Chromium (VI) Activates Ataxia Telangiectasia Mutated (ATM) Protein

The ataxia telangiectasia mutated (ATM) protein plays a central role in early stages of DNA double strand break (DSB) detection and controls cellular responses to this damage. Although hypersensitive to ionizing radiation-induced clonogenic lethality, ataxia telangiectasia cells are paradoxically deficient in their ability to undergo ionizing radiation-induced apoptosis. This contradiction illustrates the complexity of the central role of ATM in DNA damage response and the need for further understanding. Certain hexavalent chromium (Cr(VI)) compounds are implicated as occupational respiratory carcinogens at doses that are both genotoxic and cytotoxic. Cr(VI) induces a broad spectrum of DNA damage, but Cr(VI)-induced DSBs have not been reported. Here, we examined the role of ATM in the cellular response to Cr(VI) and found that Cr(VI) activates ATM. We also show that physiological targets of ATM, p53 Ser-15 and Chk2 Thr-68, were phosphorylated by Cr(VI) exposure in an ATM-dependent fashion. We found that ATM−/− cells were markedly resistant to Cr(VI)-induced apoptosis but considerably more sensitive to Cr(VI)-induced clonogenic lethality than wild type cells, indicating that resistance to Cr(VI)-induced apoptosis did not confer a selective survival advantage. However, analysis of long term growth arrest revealed a striking difference: ATM−/− cells were markedly less able to recover from Cr(VI)-induced growth arrest. This indicates that terminal growth arrest is the fate of these apoptosis-resistant cells. In summary, ATM is involved in cellular response to a complex genotoxin that may not directly induce DSBs. Our data suggest that ATM is a major signal initiator for genotoxin-induced apoptosis but, paradoxically, also contributes to maintenance of cell survival by facilitating recovery/escape from terminal growth arrest. The results also strongly suggest that terminal growth arrest is not merely an extended or even irreversible form of checkpoint arrest, but instead an independent and unique cell fate pathway.

Ataxia telangiectasia (AT) 1 is an autosomal recessive disorder characterized by progressive cerebellar degeneration, immunodeficiency, hypersensitivity to ionizing radiation (IR), and cancer predisposition (1). The gene responsible for this pleiotropic phenotype, ATM, is inactivated in the majority of AT patients. The clinical symptoms of AT are attributed to the cellular defects observed in cell lines derived from AT patients, which include chromosomal instability, telomere end-to-end fusion, as well as defects in cell cycle checkpoints (2,3), and delayed or attenuated activation of damage response pathways after exposure to IR (3). The product of the ATM gene is a 370-kDa serine/threonine kinase (4). The COOH-terminal domain of the ATM protein, which harbors the catalytic site of ATM, shows high similarity to the catalytic domain of phosphatidylinositol 3-kinases. This domain assigns ATM to a growing family of large proteins that contain the phosphatidylinositol 3-kinase-related domain (phosphatidylinositol 3-kinase-related kinases) (4). These proteins, which include ATM, ataxia telangiectasia mutation and Rad3-related kinase (ATR), and DNA-dependent protein kinase (DNA-PK), are involved, to different extents, in cell cycle progression, cellular responses to DNA damage, and maintenance of genomic integrity (5). The ATM protein has been shown to function specifically in multiple biochemical pathways linking the recognition and repair of DNA double strand breaks (DSBs) to downstream cellular processes, such as activation of cell cycle checkpoints, DNA repair, and apoptosis (6) by phosphorylating numerous substrates including p53 (7), Brca1 (8), Chk2 (9), c-Abl (10), and replication protein A (11). AT cells have a higher level of spontaneous apoptosis but appear to be deficient in their ability to undergo apoptosis in response to IR (12,13). The deficiency in apoptosis is not caused by a defect in the apoptotic machinery because these cells undergo apoptosis in response to FAS ligand (12,13). Interestingly, AT cells are hypersensitive to IR-induced clonogenic lethality (14). The contradiction between the hypersensitivity to clonogenic lethality and deficiency in apoptotic cell death in response to IR illustrates the complexity of the central role of ATM in the DNA damage response and the need for further understanding. ATM kinase has been shown to be specifically activated in response to DNA DSBs but not to UVB/UVC or base-damaging agents (7). However, recent reports have demonstrated the activation of ATM by t-butyl-hydroperoxide (15), CdCl 2 (16), UVA (17), heat shock (18), insulin (19), and MNNG (20). Therefore, the function of ATM in the cellular response to DNA damage other than DSBs also needs further investigation. Here, we studied the role of ATM in the cellular response to hexavalent chromium (Cr(VI)).
