In living cells, nuclear DNA is constantly being damaged by endogenous reactive oxidants generated during aerobic respiration in mitochondria (1–6). Both, the first- and the second-generation reactive intermediates, the byproducts of oxidative phosphorylation reactions, can diffuse from mitochondria, reach the nuclear DNA, and induce its damage (7, 8). It has been estimated that DNA exposure to endogenous oxidants during a single cell cycle of about 24-h duration results in ∼5,000 DNA single-strand lesions (SSLs) per nucleus (6). While 99% of SSLs are repaired by essentially error-free mechanisms, the remaining (∼1%) are converted, predominantly during DNA replication, to DNA double-strand breaks (DSBs). Thus, on average, about 50 DSBs (“endogenous DSBs”) are generated during a single cell cycle in human cells (6). Recombinatorial repair (also known as homologous recombination repair or template-assisted repair) and nonhomologous DNA-end joining (NHEJ) are the two major pathways of DSB repair. Recombinatorial repair takes place in cells that have already replicated some portion of their DNA that can then serve as a template, i.e., in late S and G2 phase cells. However, DNA repair in G1 and early S phase cells that lack a template relies on the NHEJ pathway. The latter is error-prone and may lead to deletion of some base pairs (9, 10). Accumulation of DNA damage by this mechanism, with each sequential cell cycle, is considered to be the predominant cause of aging and senescence as well as predisposition to cancer (4, 9–14). In fact, the strategies designed to slow aging or to prevent cancer are based on protection of DNA from oxidative damage, primarily by the use of antioxidants.
Although the presence of oxidants within the cell can be detected by cytometry (15, 16), the detection of DNA damage by endogenous oxidants in individual cells is limited. The single cell electrophoresis (comet) assay (17–19) appears to lack the desired sensitivity. Also the TUNEL assay (20, 21), while useful in labeling DNA strand breaks in apoptotic cells, is not sensitive enough to detect the constitutive DNA damage induced by endogenous oxidants. The same is true with the immunocytochemical detection of 8-oxo-7,8-dihydro-deoxyguanosine (8-oxo-dG) or apurinic/apyrimidinic sites, the products of oxidative damage to DNA bases (22). The assays used to measure DNA damage in bulk rather than in situ offer greater sensitivity but cannot provide information on individual cells, the heterogeneity of cell populations, or the relationship of DNA damage to cell cycle phase and apoptosis.
Detection of DNA damage indirectly, by immunocytochemical assessment of histone H2AX phosphorylation (23, 24) and/or activation of the Ataxia telangiectasia mutated protein kinase (ATM) (25), reporters of DNA damage particularly if it involves formation of DSBs, offers much greater sensitivity compared with other methods. These probes have already been extensively used in conjunction with cytometry to assess the extent of DNA damage induced by exogenous agents and to correlate such damage with the cell cycle phase and apoptosis (26–32, reviewed in Refs.33 and34). However, it was observed that, in untreated normal cells as well in the cells of various tumor lines, a fraction of histone H2AX molecules constantly undergoes phosphorylation (35,36). The level of this constitutive H2AX phosphorylation (CHP) was seen to vary depending on the cell type (line) as well as on cell cycle phase, being the highest in S and G2M cells (35–37). It was also observed that cells constitutively express activated ATM and that the extent of constitutive ATM activation (CAA) also varies depending on cell type and cell cycle phase (37). The level of CHP and CAA was markedly attenuated by exposure of cells to agents scavenging reactive oxygen species (ROS) such as N-acetyl-L-cysteine (37) or by lowering the cell metabolic activity by treatment with 2-deoxy-D-glucose (38, 39). In contrast, the level of CHP and CAA was elevated by suppression of synthesis of glutathione (37), the endogenous ROS scavenger. We postulated, therefore, that a significant fraction of this “background” CHP and CAA observed in untreated cells is in response to the ongoing DNA damage caused by endogenous oxidants generated during aerobic respiration in mitochondria (38). We projected that multiparameter cytometric measurement of the level of CHP and/or CAA allows one to estimate the extent of ongoing oxidative DNA damage and can be utilized to measure the DNA-protective effects of antioxidants or for that matter any agent that reduces or amplifies the generation of endogenous ROS. In the present study we provide further evidence in support of this postulate.
