DNA Replication Stress-induced Phosphorylation of Cyclic AMP Response Element-binding Protein Mediated by ATM*

The DNA damage-response regulators ATM (ataxia-telangiectasia-mutated) and ATR (ATM-Rad3-related) are structurally and functionally related protein kinases that exhibit nearly identical substrate specificities in vitro. Current paradigms hold that the relative contributions of ATM and ATR to nuclear substrate phosphorylation are dictated by the type of initiating DNA lesion; ATM-dependent substrate phosphorylation is principally activated by DNA double strand breaks, whereas ATR-dependent substrate phosphorylation is induced by UV light and other forms of DNA replication stress. In this report, we employed the cyclic AMP-response element-binding (CREB) protein to provide evidence for substrate discrimination by ATM and ATR in cellulo. ATM and ATR phosphorylate CREB in vitro, and CREB is phosphorylated on Ser-121 in intact cells in response to ionizing radiation (IR), UV light, and hydroxyurea. The UV light- and hydroxyurea-induced phosphorylation of CREB was delayed in comparison to the canonical ATR substrate CHK1, suggesting potentially different mechanisms of phosphorylation. UV light-induced CREB phosphorylation temporally correlated with ATM autophosphorylation on Ser-1981, and an ATM-specific small interfering RNA suppressed CREB phosphorylation in response to this stimulus. UV light-induced CREB phosphorylation was absent in ATM-deficient cells, confirming that ATM is required for CREB phosphorylation in UV irradiation-damaged cells. Interestingly, RNA interference-mediated suppression of ATR partially inhibited CREB phosphorylation in response to UV light, which correlated with reduced phosphorylation of ATM on Ser-1981. These findings suggest that ATM is the major genotoxin-induced CREB kinase in mammalian cells and that ATR lies upstream of ATM in a UV light-induced signaling pathway.

The ATM and ATR genes encode structurally and functionally related protein kinases belonging to the highly conserved phosphoinositide 3-kinase-related kinase gene superfamily (1). ATM encodes a 3056-amino-acid nuclear kinase that is a critical regulator of cellular responses to ionizing radiation (IR) 2 and radiomimetic agents that induce DNA double strand breaks (DSBs) (2,3). Inactivating mutations in ATM cause ataxia-telangiectasia, a syndrome of radiation sensitivity, cancer susceptibility, and neurodegeneration (3). ATM-deficient cells are characteristically defective in the IR-induced G 1 /S and G 2 /M check-points and display a hallmark abnormality termed radioresistant DNA synthesis, which reflects the failure to down-regulate DNA replicon initiation in response to DNA damage (3). In addition to cell cycle checkpoint defects, ataxia-telangiectasia cells demonstrate telomere abnormalities, subtle DNA repair defects, and genomic instability, affirming the critical role of ATM as a central regulator of the DNA damage response (3). ATR, on the other hand, is an essential gene that executes a number of critical functions during S phase in addition to a G 2 /M checkpoint function (4 -8). ATR-deficient cells exhibit high levels of spontaneous and replication stress-induced DSBs that lead to loss of viability (8 -11). ATR and its orthologs in yeasts are required to suppress premature DNA replication origin firing (12)(13)(14) and mediate the S-M checkpoint, which delays mitosis in the presence of unreplicated DNA (15)(16)(17). In humans, hypomorphic mutations in ATR are associated with a rare congenital condition known as Seckel syndrome (18).
ATM and ATR possess highly overlapping substrate specificities and demonstrate almost exclusive preference for the phosphorylation of Ser or Thr residues followed by Gln. Known substrates for ATR and ATM include p53 and BRCA1 (19 -25), the checkpoint effector kinases CHK1 and CHK2 (17, 26 -31), and a host of other proteins that participate in DNA repair and cell cycle checkpoint regulation (1,3,32). In the examples of p53, BRCA1, and CHK1, ATM is required for substrate phosphorylation in response to IR-and other DSB-inducing agents, whereas ATR promotes phosphorylation in response to UV light and other stimuli that inhibit DNA replication fork progression (19,21,22,33). This has led to the generally accepted view that ATM and ATR phosphorylate a common pool of substrates but in response to distinct types of genetic lesions. However, there appear to be exceptions to this rule, as ATM has been implicated in the UV light-induced phosphorylation of replication protein A (RPA) and CHK2 (33,34). The basis for differential substrate phosphorylation by ATM and ATR in UV irradiationdamaged cells is unclear but may reflect differences in substrate accessibility in cellulo.
