetylaminofluorene (AAF) and benzo(a)pyrene diol epoxide (BPDE) were obtained from the NCI Chemical Carcinogen Repository (Midwest Research Institute, Kansas City, MO). Caffeine and cisplatin were purchased from Sigma.

The following primary antibodies were obtained from the indicated companies and used at the indicated dilution: from Cell Signaling Technology, phospho-Chk1-Ser³¹⁷ (catalog no. 2344, 1:10,000), phospho-Chk1-Ser³⁴⁵ (catalog no. 2348, 1:10,000), phospho-p53-Ser¹⁵ (catalog no. 9284, 1:10,000), and PARP (catalog no. 9542, 1:1000); from Bethyl Laboratories, RPA70 (catalog no. A300-241A, 1:2000) and phospho-RPA2-Ser³³ (catalog no. A300-246A, 1:10,000); from Santa Cruz Biotechnology, Inc., ATM (catalog no. 2C1, 1:1000), ATR (catalog no. N-19, 1:1000), XPB (catalog no. S-19, 1:2000), Chk1 (catalog no. G-4, 1:2000), GST (catalog no. B-14, 1:1000), and Ku80 (catalog no. H-300, 1:2000); from Neomarkers, DNA-PKcs (catalog no. MS-423, 1:2000) (this antibody was also used for immunodepletion); from Calbiochem, RPA2 (catalog no. NA18, 1:2000) and the anti-ATM antibody (catalog no. PC116) used for immunodepletion.

Depletion of DNA-PKcs, ATM, and ATR from Nuclear Extract

DNA-PKcs and ATM were immunodepleted from XP-A nuclear extract (30 μg) by using two sequential 3-h incubations (4 °C) with 10 μg of antibody bound to 30 μl of recombinant Protein A/G Plus-agarose (catalog no. E1010; Santa Cruz Biotechnology, Inc.). Mock depletions were carried out in a similar manner with either mouse or rabbit IgG (Santa Cruz Biotechnology, Inc.). ATR was depleted by incubating XP-A nuclear extract (30 μg) twice (3 h, 4 °C) with 150 μg of GST-tagged TopBP1 fragment (amino acids 978–1192) bound to glutathione-Sepharose 4B (Amersham Biosciences). Incubation of extract with free glutathione resin was used as a mock depletion.

Preparation of DNA Substrates

AAF- and BPDE-damaged pUC19 DNA (pUC19) were prepared as described previously (21, 22). Cisplatin-damaged DNA was prepared by treating pUC19 plasmid DNA with 10 or 100 μm cisplatin for 18 h at 37 °C in 10 mm Tris, pH 8.0. Linearized plasmid DNAs were prepared by digestion with the restriction endonuclease EcoRI (New England Biolabs). The nicking plasmid pBC-KS⁺.nick was created by ligating oligonucleotides containing the target sequence for the nicking endonuclease Nt.Bbv.CI into the XbaI and HindIII sites of pBC-KS⁺ (Stratagene). The sequences of these oligonucleotides are available upon request. Biotinylated damaged and undamaged DNA immobilized on Dynabeads M-280 Streptavidin (Invitrogen) was prepared as described previously (36).

Kinase Assays

Kinase reactions, unless otherwise stated, contained nuclear extract (300 ng of XP-A, M059K, or M059J nuclear extract; 5 μg of CHO AA8 or V3 nuclear extract), a checkpoint substrate (10 nm Chk1, 10 nm p53, or 30 nm RPA), and 200 ng of DNA (11.5 nm) or buffer (10 mm Tris, pH 8.5) in 10 μl of 32 mm HEPES-KOH, pH 7.9, 50 mm KCl, 3.2 mm MgCl₂, 2% (v/v) glycerol, 1 mm DTT, 1 mm ATP, 10 μg/ml creatine phosphokinase, and 5 mm phosphocreatine. Reactions were incubated at 37 °C for 12 min and then stopped by the addition of 10 μl of 2× SDS-PAGE sample buffer (100 mm Tris, pH 6.8, 10% (v/v) glycerol, 200 mm DTT, 2% (w/v) SDS, 0.01% (w/v) bromphenol blue). Samples were boiled and electrophoresed on SDS gels. After transfer to nitrocellulose, blots were probed overnight with the appropriate phospho-specific antibody. After secondary antibody incubation, chemiluminescent signals were visualized with Amersham Biosciences ECL Plus or ECL Advance Western blotting detection reagents (GE Healthcare). Blots were reprobed to detect total checkpoint substrate protein levels. Blots were scanned, and signals were quantified using ImageQuant version 5.0 software (GE Healthcare). The highest phosphorylation signal on each blot was set to 100%, and the levels of phosphorylation of other samples were expressed relative to this value. Graphed values are the average and S.D. values from 2–4 independent experiments. Kinase assays with purified DNA-PK and ATM were performed using identical reaction conditions but with the purified protein kinases at a final concentration of 0.15 nm.

