Inhibition of Chk1 by CEP-3891 Accelerates Mitotic Nuclear Fragmentation in Response to Ionizing Radiation

In response to DNA-damaging agents such as ionizing radiation (IR), cells activate cell cycle checkpoints to delay cell cycle progression. Recently, much progress has been made toward understanding the signaling cascades responsible for induction of these checkpoints. The aim of the present study was to develop new tools to abrogate IR-induced checkpoints and explore the effects of checkpoint abrogation on cell survival.

Recent data showed that human Chk1 is required for both the IR-induced rapid S and G₂ checkpoints (1, 2, 3, 4). Chk1, which was known to be involved in the response to ultraviolet light and replication stress (5, 6), therefore plays a much more important role in the cellular response to IR than previously thought. We and others found that Chk1 controls the S-phase checkpoint by directly phosphorylating Cdc25A (3, 4), a human phosphatase that activates the S-phase promoting cyclin-dependent kinase (Cdk) 2 (7). When Chk1 is inhibited, the Cdc25A phosphatase and, consequently, Cdk2 kinase activity are not down-regulated after IR, and the S-phase checkpoint is impaired (3, 4). Chk1 is believed to control the G₂ checkpoint via phosphorylation-mediated negative regulation of Cdc25A, Cdc25C, and Cdc25B (4, 8). All three members of the Cdc25 family of phosphatases contribute to activation of the M phase promoting Cdk1 (9, 10, 11).

Previous studies have suggested that compounds that abrogate the G₂ checkpoint will increase the sensitivity of human cancer cells to the cytotoxic effects of IR and chemotherapeutic drugs (reviewed in ref. 12). Cancer cells, which commonly lack normal G₁ checkpoint control and may therefore rely more on the G₂ checkpoint, could likely be more affected by G₂ checkpoint abrogation than normal cells (13, 14). The classic compounds known to abrogate the G₂ checkpoint are caffeine, an inhibitor of the ATR/ATM kinases (15, 16), and UCN-01, a protein kinase C inhibitor that, in addition to other kinases, also inhibits Chk1 (17). However, although these drugs inhibit the G₂ checkpoint and increase radiosensitivity, the relationship between checkpoint abrogation and radiosensitivity is not clear (18). For example, genetic manipulations of a checkpoint-defective and radiosensitive cell line could restore the checkpoint defects without affecting radiosensitivity (19). Moreover, caffeine was shown to inhibit DNA repair (20), and its radiosensitizing effect may be due to defective repair rather than checkpoint abrogation. UCN-01, which was originally synthesized as a staurosporine analog, caused apoptosis in human cancer cells when combined with IR (21).

Cell death in response to IR is best measured as a loss of reproductive integrity (22). The predominant death mechanism for most solid tumors of nonhematopoetic origin is mitosis-linked death, which is characterized by the appearance of cells with multiple nuclear fragments or micronuclei (23, 24, 25, 26, 27). IR-induced micronucleus formation requires progression through mitosis and is due to IR-induced chromosome aberrations (24, 26). Cell death under such circumstances occurs as a consequence of chromosomal aberrations, which may finally result in loss of transcription of essential genes after cells have passed through mitosis (26, 27, 28).

Given that Chk1 is emerging as a major regulator of IR-induced checkpoints, we wanted to explore the effects of a new specific Chk1 inhibitor, CEP-3891 (3), on checkpoint activation and cell death. We show that CEP-3891 efficiently abrogates the IR-induced S and G₂ checkpoints, which results in an accelerated onset of nuclear fragmentation when IR-treated cells progress prematurely through mitosis. Treatment with CEP-3891 also increases the radiosensitivity of human cells as measured by clonogenic survival assays. These results provide new insights into cellular consequences of checkpoint abrogation by inhibition of Chk1.

Human U2-OS and U2-OS-VP16 (a subclone of U2-OS) osteosarcoma cell lines were grown in Dulbecco’s modified Eagle’s medium (DMEM) with 10% fetal bovine serum. U2-OS VP16 cells were transfected with H2B-GFP (Clontech, Palo Alto, CA) and selected in DMEM containing blastidicin (5 μg/μL) and G418 (400 μg/μL). The CEP-3891 Chk1 inhibitor was provided by Cephalon Inc. and used at a concentration of 500 nmol/L. IR was delivered by X-ray generator (Pantak, Berkshire, United Kingdom; HF160; 150kV; 15 mA; dose rate, 2.18 Gy/min).

Immunoblotting, immunoprecipitation, and in vitro kinase assays have been described previously (3). Recombinant active Chk1 and antibody 06--570 to phosphorylated histone H3 were purchased from Upstate Biotechnology (Lake Placid, NY). Antibodies to cyclin A (sc-751), cyclin B1 (sc-245), and Cdk2 (sc-163) were purchased from Santa Cruz Biotechnology (Santa Cruz, CA), and antibody to Cdk-1 (Ab-1) was purchased from Calbiochem (San Diego, CA). Fluorescein-conjugated antirabbit immunoglobulin (Alexa Fluor 488 goat anti-IgG) was purchased from Molecular Probes (Eugene, OR).

