Distinct Chk2 activation pathways are triggered by genistein and DNA-damaging agents in human melanoma cells.

Genistein, a natural isoflavone found in soybeans, exerts a number of biological actions suggesting that it may have a role in cancer prevention. We have previously shown that it potently inhibits OCM-1 melanoma cell proliferation by inducing a G(2) cell cycle arrest. Here we show that genistein exerts this effect by impairing the Cdc25C-dependent Tyr-15 dephosphorylation of Cdk1, as the overexpression of this phosphatase allows the cells to escape G(2) arrest and enter an abnormal chromatin condensation stage. Caffeine totally overrides the genistein-induced G(2) arrest, whereas the block caused by etoposide is not bypassed and that caused by adriamycin is only partially abolished. We also report that genistein activates the checkpoint kinase Chk2 as efficiently as the two genotoxic agents and that caffeine may counteract the activation of Chk2 by genistein but not by etoposide. In contrast, caffeine abolishes the accumulation of p53 caused by all the compounds. Wortmannin does not suppress the Chk2 activation in any situation, suggesting that the ataxia telangiectasia-mutated kinase is not involved in this regulation. Finally, unlike etoposide and adriamycin, genistein induces only a weak response in terms of DNA damage in OCM-1 cells. Taken together, these results suggest that the G(2) checkpoints activated by genistein and the two genotoxic agents involve different pathways.

Genistein, a natural isoflavone found in soybeans, exerts a number of biological actions suggesting that it may have a role in cancer prevention. We have previously shown that it potently inhibits OCM-1 melanoma cell proliferation by inducing a G 2 cell cycle arrest. Here we show that genistein exerts this effect by impairing the Cdc25C-dependent Tyr-15 dephosphorylation of Cdk1, as the overexpression of this phosphatase allows the cells to escape G 2 arrest and enter an abnormal chromatin condensation stage. Caffeine totally overrides the genistein-induced G 2 arrest, whereas the block caused by etoposide is not bypassed and that caused by adriamycin is only partially abolished. We also report that genistein activates the checkpoint kinase Chk2 as efficiently as the two genotoxic agents and that caffeine may counteract the activation of Chk2 by genistein but not by etoposide. In contrast, caffeine abolishes the accumulation of p53 caused by all the compounds. Wortmannin does not suppress the Chk2 activation in any situation, suggesting that the ataxia telangiectasia-mutated kinase is not involved in this regulation. Finally, unlike etoposide and adriamycin, genistein induces only a weak response in terms of DNA damage in OCM-1 cells. Taken together, these results suggest that the G 2 checkpoints activated by genistein and the two genotoxic agents involve different pathways.
Genistein, the major isoflavonoid contained in soybeans, is believed to exert a pleiotropic effect including anti-angiogenic, anti-oxidant, and anti-carcinogenic actions (1)(2)(3). Epidemiological studies, as well as work performed on animal models (4,5), suggest that it is responsible for a chemopreventive effect on breast, colon, and skin tumors. Genistein potently inhibits cell proliferation and may also induce cell differentiation or apoptosis (1,3,6). It has been shown to inhibit several tyrosinespecific protein kinases, including pp60 v-Src and epidermal growth factor receptor tyrosine kinases (7). However, it remains unknown whether the anti-proliferative effect of genistein is dependent on these inhibitions.
Genistein has also been reported to be a non-intercalative inhibitor of DNA topoisomerase II, an enzyme that interacts with DNA and catalyzes the concerted breaking and rejoining of the two DNA strands. Indeed, genistein has been shown to inhibit the relaxation activity of topoisomerase II in vitro and to stabilize topoisomerase II-DNA cleavage complexes in vitro as well as in cellulo (8 -11).
Topoisomerase II has been reported to be the primary cellular target for a series of clinically important antitumor agents, including intercalating (i.e. the anthracycline antibiotic, adriamycin, also termed doxorubicin) and non-intercalating (i.e. the epipodophyllotoxin etoposide) compounds. Inhibition of the strand passing activity of topoisomerase II by these compounds is accompanied by stabilization of the topoisomerase II-DNA cleavable complexes, which prevents the rejoining of the two DNA strands (12)(13)(14). Inhibitors of DNA topoisomerase II have been shown to arrest mammalian cells in the G 2 phase of the cell cycle (15,16), even when they do not cause direct DNA damage (17).