Materials and Methods
Human peripheral blood lymphocytes, obtained from healthy volunteers by venipuncture, were isolated by density gradient centrifugation as described in Ref.39. The cells were washed twice with phosphate-buffered saline (PBS) and resuspended in RPMI-1640 supplemented with 10% fetal bovine serum, 100 units/ml of penicillin, 100 μg/ml streptomycin, and 2 mM L-glutamine (all from GIBCO/BRL Life Technologies, Grand Island, NY) at a density of ∼ 5 × 105 cells/ml. The cells were then treated with 10 μg/ml of phytohemagglutinin (PHA, Sigma Chemical Co., St. Louis, MO) and incubated in 25-ml (12.5-cm2) polystyrene flasks (Becton Dickinson, Franklin Lakes, NJ) in a mixture of 95% air and 5% carbon dioxide at 37.5°C. Alternatively, the cells were incubated in a mixture of 92% nitrogen, 5% carbon dioxide, and 3% oxygen. Some cultures were left untreated with PHA while other lymphocyte samples were analyzed immediately after isolation, without having been cultured. Human B-cell lymphoblastoid TK6, WTK1, and NH32 cells were kindly provided by Dr. Howard Liber of Colorado State University, Fort Collins, CO (40). The cells were grown in 25-ml FALCON flasks in RPMI 1640 supplemented with 10% fetal calf serum and antibiotics, as mentioned earlier. At the onset of the experiments, there were fewer than 5 × 105 cells/ml in culture, such that the cells were in an exponential and asynchronous phase of growth. Human alveolar epithelial-like lung carcinoma A549 cells and human diploid embryonic WI-38 cells were obtained from ATCC (Manassas, VA) and grown in Dulbecco's minimum essential medium supplemented with 10% fetal bovine serum, 100 units/ml penicillin, 100 μg/ml streptomycin, and 2 mM L-glutamine (GIBCO/BRL). Normal human bronchial epithelial (NHBE) cells were purchased from Cambrex Bio Science (Walkersville, MD) and were cultured in complete bronchial epithelial cell growth medium, prepared by supplementing bronchial epithelial basal medium (BEBM) with retinoic acid, human epidermal growth factor, epinephrine, transferrin, triiodothyronine, insulin, hydrocortisone, bovine pituitary extract, and gentamicin by addition of SingleQuots™. BEBM and SingleQuots were purchased from Cambrex Bio Science. To maintain exponential growth of the cells that grow attached to the substrate, the cells were trypsinized and reseeded before reaching confluency. For experiments designed to be analyzed by laser scanning cytometry (LSC), the cells were seeded at low cell density (about 5 × 104 cells per chamber) in two-chambered Falcon Culture Slides (Becton Dickinson). Some chamber cultures were also grown in medium containing 1% serum as well as in an atmosphere consisting of 93% nitrogen, 5% carbon dioxide, and 3% oxygen. Human fibroblasts MSU1.1 (41) were cultured as described earlier (42). Further details of cell culturing are provided in figure and Table legends.
H2O2 (Sigma), 2-deoxy-D-glucose (2-DG), 3-bromopyruvate (BrPA) (Sigma) and Na ascorbate (Sigma), and dichloroacetate (DCA; Sigma) were included into cultures at concentrations and for various periods of time, as shown in the figures and Table 1 legends. Celecoxib was kindly provided Donnie W. Owens, Manager, Compound Transfer Research and Development Operations, Pfizer (Groton, CT). For treatment with celecoxib, the cells were seeded on two-well or four-well slides (BD Bioscences, Bedford, MA) and treated with various concentrations of this drug, for time periods as shown in figure legends; the medium including freshly dissolved celecoxib was exchanged every 24 h.