The differential induction of ATM and ATR signaling in response to IR and DNA replication stress reflects an apparently fundamental difference in their mechanisms of activation. Activation of ATM involves the trans-autophosphorylation of inactive ATM dimers on Ser-1981 and dissociation into catalytically active monomers (35). ATM substrate phosphorylation and autophosphorylation in response to DSBs require trimeric MRE11⅐RAD50⅐NBS1 complexes, which recruit ATM to DNA breaks through the carboxyl terminus of NBS1 (36 -41). Recent studies suggest that NBS1 is not required for ATM autophosphorylation in response to non-DSB-inducing stimuli, such as UV light, indicating that multiple upstream pathways control the ATM activation state in mammalian cells (42). The regulation of ATR is comparatively less well understood, and it is unclear whether autophosphorylation modulates ATR signaling potential. However, one established mode of ATR regulation appears to be through its targeting to stalled replication forks. ATR becomes enriched in nuclear foci following exposure of mammalian cells to DNA-damaging agents or DNA replication inhibi-tors (24). These foci contain DNA repair and checkpoint-signaling factors, including RPA-and the ATR-interacting protein, ATRIP (8,44). Hierarchical recruitment of RPA and ATRIP-ATR to regions of singlestranded DNA appears to be important for the activation of ATR toward its substrates in mammalian cells (44), although RPA-independent activation of ATR may also occur (45)(46)(47).
We previously showed that the cyclic AMP-response element-binding protein (CREB) is phosphorylated by ATM on Ser-121 in response to IR and oxidative stress (48). CREB is a transcription factor that plays key roles in cell proliferation, homeostasis, and survival (49). Identified CREB target genes include a large number of cell cycle checkpoint regulators and DNA repair factors (50,51). Here, we explore the mechanism of CREB phosphorylation by ATM and ATR in response to different forms of DNA damage. Although both kinases phosphorylate CREB in vitro, we provide evidence that ATM is the major CREB kinase in UV irradiation-damaged cells and that ATR functions upstream of ATM in a pathway leading to CREB phosphorylation. Our findings support the idea that, despite possessing nearly identical substrate specificities in vitro, ATM and ATR may exhibit distinct substrate preferences in vivo.

EXPERIMENTAL PROCEDURES
Cell Culture and Antisera-HEK293T cells and JM fibroblasts were maintained in Eagle's minimum essential medium containing 5% fetal calf serum. SV40 large T antigen-transformed ATM ϩ/ϩ and ATM Ϫ/Ϫ mouse embryo fibroblasts (MEFs) were kindly provided by Dr. John Petrini (Memorial Sloan-Kettering Cancer Center). C3ABR and AT1ABR lymphoblasts were a kind gift of Dr. Martin Lavin (Queensland Institute of Medical Research) and were maintained in RPMI FIGURE 1. Differential phosphorylation of p53 and CREB in response to UV light. A, ATR phosphorylates CREB in vitro. FLAG immunoprecipitates were prepared from HEK293T cells transfected with empty vector (V) or vectors encoding FLAG-tagged wild-type (WT) or kinase-dead (KD) ATR. The FLAG immunoprecipitates were incubated with glutathione S-transferase-p53 or His-CREB fusion protein substrates in the presence of [␥-32 P]ATP and the amount of 32 P incorporation visualized by autoradiography. B, dose dependences of p53 and CREB phosphorylation in response to UV light. HEK293T cells were exposed to the indicated fluxes of UV light, harvested 2 h later, and cell extracts analyzed by immunoblotting with antibodies recognizing total, as well as phosphorylated, forms of CREB (␣-CREB-pS121) and p53 (␣-p53-pS15). C, quantification of phospho-CREB and phospho-p53 immunoblotting results presented in B. The densitometry data are presented in arbitrary units and are normalized to total p53 and CREB levels present in each lane. medium containing 10% fetal calf serum and 10 mM HEPES (pH 7.20). UV irradiation was carried out using a UVP CL-1000 UV cross-linker with a peak emission of 254 nm. Antibody suppliers included: Gene-Tex (␣-ATM), Santa Cruz Biotechnology (␣-CHK1 (G-4)), Upstate Biotechnology (␣-tubulin, ␣-␥H2AX), Cell Signaling (␣-CREB), Novus Biologicals (␣-CREB-pS121), and R & D Systems (␣-ATM-pS1981, ␣-CHK1-pS317).