Cell cycle checkpoints are biochemical pathways that ensure the orderly and timely progression and completion of critical events such as DNA replication and chromosome segregation. Activation of checkpoints in G 1 and G 2 in response to DNA damage, in S phase upon inhibition of DNA replication, or in mitosis after disruption of the spindle leads to cell cycle arrest. Such delays provide time for repair processes or, in case of severe damage, for the activation of programmed cell death. Defects in the checkpoint regulatory network result in increased sensitivity to damaging agents and to the genomic instability that is often observed in cancer. Paradoxically, checkpoint-evading agents, such as the methylxanthine caffeine, are used in cancer therapy to sensitize cells to killing by genotoxic agents as they override the drug-induced G 2 checkpoint (18,19).
The ultimate target of the G 2 checkpoint signaling pathway is the cyclin-dependent kinase (Cdk) 1 complex Cdk1-cyclin B1 (Cdc2 in Schizosaccharomyces pombe), whose activation depends on the dephosphorylation of Thr-14 and Tyr-15 residues by the dual specificity phosphatase Cdc25C (20,21). This dephosphorylation/activation has been demonstrated to be an absolute requirement for the onset of mitosis. Recently, it has been shown that Cdc25C can be negatively regulated by phosphorylation on its Ser-216 residue, during interphase or in response to DNA damage or incomplete DNA replication (22). Phosphorylation at this residue creates a binding site for 14-3-3 proteins that is believed to be responsible for the nuclear export of Cdc25C and the subsequent impairment of nuclear Cdk1 dephosphorylation. Consistent with this model, expression of a Cdc25C mutant in which Ala was substituted for Ser-216 abrogated the interaction between Cdc25C and 14-3-3 proteins and induced premature entry in mitosis by overriding both replication and DNA damage checkpoints (22). Two checkpoint kinases have been recently identified in humans; these phosphorylate Cdc25C on Ser-216. Chk1 has been proposed to be activated in ␥-irradiated HeLa cells (23), whereas Chk2 (Cds1 in S. pombe) has been reported to be involved both in the replication and DNA damage checkpoints (24). The response to DNA damage occurred in an ataxia telangiectasia-mutated (ATM)-dependent manner (24 -26).
We have recently reported 2 that genistein arrests human melanoma cells in G 2 , as has been shown in a number of other cell types (6,(27)(28)(29)(30). Furthermore, we have demonstrated that genistein impairs the Tyr-15 dephosphorylation of Cdk1 and that Cdk1-cyclin B1 complexes from genistein-arrested cells can be reactivated in vitro by Cdc25 phosphatase. In this paper, we address the question of whether genistein exerts these effects by activating the signaling cascade involving the kinases Chk1 or Chk2.
Cell Cycle Progression Analysis-Cells were plated at a density of 4 ϫ 10 5 cells per 100-mm Petri dish. After 24 h, 60 M genistein, 0.32 M adriamycin, 30 M etoposide, or 1 mM hydroxyurea were added in the absence or presence of 2 mM caffeine for a further 24 h (except for etoposide, which was removed after 1 h). Genistein and etoposide were dissolved in Me 2 SO (final concentration 0.1%). After 24 h, approximately 10 6 cells were harvested by brief trypsinization and centrifuged at 500 ϫ g for 5 min. The cell pellet was washed in PBS and fixed by the gradual addition of ice-cold 70% ethanol. Cells were then labeled with propidium iodide, and the cell cycle distribution was determined by flow cytometry analysis using a Coulter Elite.
Immunoblotting-Whole cell lysates were prepared by directly lysing cells in sample buffer. Proteins (approximately 100 g) were separated by electrophoresis on 7.5 or 10% SDS-polyacrylamide gel electrophoresis. Gels were either stained with Coomassie Blue to control for balanced loading or electroblotted to nitrocellulose membranes (BA85 from Schleicher and Schuell) for 1 h at 20 V using a semi-dry transfer apparatus. Immunoblots were developed by using the ECL detection system (Amersham Pharmacia Biotech) according to the manufacturer's instructions. Mouse monoclonal anti-Cdk1 (SC54) and anti-p53 (clone DO1) antibodies were obtained respectively from Santa Cruz Biotechnology and PharMingen. Rabbit polyclonal anti-phospho-Cdk1 (Tyr-15) and anti-phospho-p53 (Ser-15) antibodies were from Biolabs. The anti-hemagglutinin (HA11) antibody was purchased from Babco.