DNA Binding Assays

The association of proteins with immobilized damaged and undamaged DNA was performed as described previously (36), except that proteins were incubated in 20 μl of kinase reaction buffer lacking ATP. After incubation of the indicated proteins with immobilized DNA (250 ng) for 20 min at room temperature, the DNA-beads were washed three times with wash buffer (25 mm HEPES-KOH, pH 7.9, 50 mm KCl, 10% (v/v) glycerol, 0.02% (v/v) Nonidet P-40) and then boiled in 1× SDS-PAGE sample buffer.

RESULTS

Bulky DNA Adducts Induce Phosphorylation of Multiple Checkpoint Substrate Proteins in a Cell-free System

To determine whether bulky DNA adducts can directly induce DNA damage checkpoint signals in the absence of nucleotide excision repair or stalling of DNA and RNA polymerases, we designed a simple, cell-free assay composed of mammalian cell nuclear extract, damaged plasmid DNA, and various checkpoint substrate proteins. Although reactions were supplemented with ATP and an ATP regeneration system, the reactions lacked ribonucleotides and deoxyribonucleotides required for RNA transcription and DNA replication, respectively. We initially used nuclear extract from a nucleotide excision repair-deficient XP-A cell line as a source of DNA damage checkpoint signaling factors to ensure that the response was independent of nucleotide excision repair. A highly sensitive, in vitro excision repair assay (37) confirmed that no detectable repair occurred under these reaction conditions (data not shown). The initial source of damaged DNA was circular plasmid DNA (pUC19) treated with the carcinogen AAF, which primarily forms adducts on the C8 position of guanines. The dose of AAF used was sufficient to yield ∼20–30 adducts/plasmid (38).

The read-out for the assay involved monitoring the phosphorylation status of various checkpoint substrate proteins with phospho-specific antibodies and immunoblotting. We initially used the tumor suppressor protein p53 as a substrate because its phosphorylation on Ser¹⁵ in cells treated with DNA-damaging agents stabilizes the protein and allows it to regulate transcription of specific target genes involved in DNA repair, cell cycle arrest, and apoptosis (2). As shown in Fig. 1A, p53 phosphorylation was stimulated ∼3-fold in reactions containing AAF-damaged (modified) DNA in comparison with reactions containing only undamaged (unmodified) plasmid DNA.

FIGURE 1.

The checkpoint target proteins p53, Chk1, and RPA are phosphorylated in vitro in response to bulky adduct-containing DNA. Kinase reactions were performed as described under “Experimental Procedures” and contained 300 ng of XP-A nuclear extract and 200 ng of umodified (UM) or AAF-modified (M) pUC19 plasmid DNA. Reactions were supplemented with one of the following substrates: GST-p53 (A), His-Chk1 (B), or RPA (C) and were stopped after 10 min. The phosphorylation status of each substrate was monitored using SDS-PAGE and immunoblotting with the indicated phospho-specific antibody (p53, Ser¹⁵; Chk1, Ser³⁴⁵; RPA, RPA2 Ser³³). All reactions were performed three times, and the phosphorylation signals for the reactions with AAF-modified DNA were assigned the arbitrary value of 100%. The graphs show the average and S.D. values (error bars) from three independent experiments. For each substrate, significantly more phosphorylation was induced in reactions containing AAF-modified DNA (paired t tests, p < 0.05). The band shown in the RPA2-P immunoblot represents the slower mobility form of RPA2 that is observed when probing for total RPA2.

We next tested Chk1 as a substrate because it is involved in several aspects of the cellular response to DNA damage (4, 39), including cell cycle checkpoints, transcription, DNA repair, and apoptosis. Phosphorylation of Chk1 on two residues (Ser³¹⁷ and Ser³⁴⁵) by upstream DNA damage response kinases is essential for its DNA damage checkpoint activities in vivo (40). As shown in Fig. 1B, Chk1 phosphorylation on Ser³⁴⁵ was stimulated 5-fold in reactions containing AAF-damaged DNA in comparison with reactions containing only undamaged DNA.

