The induction of DNA double-strand breaks (DSBs) in chromatin triggers histone H2AX phosphorylation (on Ser-139) by ATM-, ATR-, or DNA-dependent protein kinases (DNA-PK). Phosphorylated H2AX, denoted as γH2AX, can be detected immunocytochemically using an antibody that is specific to the Ser-139–phosphorylated epitope. We previously reported that the induction DSBs by DNA topoisomerase I or II inhibitors can be monitored in individual cells by measuring γH2AX immunofluorescence (IF) by cytometry. The present study explored whether the detection of γH2AX IF can serve as a marker of the presence of the DNA precursor bromodeoxyuridine (BrdU) that is incorporated into DNA.
HeLa cells growing on microscope slides were incubated with BrdU for 1 h, rinsed free of the precursor, and incubated for different periods for up to 12 h. The cells were then briefly incubated with Hoechst 33342 (to sensitize BrdU-labeled DNA to ultraviolet [UV] light), irradiated with 300 nm UV light to photolyze BrdU-labeled DNA, transferred back into culture for an additional hour, and fixed. Cells were concurrently immunostained for γH2AX (Alexa Fluor 633) and cyclin A (fluorescein isothiocyanate); their DNA was counterstained with 4,6-diamidino-2-phenylindole. The intensities of cellular far red (γH2AX), green (cyclin A), and blue (DNA) fluorescences were measured by laser scanning cytometry.
After a 1-h pulse of BrdU followed by exposure to UV, nearly all cells with S-phase DNA content had many-fold higher γH2AX IF than G1 or G2/M cells. The nonirradiated cells had minimal (“programmed”) expression of γH2AX, whereas the irradiated cells incubated without BrdU had uniformly elevated levels of γH2AX IF independent of the cell cycle phase. Pulse-chase experiments showed that the cohort of BrdU-labeled (γH2AX-positive) cells progressed through G2/M and into G1 phase after 8 and 12 h of growth in BrdU-free medium, respectively. Bivariate analysis of γH2AX versus cyclin A expression for the gated S-phase cells showed a correlation between these variables, suggesting that the rate of BrdU incorporation (DNA replication) correlates with expression of cyclin A.
The incorporation of bromodeoxyuridine (BrdU) by live cells is an indicator of DNA replication. Three different approaches have been developed to detect incorporation of BrdU into DNA in intact cells. The first is based on the observation that fluorescence of Hoechst 33358 is quenched in cells that incorporate BrdU (1, 2). Therefore, the BrdU-incorporating cells can be identified by their deficit in intensity of Hoechst 33358 fluorescence compared with their DNA content; DNA content is measured concurrently with another color fluorochrome that is not quenched by BrdU (2). Numerous variants of this methodology have been developed, designed primarily to measure kinetics of cell cycle progression (3–5). Another fluorochrome quenched by BrdU, acridine orange, has also been used as a marker of DNA replicating cells (6).
The second approach to detect DNA replication is based on immunocytochemical detection of the incorporated BrdU (7). In this method, DNA is partly denatured by heat or strong acid to incorporate BrdU into double-stranded DNA that is accessible to a fluorochrome-labeled (typically fluorescein isothiocyanate) antibody developed against BrdU. In this approach, the non-denatured portion of DNA is stained with fluorochrome, generally a DNA intercalator such as propidium iodide, whose red fluorescence can be discriminated from the green of the anti-BrdU antibody (8, 9). This methodology, which is more sensitive than the one based on quenching of Hoechst 33358 or acridine orange, has become widely used in various disciplines of biology and medicine for cell cycle and cell kinetics analyses (10, 11).