Table 1. Summary of the data presented in the present article (p.a.) as well as reported earlier (refs.36–39) listing different factors that either attenuated CHP and CAA, or enhanced their level, in different cell types
The data show the CHP and CAA change (Δ), either percent (or -fold) increase (+) or decrease (−), of the mean values of γH2AX or ATM-S1981P IF, estimated for the populations of S-phase cells as described in Materials and Methods, of the treated vs. untreated cells. The Δ data were approximated to 5%. While details of the experiments carried out in the present study are described in the Materials and Methods, the published data (36–39) are outlined as follows: (a) cell treatment with N-acetyl-L-cysteine (NAC), the scavenger of ROS, at concentrations and lengths as shown in the Table, distinctly lowered both CHP and CAA; (b) treatment with glutathione synthetase inhibitor buthionine sulfoximine (BSO), expected to lower the content of glutathione, the endogenous ROS scavenger, raised the level of CHP; (c) growth at increased cell density (cells/ml, as shown), known to lower metabolic activity, led to a decrease CHP and CAA; (d) 2-Deoxy-D-glucose (2-DG), antimetabolite of glucose, considered a “diet restriction mimic” that lowers aerobic metabolism (44, 45) markedly attenuated CHP and CAA; (e) mitogenic stimulation of peripheral blood lymphocytes (PBL) dramatically elevated both CHP and CAA, and this elevation was suppressed by 2-DG and by hypoxia (3% O2).
Immunocytochemical Detection of γH2AX and ATM-S1981P
The cells were fixed in suspension or on slides in a solution of 1% methanol-free formaldehyde (Polysciences, Warrington, PA) dissolved in PBS for 15 min on ice followed by suspension in 80% ethanol and stored at −20°C for 2–24 h. The cells were then washed twice in PBS and suspended in a 1% (w/v) solution of bovine serum albumin (BSA; Sigma) in PBS for 30 min to suppress nonspecific antibody (Ab) binding. The cells were then incubated in 100 μl of 1% BSA containing (1:200) diluted anti-phospho-histone H2A.X (Ser-139) mAb (Upstate, Lake Placid, NY) or anti-phospho-ATM (Ser-1981) mAb (Upstate, 1:100) and incubated for 2 h at room temperature. The cells were rinsed with 1% BSA in PBS (200g, 5 min) and then incubated in 100 μl of 1% BSA containing 1:200 diluted Alexa Fluor 488 F(ab′)2 fragment of goat anti-mouse Ab (Invitrogen, Eugene, Oregon) for 40 min at room temperature. Contribution of the nonspecific fluorescence was estimated by incubation of the cells with isotype-matched (IgG2a), irrelevant Ab. If measured by LSC (43), cellular DNA was counterstained with 1 μg/ml 4,6-diamidino-2-phenylindole (Invitrogen, San Diego, CA), while if measured by flow cytometry, DNA was stained with 5 μg/ml PI in the presence of 100 μg/ml of RNase A (Sigma).
Fluorescence of cells growing in suspension (TK6, WTK1, NH32, or lymphocytes) was measured using a FASCcan flow cytometer (Becton-Dickinson, San Jose, CA). The red (PI) and green (FITC) fluorescence from each cell were separated and quantified using the standard optics and CELLQuest software (Becton-Dickinson). At least 10,000 cells were measured per sample. Fluorescence of cells growing attached on slides was measured using a LSC (iCys; CompuCyte, Cambridge, MA), utilizing standard filter settings; fluorescence was excited with 488-nm argon ion and violet diode lasers, respectively. The intensities of maximal pixel and integrated fluorescence were measured and recorded for each cell. At least 3,000 cells were measured per sample. Gating analysis was carried out to obtain mean values (±SE) of γH2AX IF or ATM-S1981P for G1 [DNA index (DI) = 0.9–1.1], S (DI = 1.2–1.8), and G2M (DI = 1.9–2.1) cell populations in each experiment. To express the treatment-induced change in mean values of the measured cells, the means values of the untreated as well as the treated cells were compensated for the level of nonspecific fluorescence measured as described earlier. The SE was estimated based on Poisson distribution of cell populations. Each experiment was run at least in triplicate, and some experiments were additionally repeated. Other details are given in figure legends.