RPA, which functions in many DNA metabolic processes, becomes phosphorylated in genotoxin-treated cells on multiple residues within the N-terminal region of the 34-kDa RPA2 subunit of RPA (6, 7). We chose to examine Ser³³ phosphorylation because phosphorylation of this residue is important for inhibiting DNA replication in UV-treated human cells (5). As shown in Fig. 1C, phosphorylation of this residue was stimulated 3-fold in reactions with bulky DNA adducts relative to reactions containing only undamaged plasmid DNA. We also note the presence of a characteristic, slower migrating form of RPA2 observed in reactions containing AAF-treated DNA. Based on the examination of Chk1, p53, and RPA phosphorylation on physiologically important residues, we conclude that multiple checkpoint target proteins become phosphorylated in a bulky DNA adduct-stimulated manner in this cell-free system.

Characterization of the in Vitro Bulky DNA Damage Checkpoint Assay

We next optimized the in vitro checkpoint reaction to determine the factors and conditions important for the robust phosphorylation of checkpoint substrates. We chose to focus on Chk1 as a substrate in most of these experiments; however, key experiments were repeated with p53 and RPA as substrates, and similar results were observed. As shown in Fig. 2A, reactions containing AAF-damaged DNA and varying amounts of nuclear extract generated more Chk1 phosphorylation than reactions lacking DNA or containing only undamaged plasmid DNA (compare lanes 9–12 with lanes 1–8). Control reactions lacking either ATP, nuclear extract, or recombinant Chk1 failed to yield any detectable Chk1 phosphorylation (data not shown). Importantly, the degree of Chk1 phosphorylation correlated well with the amount of damaged DNA in the reaction, reaching ∼5-fold above that observed in reactions lacking DNA or containing only undamaged DNA (Fig. 2B). Moreover, as shown in Fig. 2C, phosphorylation of Chk1 occurred rapidly, peaking 10–15 min after the addition of damaged DNA to the reaction. We conclude from these results that checkpoint activation can be directly and rapidly induced by bulky adduct-containing DNA in a cell-free system by a mechanism independent of DNA repair, transcription stalling, or DNA replication stress.

FIGURE 2.

Development of a cell-free assay for studying bulky DNA damage signaling. A, kinase reactions containing either no DNA or 200 ng (11.5 nm) of unmodified (UM) or AAF-modified (M) circular pUC19 plasmid DNA were prepared with increasing amounts of nuclear extract (100 ng, 300 ng, 1 μg, and 3 μg) from XP-A cells. Reactions were stopped after 12 min, fractionated on SDS-PAGE, transferred to nitrocellulose, and then probed with the indicated antibodies. Significantly more Chk1 phosphorylation was observed in reactions containing modified DNA than in reactions lacking DNA or containing only undamaged DNA (reactions containing 0.3, 1, or 3 μg of nuclear extract; paired t tests, p < 0.05). B, unmodified or AAF-modified plasmid DNA was titrated into kinase reactions. Reactions contained 0, 0.35, 0.7, 1.4, 2.9, 5.7, or 11.5 nm DNA and were stopped after 12 min. Chk1 phosphorylation in reactions containing AAF-modified DNA was significantly different from in reactions containing unmodified DNA at DNA concentrations of 2.9, 5.7, and 11.5 nm (paired t tests, p < 0.05). C, kinase reactions containing 300 ng of XP-A nuclear extract and either no DNA or 200 ng of unmodified DNA were incubated for the indicated lengths of time. Significantly more Chk1 phosphorylation was observed at 3, 9, and 15 min in reactions containing modified DNA than in reactions lacking DNA or containing only unmodified DNA (paired t tests, p < 0.05). All graphs show the average and S.D. values (error bars) from 2–3 independent experiments.