Transfections and Protein Analysis-Transfection-ready small interfering RNA (siRNA) duplexes were purchased from Dharmacon Research. siRNAs used in this study included ATR (5Ј-AAC-CCGCGUUGGCGUGGUUGA-3Ј) and a mixture of four distinct ATM siRNA duplexes (SmartPool, Dharmacon). A scrambled siRNA sequence was used as a negative control (siCONTROL 1, Dharmacon). 3 g of siRNA was used for each transfection using the calcium phosphate DNA precipitation procedure. Cells were harvested 48 h later and extracts prepared as described previously (24). 75 g of total protein was separated on 10% SDS-polyacrylamide gels and transferred to Immobilon polyvinylidene difluoride membranes (Millipore). Membranes were blocked in Tris-buffered saline containing 0.2% Tween 20 (TBS-T) and 5% dried milk and incubated overnight at 4°C with the indicated primary antibodies diluted in blocking solution. After washing, the blots were incubated with horseradish peroxidase-conjugated sheep antimouse or goat anti-rabbit secondary antibodies (The Jackson Laboratory) and developed using SuperSignal chemiluminescent substrate (Pierce). The His-CREB and glutathione S-transferase-p53 fusion protein substrates and ATR kinase assay conditions have been previously described (24,48).
Immunofluorescence Microscopy-For the CREB and ␥H2AX immunostaining, JM fibroblasts were treated with 10 gray IR or mock-irradiated. 1 hr post-treatment, soluble nucleoplasmic protein was extracted from the cells as described previously (52). The cells were then fixed in 4% paraformaldehyde at room temperature for 15 min, permeabilized in PBS containing 0.5% Triton X-100 (PBS-T) for 10 min, washed once with PBS, and blocked for 30 min in PBS containing 3% bovine serum albumin and 2% goat serum. The cells were then incubated overnight at 4°C with ␣-␥H2AX and ␣-CREB antibodies diluted in blocking solution, washed three times in PBS-T, and incubated with fluorescein isothiocyanate-conjugated goat anti-mouse and Cy3-conjugated goat antirabbit secondary IgG (Caltag) for 1 h at room temperature. The cells were then washed twice in PBS-T, once in PBS, and then mounted onto slides using ProLong (Molecular Probes). A Carl Zeiss Axiovert 200 fluorescence microscope was used to visualize the samples.

RESULTS AND DISCUSSION
Differential Phosphorylation of CREB, p53, and CHK1 in Response to DNA Replication Stress-We previously showed that HU and UV light induce the phosphorylation of CREB on Ser-121 and that this response is partially dependent on ATR (47). Consistent with a potentially direct role for ATR as a CREB kinase, wild-type ATR (but not kinase-inactive ATR) phosphorylated recombinant His-tagged CREB in vitro (Fig. 1A). A dose-response curve revealed that CREB was phosphorylated on Ser-121 in intact cells at moderate to high fluxes of UV light; phosphorylation was weakly detected at 10 J/m 2 and increased steadily up to a UV irradiation flux of 400 J/m 2 (Fig. 1B). The phosphorylation of the ATR substrate p53 on Ser-15 following UV irradiation damage was qualitatively different. Phosphorylation of p53 was observed using a UV irradiation flux of 5 J/m 2 , and further increases in UV irradiation dose only modestly enhanced p53 phosphorylation (Fig. 1B). The distinct dose dependences of CREB and p53 phosphorylation following UV light exposure provided an initial indication that the phosphorylation of these substrates might occur through distinct mechanisms.