Fluorescence was detected using a Bio-Rad MRC1024 confocal system (Bio-Rad, Microscience, Hemel Hempstead, UK), interfaced with an inverted microscope Nikon Diaphot (Nikon, Amsterdam, The Netherlands), equipped with three fluorescence detection channels, and 15 mW KrAr laser (ALC, Salt Lake City, UT), and a 60× NA 1.4 oil immersion lens (Nikon). Alexa 488, propidium iodide, and DRAQ5 (Biostatus, Cardiff, UK) were excited simultaneously using the 488- and 568-nm laser lines. Conditions of measurements were primary triple dichroic mirror 488/568/647 (T1 block), secondary dichroic 560LP (T2A), a 620LP fixed dichroic facing PMT3, and emission filters 540/30, 580 LP, 630 LP. Eight-bit 512 × 512 images of a 41 × 41 μm field of view was collected at a rate of 0.3/s scan; z-step was 0.20 μm. 3D images were deconvolved (AutoDeblur, Autoquant), color channels merged and the maximum pixel value projection images shown.
The first experiments were designed to assess the effect of exogenous oxidants on H2AX phosphorylation and ATM activation. Toward this end, exponentially growing A549 and WI-38 cells were exposed to H2O2, subsequently fixed, immunostained for γH2AX and ATM-S1981SP, and their γH2AX and ATM-S1981P IF were compared with that of untreated (Ctrl) cells (Fig. 1). The data show an increase in expression of both, γH2AX and ATM-S1981P upon cell exposure to H2O2. The increase occurred in all phases of the cell cycle. The most pronounced rise, however, was seen in the S-phase cells, in which following exposure to H2O2, the expression of γH2AX IF and ATM-S1981P IF rose over fivefold and nearly threefold, respectively. Exposure of WI-38 cells to 100 μM H2O2 for 1 h led to a similar pattern of increase in expression of γH2AX and ATM-S1981P although the rise in γH2AX was somewhat less pronounced (Table 1).
Localization of γH2AX in the untreated-, as well as H2O2-treated human fibroblasts detected by confocal microscopy is shown in Figure 2. We observed that most untreated cells expressed relatively few (1–5) distinct γH2AX foci. Some cells, however, had a larger number of foci (Fig. 2A). These were generally cells with larger nuclei, most likely S- or G2M phase cells. The foci were heterogeneous in terms of their size and fluorescence intensity, some foci being large and very distinct, while others were of diminished size and of lower fluorescence intensity. The number of foci was clearly greater in the cells exposed to H2O2, and they were more uniform in regard to size and IF intensity (Figs. 2B and 2C).
In the next set of experiments, we explored the effect of cell growth at low concentrations of O2 on CHP and CAA. The cells were maintained in culture in an atmosphere of 3% O2, and their CHP and CAA were monitored and compared with cells growing in air (21% O2). It is quite evident that WI-38 cells growing at 3% O2 for 24 or 48 h had distinctly lower levels of CHP and CAA (Fig. 3). The reduction of CAA was more pronounced than of CHP. In fact, growth at 3% O2 for 48 h led to a decrease in the level of CAA by more than 50%. The decrease in CHP and CAA was of similar proportion in all phases of the cell cycle. Analysis of the DNA content histograms (Fig. 3 insets) revealed an increase from 45 to 64% in the percentage of cells in G1 phase of the cell cycle after growth at 3% O2 for 48 h compared with 21% O2.
The effect of hypoxia was also tested on TK6, and A549 cells, as well as mitogen-stimulated lymphocytes by growing them at 3% O2 for 24 and 48 h and comparing their CHP and CAA with their counterparts grown in air. The data showing the change in the level of CHP and CAA in TK6 cells following their growth at 3% O2 for 24 h are presented in Figure 4. In the case of TK6 cells, the hypoxia-induced reduction of CHP was less pronounced compared with the decrease of CAA, as the latter was diminished in all phases of the cell cycle by 31–36%, whereas the former decreased only minimally (7–9%). In contrast to WI-38 cells, there was no evidence of any significant change in the cell cycle distribution as a consequence of growth of TK6 cells at 3% O2 compared to growth in air (Fig. 3, insets). As mentioned, we have also tested the effect of hypoxia on A549 cells and on the mitogen-stimulated human lymphocytes. In both cases, a significant reduction in the level of CHP and CAA was observed (Table 1).