Because the immunocytochemical detection of BrdU requires harsh conditions to denature DNA, it often ruins cell morphology and damages or extracts many proteins. Therefore, although highly sensitive, this procedure is incompatible with the concurrent use of other immunocytochemical probes. Attempts have been made to increase the accessibility of BrdU antibody to the incorporated precursor by partial enzymatic hydrolysis of DNA with exonuclease III (9) rather than with its denaturation. However, the uncontrolled loss of DNA from individual cells during hydrolysis makes it difficult to distinguish their cell cycle phase after staining with the DNA-specific fluorochrome. Thus, a third approach was introduced, which does not require DNA denaturation or hydrolysis but is based on selective photolysis of DNA that contains the incorporated BrdU (12, 13). Specifically, the incorporated BrdU, especially in the presence of Hoechst 33342, sensitizes DNA to undergo photolysis after exposure to ultraviolet (UV) light. Photolysis generates an abundance of DNA double-strand breaks (DSBs) in sections of DNA that contain BrdU. The 3′-OH prime ends of these DSBs are then labeled with fluorochrome-tagged (directly or indirectly) deoxynucleotides in the reaction by using terminal deoxynucleotidyl transferase while DNA is counterstained with another color fluorochrome (14). This DNA strand-break induction by photolysis (SBIP) methodology is fully compatible with the immunocytochemical detection of intracellular protein (15).
We describe a new variant of the SBIP methodology, in which, after DNA photolysis, phosphorylation of histone H2AX caused by the induction of DSBs serves as marker of BrdU incorporation (16–18). This method detects the presence of incorporated BrdU and, by using a pulse-chase approach, can be used to follow the progression of a cohort of BrdU-labeled cells through different phases of the cell cycle to demonstrate cell kinetics. The method can also be combined with immunocytochemical detection of other proteins, such as cyclin A, to correlate its expression with BrdU incorporation.
MATERIALS AND METHODS
Cell Cultures, UV Light Irradiation
Human cervical carcinoma HeLa cells obtained from the American Type Culture Collection (Manassas, VA) were maintained in 25 ml (12 cm2) Falcon flasks (Becton Dickinson, Franklin Lakes, NJ) in RPMI 1640 supplemented with 10% fetal calf serum, 100 U/ml penicillin, 100 μg/ml streptomycin, and 2 mM L-glutamine (all from Gibco/BRL Life Technologies, Inc., Grand Island, NY) at 37.5°C in an atmosphere of 5% CO2 in air. The cultures were diluted and re-plated every 4 days to maintain them in an asynchronous and exponential phase of growth. The cells were then trypsinized, rinsed with phosphate buffered saline (PBS), and seeded at low cell density (∼2,000 cells/chamber) in two-chamber glass slides (Lab-Tek, Nunc Inc., Naperville, IL). When the cells were at 40% to 50% confluency, BrdU (Sigma Chemical Co., St. Louis, MO) at a final concentration of 30 μM was added to the cultures for 1h. During the final 20 min before irradiation with UV light, Hoechst 33342 was administered into the cultures at a final concentration of 4 μM. Some cultures were maintained in the presence of BrdU for 1 h, rinsed, supplemented with fresh, prewarmed medium free of BrdU, and transferred into a CO2 incubator for an additional 2, 4, 6, 8, or 12 h; these cultures were also treated with Hoechst 33342 (4 μM) during the final 20 min before UV irradiation. The slides bearing the cultures were then placed onto a UV light gel illuminator (Fotodyne, West Berlin, WI) and exposed to UV light at 115 mW/m2 for 5 min to a total dose 3.45 kJ/m2. The cultures were then rinsed with fresh medium and transferred into the incubator at 37.5°C for an additional hour. Cells were then fixed by transferring the slides into Coplin jars that contained 1% methanol-free formaldehyde (Polysciences, Inc., Warrington, PA) in PBS for 15 min on ice followed by suspension in 70% ethanol, in which they were stored at −20°C for 2 to 24 h.