Cr (VI) has been shown to be a complex genotoxin and induce a broad spectrum of DNA damage. The spectrum of structural DNA damage includes DNA-DNA cross-links (21)(22)(23), DNA monoadducts (24,25), and DNA-protein cross-links (25). Cr(VI)-induced DNA single strand breaks appear when DNA is isolated under alkaline conditions (26). Despite the wide range of DNA damage induced by Cr(VI), no evidence has linked Cr(VI) to direct DSB formation; however, Cr(VI) is a potent inducer of recombination (27). We have shown that Cr(VI) exposure results in activation of p53 (28) and that Cr(VI)induced apoptosis is p53-dependent (29). However, the upstream signal transducers of p53 activation remain unclear. Furthermore, in a recent study, we showed that telomeraseimmortalized human cell populations exposed to Cr(VI) exhibit a different spectrum of responses, which include clonogenic survival, terminal growth arrest or apoptosis, depending on the extent of DNA damage. The regaining of replicative potential after genotoxic exposure was attributable to either escape from, or resistance to, terminal growth arrest or apoptosis (30).
The objectives of this study were to characterize the role of ATM in the cellular response to Cr(VI) as a model complex genotoxin. We showed that ATM is activated after Cr(VI) treatment. ATMϪ/Ϫ cells were resistant to Cr(VI)-induced apoptosis compared with wild type cells. However, ATMϪ/Ϫ cells were more sensitive to Cr(VI)-induced clonogenic lethality than wild type cells, suggesting that the resistance to Cr(VI)-induced apoptosis in ATMϪ/Ϫ cells did not confer a selective survival advantage. Although ATMϪ/Ϫ cells exhibited a similar sensitivity to Cr(VI)-induced growth arrest, they were markedly less able to recover from this growth arrest than ATMϩ/ϩ cells. These results indicate that the ATM gene is required for genotoxin-induced apoptosis and also contributes to maintenance of cell survival by facilitating recovery and/or escape from terminal growth arrest.

EXPERIMENTAL PROCEDURES
Cell Culture-Normal dermal fibroblasts GM03440, AT heterozygous fibroblasts GM03397, AT fibroblasts GM03395 with homozygous ATM gene deletion (the donor of this cell line is also the son of the donor of GM03397), and GM02052 with a homozygous point mutation for the 103C-T transition in exon 5 of the ATM gene (Coriell Institute, Camden, NJ) were maintained in minimum essential medium with Earle's salts (EMEM). The EMEM contained a 2ϫ concentration of essential and nonessential amino acids, vitamins, and 20% fetal bovine serum (Hyclone Laboratories, Inc., Logan, UT). Cells were incubated in a 95% air and 5% CO 2 humidified atmosphere at 37°C, and the medium was replaced every 48 h.
Treatment of Cells with Chromium-Sodium chromate (Na 2 CrO 4 . 4H 2 O) (J. T. Baker Chemical Company) was dissolved in deionized H 2 O and sterilized by passage through a 0.2-m filter before use. Cells were treated with a final concentration of 0, 0.1, 0.5, 1, 3, or 6 M sodium chromate for the indicated times in complete medium. The cells were either collected for analysis after this treatment or rinsed with phosphate-buffered saline (PBS) and incubated in complete medium for an additional 24 h before analysis.
ATM in Vitro Kinase Assay-12 ϫ 10 6 Cells were treated with and without 6 M sodium chromate for 3 h. Cells were then lysed by sonication in TGN buffer. After centrifugation at 13,000 ϫ g, 1.5 mg of extracted proteins was incubated with 30 l of mouse IgG-agarose (Sigma) for 1 h. ATM protein was then immunoprecipitated with anti-ATM polyclonal antibody and protein A-agarose (Calbiochem). The immunocomplexes were recovered by centrifugation at 13,000 ϫ g for 20 s and washed twice with TGN buffer, twice with TGN buffer plus 0. The proteins were electrophoretically separated by SDS-PAGE, transferred to polyvinylidene difluoride membranes, and exposed to x-ray film. p53 Western blots (Ab-6 from Oncogene Research Products, Boston, MA) were conducted on the same membranes after decay of 2 half-lives to confirm the equal amount of GST-p53 1-101 substrate used in the reaction. The relative quantity of immunoreactive and 32 P-labeled proteins was determined with a Personal Densitometer SI (PDSI) and ImageQuant software (Molecular Dynamics, Sunnyvale, CA).
Phosphatidylserine (PS) Translocation Assay-This assay measures PS translocation from the inner (cytoplasmic) leaflet of the plasma membrane to the outer (extracellular) leaflet in the early stages of apoptosis. PS on the outer leaflet is available for binding to fluorescently labeled annexin V (31). This assay was conducted as described previously (28). Briefly, cells were seeded at 10 5 cells/60-mm 2 dish and incubated for 24 h prior to sodium chromate exposure. After the sodium chromate treatment, cells were gently harvested by trypsinization and combined with nonadherent cells from the culture medium. The cells were centrifuged at 600 ϫ g for 5 min. Cell pellets were washed once with PBS and resuspended in 100 l of binding buffer (10 mM HEPES (pH 7.4), 150 mM NaCl, 5 mM KCl, 1 mM MgCl 2 , 1.8 mM CaCl 2 ) containing 2 l of annexin V-FLUOS (Roche Applied Science). Samples were incubated in the dark at room temperature for 15 min. 30 l was loaded on a microscope slide, and the percentage of annexin V-FLUOS-stained cells was determined using an Olympus AX70 microscope (Olympus, Lake Success, NY) with a fluorescent filter set suitable for FLUOS analysis (excitation 460 -490, emission 515).