Figure 5 illustrates the effect of cell growth in the presence of the hexokinase II inhibitor BrPA. Similar to 2-deoxy-D-glucose, BrPA inhibits glycolysis (47), and since part of glycolysis is along the aerobic pathway (48), exposure of cells to BrPA is expected to decrease the generation of free radicals in mitochondria. We observed that, indeed, the expression of both γH2AX and ATM-S1981P was dramatically reduced in TK6 cells maintained in the presence of 0.1 or 0.3 mM BrPA for 4 h (Fig. 5). The reduction in the level of CHP was much more pronounced compared with CAA, and was greater for S- and G2M-phase cells (over fourfold) than for G1 phase cells.
The antioxidative properties of vitamin C (Asc; sodium ascorbate) are well recognized (49–51). In the next experiments, therefore, we attempted to measure the possible effect of Asc on CHP and CAA. The data (Fig. 6) clearly indicate that the growth of A549 cells in the presence of 2 mM Asc for 24 h led to a distinct reduction in the level of both CHP and CAA. The degree of decrease was similar for CHP and CAA and was within a range of 22–38%, being somewhat more pronounced for S-phase cells.
There is some evidence in the literature suggesting that cyclooxygenase-2 (COX-2) inhibitors may exert a protective effect on DNA against oxidative damage (52–55). We explored, therefore, whether the COX-2 inhibitor celecoxib might affect the level of CHP and CAA. Toward this end, the A549 as well as WI-38 cells were treated in cultures with 5 μM celecoxib for 3 days and then subjected to analysis of γH2AX and ATM-S1981P expression. The data (Fig. 7; Table 1) show that the level of CHP and CAA of celecoxib-treated cells was only minimally (∼5%) affected. However, when the celecoxib-treated cells were subsequently exposed to 100 μM H2O2 for 30 min, the expression of γH2AX and ATM-S1981P was significantly reduced (20–35%) compared with the cells treated with H2O2 but not pretreated with the COX-2 inhibitor. The pattern of response to celecoxib was similar, both for the A549 cells (Fig. 7) as well as for WI-38 fibroblasts (Table 1).
We have also tested the effect of DCA on CHP and CAA. Through its activating effect on the pyruvate dehydrogenase kinase, DCA shifts the anaerobic glycolytic pathway towards glucose oxidation (56, 57). Because tumor cells are characterized by relatively high levels of anaerobic glycolysis (58, 59), it might be expected that cell growth in the presence of DCA would lead to increased levels of CHP and CAA. Indeed, we observed a distinct and reproducible increase in the level of CHP and CAA, particularly in S- and G2M-phase A549 cells (Fig. 8). Interestingly, a similar degree of CHP and CAAenhancement was also observed in WI-38 cells (Fig. 8, Table 1).
The full list of different agents analyzed for their effect on the level of CHP and CAA on various cell types is presented in Table 1. For completeness, this list includes both the agents investigated in the present study as well as data reported earlier (36–39, 60).
The present data are consistent with our earlier observations (36–39, 60) and collectively provide strong evidence that a large portion of CHP and CAA, detected immunocytochemically and measured by multiparameter flow- or LSC, represents a response to oxidative DNA damage caused by metabolically generated oxidants in mitochondria. These results, obtained using a variety of cell types, whether normal such as human lymphocytes and NHBE cells, normal-like WI-38 fibroblasts, or tumor cells either of hematopoietic (TK6, NH32, WTK1) or of solid tumor (A549) lineage, are summarized in Table 1.
It is quite evident that short (1 h) exposure to the external oxidant (H2O2) triggered a strong rise in the level of H2AX phosphorylation concurrent with ATM activation (Fig. 2). The response of S-phase cells was the most prominent as the increase in expression of γH2AX and ATM-S1981P in the treated cells distinctly exceeded the increase of γH2AX and ATM-S1981P in G1- or G2M-phase cells (Fig. 1). Treatment with H2O2 for even shorter (30 min) periods of time led to quite similar results (data not shown). The data are compatible with the known mechanism of induction of DNA damage by oxidants, which initially led to the formation of DNA SSLs, predominantly consisting of 8-oxo-dG, a fraction of which are converted into DSBs during DNA replication (1–8, 61). Since DSBs are the most potent inducers of ATM activation and H2AX phosphorylation (23–25) it is not surprising, therefore, that the S-phase cells responded maximally to the short pulse of the oxidant (Fig. 1).