Bulky DNA Adducts Induce Checkpoint Activation Independent of DNA Ends or Nicks

To determine whether other forms of DNA damage are capable of directly inducing checkpoint activation in this cell-free system, we treated circular pUC19 plasmid DNA with BPDE, an environmental carcinogen and metabolic derivative of the polyaromatic hydrocarbon benzo(a)pyrene. Like AAF, BPDE forms bulky base lesions on guanine residues. The treatment conditions we used resulted in ∼20 adducts/plasmid (41). The addition of BPDE-modified DNA to reactions containing nuclear extract and Chk1 induced ∼5-fold more phosphorylation of Chk1 than reactions lacking DNA or containing only undamaged DNA (Fig. 3A). Importantly, phosphorylation was observed on both Ser³¹⁷ and Ser³⁴⁵, two residues critical for Chk1 function in DNA damage checkpoints. We also examined plasmid DNA treated with the chemotherapeutic agent cisplatin. However, even at the highest concentration of cisplatin, we observed only a small, ∼2-fold increase in Chk1 phosphorylation relative to reactions containing undamaged DNA (Fig. 3B). Although cisplatin induces a variety of intra- and interstrand cross-links in DNA, the major cisplatin adduct 1,2-d(GpG) occurs between adjacent guanine residues and does not significantly distort the DNA duplex. The 1,2-d(GpG) adduct is a relatively poor substrate for nucleotide excision repair in comparison with AAF-guanine and 1,3-d(GTG) cisplatin adducts (42, 43). These results therefore indicate that only bulkier, helix-distorting adducts, such as AAF and BPDE, are capable of efficiently inducing checkpoint activation in this in vitro system.

FIGURE 3.

Characterization of damaged DNA templates in the cell-free assay. A, kinase reactions containing either no DNA or unmodified (UM) or BPDE-modified (M) circular plasmid DNA were performed as described. The graphed data for Chk1 phosphorylation was based on phosphorylation of Ser³⁴⁵, although similar results were observed for phosphorylation of Ser³¹⁷. Significantly more Chk1 phosphorylation was observed in reactions containing modified DNA than in reactions lacking DNA or containing only unmodified DNA (paired t tests, p < 0.01). A portion of the unmodified and BPDE-modified plasmid DNA was electrophoresed on an agarose gel and stained with ethidium bromide to demonstrate the slower mobility caused by BPDE modification. B, kinase reactions were prepared with either no DNA, unmodified DNA, AAF-modified DNA (AAF), or cisplatin-treated DNA (cisplatin; 1 or 10 μm). Only the 100 μm cisplatin and AAF treatments induced significantly more Chk1 phosphorylation than reactions lacking DNA or containing unmodified DNA (paired t tests, p < 0.05). An ethidium bromide-stained agarose gel is shown to indicate the aberrant migration of the cisplatin- and AAF-modified DNA. C, circular or EcoRI-linearized pUC19 DNA were left unmodified or were treated with AAF and then used in the in vitro kinase assay. Significant differences in Chk1 phosphorylation were only observed in reactions containing modified DNA in comparison with reactions lacking DNA or containing unmodified DNA (paired t tests; p < 0.05). D, a pBluescript plasmid derivative containing two target sites for the nicking endonuclease Nt.BbvCI was left in circular form (Circ), nicked with Nt.BbvCI (Nick), or linearized with EcoRI (Lin) before being treated or not with AAF. Significant differences in Chk1 phosphorylation were only observed in reactions containing modified DNA in comparison with reactions lacking DNA or containing unmodified DNA (paired t tests; p < 0.05). The plasmids were analyzed on an ethidium bromide-stained agarose gel to monitor mobility and then used in kinase assays. The graphs show the average and S.D. values (error bars) from 2–3 independent experiments.

Because double strand breaks in DNA are a potent inducer of DNA damage checkpoint responses in vivo (1, 2), we wanted to ensure that the checkpoint responses to bulky adduct-containing plasmid DNA were independent of DNA ends. We therefore prepared AAF-modified, EcoRI-linearized plasmid DNA and compared the Chk1 phosphorylation response to both circular and undamaged DNA. As shown in Fig. 3C, an identical, 4–5-fold stimulation of Chk1 phosphorylation was observed in reactions containing either linearized or circular AAF-treated DNA relative to undamaged DNA. To further verify these findings and to determine whether other forms of damage, such as DNA nicks, could induce Chk1 phosphorylation, we prepared AAF-treated DNA with a pBluescript derivative modified to allow nicking with the endonuclease Nt.BbvCI. As shown in Fig. 3D, with each form of DNA (circular, nicked, or linear), stimulation of Chk1 phosphorylation was only observed in reactions containing AAF-treated DNA. We conclude from these results that bulky DNA adducts formed by AAF and BPDE can directly induce checkpoint responses independent of DNA ends or nicks.

Identification of Bulky DNA Adduct-stimulated Kinases in Nuclear Extracts

The PIKKs ATR, ATM, and DNA-PKcs, are the major kinases activated by DNA damage in mammalian cells. To identify whether one or more PIKKs was responsible for the stimulation of Chk1 phosphorylation by bulky adduct-containing DNA in these reactions, we initially took a pharmacological approach to inhibit one or more of the PIKKs. We first used caffeine, which at high concentrations inhibits all three PIKKs (44, 45). As shown in Fig. 4 (compare lanes 1–3 with lanes 4–6), caffeine completely prevented damaged DNA-stimulated Chk1 phosphorylation.