Clonogenicity Analysis-This assay was conducted as described previously (30). Briefly, cells were seeded at 10 5 cells/T25 flask and treated with 0, 0.1, 0.5, and 1 M sodium chromate. After 24 h, cells were collected by trypsinization, counted, and reseeded at 2 ϫ 10 3 /60-mm 2 dish in triplicate. The cells were incubated for 12 days, then rinsed with PBS and incubated with crystal violet stain (80% methanol, 2% formaldehyde, 2.5 g/liter crystal violet) for 15 min at room temperature. The plates were rinsed thoroughly with distilled H 2 O and allowed to dry. Colonies with 20 or greater cells were counted, the triplicates were averaged, and clonogenicity was determined as percent of control.
Cell Growth Analysis-Cells were seeded at 10 5 cells/100-mm dish and treated with 0, 0.1, 0.5, and 1 M sodium chromate. A group of nine replicate dishes were seeded for each dose tested, and all of the replicates within the group received the same treatment. One replicate was taken at day 1-6, 8, 12, and 16 after Cr(VI) treatment and counted to determine total cell number at each dose and time. Cells were gently harvested with trypsin, centrifuged at 600 ϫ g for 5 min, and the cell pellets were resuspended in 1 ml of PBS. The total cell number was determined using a hemacytometer (Hausser Scientific, Horsham, PA). Data were normalized to the same day nontreated control. PS translocation assay was conducted at days 1, 8, 12, and 16 to determine the possible cell death induced by 1 M Cr(VI) treatment.
Cell Cycle Analysis Using Bromodeoxyuridine (BrdUrd) and Propidium Iodide (PI) Double Staining-Cells were seeded at 4 ϫ 10 5 cells/150-mm dishes and incubated with or without 1 M Cr(VI) for 24 h and then washed and maintained in complete medium. At the indicated time after Cr(VI) treatment, BrdUrd (Molecular Probes, Eugene, OR) was added to the medium to a final concentration of 10 M, and cells were incubated for an additional 1 h. Cells were then harvested and fixed in 70% ethanol. After fixation, cells were resuspended in 2 N HCl for 20 min and neutralized in 0.1 M sodium borate. Cells were then incubated with a 1:40 dilution of anti-BrdUrd antibody for 30 min. Alexa 488-conjugated anti-mouse IgG antibody was used as a secondary antibody (Molecular Probes). Cells were stained with 5 g/ml PI (Sigma) in PBS and analyzed using a FACSort flow cytometer (BD Biosciences, Palo Alto, CA). Excitation was 488 nm, and the emission filters used were 530 nm for Alexa 488 and 620 nm for PI. The PI fluorescence signal (fluorescence pulse area versus pulse width) was used to exclude doublets and aggregates from analysis. The percentage of cells in the S, G 0 /G 1 , and G 2 -M regions were determined with 20,000 cells.
Immunofluorescent Detection of Phosphorylated Histone H3-Cells were seeded at 4 ϫ 10 5 cells/150-mm dishes and incubated with or without 1 M Cr(VI) for 24 h and then washed and maintained in complete medium. The cells were harvested at various time points after treatment (1-7 days) and fixed in 70% ethanol at Ϫ20°C. The cells were resuspended in 0.25% Triton X-100 in PBS and incubated on ice for 15 min. The cells were pelleted by centrifugation and resuspended in PBS containing 1% bovine serum albumin with 0.25 g/ml of a polyclonal antibody that specific recognizes the phosphorylated form of histone H3 (Upstate Biotechnology, Lake Placid, NY) and incubated for 3 h at room temperature. The cells were then rinsed with PBS containing 1% bovine serum albumin and incubated with Alexa 488-conjugated goat antirabbit secondary antibody (Molecular Probes) diluted at a ratio of 1:100 in PBS containing 1% bovine serum albumin. The cells were then stained with PI and analyzed using a FACSort as described above for BrdUrd and PI. The percentage of cells containing phosphorylated histone H3, undergoing mitosis, was determined with 20,000 cells.