Immunocytochemical Detection of H2AX Phosphorylation and Expression of Cyclin A vis-á-vis Cellular DNA Content
Cells growing on two-chamber slides were incubated in the presence of 30 μM BrdU for 1 h, exposed to UV light, and incubated for 1 or 2 h before being fixed as described above. The slides were rinsed twice in PBS, immersed in 0.2% Triton X-100 (Sigma Chemical Co.) in a solution containing 1% (w/v) bovine serum albumin and PBS for 30 min, and then incubated in 50 μl of 1% bovine serum albumin that contained a 1:200 dilution of antiphosphohistone H2AX (Ser-139) murine monoclonal antibody (Upstate Biotechnology, Lake Placid, NY) and a 1:100 dilution of anti–cyclin A rabbit polyclonal antibody (Santa Cruz Biotechnology, Santa Cruz, CA) overnight at 4°C. On the following day, the cells were washed twice with PBS, incubated in 50 μl of a 1:100 dilution of a Alexa Fluor 633–conjugated F(ab′)2 fragment of goat anti-mouse immunoglobulin G (H + L; Molecular Probes, Eugene, OR) and a 1:30 dilution of a fluorescence isothiocyanate–conjugated F(ab′)2 fragment of swine anti-rabbit immunoglobulin (DAKO, Carpinteria, CA) for 45 min at room temperature in the dark. The cells were counterstained with 1 μg/ml 4,6-diamidino-2-phenylindole (DAPI; Molecular Probes) in PBS for 5 min before measurement.
Fluorescence Measurements by Laser Scanning Cytometry
Cellular green (cyclin A), far red (histone γH2AX), and blue (DAPI) fluorescence emissions were measured simultaneously in the same cells by using a laser scanning cytometer (CompuCyte, Cambridge, MA) with standard filter settings (19, 20); fluorescence was excited with 488-nm argon ion, helium neon, 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. Other details of fluorescence measurement are described in detail elsewhere (18).
Figure 1 shows fluorescence photomicrographs of HeLa cells that were incubated in the presence of BrdU for 1 h, exposed to UV light, cultured for an additional hour, and immunostained for γH2AX; their DNA was counterstained with DAPI. It is quite evident that only a fraction (∼30%) of cells were labeled with γH2AX under these conditions. The γH2AX immunofluorescence (IF) of the remaining cells was too faint to be detectable under the photographic exposure conditions aimed to demonstrate IF of the strongly labeled cells. To obtain the image of the weakly labeled cells, one had to overexpose the photomicrograph to the point that the strongly labeled cells appeared all white rather than green (not shown).
The bivariate analysis of γH2AX IF versus DNA-associated DAPI fluorescence of cells that were not incubated or incubated with BrdU for 1 h, unexposed or exposed to UV light, and returned to culture for an additional 1, 2, or 4 h is shown in Figure 2. The sensitivity of the photomultiplier detecting the red fluorescence (γH2AX) was identical for all these samples and was set up for optimal display of the maximally labeled cells. Under these conditions, the γH2AX IF of the cells that were untreated with BrdU and not exposed to UV and of the cells that were treated with BrdU but not exposed to UV light was minimal (Fig. 2A,B). The cells that were exposed to UV light but not preincubated with BrdU had rather strong γH2AX IF, but no cell cycle phase specificity was apparent among them: the higher level of γH2AX IF in G2/M and S compared with G1 cells is a reflection of the differences in their DNA content (Fig. 2C). In contrast, the cultures that were incubated with BrdU and then exposed to UV light demonstrated a very characteristic, cell cycle phase-specific pattern of γH2AX IF expression. Namely, the S-phase cells in these cultures were selectively labeled, whereas G1 and G2/M cells had distinctly low fluorescence (Fig. 2D–F). Further increase in intensity of γH2AX IF in the S-phase cells was seen after incubation with BrdU and UV irradiation of the cells that had been incubated for an additional 2 or 4 h. The apparent lack of the progression through the cycle of the BrdU-labeled cohort of S-phase cells during this time is a reflection of their cell cycle arrest induced by the massive irradiation dose.
The BrdU pulse-chase experiment, in which the BrdU-labeled cells were allowed to grow in BrdU-free medium for up to 12 h, is shown in Figure 3. The sequential scattergrams demonstrate that, with increasing duration of the chase period, the cells expressing H2AX IF moved from S phase to G2/M phase after 8 h, and most of them appeared in G1 phase after 12 h. Thus, it is quite evident that progression of the BrdU pulse-labeled cohort of cells through the cycle can be monitored by photolysis of their DNA, which induces phosphorylation of H2AX.
