Hedamycin, a DNA alkylator, induces γH2AX and chromosome aberrations: Involvement of phosphatidylinositol 3-kinase–related kinases and DNA replication fork movement

Treatments that induce replication stalling, albeit by diverse mechanisms, such as UV (1), hydroxyurea (2), or camptothecin (3), typically activate an ataxia and Rad-related kinase (ATR) to induce a phosphatidylinositol 3-kinase–related kinase (PIKK)–regulated checkpoint response (4) and also induce DNA breaks (5). Generally, loss of ATR impedes the DNA damage response to DNA double-strand breaks (DSB) and can result in increased sensitivity to agents that arrest replication because cell division can occur in the presence of damaged DNA (6, 7). Ultimately, failure to suppress cell cycle progression in the presence of sustained DNA DSBs can generate chromosomal abnormalities associated with genomic instability (8).

Following induction of DSBs, H2AX, a variant of a core histone, is almost immediately phosphorylated by PIKKs on Ser¹³⁹ at its COOH-terminal tail and rapidly localizes to sites of breaks (9, 10) and promotes recruitment of other DNA damage response proteins, such as Mre11, Rad51, and Nbs1 (11–13). Because γH2AX formation is observed within seconds of induction of DNA breaks and the level of phosphorylation and intensity of immunolabeling increase linearly with the amount of damage, it is now a widely accepted marker of DSBs (9, 10).

DNA lesions, whether induced by depletion of deoxynucleotide triphosphate pools or photoproducts, activate an intra-S-phase checkpoint response, resulting in the blockage of late-firing origins and stabilization of stalled replication forks (14). However, this intra-S-phase checkpoint is unable to stop fork progression of early-firing origins, which were triggered before checkpoint activation (15, 16). Consequently, collisions between early-firing replication forks and hydroxyurea or UV-induced DNA lesions can result in DSBs and γH2AX formation and its nuclear foci (17). Unlike ionizing radiation, H2AX phosphorylation is regulated prominently by ATR, consistent with its role in sensing stalled replication forks, such as those generated by UV type lesions (17, 18).

The cellular responses to DNA-targeted alkylating drugs that alter DNA structure and impede its replication while indirectly inducing DNA DSBs remain relatively unknown. Typically, ATR activation is also involved in response to various types of bulky DNA adducts that lead to arrested replication forks (19). However, growing evidences revealed that lesion recognition is not sufficient for ATR activation and that single-strand DNA, resulting from replication or DNA repair process, is required for recruiting checkpoint proteins to damaged DNA (7, 20). Furthermore, based on H2AX phosphorylation, many nonradiomimetic DNA-damaging agents that do not directly induce breaks induce DSBs stemming from encounters of replication forks with preexisting nicks, gaps, or adducts (5, 21). For example, camptothecin, which induces cleavable complexes of topoisomerase I and DNA, also induces DNA breaks resulting from collisions between advancing replication forks and these complexes (22, 23). Similarly, DNA adducts induced by the monofunctional DNA alkylator adozelesin are not detected by the checkpoint response machinery until collisions of replication forks with the DNA lesions result in DSBs (24). For both agents, H2AX phosphorylation is suppressed by aphidicolin, a DNA polymerase inhibitor that blocks replication fork movement (23, 24).

We recently delineated the DNA damage responses induced by a pluramycin antibiotic, hedamycin (25). Hedamycin, like adozelesin, is a highly cytotoxic monofunctional alkylator (the concentration of drug required to kill 90% of the cells in a clonogenic survival assay is 0.38 nmol/L), but it also significantly distorts the DNA helix by threading intercalation (26). Like other DNA-damaging agents, hedamycin rapidly and strongly blocks DNA replication (25). However, it is not known whether its DNA adducts and stalled replication forks are converted into DNA strand breaks. One would speculate that hedamycin-induced DNA damage responses would be similar to other agents that induce bulky adducts; yet unexpectedly, DNA damage responses of hedamycin exhibit several atypical characteristics. For example, it does not induce G₂ arrest but primarily S-phase accumulation and G₁ arrest (25). What is particularly striking is that, unlike adozelesin or camptothecin, the cytotoxicity and cell cycle effects of hedamycin are not affected by the loss of ATR (25).