FIGURE 4.

Chemical inhibitors indicate a role for DNA-PK in the stimulation of checkpoint substrate phosphorylation by bulky adduct-containing DNA. Kinase reactions were supplemented with the indicated chemical inhibitors and performed as described previously. Reactions lacked DNA (−) or were supplemented with either unmodified (UM) or AAF-modified (M) pUC19 DNA. For reactions containing modified DNA and supplemented with caffeine, NU7026, or KU-55933 (10 μm), Chk1 phosphorylation was significantly different compared with reactions with the solvent control (paired t test, p < 0.05). Error bars, S.D.

To determine which of the three PIKKs was responsible for the effect of damaged DNA on Chk1 phosphorylation, we then tested the DNA-PK-specific chemical inhibitor NU7026, which competes with ATP for binding to DNA-PK (46). As shown in Fig. 4 (lanes 7–9), this inhibitor nearly completely abolished Chk1 phosphorylation by damaged DNA, indicating a major role for DNA-PK in this phosphorylation event. We also used the ATM inhibitor KU-55933, which, although very selective for ATM (IC₅₀ = 12.9 nm), also inhibits DNA-PK at higher concentrations (IC₅₀ = 2.5 μm) (47). As shown in Fig. 4 (lanes 10–15), the highest does of KU-55933 (10 μm) led to a 70% reduction in Chk1 phosphorylation in comparison with reactions with DMSO as a control. The dominant role for DNA-PK in checkpoint substrate phosphorylation in these reactions was also observed when either p53 or RPA was used as substrate in the reaction (data not shown). We conclude from the results with these pharmacological agents that DNA-PK plays a major role in the stimulation of checkpoint substrate protein phosphorylation by bulky adduct-containing DNA in this system.

We next took a genetic approach to identify the bulky DNA adduct-stimulated kinase(s), by using nuclear extract from cells lacking either DNA-PKcs or ATM. In Fig. 5A, we show results of immunoblotting the nuclear extracts from XP-A cells, the glioblastoma cell lines M059K and M059J, and the CHO cell lines AA8 and V3. The M059K and M059J cell lines were derived from a single patient glioblastoma, but the M059J line was found to be radiosensitive (48). It was later found that the M059J cell line lacks DNA-PKcs (49) and, due to ATM gene truncation (50), has a significant reduction in ATM protein expression (51, 52). Defects in DNA-PKcs and ATM protein expression in these cell lines were corroborated in Fig. 5A. We then used these nuclear extracts to test for BPDE-damaged DNA-stimulated Chk1 phosphorylation. As shown in Fig. 5B, although Chk1 phosphorylation was induced by damaged DNA in reactions containing nuclear extract from M059K cells, the response was completely absent when nuclear extract from the M059J line was used.

FIGURE 5.

The use of extracts deficient in one or more PIKKs supports a role for DNA-PK and ATM in the stimulation of checkpoint substrate phosphorylation by bulky adduct-containing DNA. A, nuclear extract (10 μg) from the indicated cell lines was separated by SDS-PAGE, transferred to nitrocellulose, and then probed for the indicated proteins. B, kinase reactions containing 300 ng of M059J or M059K nuclear extract. Reactions lacked DNA (−) or were supplemented with either unmodified (UM) or AAF-modified (M) pUC19 DNA. The level of Chk1 phosphorylation in the M059K extract reactions containing AAF-modified DNA was significantly different from reactions lacking DNA or containing unmodified DNA (paired t test, p < 0.05). C, kinase reactions containing 5 μg of nuclear extract from either AA8 or V3 cells. A significant difference in Chk1 phosphorylation was observed when comparing the response to modified DNA between the AA8 and V3 extract-containing reactions (paired t test, p < 0.05). D, kinase reactions with V3 nuclear extract were supplemented with either DMSO or the ATM inhibitor KU-55933 (10 μm). A statistically significant effect of modified DNA on Chk1 phosphorylation was only observed in reactions containing DMSO (paired t test, p < 0.05). Error bars, S.D.