Overexpression of Cdc25C and Abnormal Chromatin Condensation-HeLa cells were grown on glass coverslips in Dulbecco's modified Eagle's medium supplemented with glutamine, antibiotics, and 10% fetal calf serum. After treatment with 60 M genistein for 24 h, the cells were transfected with the HA-tagged Cdc25C pcDNA using linear ethyleneimine polymers (Exgen 500, Euromedex), according to the instructions of the manufacturer. The transfection efficiency was usually around 10%. Twenty four hours later, the cells were recovered, washed once with PBS, fixed in 3.7% formaldehyde in PBS at 4°C for 20 min, and then washed 3 times with PBS. They were then permeabilized with 0.25% Triton X-100 in PBS for 5 min at room temperature and with 100% cold methanol for 10 min at Ϫ20°C. The cells were then washed 3 times in PBS, 1% bovine serum albumin (fraction V, Euromedex) for 15 min at room temperature, and then incubated for 1 h at 37°C with polyclonal anti-HA antibody (1:400 dilution in PBS, 1% bovine serum albumin). After 3 washes with PBS, the cells were incubated with secondary antibodies (Alexa 594, 1:400 dilution in PBS, 1% bovine serum albumin) for 45 min at room temperature and then in PBS/Hoechst 33342 (1 g/ml) for 10 min at room temperature. After 3 washes with PBS and a 4th wash with water, coverslips were mounted for microscopic observation.
DNA Damage Measurement Using the Alkaline Single Cell Gel Electrophoresis (Comet) Assay-Comet assays were performed according to Singh et al. (31) with slight modifications. After treatment with the different compounds for 1-4 or 24 h, OCM-1 cells were trypsinized. Immediately, 3.5 ϫ 10 4 cells were embedded in 0.35 ml of low melting agarose (1% in distilled water) that was layered onto microscope slides. After solidification of agarose, the slides were immersed in cold lysing buffer (1% sodium sarcosinate, 2.5 M NaCl, 100 mM Na 2 EDTA, 10 mM Tris, pH 10, and 1% Triton X-100 added fresh) for 1-2 h at 4°C. The slides were then removed from the lysing solution, gently rinsed with distilled water, and placed on a horizontal gel electrophoresis unit, which was filled with freshly made alkaline buffer (1 mM Na 2 EDTA, 300 mM NaOH, pH Ͼ12.5). After immersion of the slides for 20 min, electrophoresis was performed in the same buffer for the next 20 min at 2 V/cm. The slides were then gently washed with 0.4 M Tris, pH 7.4, and DNA was stained by immersion in 2.5 g/ml propidium iodide for 10 min at room temperature in the dark. After 4 gentle rinses in distilled water, the slides were mounted for observation with an epifluorescence microscope (LSM 410 invert laser scan microscope, Zeiss).
The DNA damage in each cell was quantified using an image analysis system (Komet 4.0, Optilas system). Twenty five to 76 comets were analyzed for each sample. The "tail moment" accounted for DNA damage and was defined as the product of the percentage of DNA in the comet tail and of the distance between the means of the head and tail DNA distributions (32).

Caffeine Overrides the G 2 Cell Cycle Checkpoint Caused by
Genistein but Not by Etoposide-We first compared the effects of genistein and the DNA-damaging agents adriamycin and etoposide on cell cycle distribution in human melanoma cells. As shown in Fig. 1, treatment with genistein, adriamycin, and etoposide led to a clear-cut G 2 cell cycle arrest in OCM-1 cells. Interestingly, the G 2 block induced by genistein was bypassed by a concomitant treatment of the cells with 2 mM caffeine, whereas the arrest caused by etoposide was not. An intermediate sensitivity to caffeine was observed with adriamycin. These cell cycle events mirrored fairly well the effects of the drugs on cell proliferation (data not shown).
As the status of Cdk1 phosphorylation on Tyr-15 has been shown to be a key event in the control of the G 2 /M transition (the kinase remaining inactive until it is dephosphorylated), we investigated the effects of the different compounds on the phosphorylation of this residue. As shown in Fig. 2, all three agents resulted in an increase in the amount of the slower electrophoretic mobility form of Cdk1 (upper panel). That this band corresponded to the Tyr-15-phosphorylated form was confirmed by using a specific anti-phospho-Cdk1(Tyr-15) antibody (lower panel). Remarkably, the phosphorylation of Cdk1 induced by genistein and DNA-damaging agents was abolished by caffeine (Fig. 2).