Chromium Uptake-Cells were seeded at 1 ϫ 10 6 cells/60-mm dishes and incubated for 24 h prior to sodium chromate exposure. One extra dish of each cell type was seeded for determining the final cell number. Sodium chromate was prepared as above and spiked with Na 2 51 CrO 4 (ICN, Irvine, CA). Cells were treated with a final concentration of 3 M sodium chromate for 5 h at 37°C. After sodium chromate treatment, cells were washed twice with PBS and lysed in 50 mM NaOH for 15 min. Cells were collected and combined with 5 ml of Ecolite scintillation mixture (ICN). Cell-associated radioactivity was determined on a Beckman LS3801 scintillation counter. Data were normalized to cell number.

Kinase Activity of ATM Was Enhanced after Cr(VI) Treatment-
The lack of ATM protein in the ATMϪ/Ϫ and mutant cell lines was confirmed by immunoblotting analysis. As expected, the ATMϩ/ϩ and ATMϩ/Ϫ cell lines contained similar levels of ATM protein (Fig. 1). It has been shown that ATM isolated from cells exposed to IR exhibits an increased ability to phosphorylate p53 in vitro on a single residue, Ser-15, in response to IR (7). To investigate whether the kinase activity of ATM was enhanced after exposure of cells to a complex genotoxin, we measured ATM kinase activity in ATMϩ/ϩ and ATMϪ/Ϫ cell lines after a 3-h treatment with 0 or 6 M Cr(VI). ATM immunoprecipitates were used to phosphorylate GST-p53 1-101 in vitro. ATM protein kinase activity toward GST-p53 1-101, expressed as the amount of GST-p53 1-101 phosphorylated by ATM normalized to the total GST-p53 1-101 , was increased from 0.36 to 0.69 ( Fig. 2B) (p Ͻ 0.05). This increased kinase activity is attributed specifically to the activation of ATM protein. No ATM-immunoreactive protein was present when preimmune serum was used for immunoprecipitation. Furthermore, no protein kinase activity was found in immunoprecipitates from ATM-deficient cells or in Cr(VI)-treated ATMϩ/ϩ cells immunoprecipitated with preimmune serum (data not shown). Therefore, the ATM kinase appears to be activated in response to Cr(VI) treatment, and it phosphorylates p53 in vitro.
Cr(VI)-induced Chk2 Thr-68 Phosphorylation Is ATMdependent-Chk2 was found to phosphorylate Ser-20 on p53, and it is required for increased stability of p53 in response to IR (34). Phosphorylation and activation of Chk2 have been shown to be ATM-dependent in response to IR, whereas Chk2 phosphorylation is ATM-independent when cells are exposed to UV or hydroxyurea (35). To investigate whether Cr(VI) exposure can induce Chk2 phosphorylation and whether this phosphorylation is ATM-dependent, we studied the Chk2 Thr-68 phosphorylation in ATMϩ/ϩ (GM03440) and ATMϪ/Ϫ (GM03395) cells after 0, 30, and 60 min of continuous exposure to 6 M Cr(VI). As shown in Fig. 4, A and B, a 2-fold increase of phosphorylation of Chk2 Thr-68 was observed within 30 min after a 6 M Cr(VI) treatment in ATMϩ/ϩ cells, whereas no increased phosphorylation of Chk2 Thr-68 was detected within FIG. 1. ATM status. Total cellular protein was extracted from normal fibroblasts (GM03440) and fibroblasts with either heterozygous (GM03397) or homozygous ATM deletions (GM03395 and GM02052). The protein was separated by SDS-PAGE, and ATM protein was detected by immunoblotting. 30 min of Cr(VI) exposure in ATMϪ/Ϫ cells. After a 60-min treatment, the phosphorylation of Chk2 Thr-68 increased 2.4fold in ATMϩ/ϩ cells, but again, no phosphorylation was detected in ATMϪ/Ϫ cells. Furthermore, total Chk2 expression, as determined by Western blotting, was not changed after Cr(VI) exposure because equal levels of Chk2 were found before and after Cr(VI) treatment in both ATMϩ/ϩ and ATMϪ/Ϫ cells (Fig. 4A). Therefore, the results show that Cr(VI) exposure induces phosphorylation of Chk2 Thr-68, and this phosphorylation is ATM-dependent.