To further explore the mechanism of CREB phosphorylation, we compared the time course of its phosphorylation to that of the canonical ATR substrate CHK1 in response to HU and UV light. HU-induced maximal phosphorylation of CHK1 on Ser-317 within 15-30 min ( Fig.  2A). In contrast, HU-induced phosphorylation of CREB was weakly detected at 60 min and reached its maximum at 4 h, the last time point examined (Fig. 2B). The UV light-induced phosphorylation of CREB was also temporally delayed in comparison to CHK1. Although UV light-induced CHK1 phosphorylation was maximal within 15-30 min, the phosphorylation of CREB was barely detectable at 30 min but increased steadily up to 4 h post-UV irradiation, at which time phospho-CREB levels were comparable with those observed in IR-treated cells (Fig. 2). In comparison to HU and UV light, IR-induced CREB phosphorylation was much more rapid, with maximal phosphorylation observed within 15-30 min (Fig. 2B). Because CHK1 phosphorylation in response to UV light is tightly linked to ATR activation (17,26,27), these results imply that the ATR activation status at early time points following UV irradiation exposure does not directly correlate with the level of CREB phosphorylation.
ATM Is Required for CREB Phosphorylation in Response to UV Light-The delayed kinetics of CREB phosphorylation in UV irradiation-damaged cells and the apparent lack of correlation between CREB phosphorylation and ATR activation at short time points after UV light exposure caused us to consider the potential role of ATM in this process. The kinetics of ATM autophosphorylation on Ser-1981 were similar to the kinetics of CREB phosphorylation in UV irradiation-damaged HEK293T cells, suggesting the two processes are linked (Fig. 3A). To directly test the contributions of ATM and ATR to UV light-induced CREB phosphorylation, we transfected HEK293T cells with ATM-or ATR-specific siRNAs and measured CREB phosphorylation 2 h after exposure to UV light. An ATM siRNA strongly suppressed CREB phosphorylation in response to UV light, whereas an ATR siRNA partially suppressed CREB phosphorylation (Fig.  3B). The converse result was obtained when measuring the UV light-induced phosphorylation of CHK1 on Ser-317 and p53 on Ser-15. In both cases, UV light-induced phosphorylation was strongly suppressed by ATR siRNA, but only weakly inhibited by transfection with ATM siRNA (Fig. 3,  B and C). These findings suggest that the UV light-induced phosphoryla- tion of CREB proceeds by an ATM-dependent mechanism that is distinct from that of the canonical ATR substrates p53 and CHK1.
The above results suggested that ATM is the major CREB kinase in UV irradiation-damaged cells. To further test this idea, we measured UV light-induced CREB Ser-121 phosphorylation in control (ATM ϩ/ϩ ) and ATM-deficient (ATM Ϫ/Ϫ ) MEFs. Whereas CREB Ser-121 phosphorylation was induced within 2 h after exposure of ATM ϩ/ϩ MEFs to UV irradiation, no detectable phosphorylation of this residue was observed in UV-irradiated ATM Ϫ/Ϫ MEFs (Fig. 3D). This finding is consistent with the RNAi data and strongly suggests that ATM is the predominant CREB Ser-121 kinase activated by UV light in mammalian cells.