This study revealed that, whereas hedamycin induces DNA strand breaks and γH2AX formation, ATR suppression does not diminish either. Moreover, blockage of replication fork movement significantly reduced neither DNA breaks nor γH2AX formation. Hedamycin-induced DNA breaks are accompanied by chromosome abnormalities likely resulting from nonhomologous end joining (NHEJ). Hedamycin seems to be a unique agent for exploring cellular DNA damage response mechanisms that are associated with replication stalling and induction of DNA breaks and chromosomal aberrations.

The HCT116 human colon carcinoma cell line (a gift from Dr. B. Vogelstein, Johns Hopkins Oncology Center, Baltimore, MD) was grown in McCoy's medium supplemented with 10% fetal bovine serum. Human glioma cell lines M059K [wild-type and DNA-protein kinase (DNA-PK) +] and M059J (DNA-PK−) were cultured in DMEM/F-12K (1:1) supplemented with 10% fetal bovine serum and 2× nonessential amino acids. Human lymphoblast cell lines GM00536B [wild-type and ataxia telangiectasia mutated (ATM) +] and GM01526E (ATM−) were maintained in RPMI supplemented with 15% heat-inactivated fetal bovine serum. Both glioma and lymphoblast cell lines were purchased from Coriell Institute for Medical Research (Camden, NJ), and all cell lines were kept at 37°C in a 5% CO₂ incubator.

Hedamycin was provided by the National Cancer Institute. Hedamycin stock solutions (10 mmol/L) were prepared in water, aliquoted, and stored at −20°C. Subsequent dilutions were made in growth medium. Adozelesin (supplied by Pfizer, Inc., New York, NY) was dissolved in dimethylacetamide (2 mg/mL) and further diluted in DMSO to a final concentration of 1 nmol/L. Camptothecin, aphidicolin, caffeine, colcemid, Hoechst 33258, and ethidium bromide were purchased from Sigma Co. (St. Louis, MO). Camptothecin was aliquoted and stored frozen at 10 mmol/L in DMSO and further diluted in growth medium to a final concentration of 1 μmol/L before use. Aphidicolin was stored frozen at 1 mg/mL in ethanol and further diluted in growth medium before use. Caffeine was dissolved in growth medium at a stock concentration of 80 mmol/L before use. UV was given with a Stratalinker (Stratagene, La Jolla, CA) after cells were washed once with PBS. Oligofectamine was purchased from Invitrogen (Carlsbad, CA). ATR small interfering RNA (siRNA; ref. 27) was purchased from Dharmacon Research, Inc. (Lafayette, CO).

Anti-γH2AX and anti–proliferating cell nuclear antigen (PCNA) mouse monoclonal antibodies were purchased from Upstate Biotechnology, Inc. (Altham, MA). Anti-ATR goat polyclonal antibody was purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Anti-Ku80 monoclonal antibodies were purchased from Sigma. Texas red–conjugated goat anti-mouse IgG was purchased from Pierce (Rockford, IL).

Cells were treated with hedamycin for 4 hours and either harvested directly or incubated in drug-free medium for another 24 hours after drug removal before harvesting for further evaluation. In some experiments, cells were pretreated with 16 mmol/L caffeine for 30 minutes or 0.5 μmol/L aphidicolin for 5 minutes before addition of hedamycin. If not indicated, cells were harvested 1 hour after exposure to 10 J/m² UV, 4 hours after exposure to 5 nmol/L hedamycin, or 3 hours after exposure to 1 μmol/L camptothecin.