ATM Contributes to Cr(VI)-induced Apoptosis-ATM is responsible for sensing and effecting certain DNA damage response pathways that converge on p53 (33). Cr(VI)-induced apoptosis has been shown to depend on activation of p53 (29). We used cells with heterozygous or homozygous deletions in the ATM gene and studied their sensitivity to Cr(VI)-induced apoptosis using the PS translocation assay. As shown in Fig.  5A, the basal apoptotic level in the untreated cells was around 2% in ATMϩ/ϩ and ATMϩ/Ϫ cells but was 2-3-fold higher in ATMϪ/Ϫ cells. Treatment with 3 M Cr(VI) caused a modest increase in apoptosis in both ATMϩ/ϩ and ATMϩ/Ϫ, but not ATMϪ/Ϫ cells. Treatment with 6 M Cr(VI) induced apoptosis in 40% of both ATMϩ/ϩ and ATMϩ/Ϫ cells, but the percentage of apoptosis (17-23%) was significantly lower in ATMϪ/Ϫ cells. When expressed as a -fold of untreated control, treatment with 3 M Cr(VI) induced a 2.6 -4.3-fold increase in apoptosis in both ATMϩ/ϩ and ATMϩ/Ϫ cells but only a 1.3-fold increase in ATMϪ/Ϫ cells. ATMϩ/ϩ and ATMϩ/Ϫ cells exhibited a 20-fold increase in apoptosis after 6 M Cr(VI) exposure, whereas ATMϪ/Ϫ cells showed marked resistance to Cr-induced apoptosis (p Ͻ 0.05). Furthermore, we confirmed the results of PS translocation assay by studying the cellular level of cleaved caspase-3. Caspase-3 is one of the key executioners of apoptosis. Activation of caspase-3 requires proteolytic processing of its inactive zymogen into activated P12 and P17 subunits (36). Western blotting with an antibody to cleaved caspase-3 showed that treatment with 6 M Cr(VI) produced a significant 8-fold induction of cleaved caspase-3 in ATMϩ/ϩ cells (p Ͻ 0.05) but no change in ATMϪ/Ϫ cells (Fig. 5, B and C). The results of these studies suggest that ATM is required for the majority of Cr(VI)-induced apoptosis. We also explored the possibility that the decreased apoptotic response of the ATMϪ/Ϫ cell lines to Cr(VI) exposure may be caused by an impaired Cr(VI) uptake mechanism. Therefore, we studied the uptake of Na 2 51 CrO 4 at a concentration of 3 M for 5 h in all cell lines studied. The average uptake was 2.5 nmol/10 6 cells, with no difference among the different cell lines (data not shown).
ATMϪ/Ϫ Cells Are Hypersensitive to Cr(VI)-induced Clonogenic Lethality-Because the lack of ATM conferred resistance to Cr(VI)-induced apoptosis, we determined clonogenic survival after Cr(VI) exposure in ATMϩ/ϩ, ATMϩ/Ϫ, and ATMϪ/Ϫ cells. Clonogenicity is an indicator of long term cell survival and replicative potential after exposure to a genotoxic agent. By determining the clonogenicity of Cr(VI)-exposed cells we were able to examine the cumulative effect of Cr(VI)-exposure on a cell population. The results are shown in Fig. 6 and are expressed as the percentage of the respective control, i.e. in the absence of Cr(VI). The ATMϪ/Ϫ cells showed significantly less clonogenic survival after 0.1, 0.5, or 1 M Cr(VI) treatment compared with ATMϩ/ϩ and ATMϩ/Ϫ cells (p Ͻ 0.05). The percent clonogenic survival for 0.1, 0.5, and 1 M Cr(VI) was 98, 78, and 51-59% for ATMϩ/ϩ and ATMϩ/Ϫ cells compared with 60 -84, 28 -51, 7-14% for the two ATM-deficient cell lines, respectively. Therefore, despite the resistance to Cr(VI)-in-duced apoptosis, these results indicate that lack of the ATM gene results in increased clonogenic lethality.

ATM Is Not Required for Cr(VI)-induced Growth Arrest, but It Is Required for Cells to Recover from Cr(VI)-induced Growth
Arrest-The single time point determinations of Cr(VI)-induced apoptosis could not explain the markedly increased sensitivity of ATMϪ/Ϫ cells to Cr(VI)-induced clonogenic lethality. Therefore, we examined both ATMϩ/ϩ (GM03440) and ATMϪ/Ϫ cell (GM03395) growth by counting the cell number over a 16-day period in populations of cells exposed to 0 -1 M Cr(VI) for 24 h. Data were expressed as -fold change compared with untreated controls (Fig. 7). In both cell lines, Cr(VI) induced a similar dose-dependent decrease in cell growth, which lasted up to 8 days after treatment. However, after 8 days a striking difference emerged. The Cr(VI)-treated ATMϩ/ϩ cells  7). We further explored the possibility that the persistent decreased number of ATMϪ/Ϫ cells after Cr(VI)-exposure may be due, in part, to a low level of cell death induced by Cr(VI). Therefore, we studied apoptotic cell death by PS translocation assay at days 1, 8, 12, and 16 after 24 h 1 M Cr(VI) exposure in ATMϩ/ϩ and ATMϪ/Ϫ cells. Our results showed that no apoptotic cell death was induced by 1 M Cr(VI) in either ATMϩ/ϩ or ATMϪ/Ϫ cells at any time point. The respective proportion of apoptotic cells after 1 M Cr(VI) exposure ranged from 0.9-to 1.1-fold of the nontreated control in ATMϩ/ϩ cells and from 0.8-to 1.2-fold in ATMϪ/Ϫ cells (data not shown). The results indicate that the decreased cell number observed in ATMϪ/Ϫ cells was caused by growth arrest and not apoptotic cell death induced by Cr(VI) treatment. These data strongly indicate that the initial growth arrest after Cr(VI) treatment is not dependent on ATM; however, ATM is either required for cells to recover from Cr(VI)induced growth arrest or for preventing the growth arrest from becoming terminal.