UV Light-induced ATM Autophosphorylation Is Partially ATRdependent-Because the UV light-induced phosphorylation of CREB appeared to be partially ATR-dependent but was completely dependent upon ATM, we hypothesized that ATR may lie upstream of ATM acti-vation in response to UV irradiation damage. To explore this possibility, we transfected HEK293T cells with control or ATR siRNAs and then examined ATM phosphorylation at Ser-1981 after exposure to 25 or 50 J/m 2 of UV light. We found that lowered levels of ATR correlated with a partial inhibition of ATM Ser-1981 autophosphorylation in response to either dose of UV light (Fig. 4A). The basal autophosphorylation of ATM was also partially suppressed by the ATR siRNA, and similar results were obtained using a vector-based ATR RNAi strategy (data not shown).
The partial ATR dependence of ATM autophosphorylation suggested that ATR might directly phosphorylate ATM on Ser-1981 in cellulo. If so, ATR-dependent phosphorylation of Ser-1981 should be observed in the absence of ATM catalytic activity. To test this hypothesis, we measured UV light-and IR-induced ATM autophosphorylation in AT1ABR lymphoblasts, which produces a catalytically inactive mutant ATM protein lacking amino acids 2546 -2548 (ATM ⌬SRI). . ATM is required for CREB phosphorylation in response to HU and UV light. A, time courses of ATM activation and CREB phosphorylation following exposure to UV light. HEK293T cells were exposed to UV light (100 J/m 2 ) and harvested at the indicated times post-irradiation. Cell extracts were immunoblotted with ␣-ATM-pS1981, ␣-CREB-pS121, ␣-ATM, and ␣-CREB antibodies. B, relative contributions of ATM and ATR to UV light-induced substrate phosphorylation. HEK293T cells were transfected with scramble control (Co) siRNA or siRNAs specific for ATM or ATR. The cells were then exposed to UV light (25 J/m 2 , 2 h) or mock-treated. Cell extracts were prepared and immunoblotted with the indicated antibodies. C, densitometric quantification of the UV light-induced CHK1 and CREB phosphorylation data shown in B. The phosphorylation data were normalized for slight differences in CHK1 and CREB levels between samples. SCR, scrambled. D, defective phosphorylation of CREB in ATMdeficient MEFs. ATM ϩ/ϩ or ATM Ϫ/Ϫ MEFs were exposed to UV light (100 J/m 2 ) and harvested 1, 2, or 4 h later. Cell extracts were analyzed by SDS-PAGE and immunoblotting using ␣-CREB-pS121, ␣-CREB, and ␣-tubulin antibodies. Phospho-Ser-121-CREB, as well as a nonspecific protein that cross-reacts with the ␣-CREB-pS121 antibody, are denoted by arrows.
Despite being catalytically inactive, ATM ⌬SRI is targeted to the nucleus and associates with ATM effectors, such as p53, suggesting that this mutant retains a biologically active conformation (53). Although the mutant ATM was expressed at a lower level than wild-type ATM, no detectable phosphorylation of ATM ⌬SRI was observed following exposure to either IR or UV light, suggesting that ATM kinase activity is absolutely required for phosphorylation of Ser-1981 (Fig. 4B). This finding argues against a direct role for ATR as an ATM Ser-1981 kinase in mammalian cells and suggests that the effects of ATR on ATM activation following UV irradiation damage are indirect.
In this report, we have explored the mechanism of CREB phosphorylation by ATM/ATR kinases in response to genotoxic stress. Using a combination of RNAi and ATM-deficient cell lines, we showed that the phosphorylation of CREB in response to UV light is mediated by a non-canonical, ATM-dependent pathway. This finding challenges the widely held view that ATM and ATR phosphorylate the same substrates in response to DSBs and DNA replication stress, respectively. Although numerous ATM/ATR substrates, including p53, BRCA1, and RAD17, seem to obey these rules (19, 23-25, 54 -56), the delayed phosphorylation of CREB on Ser-121 in response to UV light and HU provided an initial clue that its phosphorylation mechanism might be different. Although ATR phosphorylates CREB in vitro (Fig. 1A), the finding that CREB is not substantially phosphorylated at short time points after HU or UV irradiation exposure, when ATR is fully activated toward CHK1, casts doubt on whether ATR is a relevant CREB kinase in cellulo. ATR RNAi partially suppressed CREB phosphorylation on Ser-121 in response to UV light and HU (Fig. 3B); however, this may reflect an indirect effect of ATR on ATM activation (see below). Thus, although we cannot entirely rule out that ATR contributes to the direct phosphorylation of CREB in DNA-damaged cells, our results implicate ATM as the major CREB Ser-121 kinase. CREB may therefore represent an example of substrate discrimination by ATM and ATR in cellulo.