SBIP methodology is compatible with the concurrent immunocytochemical detection of other intracellular proteins. Therefore, in the subsequent experiments, through multiparameter analysis, we attempted to correlate within the same cells the level of γH2AX IF induced by photolysis of BrdU-labeled DNA with the expression of cyclin A. The aim of this analysis was to assess whether there is a correlation between intensity of γH2AX IF, presumed to reflect the frequency of DSBs induced by photolysis and therefore the extent of BrdU incorporation during the pulse incubation with the precursor, and abundance of cyclin A in S-phase cells. Because the extent of BrdU incorporation under these conditions reflects the rate of DNA replication, the quest was to find a possible correlation between the level of cyclin A in individual S-phase cells and their rate of DNA replication. Toward this end, using three-laser excitation, the expressions of γH2AX and cyclin A were measured in conjunction with measurement of DNA content in cells that had been incubated with BrdU for 1 h, irradiated with UV light, incubated for another 1 or 2 h, and fixed. The scatterplot representing cyclin A IF versus DNA content (Fig. 4 A) shows a typical pattern of cyclin A expression, as previously described (15, 21), with the G1 and M cells being cyclin A negative and with the distinctly increasing level of cyclin A during progression of cells through S and G2 phases. This pattern of expression of cyclin A was not altered after cell incubation with BrdU or irradiation with UV light (not shown).
Two approaches have been used to measure the correlation between cyclin A expression and intensity of γH2AX IF in the selected population of S-phase cells. In the first approach, the S-phase cells were gated, based on their DNA content values ranging from DNA index (DI) 1.2 to 1.8, on the scatterplots representing cyclin A versus DNA content (Fig 4A, dashed-line thresholds). This subpopulation was then plotted as cyclin A versus γH2AX IF distributions (Fig. 4B,C). The data were transferred to Excel (Microsoft, Redmond, WA) to obtain the correlation coefficient between these variables as described in detail elsewhere (15). Briefly, the maximal threshold fluorescence intensities that characterized cyclin A–negative (G1 and M) and γH2AX-negative (G1 and G2/M) cells (dashed lines in Fig. 4D,E) were subtracted from the respective cyclin A and γH2AX IF of the S-phase selected cells, and the product of the subtraction was used in the Excel regression analysis. A rather strong correlation in the expression of cyclin A and the intensity of γH2AX IF was detected, with correlation coefficients of 0.88 and 0.83 for the cultures that were incubated for 1 or 2 h after labeling with BrdU followed by UV light irradiation, respectively. In the second approach, all measured cells were plotted as the cyclin A versus γH2AX IF distributions (Fig. 4D,E), but only the cells that were positive for cyclin A and γH2AX IF (a feature characteristic of the S-phase cells) were selected (right quadrant) for regression analysis. The regression was also estimated after the subtraction of the threshold values (see above). Similar to what is shown in Figures 4B and 4C, a correlation between cyclin A expression and intensity of γH2AX IF for these cells was detected, with correlation coefficients of 0.82 and 0.85 for cells incubated for 1 and 2 h after pulse of BrdU, respectively. It should be noted the cyclin A–positive and γH2AX IF–negative cells located on these scatterplots (Fig. 4D,E, bottom right) have characteristics of G2 cells, whereas the cyclin A–negative cells (Fig. 4, left quadrants) are typical of G1 and M cells.
It should be stressed that, in all the experiments described above, the cells were tested for the presence of apoptosis-associated H2AX phosphorylation, which is induced by DNA fragmentation by the caspase-activated endonucleases (18). This was done by screening for caspase-3 activation, as we described previously (18). No evidence of caspase-3 activation was detected for up to 4 h of incubation after exposure of BrdU-labeled cells to UV light.