ATM-deficient Cells Fail to Exhibit Checkpoint Arrest in Early Response to Cr(VI) Exposure, but Showed Latent Blockage at S and M Phase Entry at Later Times-To understand fully the nature of Cr(VI)-induced growth arrest in ATMϩ/ϩ and ATMϪ/Ϫ cells, we evaluated the G 1 /S and G 2 /M checkpoint transitions in ATMϩ/ϩ and ATMϪ/Ϫ cells at days 1, 2, 4, and 7 after 24-h 1 M Cr(VI) exposure. G 1 checkpoint function, as reflected by delay or suppression of entry into S phase, was analyzed by BrdUrd and PI double staining (Fig. 8). Exposure of ATMϩ/ϩ cells to 1 M Cr(VI) resulted in a suppression of S phase entry, as measured by flow cytometry. The percentage of S phase cells decreased to 53% of that of the nontreated control at day 1. By day 7, ATMϩ/ϩ cells were released from the G 1 checkpoint and reentered S phase, as the S phase population increased to 73% of nontreated control. When ATMϪ/Ϫ cells were subjected to the same treatment, comparatively little inhibition of S phase entry was observed at day 1, as S phase population remained at 88% of the untreated control. However, at day 7, ATMϪ/Ϫ cells were markedly less able to enter S phase, as the S phase population decreased to 42% of that in the untreated control. G 2 /M checkpoint function, as reflected by suppression of entry into mitosis, was measured by flow cytometric assessment of histone H3 phosphorylation. Histone H3 is cyclically phosphorylated during mitosis (37). After Cr(VI) exposure, ATMϩ/ϩ cells were blocked from entering mitosis at day 1 (data not shown); however, these cells were released from the G 2 checkpoint by day 7, as the mitotic population increased from 1.6% of the total G 2 /M population at day 4 to 4.1% at day 7. In contrast, ATMϪ/Ϫ cells were unable to enter mitosis at day 7, as the mitotic cells remained at 0.6% of G 2 /M population. Therefore, despite the deficiency of checkpoint arrest in the early response to Cr(VI), these results suggest that ATMϪ/Ϫ cells were arrested and impaired in the ability to enter both S and M phase at later times after Cr(VI) exposure. DISCUSSION ATM kinase has been shown to be activated in response to DNA DSBs, which can be induced by IR, radiomimetic agents, and during recombinational repair, but not to UVB/UVC or base-damaging agents (7), indicating that ATM may act specifically in response to DSBs. However, the existence of various abnormalities in AT patients other than hypersensitivity to IR, such as premature aging, cancer predisposition, and progressive cerebellar degeneration, has raised the possibility that ATM may be involved in cellular processes in addition to the DSB response (5). This has been supported by recent reports, which showed the activation of ATM kinase by t-butylhydroperoxide (15), CdCl 2 (16), UVA (17), heat shock (18), insulin (19), and MNNG (20). Furthermore, a recent study by Bakkenist et al. (38) suggested that the activation of ATM might be the result of DNA damage-induced changes in chromatin structure. In this study, we showed enhanced ATM kinase activity in ATMϩ/ϩ cells after Cr(VI) treatment, using p53 as a substrate (Fig. 2). Furthermore, the respective phosphorylation of p53 Ser-15 and Chk2 Thr-68, which have been shown to be physiological targets of ATM after DNA damage (6,28), was either delayed or abrogated in ATM-deficient cells after Cr(VI) exposure (Figs. 3 and 4). Therefore, the results of the present study suggest that Cr(VI) also activates ATM kinase despite no direct evidence for Cr(VI) DSB induction.