Cells were harvested by trypsinization and resuspended in PBS (28). Cell suspensions (1 × 10⁵ cells) were mixed with 0.5% LMP agarose and spread on 1% agarose-precoated slides. After 30-minute incubation at 4°C, cells were lysed for 1 hour in alkaline buffer [2.5 mol/L NaCl, 0.1 mol/L EDTA, 10 mmol/L Tris-HCl (pH 10), 1% sarcosyl, 10% DMSO, 1% Triton X-100 (pH 10)]. After 20-minute equilibration in electrophoresis buffer [0.3 mol/L NaOH, 1 mmol/L EDTA (pH 13)], cells were subjected to electrophoresis at 28 V for 25 minutes under alkaline conditions at 4°C. After electrophoresis, slides were neutralized with 0.4 mol/L Tris (pH 7.4), fixed with methanol and ethanol, and dried overnight. Cells were then stained with 1 μg/mL ethidium bromide and analyzed using a fluorescence microscope Olympus BX40 (Melville, NY) with a Spot-RT digital camera and software (Webster, NY). At least 50 cells were evaluated per slide. Visual scoring of comet images using fluorescence microscopy has been described.³

Oligofectamine was used for transfection of siRNAs into HCT116 cells according to the protocols provided by Dharmacon. Expression of ATR was monitored by Western blotting. Following suppression of ATR (48 hours after transfection with siRNA), cells were exposed to hedamycin or UV irradiation.

Cells were washed twice with PBS and lysed at 4°C in a lysis buffer [50 mmol/L HEPES (pH 7.4), 4 mmol/L EDTA, 2 mmol/L EGTA, 1% Triton X-100, 50 mmol/L phenylmethylsulfonyl fluoride, 20 μg/mL leupeptin, 100 mmol/L Na₃VO₄]. Lysates were centrifuged at 14,000 rpm for 20 minutes. The protein concentration in the lysates was determined using the Bio-Rad (Hercules, CA) protein assay kit. Cell lysates were subjected to SDS-PAGE electrophoresis and transferred to a polyvinylidene difluoride membrane (Hybond-P, Amersham Pharmacia Biotech, Piscataway, NJ). The membrane was blocked with 4% nonfat milk in TBS [25 mmol/L Tris (pH 7.4), 125 mmol/L NaCl] for 1 hour at room temperature followed by incubation with primary antibodies against γH2AX, ATR, or PCNA. The membranes were washed thrice with TBS and then incubated with horseradish peroxidase–conjugated anti-goat or anti-mouse IgG for 1 hour. The immunoblot was then washed five times with TBS and developed using an enhanced chemiluminescence Western lightning kit (Perkin-Elmer, Boston, MA). Experiments were repeated at least thrice.

Cells were grown on a coverslip, which was precoated with fetal bovine serum for 30 minutes before cell seeding. After drug treatment, the cells were washed with PBS, fixed with 2% formaldehyde for 20 minutes at room temperature followed by permeabilization with acetone for 3 minutes at −20°C, and then washed with PBS. The primary antibodies anti-γH2AX (1:300) or anti-Ku80 (1:50,000) monoclonal antibody were diluted in PBS containing 1% bovine serum albumin, applied to the coverslip, and kept overnight at 4°C. Unbound primary antibodies were washed off with PBS, and Texas red–labeled (1:600) secondary antibody with 10 mmol/L Hoechst 33258 (1:500) was applied for another 1 hour at room temperature. After a series of washes, the stained cells were mounted on slides, viewed using appropriate filters, and photographed by fluorescence microscopy with a Spot-RT digital camera. For Ku80 foci detection, cells were extracted with a lysis buffer containing 50 mmol/L HEPES (pH 7.4), 150 mmol/L NaCl, 1 mmol/L EDTA, and 0.5% NP40 supplemented with 10 mmol/L NaF, 1 mmol/L Na₃VO₄, and 10 μg/mL leupeptin for 30 minutes at 4°C and washed once with PBS before fixation. Each treatment was duplicated and the experiment was repeated at least thrice.