Incorporation of BrdU by individual cells was monitored by selective photolysis of their DNA followed by immunocytochemical detection of γH2AX. The data indicate that the UV light–induced photolysis of DNA, in which thymidine bases were substituted for by BrdU-generated DSBs, whose appearance triggered phosphorylation of histone H2AX. The DNA photolysis was enhanced by the presence of Hoechst 33342, which was added 20 min before cell irradiation. We previously used this strategy to sensitize DNA with halogenated bases to UV-induced photolysis, by using fluorescence resonance energy transfer between the Hoechst 33342 or Hoechst 33358 fluorochrome and the halogen atom, during the development of the SBIP methodology (12, 13). It should be pointed out that, through a series of pilot experiments (data not shown) in which the cells were exposed for 1 to 10 min (0.69 to 6.9 kJ/m2), the dose of 3.45 kJ/m2 was found to be the optimal one, resulting in maximal expression of γH2AX IF and the greatest difference in γH2AX expression between minimally labeled G1 and maximally labeled S-phase cells, and this dose was used throughout the present experiments. The pilot experiments also provided evidence that the inclusion of Hoechst 33342 for 20 min before cell irradiation with UV significantly enhanced induction of γH2AX IF.
Extensive H2AX phosphorylation was already evident 1 h after cell irradiation with UV, well before activation of caspase-3 and the appearance of apoptosis-associated DNA strand breaks. The latter events were observed significantly later, generally 6 h after irradiation (data not shown). Therefore, it is evident that phosphorylation of H2AX was triggered by DNA photolysis rather than by the secondary apoptosis-associated DSBs. As we reported previously (17, 18), DSBs result from DNA fragmentation during apoptosis, which triggers H2AX phosphorylation occur rather late, concurrently to and after but not before, caspase-3 activation. Thus, these data indicate that the UV-induced photolysis of DNA with halogenated bases caused DNA breakage that manifested as a multiplicity of DSBs.
Phosphorylation of H2AX appeared to progress up to 4 h after UV irradiation because the intensity of γH2AX IF continued to increase during this interval (Fig. 2D–F). Therefore, the rate of H2AX phosphorylation had to exceed its dephosphorylation rate. In our previous study, in HL-60 cells treated with the DNA topoisomerase I inhibitor topotecan, we observed significant dephosphorylation that was preventable by the phosphatase inhibitor calyculin A, which occurred as soon as 3 h after induction of DSBs (18). In a Drosophila histone H2AX variant, the decrease in the extent of phosphorylated H2Av was observed as soon as 30 min after induction (22). No attempts were made in the present study to investigate in more detail the respective rates of H2AX phosphorylation and dephosphorylation by using phosphatase inhibitors such as calyculin A or okadaic acid.
By using DNA photolysis as an inducer of DSBs and H2AX phosphorylation (γH2AX IF) as the marker of incorporated BrdU, we observed the progression of a cohort of cells labeled with this precursor during S phase, through G2/M, and their entrance to G1 (Fig. 3). Therefore, this approach, similar to other methods that detect the incorporated DNA precursor and concurrently identify the cell cycle position based on the measurement of cellular DNA content (10, 11), can be used to analyze the rate of cell progression through the cell cycle. However, it should be noted that the cells were irradiated with UV light 1 h before culture termination (assessment of H2AX phosphorylation). The irradiation, particularly of the BrdU-labeled cells, because it causes extensive DNA damage, is expected to activate the cell cycle checkpoints and arrest cell cycle progression, thereby “freezing” the cell cycle distribution at that time point. This event should be accounted for during analysis of cell cycle kinetics. Namely, the time of cell irradiation rather than the time of cell harvesting after the incubation should be considered as the endpoint when estimating the rate of cell cycle progression.
