The current study identifies ATF2 as a substrate for ATM and reveals its role as a participant in the DNA damage response. Concomitant to IR induced activation of ATM, ATM phosphorylates ATF2, resulting in its rapid recruitment into IRIF and pointing to its possible role as a sensor/adaptor in very early stages of the DNA damage response. Further, our data show that ATM-phosphorylated ATF2 is contributing to the recruitment of Mre11 and Nbs1 to IRIF, a finding that points to its role in coordinating the DNA damage response. ATF2 also affects the S phase checkpoint in response to IR and affects radiosensitivity, similarly to what is seen in cells that harbor mutant NBS1, MRE11, or A-T genes. Our data also demonstrate that ATF2 is important for activation of ATM as well as for concomitant activation of Chk1 and Chk2. Further studies will delineate the mechanism underlying ATF2 activation of ATM, which is likely to contribute to maintenance of active ATM at IRIF. Lastly, ATF2's role in the DNA damage response is distinct from its transcriptional activities, an observation that underscores the significance of our findings and establishes a paradigm for uncoupling transcription from DNA damage control.
Of interest is to address how ATF2 could contribute to ATM activation. Two of the possibilities currently considered relate to ATF2 being part of an upstream signal for ATM or for ATF2 ability to be part of mechanism that serves to maintain active ATM at IRIF. The link between ATF2 and chromatin organization is supported by independent studies from yeast and from mammalian systems. ATF2 associates with TIP49b (Cho et al., 2001
), which is part of the TIP60 histone acetylase complex implicated in DNA repair and chromatin organization (Kanemaki et al., 1999
; Ikura et al., 2000
). Further, binding of TIP60 to phosphorylated H2AX was recently shown to be an important step in subsequent modifications that are part of the DNA damage response (Morrison et al., 2004
). Atf1 and pcr1 (pombe
homologs of ATF2) were shown to contribute to deacetylation of certain lysines on histone H3 and H4, a prerequisite for heterochromatin assembly (Jia et al., 2004
; Kim et al., 2004
). The latter changes are consistent with the notion that activation of ATM could be induced upon changes in chromatin organization (Bakkenist and Kastan, 2003
). At this point, we equally entertain the second possibility, which would position ATF2 as part of mechanism to maintain active ATM at the damaged sites. Preliminary results revealed that for its ability to activate ATM, ATF2 needs to be phosphorylated on residues 490 and 498, but not on aa 69 and 71 (data not shown). Thus, either initial increase in ATM activity would suffice to promote ATF2 contribution to DNA damage response, and in turn to maintain ATM active at the sites of DSB, or, other PIKK may mediate such activation, which would consequently promote activation of ATM. Regardless of the initial signal, such changes are expected to result in maintaining active ATM at the site of DSB.
Upon its phosphorylation by ATM, ATF2 is recruited to DSB repair foci as one of the immediate early events (3 min). While ATF2 colocalizes with components of the MRN complex and γ-H2AX, it also affects recruitment of Mre11 and Nbs1 into repair foci. Of interest is that ATF2 did not affect localization of 53BP1 into repair foci, suggesting that the latter are subject to a different regulation, somewhat similar to the parallel interacting pathways shown for 53BP1 and Mdc1/NFBD1 in ATM activation (Mochan et al., 2003
Importantly, the contribution of ATF2 to the DNA damage response is mediated by the transcriptionally inactive form of the protein. This conclusion is supported by the following observations: (1) transcriptionally inactive ATF2 (69/71 mutant) is recruited to IRIF as efficiently as wt ATF2; (2) transcriptionally inactive ATF2 mediates the IR-induced S phase checkpoint as well as wt ATF2; (3) MEFs of ATF2
mutant mice in which the ATF2 DNA binding domain and part of the dimerization domains were deleted exhibit physiologic localization of this transcriptionally dead form of ATF2 into IRIF; (4) p38/JNK inhibitors do not interrupt localization of ATF2 into IRIF; (5) ATF2 is found on promoters of RAD50 and ATM (4 hr after exposure to genotoxic stress) without altering their transcription (Hayakawa et al., 2004
); and lastly (6) ATF2 localization into repair foci does not require its primary transcriptional heterodimeric partner c-Jun, as evident from analysis in c-Jun−/−
cells. The latter is of further interest in light of the report that c-Jun is also localized into DSB repair foci (MacLaren et al., 2004
). Since c-Jun was not shown to be associated with, or phosphorylated by, ATM, its localization in IRIF may be ATF2 dependent.