HCT116 cells were treated with hedamycin for 4 hours followed by removal of free drug and further cultured in drug-free medium for 24 hours. At 2 hours before harvest, colcemid was added to a final concentration of 50 ng/mL. Cells were harvested by trypsinization, treated with hypotonic KCl solution (75 mmol/L) for 7 minutes, and fixed with methanol/glacial acetic acid (3:1). Slides were prepared by standard techniques and chromosomes were stained with 4′,6-diamidino-2-phenylindole and inverted images were processed using Applied Spectral Imaging's EasyFISH software (29). One hundred metaphases per treatment condition were evaluated. Because untreated normal cells often display one to two gaps and/or breaks (mostly chromatid type), cells with more than two gaps/breaks were considered as metaphases with aberrations.

DSBs stemming from replication arrest, such as those generated by treatments that induce bulky DNA lesions, such as UV and camptothecin, are generally detected by their ability to trigger H2AX phosphorylation and focalization at DNA damage sites (17, 30–32). To test if hedamycin, which also induces bulky DNA adducts, induces DNA DSBs based on phosphorylation of H2AX (γH2AX), cell lysates from HCT116 cells treated with hedamycin for 4 hours were analyzed by Western blotting. Figure 1A shows that the phosphorylation level of H2AX was increased in a concentration-dependent manner. A minimum induction was detected even at subnanomolar concentration (0.1 nmol/L) and phosphorylation levels approach a plateau at concentrations above 2.5 nmol/L. Whether phosphorylation of H2AX was accompanied by γH2AX foci formation was examined by immunofluorescence under the same treatment condition. γH2AX was stained with γH2AX antibody and chromosomes were stained with Hoechst 33258 (Fig. 1B shows representative images). The intensity of γH2AX foci seemed consistent with its phosphorylation levels.

The phosphorylation of H2AX in response to replication arrest-mediated DNA DSBs is dependent on the PIKK, ATR (33). To test the ATR dependency for hedamycin-induced phosphorylation of H2AX, siRNA interference was used to reduce ATR levels in HCT116 cells. Based on Western blotting, decreased levels of ATR were observed at 48 hours after transfection of a siRNA against ATR (Fig. 2A). Although H2AX phosphorylation levels increased albeit to a small degree in untreated siRNA-transfected cells, hedamycin-induced phosphorylation of H2AX was substantially increased compared with mock-transfected cells (Fig. 2A). Another cellular protein PCNA, involved in DNA replication, was unaffected by either drug treatment or siRNA transfection. The unanticipated increase in hedamycin-induced H2AX phosphorylation in the ATR-reduced cells seemed unrelated to the method of achieving ATR reduction or to the nature of HCT116 cells, because with UV, as shown by others, H2AX phosphorylation was greatly reduced in the absence of ATR compared with mock-transfected cells (Fig. 2B).

It was unknown whether the diminished ATR levels, which lead to an increase in H2AX activation, were accompanied by an increase of DNA strand breaks. Similarly for UV, the level of DNA breaks could also increase despite the drop in H2AX phosphorylation. To confirm these possibilities, DNA strand breaks were measured in a single-cell assay using alkaline comet analysis to detect both single-strand breaks and DSBs [for hedamycin, parallel neutral comet assays were also carried out and resulted in similar damage scores (at 5 nmol/L was 0.75 versus 0.8 in neutral and alkaline comet assay, respectively)]. In the absence of strand breaks, cellular DNA maintains a compact form during electrophoresis, whereas when DNA strand breaks are induced fragmented DNA migrates away from the nucleus and forms comet tails during electrophoresis. The level of drug-induced DNA strand breaks can be quantitated on a scale from 0 to 4 based on the length of the comet tails and expressed as an average level of DNA damage per cell as described in Materials and Methods (Fig. 2C). Following a 4-hour treatment, hedamycin-induced breaks are increased under ATR suppression condition, consistent with the increase of H2AX phosphorylation. In contrast, whereas UV-induced strand breaks also increased, H2AX phosphorylation decreased, suggesting that, unlike hedamycin, H2AX activation under conditions of ATR suppression does not reflect DNA strand breaks (Fig. 2C).