The presence of IF foci of γH2AX is considered to be a marker of cell response to the formation of DSBs (23–25). With confocal microscopy, we observed that the number of foci in untreated cells varied significantly; while some cells had very few—or even none—others had several foci. Generally, the nuclei of cells with the greater number and density of foci were larger in size than the nuclei with none or a few foci. It is likely, therefore, that the foci were more numerous in S- or G2-phase cells. This would be expected, since as mentioned, DSBs are generated predominantly during DNA replication, and thus foci, which upon induction of DSBs last for up to 1–3 h (23, 24, 27) would be more numerous in the S- and G2-phase cells. In the untreated cells, the foci were more heterogeneous in size compared with the cells treated for 1 h with H2O2. This is consistent with the greater heterogeneity of the “age” of individual foci in these cells. Some foci would have been generated shortly after induction of DSBs and, thus, have maximal γH2AX expression, while others having H2AX partially dephosphorylated or replaced by unphosphorylated H2AX (62) would have diminished γH2AX IF. Such heterogeneity is expected considering that DSBs occur asynchronously throughout S-phase, which generally lasts for 6–8 h.
The question may be asked as to whether the overall intensity of γH2AX IF integrated over the nucleus and measured by cytometry correlates with the number of foci, considered to be bona fide markers of individual DSBs? It appears that under some circumstances these two markers do correlate with each other quite well. For example Rogakou et al. (23) observed that the γH2AX IF integrated on Western blots strongly correlated with the numbers of DSBs induced by X-radiation. Also the data of Banath and Olive (64) by providing evidence of a strong correlation between the doses of radiation, cell survival, and intensity of γH2AX IF measured by cytometry point in the same direction. Likewise, we observed that within certain dose ranges of DNA damaging agents that induce DSBs, the intensity of γH2AX IF measured by cytometry correlated with their dose (27–29). Also reassuring were the data of Bouquet et al. (63), who demonstrated at low levels of DNA damage (<150 foci per nucleus) a good correlation between the number of γH2AX foci and the overall γH2AX IF measured by cytometry. It is likely, thus, that when the overall level of DNA damage (number DSBs) is rather low, as it is in the case of CHP, the intensity of γH2AX IF measured by cytometry is a good reflection of the extent of DNA damage. The attenuation or enhancement of CHP by exogenous factors known to affect cell metabolic (aerobic) activity, therefore, do indeed reflect, respectively, protection or enhancement of oxidative DNA damage by these factors.
A good correlation was observed between the change (Δ) in expression of both γH2AX and ATM-S1981P induced by the various factors in this study: in every case, when the expression of γH2AX decreased the expression of ATM-S1981P also decreased, and vice versa (Table 1). This suggests that CHP, triggered by endogenous oxidants, is mediated by ATM. The data also imply that CHP is not a reflection of replication stress, during which phosphorylation of H2AX is known to be mediated not only by ATM but also by ATR and/or DNA PKcs (30, 65, 66). The concurrent activation of ATM and phosphorylation of H2AX is considered to be a reliable marker of DNA damage that involves formation of DSBs (67–71).
Apart from the theoretical implications of the present findings, which may address mechanisms by which particular factors investigated in this study affected the CHP and CAA, our data provide for the first time evidence that cytometry of γH2AX and ATM-S1981P can be used to detect and measure the potential DNA protective properties of a variety of external and endogenous agents. The methodology offers a sensitive, quantitative tool that can be used to assess the effectiveness of dietary supplements or diet restrictions (“diet restriction mimetics”; 44,45) aimed to protect DNA from oxidative damage and thus slow down the aging process and decrease the propensity of cells to undergo neoplastic transformation.
We thank Donnie W. Owens, Manager, Compound Transfer Research and Development Operations, Pfizer Inc., for providing samples of celecoxib. We also thank Millipore Corporation for providing samples of antibodies used in this study.