To further examine whether DNA-PKcs and/or ATM are important for the response to damaged DNA, we compared the CHO cell lines AA8 (wild-type) and V3 (lacking DNA-PKcs). As shown in Fig. 5C, although reactions containing AA8 nuclear extract showed a stimulation of Chk1 phosphorylation in response to bulky DNA adducts, the response was inhibited by ∼60% in reactions containing V3 nuclear extract. To determine the kinase responsible for the remaining Chk1 phosphorylation in the V3 nuclear extract, we added the ATM inhibitor KU-55933 to reactions and found that the residual damaged DNA response was completely eliminated (Fig. 5D). We conclude from these results that both DNA-PKcs and ATM are involved in inducing Chk1 phosphorylation in response to bulky DNA adducts.

We then used biochemical methods to remove specific PIKKs from the XP-A nuclear extract to further confirm the identity of the bulky DNA damage-stimulated protein kinases. We first depleted ATR by incubating XP-A nuclear extract with a small GST-tagged fragment of the protein TopBP1, which is known to directly bind to ATR (53). As shown in Fig. 6A, this method specifically removed ATR but neither ATM or DNA-PKcs from the extract. We then used the mock- and ATR-depleted nuclear extract in kinase assays to monitor the phosphorylation status of Chk1. Identical results were obtained with both extracts, indicating that ATR was not responsible for inducing Chk1 phosphorylation in these reactions. We next specifically depleted ATM from XP-A nuclear extract by incubating the extract with either nonspecific anti-rabbit IgG or an anti-ATM antibody. As shown in Fig. 6B, removal of ATM from the extract was efficient and specific and led to an approximately 40% reduction in AAF-DNA-induced Chk1 phosphorylation. These results suggest a role for ATM in the response to AAF-damaged DNA. We then depleted DNA-PKcs from XP-A nuclear extract with antibodies against DNA-PKcs. Depletion of DNA-PKcs led to a complete loss of Chk1 phosphorylation (Fig. 6C). Importantly, identical results were obtained with BPDE-damaged DNA and with p53 or RPA as the substrate (data not shown). These results show that DNA-PKcs plays a major role in inducing Chk1 phosphorylation in response to bulky DNA damage and that ATM has a smaller role that is possibly dependent, in part, on DNA-PK activity.

FIGURE 6.

Depletion of DNA-PK and ATM from XP-A nuclear extract inhibits the stimulation of checkpoint substrate phosphorylation by bulky adduct-containing DNA. A, ATR was depleted from XP-A nuclear extract with a fragment of the protein TopBP1, as described under “Experimental Procedures.” Mock- and ATR-depleted nuclear extract were separated by SDS-PAGE, transferred to nitrocellulose, and then probed to detect the indicated proteins (top). Mock- and ATR-depleted nuclear extracts were used in kinase assays using standard reaction conditions (bottom). B, ATM was immunodepleted from XP-A nuclear extract using anti-ATM antibody. Mock depletion was performed with rabbit IgG. Mock- and ATM-depleted nuclear extracts were analyzed by Western blotting and probed for the indicated proteins (top). Mock- and ATM-depleted nuclear extracts were then used in kinase assays (bottom). The level of Chk1 phosphorylation in reactions with mock- and ATM-depleted extract upon the addition of modified DNA was significantly different (paired t test, p < 0.05). C, DNA-PKcs was immunodepleted from XP-A nuclear extract with antibodies against DNA-PKcs. Mock depletion was performed with mouse IgG. Nuclear extracts from the mock and DNA-PKcs depletions were analyzed by Western blotting and probed with antibodies against the indicated proteins (top). Mock- and DNA-PKcs-depleted extracts were then used in kinase assays (bottom). The levels of Chk1 phosphorylation in all three DNA-PKcs-depleted reactions were significantly different from those in the mock-depleted reactions (paired t test, p < 0.05). Error bars, S.D.

Purified DNA-PK and ATM Are Directly Stimulated by Bulky Adduct-containing DNA

Taken together, the pharmacological, genetic, and biochemical approaches utilized in this in vitro checkpoint assay indicated that DNA-PK and ATM may be directly stimulated by bulky DNA adducts to phosphorylate the checkpoint targets p53, Chk1, and RPA. To test this directly, we used purified DNA-PK that was free of any contaminating ATM or ATR (Fig. 7A). We then used this purified DNA-PK in kinase reactions as described above, where the purified DNA-PK was used in place of nuclear extract. Similar to reactions with nuclear extract, Chk1 phosphorylation by DNA-PK was stimulated 4–5-fold by AAF-damaged DNA in comparison with reactions containing undamaged DNA (Fig. 7B). Similar results were obtained when p53 was used as a substrate for DNA-PK (data not shown), indicating that the substrates are largely interchangeable in these reactions. This response may be independent of the Ku subunits of DNA-PK because the stimulation by AAF-modified DNA is significantly less than the canonical response to linearized DNA (supplemental Fig. 1).