Because suppression of ATR did not reduce phosphorylation of H2AX by hedamycin, we next examined whether H2AX phosphorylation would be altered in ATM and DNA-PK mutant cell lines. We also tested camptothecin, which is known to induce γH2AX in an ATR- and DNA-PK-dependent but ATM-independent manner to verify that our assay system produces similar findings (23). As expected, in response to camptothecin in ATM-deficient lymphoblasts, the level of H2AX phosphorylation was similar to that in wild-type cells (GM00536B). In contrast, hedamycin-induced phosphorylation of H2AX like in ATR-deficient cells increased in ATM-deficient compared with wild-type cells (Fig. 3A).

In DNA-PK-deficient glioblastoma cells (M059J), as shown by others, camptothecin-induced phosphorylation of H2AX was reduced compared with wild-type cells (M059K; Fig. 3B). However, hedamycin-induced phosphorylation of H2AX was relatively similar in both cell lines.

It seems that, unlike other agents that arrest replication, any one of the three PIKKs does not reduce H2AX phosphorylation in response to hedamycin. To test whether H2AX phosphorylation would occur when the activity of multiple PIKKs are reduced, HCT116 cells were pretreated with 16 mmol/L caffeine for 30 minutes, which is sufficient to significantly reduce ATM and ATR activity before drug exposure (34). Figure 3C shows that, as expected, camptothecin-induced phosphorylation of H2AX was reduced by caffeine pretreatment. In contrast, hedamycin-induced phosphorylation of H2AX was increased compared with that of non-caffeine-treated cells. Similar results were also obtained when HCT116 cells were pretreated with wortmannin, which would effectively block ATM and DNA-PK activity (data not shown).

To further test how lack of PIKKs effects hedamycin-induced H2AX phosphorylation, DNA-PK-deficient M059J cells, which are known to also have low ATM levels, were transfected with a siRNA against ATR. At 24 hours after transfection, cells were treated with hedamycin for 4 hours. Figure 3D shows that hedamycin-induced H2AX phosphorylation was not decreased but increased under a condition where cells lacked or contained reduced levels of each of the PIKKs. Treatment of M059J cells with caffeine to reduce ATM/ATR activity also resulted in no reduction, but again an increase of hedamycin induced H2AX phosphorylation (Fig. 3E).

Bulky DNA adducts, such as those induced by camptothecin and adozelesin, are generally converted into DSBs when active replication forks collide with drug adducts, leading to γH2AX formation (23, 24). To determine the dependency of hedamycin-induced γH2AX on active replication fork movement, HCT116 cells were pretreated with aphidicolin for 5 minutes to stall replication forks (as expected, [³H]thymidine incorporation was reduced by >90%; data not shown) and then incubated with either hedamycin or adozelesin for 1 hour or camptothecin for 3 hours. Subsequently, cell lysates were prepared and analyzed for H2AX phosphorylation by Western blotting. In agreement with other studies (23), the levels of H2AX phosphorylation induced by camptothecin and adozelesin in the presence of aphidicolin were significantly reduced 75% and 84%, respectively (Fig. 4A). In contrast, hedamycin-induced phosphorylation of H2AX was only moderately affected (30% reduction) in the presence of aphidicolin. We also examined γH2AX focalization under the same treatment condition (Fig. 4B shows representative images). Consistent with the Western blot results, adozelesin-induced γH2AX foci were substantially decreased in the presence of aphidicolin. In contrast, in hedamycin-treated cells, γH2AX foci were notably less dependent on replication fork movement. In total, these results show that hedamycin-induced strand breaks are not strongly dependent on DNA replication fork movement.