In all, ATF2 appears to play distinct functions in the cell cycle and the DNA damage response. First is its contribution to the cell cycle under nonstressed conditions. Transcriptional activity of ATF2 is necessary for maintaining physiologic cell cycle control under normal nonstressed growth conditions, probably through its established effects on cyclin A and cyclin D (Beier et al., 1999
; Shimizu et al., 1998
). Second is the phosphorylation of ATF2 by ATM required for its recruitment to IRIF, as for its recruitment of MRN components and for its role in IR-induced checkpoint control. For this function, ATF2 does not require its transcriptional activities. Third is ATF2's contribution to ATM activation, which is likely to contribute to maintenance of active ATM at sites of DNA damage. These observations raise the possibility that the balance between transcriptionally active and nonactive ATF2 (i.e., phosphorylated on the 69 and 71 sites versus the aa 490 and 498 sites) may affect the cell's ability to respond to DNA damage by means of altered RDS and radioresistance. These findings offer insight into our understanding of the immediate DNA damage response and into the relationship between transcription and DNA damage control. In as much, the present study establishes a paradigm for a function of a transcription factor in the DNA damage response that is independent of its transcriptional activities.
Cultures of HeLa, 293, IMR90, GM00637, and A-T
(GM05849) cells were obtained from Coriell repository or ATCC and maintained according to the supplier's recommendations. NBS-ILB1
cells (Kraakman-van der Zwet et al., 1999
) were kindly obtained from T. Halazonetis; MeWo, LU1205, and WM793 melanoma cells were kindly provided by M. Herlyn; and the A-T
cells (AT22IJE-T) as well as those that were reconstituted for ATM
were obtained from Y. Shiloh (Ziv et al., 1997
mutant mice were generated by insertion of loxP
sites into genomic sequences flanking exons 8 and 9 of the ATF2
gene (encoding the whole DNA binding and most of the leucine zipper domain) and induction of recombination by transiently expressing Cre
recombinase in ES cells (W.B., unpublished data). Primary mouse embryonic fibroblasts (MEFs) were generated from E13.5 embryos derived from crossing heterozygous animals carrying a germline allele of the mutant ATF2
ATF2 wt and mutant forms on aa 69, 71, 490, and 498 (generated with the aid of a QuikChange site-directed mutagenesis kit, Stratagene) were cloned into BamHI and NotI sites within the mammalian expression vector pEF HA or bacterial expression vector pGEX-4T. Purification of GST-ATF2 was performed under standard conditions. ATM
constructs were previously described (Bakkenist and Kastan, 2003
In vitro kinase assays were performed using G361, 293T, or U2OS cells transfected with either FLAG-ATM wt or FLAG-ATM-KD (10 μg). Bacterially expressed and purified GST-ATF2 (full length or the spliced form containing aa 1–48 and aa 310–505) or GST p53 1–80 aa were incubated with immunopurified endogenous or exogenous ATM coupled to protein G beads in the presence of kinase buffer (30 μl) containing 10 μCi of [γ32P]ATP. The reaction mixtures were incubated at 30°C for 20 min before separation on SDS-PAGE, electroblotting, and analysis on a phosphorimager.
Phosphoantibodies to ATF2 Amino Acids 490 and 498
Antibodies to phosphorylated aa 490 and 498 on ATF2 antibody were produced in rabbits immunized with keyhole-limpet hemocyanin-conjugated phosphopeptide (TEPALpSQIVM and APSpSQSQPSG), derived from aa 485–494 and aa 495–502 of ATF2, respectively. The phosphospecific antibodies were affinity purified (Phosphosolutions).