FIGURE 7.

Purified DNA-PK is stimulated by bulky adduct-containing DNA. A, purified DNA-PK was fractionated by SDS-PAGE and then stained with silver (left) or analyzed by Western blotting (right) with antibodies against the indicated proteins. B, kinase assays were performed with purified DNA-PK and either no DNA (−), unmodified plasmid DNA (UM), or AAF-modified plasmid DNA (M). C, unmodified or AAF-modified pUC19 was immobilized on magnetic beads and then incubated in kinase reactions (20 μl) containing purified DNA-PK but lacking ATP. The relative amount of DNA-PKcs retained on the beads was quantified and normalized to the unmodified DNA sample. A portion of the bead-immobilized DNA was also electrophoresed on an ethidium bromide-stained agarose gel to serve as a loading control. Error bars, S.D.

To test whether the phosphorylation of checkpoint substrates by DNA-PK involved the specific recognition of bulky adduct-containing DNA by DNA-PK, we set up an in vitro DNA pull-down assay in which purified DNA-PK was incubated with either unmodified or AAF-treated plasmid DNA that had been linearized, biotinylated on both ends, and immobilized on streptavidin-coupled magnetic beads. As shown in Fig. 7C, ∼3-fold more DNA-PK was retained on the AAF-damaged DNA than on the undamaged DNA. Together, these results show that DNA-PK can directly recognize bulky DNA adducts to induce phosphorylation of checkpoint substrate proteins.

Because the pharmacological, genetic, and biochemical approaches also indicated a smaller role for ATM in inducing Chk1 phosphorylation, we purified a FLAG epitope-tagged version of ATM free of DNA-PKcs and ATR (Fig. 8A) and tested its ability to phosphorylate Chk1. As shown in Fig. 8B, Chk1 phosphorylation was stimulated ∼3-fold by AAF-damaged DNA when compared with reactions lacking DNA or containing undamaged plasmid DNA. As was observed for DNA-PK, ATM showed a slight binding preference for DNA containing bulky AAF adducts (Fig. 8C). We conclude that both ATM and DNA-PK can be directly stimulated by bulky adduct-containing DNA.

FIGURE 8.

Purified ATM is stimulated by bulky adduct-containing DNA. A, purified, FLAG-tagged ATM was fractionated by SDS-PAGE and then stained with silver (left) or analyzed by Western blotting (right) with antibodies against the indicated proteins. B, kinase assays were performed with purified FLAG-ATM and either no DNA (−), unmodified plasmid DNA (UM), or AAF-modified plasmid DNA (M). Significantly more Chk1 phosphorylation was observed in reactions containing modified DNA than in reactions lacking DNA or containing unmodified DNA (paired t test, p < 0.05). C, unmodified or AAF-modified pUC19 was immobilized on magnetic beads and then incubated in kinase reactions (20 μl) containing purified ATM but lacking ATP. The relative amount of ATM retained on the beads was quantified and normalized to the unmodified DNA sample. A portion of the bead-immobilized DNA was also electrophoresed on an ethidium bromide-stained agarose gel to serve as a loading control. Error bars, S.D.

DISCUSSION

Here we described a mammalian cell-free system for studying DNA damage checkpoint responses to bulky DNA adducts. Importantly, this system allowed an examination of DNA damage checkpoint activation in the absence of DNA repair, transcription stalling, and replication stress. Similarly, these nuclear extracts lack the histone variant H2AX (data not shown), which through phosphorylation and interaction with additional DDR proteins is involved in early DNA damage responses and in amplifying DNA damage response signaling (54). Thus, the checkpoint protein phosphorylation events we observed in our in vitro reactions occurred independent of the H2AX-dependent responses that take place in vivo. Ultimately, this cell-free system should be useful to examine how other protein factors, including H2AX and other DNA-binding proteins, impact the bulky DNA damage response. Similarly, other types of DNA damage and alternative DNA structures can also be tested with this system.