Cr(VI) exposure results in a broad spectrum of genetic damage. One lesion in particular, Cr-DNA interstrand cross-links (Cr-DDC) could potentially contribute to the cytotoxicity of Cr(VI) (22) and may lead to the activation of ATM. Our recent studies suggest that recombinational repair is critical for the processing of potentially lethal genetic lesions resulting from the intracellular reduction of Cr(VI) (39). Indeed, studies by Morrison et al. (40) on recombinational repair in AT cells confirmed the central role of ATM in homologous recombinational repair of DNA damage. Misrepair of the recombining DNA appears to be a feature of recombination in AT cells (41). Therefore, the activation of ATM kinase by Cr(VI) may be the result of (i) ATM directly sensing Cr-DDC or other lesions or (ii) chromatin structure changes induced by Cr(VI) or (iii) ATM sensing DSBs generated during recombinational repair of DNA-DNA cross-links. The last hypothesis is supported by recent studies with MNNG, which suggested that activation of ATM by the alkylating agent MNNG was caused by the accumulation of strand breaks induced by the activity of repairassociated endonucleases (20). Alternatively, ATM has been reported to be a sensor of oxidative damage. Indeed, deregulation of the oxidative stress response is part of the A-T phenotype (16). Some studies have shown reactive oxygen species generated during Cr(VI) reduction; however, the experiments were conducted using extraordinarily high Cr(VI) doses (100 M-2 mM) and already malignant cells (42)(43)(44)(45)(46). The doses employed in this study (0.1-6 M) have not been shown to produce reactive oxygen species and are more relevant to human exposure.
ATM has been implicated in regulating phosphorylation and induction of p53 in cells exposed to IR through several mechanisms (47). ATM induces p53 by directly phosphorylating Ser-395 of murine double minute clone 2 protein (MDM2) (48), which decreases the ability of MDM2 to shuttle p53 from the nucleus to the cytoplasm, and by indirectly phosphorylating Ser-20 of p53 through the activation of Chk2 (35) (34). ATM is also implicated in the direct phosphorylation of Ser-15 of p53 (33) (7). Although the latter phosphorylation has little effect on the half-life of p53 or on its ability to cause cell cycle arrest or apoptosis, it has been suggested that phosphorylation of p53 Ser-15 might facilitate other post-translational modifications of p53, such as acetylation of Lys-382, which in turn facilitates the sequence-specific DNA binding (47). Indeed, poor p53 induction has been found in ATM-deficient cells after IR exposure (32). We have shown previously that p53 is activated in normal human fibroblasts in response to Cr(VI) exposure (28). After Cr(VI) treatment of human lung epithelial A549 cells, p53 protein became phosphorylated at Ser-15 and acetylated at Lys-382, whereas MDM2 protein was dissociated (49). Our additional studies showed that Cr(VI) exposure was accompanied by induction of both p53-transactivated and p53-independent proapoptotic genes in human lung fibroblasts (50). Here, we show that p53 Ser-15 and Chk2 Thr-68 phosphorylations were either delayed or abrogated in ATM-deficient cells (Figs. 3 and 4). The phosphatidylinositol 3-kinase-related kinases other than ATM, i.e. DNA-PK and ATR, are implicated in DNA repair and signal transduction pathways induced by DSBs. DNA-PK is implicated in nonhomologous end-joining repair (6), whereas ATR is involved in the later stages of the cellular response to DSBs (51). These other phosphatidylinositol 3-kinase-related kinases share many targets that are phosphorylated by ATM following DSBs (52). This functional redundancy may explain the delayed phosphorylation of p53 Ser-15 in ATM-deficient cells in response to Cr(VI).
Studies by us and others have shown that Cr(VI)-induced apoptosis is p53-dependent (29,53). Indeed, Cr(VI)-induced apoptosis in human lung fibroblast cells is dependent on the presence of p53, which is activated in response to Cr(VI) exposure (28,29). Cr(VI) was shown to increase p53 protein stability in human lung fibroblast cells, by a 4 -6-fold increase in p53 protein levels, and was accompanied by p53 translocation to the nucleus (29). The p53 requirement for Cr(VI)-induced apoptosis was confirmed in two models: (i) mouse dermal fibroblasts heterozygous and homozygous for p53 deletion, and (ii) human lung fibroblast cells transiently transfected with the human papilloma virus E6 gene, which targets p53 for degradation, thus creating a functional p53 deletion (28). Previous studies have shown that IR-induced apoptosis was reduced in AT cells compared with normal cells (54), and the failure to undergo apoptosis has been associated with an altered pattern of p53 induction after IR exposure (55). In the present study, ATM-deficient cell lines were significantly resistant to Cr(VI)induced apoptosis (Fig. 5), which might be because of the attenuated p53 induction after Cr(VI) exposure. These results indicate that ATM is a major upstream activator of p53-dependent Cr(VI)-induced apoptosis.
The altered induction of p53 does not explain AT radiosensitivity. It has been shown by numerous investigators that ATM-deficient cells are hypersensitive to IR-induced clonogenic lethality. Shackelford et al. (15) showed that GM03395 ATMϪ/Ϫ and GM02052 ATM mutant cells exhibit increased sensitivity to IR-induced clonogenic lethality compared with normal human fibroblasts. Furthermore, using neocarzinostatin (generous gift of Dr. Irving Goldberg, Harvard Medical School), a radiomimetic agent, we found that 10 and 25 ng/ml neocarzinostatin caused 82 and 35% clonogenic survival in GM03440 ATMϩ/ϩ cells, but only 40 and 7% in GM03395 ATMϪ/Ϫ cells (data not shown). Indeed, in primary fibroblasts, which exhibit an attenuated degree of apoptosis after a genotoxic treatment, the difference in radiosensitivity between ATM-deficient cells and normal cells is still observed (12).