The failure of ATR to phosphorylate CREB at short time points following UV irradiation or HU exposure may reflect different compartmentalization of these two proteins within the nucleus. ATR colocalizes with RPA to DNA damage-induced nuclear foci, which contain DNA repair factors, checkpoint regulators, and ATR substrates, such as BRCA1 (24,44,57). In contrast, the majority of cellular CREB may be constitutively associated with cognate promoter elements. Immunofluorescence experiments with ␣-CREB antibodies showed that CREB does not detectably target to ␥H2AX-positive foci in response to IR treatment (Fig. 5). The phospho-CREB antibodies used in this study are not suitable for immunofluorescence experiments, and we cannot rule out that CREB transiently associates with DNA damage foci prior to its phosphorylation by ATM and/or ATR. 3 However, because its activation mechanism involves the release of catalytically active monomers into the nucleoplasm (43), it seems more plausible that ATM phosphorylates CREB distal to sites of DNA damage. On the other hand, ATR activity toward substrates, such as p53, BRCA1, and CHK1, might be spatially restricted to sites of stalled DNA replication as has been proposed previously (24). This speculative model will require further testing.
The results presented here suggest that autophosphorylation of ATM in UV irradiation-damaged cells is partially dependent on ATR, and this likely explains the partial ATR dependence of CREB phosphorylation on Ser-121 (Fig. 3B). The mechanistic basis for the observed ATR dependence is unknown. In principle, ATR could directly phosphorylate ATM on Ser-1981 in vivo. However, the finding that the catalytically inactive ATM ⌬SRI mutant was devoid of Ser-1981 phosphorylation in UV irradiation-damaged cells argues against this possibility. Attempts to validate this finding using a catalytically inactive ATM point mutant were hampered by the low transfection efficiency of ATM Ϫ/Ϫ cells. 4 Nevertheless, the results using the ATM ⌬SRI mutant suggest that ATM mediates Ser-1981 phosphorylation in response to UV light and that the effects of ATR on the ATM activation process are indirect.
Although the mechanism of ATM activation by UV light is unclear, the relatively delayed kinetics of CREB phosphorylation in UV irradiation-damaged cells suggests that one or more intermediate steps are involved. A recent study demonstrated that the MRE11⅐RAD50⅐NBS1 complex is not required for ATM activation in response to UV irradiation (42), which implies that other DNA damage sensors may activate 3 R. Tibbetts, unpublished data. 4 G. Dodson and R. Tibbetts, unpublished data. . ATM autophosphorylation is partially dependent on ATR. A, HEK293T cells were transfected with control (Co) or ATR siRNA and then left untreated or exposed to UV light (25 or 50 J/m 2 ). Extracts were prepared 1 h later and analyzed by immunoblotting with ␣-ATR, ␣-ATM-pS1981, and ␣-ATM antibodies. B, autophosphorylation of wild-type ATM and kinase-inactive ATM in UV irradiation-or IR-damaged cells. Control lymphoblasts expressing wild-type ATM (WT) or AT1ABR lymphoblasts expressing catalytically inactive ATM ⌬SRI (AT) were exposed to HU (3 mM), UV light (50 J/m 2 ), or IR (10 gray) and harvested 1 h later. The extracts were then analyzed by immunoblotting with ␣-ATM-pS1981 and ␣-ATM antibodies FIGURE 5. Subnuclear localization patterns of CREB and ␥H2AX in untreated and IR-treated human fibroblasts. JM fibroblasts were exposed to 10 gray IR or left untreated (UT) and then stained with CREB and ␥H2AX-specific antibodies as described under "Experimental Procedures."