Replication-mediated DSBs are generally repaired after release from replication stress (35). To test if hedamycin-induced DNA strand breaks persist, hedamycin was removed from HCT116 cells following a 4-hour drug treatment and cells were incubated in drug-free medium for another 24 hours. Cells were then collected and subjected to alkaline comet analysis (whereas alkaline comet analysis provides a sensitive way to detect both single-strand breaks and DSBs, induction of DSBs by neutral comets revealed similar profiles). As shown in Fig. 5A and B, more extensive comets appeared at all levels of hedamycin treatments compared with 4-hour treatments (data not shown). For example, the DNA strand break score increased from 0.2 (4 hours) to 1.5 (24 hours) at 0.25 nmol/L and from 0.8 (4 hours) to 2.1 (24 hours) at 1 nmol/L. Notably, the extent of strand breaks reached a peak at 1 nmol/L and actually declined by 5 nmol/L.

Hedamycin-induced DNA strand breaks seem to persist in interphase nuclear DNA. It is possible that such unrepaired strand breaks would lead to chromosome aberrations in mitotic cells (36). To test if hedamycin-induced DNA damage is associated with chromosomal aberrations, hedamycin was removed from HCT116 cells after a 4-hour drug treatment and cells were incubated in drug-free medium for another 24 hours. At 2 hours before harvest, the mitotic cell population was enriched by the addition of colcemid and then analyzed for chromosomal aberrations using spectral karyotyping analysis. Images reveal that hedamycin induces a significant amount of chromosomal misjoining and fragmentation events compared with untreated cells (Fig. 6). The most frequent types of chromosomal aberration induced by hedamycin were chromatid/chromosome breaks, gaps (Fig. 6A, c-f, arrowheads), end joining (Fig. 6A, c and d, open arrows), and nonhomologous recombination forming quadriradial chromosomes (Fig. 6A, c-f, arrows). At 0.25 nmol/L, the major type of aberrations was end-to-end association.

Table 1 summarizes the effects of hedamycin on the numbers and types of chromosomal aberrations. Chromosomal aberrations increased with drug concentration, and profound chromosomal aberrations occurred at 1 to 2.5 nmol/L where cells progressing through S phase were slowed and the highest levels of interphase DNA breaks were detected by comet assays. Although the magnitude of DNA damage reached near saturation at >1 nmol/L, it is of special interest to note that only 21% of mitotic cells contained chromosome aberrations. However, when the drug concentration was increased to 1 to 2.5 nmol/L, >60% of the cells with damaged chromosomes contained gaps/breaks and fragmentation (Fig. 6B; Table 1).

The significant numbers of NHEJ events detected by spectral karyotyping analysis suggested that proteins involved in NHEJ repair might be recruited to the site of hedamycin-induced DSBs. Ku80 is a DNA end-binding protein that preferentially interacts with chromosomal breaks (e.g., ionizing radiation induced), which not only protect DNA ends from nuclease attack but also is an essential component of the NHEJ machinery used in DSB repair (37).

Because Ku80 is an abundant protein that distributes in both nuclear and cytosolic fractions, detection of distinct nuclear foci requires detergent extraction (38, 39). HCT116 cells treated with hedamycin were extracted to remove nucleoplasmic protein while retaining detergent-resistant chromatin-bound Ku80. Localization of Ku80 was examined using immunofluorescence microscopy. Figure 7 shows representative images. In untreated control cells, Ku80 aggregated in the nucleolus but did not tightly associate with chromatin. Following 1 nmol/L hedamycin treatment, a concentration that induced significant NHEJ events, most of the Ku80 was redistributed and tightly bound to chromatin while being lost from the nucleolus. Like γH2AX, H2AX, the Ku80 foci persisted at the DNA damage site because it was also observed at both the 4-hour treatment and even at 24 hours after drug removal.