Overexpression of Cdc25C Bypasses the Genistein-induced G 2 Arrest-We have previously shown that Cdk1-cyclin B1 complexes from genistein-treated OCM-1 cells can be reactivated in vitro by recombinant Cdc25 phosphatase. 2 We also reported (33) similar results with Cdk1-cyclin B1 complexes isolated from HeLa cells exposed to etoposide. To confirm that the inhibition of Cdk1 was due to the inactivation of the phosphatase in treated cells, we performed transient overexpression of Cdc25C in genistein-arrested HeLa cells. As shown in Fig. 3, when these cells were forced to overexpress Cdc25C, most of them (56%) left the G 2 block and entered an abnormal chromatin condensation stage. In contrast, under our experimental conditions, cells that were not exposed to genistein but that were transfected with Cdc25C did not display any significant sign of chromatin condensation (not shown). These in vivo results strongly suggest that the genistein-induced G 2 block was due to an impairment of Cdc25C activity in genisteintreated cells.
Caffeine Suppresses the Activation of Chk2 by Genistein but Not by Etoposide-Activation of the Chk1 and/or Chk2 kinases has been proposed to be a key event in the DNA damage G 2 checkpoint pathway. Phosphorylation/activation of these kinases can be revealed by the appearance of a slower electrophoretic mobility form (23,24). As shown in Fig. 4a, none of the drugs tested induced any modification of Chk1 electrophoretic mobility. In contrast, genistein activated the checkpoint kinase Chk2, as did etoposide and adriamycin. Indeed, all these agents caused a discrete but significant shift in the electrophoretic mobility of Chk2. Caffeine abolished the genistein-induced and, to some extent, the adriamycin-induced Chk2 activation, but it was unable to counteract the etoposide effect. Whereas caffeine was able to suppress the activation of Chk2 resulting from short term adriamycin treatment (2-4 h, Fig. 4b), it was less efficient in long term treatment (24 h, Fig. 4a).
Genistein and DNA-damaging agents induced p53 accumulation in a somewhat different manner, with adriamycin having a much greater effect, followed by etoposide; genistein treatment led to only a slightly increased level of p53 (Fig. 4a). Similar results were obtained with respect to p21 induction, except that the differences were attenuated (not shown). Caffeine counteracted the accumulation of both p53 (Fig. 4) and p21 (not shown) induced by all the drugs tested.
As shown in Fig. 5a, the effect of genistein on Chk2 activation was rather rapid, as a 1-2-h cell stimulation led to a significant shift of the protein. The maximal response was reached at 4 h and was maintained at least until 24 h. A dose-response study indicated that Chk2 activation could already be observed with 15-30 M genistein and that the max-  Fig. 1 (a) or exposed to the drugs for 4 h (b). Total protein extracts were directly fractionated on polyacrylamide gels and electroblotted proteins were probed with polyclonal anti-Chk1 and anti-Chk2 antibodies as well as with monoclonal anti-p53 antibody. The slowest electrophoretic (activated) form of Chk2 is marked with an asterisk (a). Cont, control.
imal effect was obtained with concentrations of 45-60 M (Fig.  5, b and c). As expected, p53 accumulation occurred during the long term exposure and, interestingly, increased significantly for slightly higher genistein doses than those required to activate Chk2 (60 M instead of 30 M, Fig. 5c). A similar observation was made when the effect of increasing concentrations of adriamycin on Chk2 and p53 was investigated. Indeed, Chk2 was already maximally activated by 0.017 M adriamycin, whereas p53 progressively accumulated when adriamycin concentration increased from 0.043 to 0.34 M (not shown).
Genistein Causes Only Marginal DNA Damage in Melanoma Cells-Because genistein resulted in events similar to those caused by DNA-damaging agents in human melanoma cells (i.e. Chk2 activation, Cdk1 inhibition, and G 2 block) and because genistein has been shown to inhibit DNA topoisomerase II and to stabilize the topoisomerase II-DNA cleavage complexes in other cell types, we wondered whether it could induce DNA damage in OCM-1 cells. Using the so-called "Comet" assay, we compared the ability of genistein, adriamycin, and etoposide to damage DNA in melanoma cells. As illustrated in Fig. 6, genistein was only weakly effective in causing DNA damage compared with the two other drugs. Nevertheless, as shown in Table I, 60 M genistein significantly increased the tail moment parameter after both short term (1-4 h) and long term (24 h) exposure. Higher concentrations (180 M) of genistein did not lead to a major increase (not shown).