Cells were plated on cover slips and irradiated, fixed at indicated time points, and processed as described (Maser et al., 1997
; Carney et al., 1998
). Antibodies used were monoclonal or polyclonal γ-H2AX (1:500), pSmc1, Smc1 (Upstate Biotechnology), Mre11, Nbs, Rad50 (1:500), ATM, 53BP1, Chk2 (Genetex Inc.), ATF2 (Santa Cruz), p-ATF2 (1:500; Phosphosolutions), ATM-pSer1981 (Rockland Immunochemicals), Mdc1/NFBD1 (Bethyl Labs) and pChk1-S317, pChk2-T68 (Cell Signaling).
For RNAi of ATF2
expression, the p-Super vector system that directs synthesis of siRNAs in mammalian cells was used (Brummelkamp et al., 2002
). The two targets chosen for RNAi of ATF2 were as follows: 5#UTR192
) and within the coding sequence 1207
) of the human ATF2
gene (National Center for Biotechnology Information [NCBI] accession number NM_001880).
Reconstitution experiments were carried out by transfection of ATF2 plasmids (1 μg) with lipofectamine in MeWo cells that were infected (48 hr earlier) with ATF2 RNAi.
Cell Cycle Checkpoint Analysis
Cells (293 or MeWo) were infected with RNAi and 72 hr later were labeled (24 hr) with 10 nCi/ml of [methyl
C]-Thymidine (Perkin Elmer) followed by incubation (8 hr) with nonradioactive medium. The cells were than transfected and 48 hr later were irradiated (0, 5, or 15 Gy) using a 137
Cs source (dose rate: 5.1 Gy/min). 45 min after irradiation cells were pulse-labeled (2.5 μCi/ml of [methyl
H]Thymidine; Perkin Elmer) for 15 min. Inhibition of DNA synthesis was measured as described (Lim et al., 2000
). The ratios obtained for 3
H (counts/min) over 14
C (counts/min) (3
C ratio) were corrected for channel crossover. The ratio of DNA synthesis after exposure to ionizing radiation was calculated as 3
C ratio in irradiated over the 3
C ratio in unirradiated cells.
For FACS analysis cells that were about 50% confluent were treated with thymidine (2 mM in DMEM and FBS) for 16 hr. Cells were washed with PBS three times and incubated for 8 hr in DMEM and FBS in the absence of excess thymidine. Cells were then treated again with thymidine (2 mM) in DMEM and FBS (16 hr) to result in cells that were arrested at the G1/S phase boundary of the cell cycle. Once synchronized in G1/S phase, cells were washed with PBS, thereby allowing their release into the cell cycle. Cells were than harvested at various time points after G1/S phase, which corresponded to G2 and M phase as determined by flow cytometry analysis.
U2OS, or early passage A-T cells were infected with control RNAi or ATF2 RNAi . 72 hr later cells were resuspended to reach same density (5 × 104 cells/ml), irradiated, and subsequently plated in triplicates within the indicated range of cell densities. Cells were monitored for colony formation for 12 days (18 days for A-T cells) at 37°C. Colonies (containing 50 or more cells) were stained with 2% crystal violet, 50% ethanol, and counted. Each point represents a triplicate sample in experiments reproduced twice.
Wt and mutant MEFs were treated with anisomycin (25 μg/ml for 30 min) and fixed in formaldehyde (1%). Whole cell extracts were sonicated and immunoprecipitated (IP) with ATF2 antibodies C19 (Santa Cruz Biotechnology) or with IgG control antibodies. Afterward, reverse crosslinking DNA was purified using a PCR Purification Kit (Qiagen) and amplified by PCR (35 cycles) using primers GCGAGGAACGCAGGACGCGCCGTG (5′) and GCCCTCGCGTTGGCAGGGAGCCCG (3′) for ATF3, CTCCTCTGCGCAGGCGCGTCCTC (5′) and GTAGAGCCCAGGAGCCGCGAGCTG (3′) for Cyclin A (CcnA) promoter sequences containing ATF binding sites, and GGGAAGCCCATCACCATCTTC (5′) and CACCAGTAGACTCCACGACATACTCA (3′) for the GAPDH control sequence. Control PCR reactions were carried out using whole cell extracts (WCE) as template DNA.