Recent studies suggest that PIKK-regulated DNA damage checkpoint responses to DNA adducts may be triggered by the subsequent generation of single-strand DNA or DNA breaks resulting from sustained DNA replication arrest (7, 40). This study revealed that hedamycin is similar to other treatments, such as UV or camptothecin, which induce bulky DNA lesions accompanied by DNA strand breaks. Initially, UV and camptothecin bulky DNA lesions activate chk1 via ATR and subsequently, also through ATR, activate chk2 to cope with strand breaks (41, 42). Similarly, hedamycin under conditions that induce strand breaks also induces checkpoint protein phosphorylation, such as chk1 and chk2 (25), suggesting that DNA strand breaks likely contribute to the DNA damage responses. Following DSB induction, H2AX undergoes phosphorylation at a conserved PIKK motif before focalization at DNA break sites (43).

Replication stress induced by diverse genotoxins, including hydroxyurea, camptothecin, and monofunctional alkylating agents, such as adozelesin, and methylmethanesulfonate all generate DSBs based on induction of the DNA damage response protein γH2AX (17, 23, 24). Western and immunofluorescence microscopy analyses revealed that γH2AX also appears in hedamycin-treated cells and increases in parallel with DNA breaks, suggesting that hedamycin-DNA adducts are also converted to DSBs (Fig. 1).

Typically, H2AX is phosphorylated by ATR in response to replication-mediated DSBs, such as those induced by UV and camptothecin (33). However, despite the many similarities to these agents, hedamycin-induced DSBs result in extensive phosphorylation of H2AX when ATR is suppressed. Hedamycin-induced H2AX phosphorylation actually increases under ATR suppression conditions, which is in striking contrast to many other genotoxins, such as UV, hydroxyurea, and camptothecin, which all show significant decreases. Moreover, for hedamycin, the assessment of DNA strand breaks by comet assay in ATR-reduced cells (Fig. 2C) parallels the increase in γH2AX. Intriguingly, UV-induced breaks, as determined directly by comet analysis, also increased in ATR-reduced cells (Fig. 2C), although we observed like others a marked reduction of H2AX phosphorylation (Fig. 2B). Although H2AX phosphorylation is widely used as a measure of induction of DSB, our analysis indicates its limitations. For hedamycin, DSBs and H2AX are concordant, but when ATR is required to activate H2AX, such as the case with UV, it is difficult to distinguish whether decreased γH2AX stems from loss of signaling or truly reflects diminished DSBs. Overall, when ATR is defective, replication may be less likely to stop before a collision with a DNA lesion and/or stalled replication fork collapse, both of which can elevate DNA breaks (44).

Loss of other individual PIKKs, ATM and DNA-PK, and suppression of PIKK activity by high concentrations of caffeine also do not decrease hedamycin-induced γH2AX (Figs. 2 and 3). Currently, we are evaluating whether kinases other than ATM, ATR, and DNA-PK, such as the recently described hSmg-1 (ATX), plays a role in regulating hedamycin.

Characteristically, replication is required for DSB generation, such as with camptothecin and adozelesin, because breaks are generated as a consequence of collisions between sites of stalled replication and advancing forks (23, 24). In these cases, the addition of aphidicolin and inhibitor of DNA chain elongation nearly eliminates H2AX phosphorylation and its focalization. Although hedamycin is unlike these genotoxic agents in that ATR is not the predominant PIKK responding to DNA adduct, it was unexpected that H2AX phosphorylation and focalization were also rather independent of fork movement (Fig. 4). Consistent with the finding that aphidicolin does not significantly prevent hedamycin-induced γH2AX, the strong intensity of γH2AX foci in >90% of hedamycin-treated cells after 4-hour treatment suggests that hedamycin-induced DSBs, unlike camptothecin and adozelesin, are not limited to S-phase cells. However, treatment of G₁-S synchronization cells with hedamycin resulted in an elevation of strand breaks and γH2AX compared with asynchronous cultures (data not shown), indicating perhaps that ongoing replication can still contribute to the level of DNA breaks.