Using this in vitro assay, we have discovered a new mechanism by which bulky DNA damage checkpoint signaling may be initiated: the direct recognition of bulky DNA adducts by the protein kinases DNA-PK and ATM. In vivo, checkpoint protein phosphorylation induced by bulky DNA adducts is generally thought to only occur during the processing of bulky DNA adducts by nucleotide excision repair (12–17), by the stalling of RNA and DNA polymerases at bulky DNA base adducts (1, 2, 8, 9), or by the repriming of DNA replication downstream of bulky lesions (10, 11). However, due to dynamic regulation by other regulatory factors, including various protein phosphatases (55), cell signaling pathways and protein phosphorylation events may occur very rapidly and only transiently in vivo. Thus, the ability to observe the direct effects of bulky DNA adducts in cells is severely limited by the many regulatory systems and numerous DNA metabolic processes that occur simultaneously. The use of defined, cell-free or reconstituted systems therefore may allow for the identification of novel mechanisms regulating the cellular response to DNA damage.

Our previous reports that ATR can be activated by bulky DNA adducts in vitro in the absence of additional DNA metabolic processes (21, 22, 56) similarly indicate that there are alternative mechanisms for activating ATR in the absence of ssDNA and RPA. Our current findings that bulky DNA adducts similarly stimulate DNA-PK and ATM simply indicates that additional mechanisms for inducing checkpoint protein phosphorylation may exist in vivo. Moreover, the observation that our in vitro cell-free system results in protein phosphorylation in the absence of ATR indicates that DNA-PK and ATM may play a role in the direct recognition of bulky DNA adducts that are not recognized by DNA repair, transcription, or DNA replication in vivo. Consistent with this hypothesis, we found that inhibition of ATM and DNA-PK activity in XP-A cells leads to an approximate 30% reduction in AAF-induced Chk1 phosphorylation (supplemental Fig. 2).

Interestingly, DNA-PK and ATM are primarily thought to function in the response to a different physical form of DNA damage, the DNA double-strand break. However, as is the case for ATR activation, many studies on DNA-PK and ATM activation have focused on the protein-protein interactions that aid the recruitment of the kinases to sites of DNA damage. For example, just as ATR can be stably recruited to single-stranded DNA regions through interaction of ATRIP with RPA, DNA-PK and ATM are recruited to double strand breaks through interactions with the Ku80/70 and MRN (Mre11-Rad51-Nbs1) complexes, respectively (1, 2, 20). Although these interactions are undoubtedly important for robust responses to certain forms of DNA damage, it remains unclear how the physicochemical properties of these kinases and their direct interactions with damaged DNA may affect their catalytic activities.

The fact that DNA-PKcs, ATM, and ATR can all be stimulated by bulky DNA adducts to phosphorylate checkpoint substrates indicates that a common biochemical mechanism may be responsible for kinase activation by bulky DNA damage. Indeed, all three kinases have previously been shown to directly bind to double-stranded DNA independent of DNA ends or accessory proteins (24, 57–59). Moreover, examination of their primary and higher order structures indicates potentially important features that may control their kinase activities. DNA-PKcs, ATM, and ATR are all very large polypeptides (300–470 kDa), of which only 5–10% of the total sequence comprises the kinase domain. A significant fraction of each protein is composed of 40–55 repeats of the HEAT (huntingtin, elongation factor 4A, protein phosphatase 2A (PP2A) subunit, mTor) motif (60), which is a simple pair of anti-parallel α-helices linked by a short intraunit turn. Individual HEAT repeats are linked by short interunit turns, which can then stack to form higher order scaffolding structures. Recent cryoelectron microscopy and x-ray crystallography studies of DNA-PKcs have indeed confirmed that a large open ring cradle structure is formed in large part by these HEAT repeats (61, 62).

Although HEAT motifs often mediate direct protein-protein interactions, additional data suggest that they may also be utilized for protein association with DNA. Along these lines, we note that the Bacillus cereus DNA glycosylase AlkD, which is composed entirely of HEAT repeats, was recently shown to recognize and bind alkylated and mismatch-containing DNA via its HEAT repeats (63). Furthermore, recent molecular dynamic simulations of PR65, the HEAT repeat-containing subunit of PP2A, have indicated that HEAT repeat scaffolds may be flexible and responsive to force, such that changes in the HEAT repeat structure of PR65 may lead to alterations in PP2A substrate binding or catalysis (64). Thus, through their HEAT repeats and alterations in higher order structure upon binding bulky adduct-containing DNA, we speculate that the kinase activities of ATR, ATM, and DNA-PK may all be similarly stimulated to phosphorylate substrate proteins. Regardless of the mechanism by which DNA-PK, ATM, and ATR are stimulated by bulky adduct-containing DNA, our observations that all three kinases can be stimulated by similar DNA substrates indicate that these kinases may have partially redundant or overlapping roles in the cellular response to bulky DNA adducts.