Here, we showed that ATM-deficient cells were hypersensitive to Cr(VI)-induced clonogenic lethality (Fig. 6), which indicates that the resistance to Cr(VI)-induced apoptosis in ATM-deficient cells did not confer a selective survival advantage.
The population-wide cell cycle arrest in response to DNA damage provides the opportunity for a cell to regulate its own proliferation, thereby avoiding the propagation of damaged DNA. Depending on the extent of the genotoxic insult, an arrested cell could either regain its replicative potential by repairing the damaged DNA or be removed from the dividing population by undergoing either terminal growth arrest or apoptosis (56,57). Studies by us and numerous investigators have shown that ATM-deficient cell lines are impaired in their activation of cell cycle checkpoints (Fig. 8) (5), which suggests that ATM-deficient cells may lose the opportunity to regulate proliferation, thereby causing the propagation of damaged DNA. Furthermore, the ATM gene is involved in the DNA damage repair pathway by phosphorylating Brca1 (8), c-Abl (10), NBS1 (8), Gadd45 (55), etc. Collectively, the deficiency in cell cycle checkpoint arrest and DNA damage repair may contribute to the genetic instability of ATM-deficient cells. Indeed, ATM-deficient cells exhibit higher numbers of chromosomal gaps and breaks that persist for longer periods of time after IR (2) (58). Therefore, the accumulation of unrepaired damaged DNA may contribute to the hypersensitivity to Cr(VI)-induced clonogenic lethality observed in ATM-deficient cell lines. In the present study we showed that Cr(VI)-exposed normal human fibroblasts undergo an extended, but not terminal, form of growth arrest whereas they presumably slowly repaired the damaged DNA and recovered from this growth arrest after a considerable period. This is supported by our previous studies (30), which also found a similar transient cell growth arrest followed by replicative growth recovery in telomerase-immortalized normal human fibroblasts after a 24-h exposure to 1 M Cr(VI), a concentration at which no apoptosis was detected. In contrast, ATM-deficient cells were unable to recover from this growth arrest. Cell cycle analysis of ATMϩ/ϩ and ATMϪ/Ϫ cells showed that ATMϩ/ϩ cells undergo Cr(VI)-induced G 1 /S and G 2 /M checkpoint arrest (Figs. 8 and 9), which presumably allows an opportunity for DNA repair before replication. In contrast, ATMϪ/Ϫ cells failed to activate checkpoints in early response to Cr(VI) (Fig. 8). Therefore, the combination of the failure to activate checkpoints and the deficiency in DNA damage repair jointly contribute to the accumulation of unrepaired DNA damage in ATM-deficient cells. This, in turn, causes ATMϪ/Ϫ cells with accumulated unrepaired DNA damage to be removed from the proliferative population by undergoing terminal growth arrest. This is supported by our data, which showed that ATMϩ/ϩ cells resumed proliferation at later time points; however, ATMϪ/Ϫ cells remained in an arrested state in which few ATMϪ/Ϫ cells were able to enter S or M phase. Therefore, these results indicate that terminal growth arrest is the fate of these apoptosis-resistant cells, and ATM is required for recovery from genotoxin-induced cell growth arrest.
In summary, our data suggest that ATM is activated after Cr(VI) exposure, and ATM is a major signal initiator for Cr(VI)induced apoptosis. Paradoxically, ATM is also necessary for clonogenic survival, specifically for the cells to recover from the protracted growth arrest or escape from the terminal growth arrest induced by Cr(VI). We propose a model for the role of the ATM gene in the cellular response to a complex genotoxin (Fig.  10). Exposure of ATMϩ/ϩ cells to a complex genotoxin, such as Cr(VI), will cause the activation of the ATM protein (Fig. 10A), which causes the majority of the cells to undergo a transient (albeit protracted) cell cycle checkpoint arrest by activating downstream proteins such as p53 and ChK2, presumably to allow an opportunity for DNA repair before replication. Cells in which DNA repair is complete will regain replicative potential, whereas cells in which DNA repair is incomplete will be eliminated from the cell cycle by apoptosis or possible terminal cell growth arrest. In contrast, because of the lack of the ATM gene, ATM-deficient cells fail to activate cell cycle checkpoint arrest after complex genotoxin insult (Fig. 10B). The majority of the cells undergo terminal growth arrest. A stochastic fraction of cells may survive with unrepaired DNA damage and exhibit a predisposition to genomic instability and neoplasia. Taken together, this paradigm also strongly suggests that terminal growth arrest is not merely an extended or irreversible form of checkpoint arrest, but instead an independent and unique cell fate pathway.