Our previous study reported that hedamycin-induced DNA damage checkpoint activation, including chk1, chk2, and p53 phosphorylation, is maintained for at least 24 hours after drug removal (25), suggesting that hedamycin lesions may be persistent. The high levels of hedamycin-induced strand breaks (Fig. 5) observed 24 hours after drug removal are consistent with both the DNA lesions being refractory to removal and a sustained DNA damage checkpoint response that triggers a significant decrease in S-phase progression. Interestingly, the extent of strand breaks somewhat diminished as the drug concentration was raised from 1 to 5 nmol/L (Fig. 5). Previously, we showed that 1 nmol/L hedamycin is not sufficient to promote cdc25A degradation after a 4-hour treatment (25), allowing cells to enter S phase and replicate damaged DNA, which would likely contribute to strand breaks. By contrast, hedamycin at 5 nmol/L promotes cdc25A degradation, blocking entry into S phase to minimize the potential of generating breaks due to replication of damaged DNA. Also, damaged DNA in cells progressing through S phase would be subject to collision with adducts, stalled replication fork collapse, and induction of repair intermediates under conditions where many of the cells cannot complete the transit out of S phase (21, 35). In contrast, cells treated at 5 nmol/L hedamycin never exit from the G₁ arrest and strand breaks would more likely result from accumulation of repair intermediates.

Possibly, hedamycin-induced apoptotic responses could also contribute to DNA strand breaks (25). However, apoptotic contribution to DNA breaks likely is minimal. First, significant levels of DNA strand breaks are observed at concentrations that induce barely detectable levels of apoptosis. Second, cells displaying γH2AX foci were not undergoing apoptosis based on microscopic examination. Moreover, γH2AX does not form foci in apoptotic cells (45). Third, fragmented DNA in the form of DNA laddering that is associated with apoptosis was not observed when DNA strand breaks were analyzed using pulse-field electrophoresis.⁴

The persistently elevated levels of DNA damage observed in interphase cells may translate into chromosomal aberrations (46). Elevated chromosomal aberrations were observed in those hedamycin-treated cells able to progress into metaphase. Among the chromosomal aberrations induced following moderate DNA damage (e.g., Fig. 6A, a-d), hedamycin induces a predominance of NHEJ events, indicating that NHEJ was a mechanism activated in the repair of hedamycin-induced DSBs. Notably, hedamycin produces many quadriradial chromosomes, composed from two to three chromosomes, which had been reported only in bifunctional DNA cross-linker–treated cells (47). Collectively, these results imply the following: (a) that some of the initial damage induced by hedamycin was effectively repaired during the cell cycle phase, which could promote cell entry into mitosis; (b) that end-to-end association forming quadriradial chromosomes stem from some form of repair response; and (c) that excessive DNA damage can overwhelm the capacity of repair systems resulting in gaps/breaks and fragmentation of chromosomes.

Evidence supporting a role for NHEJ in processing hedamycin-induced DSBs is that enhanced Ku80 relocalization and focalization in tight-binding chromatin-bound complexes was detected following short (1 hour) and long-term (24 hours) hedamycin treatments (Fig. 7). Ku80 is a DSB sensor, which subsequently conducts DSB repair through activation of DNA-PK and recruitment of ligase IV (48). Once repair is complete, tight binding of Ku80 should diminish. Notably, for hedamycin, Ku80 foci continued to increase over time and associate with drug-induced DSBs, consistent with the idea that the misjoined chromosomes observed in metaphase cells could have resulted from a failure to complete NHEJ. Alternatively, Ku80/DNA-PK can activate a p53-mediated apoptotic pathway when repair failure leads to persistence of DSBs (49), yet hedamycin does not induce significant levels of apoptosis (25).

Hedamycin induces DNA damage responses that are atypical of most genotoxic agents examined to date. Studies are currently under way to assess how hedamycin interferes with DNA replication and to determine whether its mechanism of disruption of replication relates to its unique DNA damage response.

We thank Dr. Gokul Das for the advice and comments on the article.