It should be noted that sensitivity of detection of BrdU labeled cells with the SBIP γH2AX IF approach appears to be lower than that of the method that relies on immunocytochemical detection of BrdU after partial DNA denaturation (7–10). This is reflected by the fact that, with the comparable extent of BrdU incorporation (e.g., 1-h pulse labeling with 30 μM BrdU), a logarithmic scale of BrdU IF is generally used on the scatterplots with the latter method (10), whereas the intensity of γH2AX IF is projected on a linear scale (Figs. 2–4). However, the sensitivity of the SBIP γH2AX IF approach appears to be higher than those of methods that rely on quenching of Hoechst 33358 (1–5) or acridine orange (6) fluorescence and comparable to that of the direct SBIP-terminal dUTP nick end labeling assay (12–15).
As with the original SBIB methodology (12, 13), the present approach for the detection BrdU incorporation by photolysis followed by analysis of H2AX phosphorylation was compatible with the concurrent immunocytochemical detection of another protein (cyclin A). This allowed us to carry out multiparameter analysis of γH2AX and cyclin A expressions combined with cellular DNA content analysis (Fig. 4). On the bivariate scattergrams representing cyclin A versus γH2AX IF, a positive correlation was observed between their expressions in S-phase cells. One can assume that the extent of H2AX phosphorylation (intensity of γH2AX IF) in these cells correlates with the number of DSBs. At a constant dose of UV radiation, the frequency of DSBs is expected to depend on the extent of BrdU incorporated into DNA. The latter in turn is likely to depend on the rate of DNA replication, which is known to differ across cells in S phase (15). For these reasons, evidence of a correlation between cyclin A and γH2AX IF (Fig. 4) indicates a correlation between expression of cyclin A and DNA replication rate. When the incorporation of BrdU was measured in a more direct way by the SBIP approach, a very strong correlation (r = 0.98) between these variables was observed for a subpopulation of cells in mid-S phase (15). Such a strong correlation was consistent with cyclin A being a cofactor in the DNA replication machinery, thus limiting the replication rate. Hence, our observation is in concordance with the reported roles of cyclin A/E complexes in driving DNA (23).
Exposure of HeLa cells that were not incubated with BrdU and exposed to UV light led to a significant level of H2AX phosphorylation (Fig. 2C). At that large dose of UV irradiation (3.45 kJ/m2), H2AX phosphorylation was not specific to the cell cycle phase. Because the primary DNA lesions induced by UV light are dipyrimidine photoproducts rather than by DSBs (24) and they are repaired predominantly by the nucleoside excision repair mechanism that generates single-strand DNA nicks and does not involve formation of DSBs (25–27), the detection of H2AX phosphorylation was somewhat unexpected. However, several observations have indicated that repair of DNA damaged by UV light, through involvement of the nonhomologous end-joining repair mechanism (28) or through replication and transcriptional machinery, can be associated with formation of DSBs and H2AX phosphorylation (28–32). In addition to the data shown in Figure 2C, the increased γH2AX IF in the cells that were exposed to UV light with no prior incubation with BrdU was seen by us in numerous experiments on HeLa cells and other cell types. Further, we observed that, at doses of UV irradiation lower than those used in the present study, the H2AX phosphorylation was specific to the cell cycle phase, being much more pronounced in S than in G1 and G2/M cells (unpublished observations). However, it is interesting to note that, in the cells that were incubated with BrdU, the extent of the UV-induced H2AX phosphorylation in most of G1 and G2/M cells was markedly lower than that in the cells that were not incubated with BrdU (Fig. 2D vs. 2C). The processes associated with DNA repair that generate DSBs and/or with H2AX phosphorylation thus appear to be suppressed in the BrdU-treated cells. The mechanism of the suppression is unclear, but it is a subject of an ongoing investigation.
In conclusion, the selective photolysis of BrdU-labeled DNA followed by analysis of H2AX phosphorylation allows one to use this approach to detect the presence of this DNA precursor in the cells and can be combined with the multiparametric analysis of the cell cycle progression and with the concurrent immunocytochemical detection of other proteins. However, because DNA fragmentation that occurs during apoptosis also induces H2AX phosphorylation, caution should be exercised to distinguish formation of primary DSBs induced by photolysis from apoptosis-associated DNA fragmentation.
Malcolm A. King was supported by a Ramsay Health Care Study Fellowship.