Chk2 Activation during G 2 or Replication Blocks Was Wortmannin-insensitive-Genistein caused only slight DNA damage in OCM-1 cells. Because we have shown that it induced transient accumulation of cells in S-phase, 2 we wondered whether the Chk2 activation might also be triggered by a replication block. Accordingly, we investigated the effect of hydroxyurea (HU) on melanoma cells to compare its effect with that exerted by genistein. Fig. 7 shows that HU activated Chk2 after both short term and long term stimulation. Interestingly, caffeine did not counteract Chk2 activation (Fig. 7b) nor did it disrupt the replication block (not shown) caused by HU.
ATM, a member of the PI 3-kinase superfamily, is believed to be the upstream kinase responsible for Chk1/2 activation. High concentrations of wortmannin (the well known PI 3-kinase inhibitor) have been shown to impair the activation of ATM by ␥-irradiation (34). As shown in Fig. 8, in the presence of wort-mannin, Chk2 activation was still detected; in contrast, phosphorylation of p53 on Ser-15 was almost abolished. Thus, despite the fact that phosphorylation of a physiological substrate of ATM was inhibited by wortmannin, the stimulatory effect of all the tested compounds on Chk2 remained unchanged, suggesting that ATM was not involved in that process.

Is the G 2 Checkpoint Activated by Genistein Distinct from
Those Induced by DNA-damaging Agents?-We demonstrate here that genistein exerts its effects on cell cycle somewhat differently from those of the DNA-damaging agents adriamycin and etoposide. (i) Although genistein induced only a weak response in terms of DNA damage in OCM-1 melanoma cells, it caused a complete G 2 cell cycle arrest and was as efficient as the two DNA-damaging agents in activating Chk2. (ii) Caffeine completely overrode the genistein-induced G 2 cell cycle arrest. In contrast, the blocks caused by etoposide and adriamycin were either not bypassed (etoposide) or only partially abolished (adriamycin) in the presence of caffeine.
These results suggest that the G 2 checkpoints activated by genistein and the two DNA-damaging agents are likely to be different in nature. They are compatible with the existence of two different G 2 checkpoint pathways in mammals, as proposed by Downes et al. (17) as follows: one sensitive to DNA damage and the other to a decatenation state of DNA. Indeed, these authors reported that the bisdioxopiperazine ICRF-193, a potent topoisomerase II inhibitor, induced a G 2 cell cycle block without inducing DNA damage. Unlike other inhibitors, which arrest the enzyme at its transition state after DNA-strand breakage, bisdioxopiperazines trap DNA topoisomerase II as a closed protein clamp, with no associated DNA break. Accordingly, although genistein has been demonstrated to be a nonintercalative topoisomerase II inhibitor in several cell types (8 -10), we show here that genistein is a very weak inducer of DNA damage in OCM-1 cells. However, treatment of other cell types with genistein has been shown to result in the occurrence of protein-linked DNA strand breaks (9,11).
Activation of Chk2, but Not Chk1, Occurs during S and G 2 Cell Cycle Checkpoints in Melanoma Cells-Our data demonstrate a close correlation between the activation of Chk2 by genistein or DNA-damaging agents and the induced cell cycle arrest in G 2 . When caffeine was able to override completely the G 2 checkpoint (in genistein-treated cells), it was also found to abolish Chk2 activation. In contrast, in etoposide-treated cells, caffeine was inefficient both in counteracting Chk2 activation and in bypassing G 2 arrest. This close correlation between the two events is further emphasized by the fact that caffeine had an intermediate effect both on Chk2 and G 2 block in adriamycin-treated cells. Moreover, when DNA replication was inhibited by hydroxyurea treatment, caffeine was shown to be unable to counteract either the Chk2 activation or the replication block. At first glance, this is surprising, because caffeine has been shown to cause premature mitosis in hamster fibroblasts arrested in early S phase (35). However, in contrast to hamster cells, it has been shown that the S phase block induced by hydroxyurea in murine and human cells is not released by caffeine (36).
Cdk1 Phosphorylation Status Does Not Systematically Correlate with G 2 Arrest and Chk2 Activation-We show that Cdk1 is phosphorylated on Tyr-15 in OCM-1 cells treated either with genistein or the DNA-damaging agents adriamycin and etoposide. Surprisingly, caffeine treatment led to Cdk1 dephosphorylation in all cases, regardless of its ability to bypass the G 2 block. In particular, caffeine caused dephosphorylation/activation of Cdk1 in etoposide-treated cells, although the block was maintained. Such an inability of caffeine to bypass the G 2 arrest caused by etoposide has been already reported (17). Recently, Toyoshima et al. (37) demonstrated that, in etoposide-treated HeLa cells, Cdk1 was dephosphorylated (and activated) in the presence of caffeine but that cyclin B1 remained cytoplasmic. When cyclin B1 was forced to accumulate in the nucleus either by treatment with leptomycin B, a specific inhibitor of the nuclear export signal-dependent intracellular transport, or by expressing a nuclear export signal-disrupted cyclin B1, caffeine became able to override the G 2 block (37). These results suggest that the etoposide-induced G 2 arrest is under the control of at least two different mechanisms, a caffeine-sensitive mechanism (the inhibition of Cdk1 dephosphorylation) and a caffeine-insensitive, nuclear export signal-mediated transport mechanism (leading to the nuclear exclusion of cyclin B1).
In any case, we have shown that caffeine is unable to disrupt not only the G 2 block but also the Chk2 activation pathway in etoposide-treated cells. Cdk1 was dephosphorylated in these cells, although Chk2 was still activated. This could suggest that Chk2 is not really involved in the regulation of Cdk1. One can rather speculate that Cdk1 is mainly cytoplasmic in these cells due to the impairment of nuclear localization of cyclin B1. As a consequence, the kinase might be dephosphorylated in the cytoplasm by Cdc25C that is itself prevented from localizing in the nucleus because of the maintained Chk2 activation.
Is the Activation of Chk2 ATM-independent?-In this study we show that wortmannin does not inhibit the Chk2 activation caused by all the drugs we have tested. This is rather surprising with regard to the agents causing important DNA damage like adriamycin and etoposide, since the PI 3-kinase inhibitor has been previously reported to abolish the activation of Chk2 caused by ␥-irradiation in HeLa cells (34). This inhibition was believed to reflect the ATM dependence of Chk2 activation (24 -26). As the concentration of wortmannin used in our experiments was shown to inhibit the phosphorylation of p53 on Ser-15, the Chk2 activation we observed was very likely ATMindependent. Indeed, the phosphorylation at that specific site has been reported to reflect cellular ATM activity (38 -40).  Although ATM-and Rad3-related (ATR) and DNA-dependent protein kinase are also candidates for phosphorylating p53 on Ser-15, there is evidence suggesting that, in vivo, this phosphorylation is primarily under the control of ATM. Indeed, DNAdependent protein kinase has been reported not to phosphorylate p53 in cellulo (41,42), and p53 has been reported to be a poor substrate for ATR compared with ATM (40).
Our results indicate that distinct pathways might be involved in the activation of Chk2 depending on the nature of DNA-damaging agents. They are reminiscent of reports indicating that the activation of Chk2 in response to DNA replication inhibitors is ATM-independent (25,26). This could indicate that Chk2 activation involves other members of the PI 3-kinase superfamily. We propose a model that takes these features into account (Fig. 9). As the Chk2 activation caused by genistein is wortmannin-insensitive but caffeine-sensitive, we postulate the involvement of an ATM homolog (X in Fig. 9), such as ATR. Interestingly, ATR has been shown to be much less sensitive (20-fold) to wortmannin than ATM (43), but at least in vitro, ATR was inhibited by caffeine in the millimolar range (38). From this point of view, it is interesting to note that if ATM plays a key role in the response to ionizing radiations, ATR has been proposed to be an important component of multiple DNA damage response pathways and has also been suggested to be involved in the DNA replication checkpoint (44,45). However, our results do not support an implication of ATR in the HUinduced Chk2 activation, since we found it to be insensitive not only to wortmannin but also to caffeine. The same is true for the Chk2 activation caused by etoposide. We propose in such cases the involvement of an unknown homolog of ATM/ATR (Y in Fig. 9) that is insensitive to both wortmannin and caffeine. In any case, the difference between the biochemical pathways activated by ␥-irradiation and other DNA-damaging agents (as well as DNA replication inhibitors) was further emphasized by the recent report that caffeine abolished the Chk2 activation caused by ionizing radiations (46).
In contrast with distinct effects on Chk2, we have shown that caffeine strongly decreases the accumulation of p53 induced by all the compounds tested. Consistent with this observation, caffeine was found to inhibit the p53 phosphorylation on Ser-15 (not shown). These results and those obtained with wortmannin suggest a dissociation between Chk2 activation and p53 phosphorylation/accumulation. As discussed above, ATM is very likely the main kinase that phosphorylates p53 on Ser-15 in response to DNA damage. As a consequence, we propose that the different compounds we have studied induced phosphorylation/stabilization of p53 via ATM (Fig. 9). Since we have also demonstrated the ATM independence of the Chk2 activation caused by these agents, this suggests that ATM is unable to activate Chk2 in cellulo by direct phosphorylation.