To assay cell cycle distribution by DNA content, trypsinized cells were fixed in 70% EtOH, stained with 0.1 mg/mL propidium iodide, and analyzed by a FACSCalibur flow cytometer (BD Biosciences, Stockholm, Sweden) using Cellquest software. For two-parameter flow cytometry analysis to assay the number of mitotic cells (19), fixed cells were incubated with PBS/0.25% Triton X-100 for 15 minutes on ice and stained with antibody to phosphorylated histone H3 (1:500) for 1 hour at room temperature, followed by a 30-minute incubation with fluorescein conjugated antirabbit immunoglobulin antibody (1:500). Cells were then stained with propidium iodide and analyzed as described above.

Inhibition of DNA synthesis was assessed by standard [³H]thymidine/[¹⁴C]thymidine assay as described previously (3).

U2-OS-GFP-H2B cells were grown in Lab-Tek chambered cover glass (Nalge Nunc International, Rochester, NY) and incubated at 37°C in phenol red-free, CO₂-independent medium (Invitrogen, Paisley, United Kingdom) overlaid with mineral oil. Fluorescence and differential interference contrast images were acquired using a charge-coupled device camera (CoolSnap/VP; Roper Scientific, Tuscon, AZ) mounted on an Axiovert 200 microscope [Zeiss; 12 frames per hour; exposure times, 10 (differential interference contrast) and 15 milliseconds (fluorescence); ×40 objective (unless otherwise stated in the figure legend)] and subsequently processed by the Metamorph software package.

Cells cultured on glass coverslips were treated with CEP-3891 (0 or 500 nmol/L), exposed to IR (0 or 6 Gy) for a time period specified in the figure legends, and stained by ToPro3 (Molecular Probes, Eugene, OR) to mark nuclear DNA. Confocal images were obtained by using a Zeiss 510 laser scanning microscope (Axiovert) and operated by LSM510 software. For quantification of nuclear fragmentation, at least 200 cells were scored for each sample.

Cultured U2-OS-VP16 cells were incubated with a mixture of the membrane-permeable dye Hoechst 33342 (500 ng/mL) and the membrane-impermeable dye Sytox green (500 nmol/L) for 5 minutes at 37°C, and images were obtained using an epifluorescence microscope as described previously (29).

Between 250 and 2,000 cells (depending on radiation dose to yield 50–100 colonies per dish) were seeded to 6-cm–diameter dishes, incubated for 20 to 24 hours, and treated with CEP-3891 (0 or 500 nmol/L) and IR (0, 2, 4, or 6 Gy). After 24 hours, the medium was removed, cells were washed once with PBS, and regular DMEM was added. Cells were cultivated for another 12 to 13 days and then stained with crystal violet. Colonies containing more than 50 cells were scored as survivors. Survival fractions were calculated in each experiment as the average cloning efficiency (from three parallel dishes) after IR divided by the average cloning efficiency for nonirradiated cells. The cloning efficiencies for nonirradiated U2–0S-VP16 cells were about 0.6 and 0.5 in the absence and presence of CEP-3891 (500 nmol/L, 24 hours), respectively.

To measure total caspase activity, subconfluent cells were treated with an extraction buffer [250 mmol/L sucrose, 20 mmol/L HEPES, 10 mmol/L KCl, 1.5 mmol/L MgCl₂, 1 mmol/L EDTA, 1 mmol/L EGTA, and 1 mmol/L pefablock (pH 7.5)] containing 200 μg/mL digitonin for 15 minutes on ice. The effector caspase activities were estimated by adding 1 volume of 20 μmol/L N-acetyl-Asp-Glu-Val-Asp-7-amino-trifluoromethylcoumarin (Biomol, Plymouth, PA) in caspase reaction buffer [100 mmol/L HEPES, 20% glycerol, 0.5 mmol/L EDTA, 0.1% 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonic acid, 5 mmol/L dithiothreitol, and 1 mmol/L pefablock (pH 7.5)]. The V_max of the liberation of AFC (excitation, 400 nm; emission, 489 nm) was measured over 20 minutes at 30°C with a Spectramax Gemini fluorometer (Molecular Devices, Sunnyvale, CA). Lactate dehydrogenase activity of the cytosol determined by a cytotoxicity detection kit (Roche A/S Diagnostic, Hvidovre, Denmark) was used as an internal standard to which protease activities were normalized.

CEP-3891, a new inhibitor of Chk1 kinase, is more specific than existing Chk1 inhibitors such as UCN-01 (ref. 3; see Supplementary Table of Ref. 3 for in vitro inhibitory effects of Cep-3891 on a number of human kinases). Here, we first verified that CEP-3891 can inhibit the activity of recombinant Chk1 in vitro (Fig. 1,A). Next, we examined whether this inhibitor could abrogate the in vivo induction of the Chk1-dependent S and G₂ checkpoints in response to IR. As seen in Fig. 1,B, bottom panels, CEP-3891 significantly abrogated the rapid IR-induced G₂ checkpoint measured at 1 hour after 6 Gy. CEP-3891 also caused an increased entry of cells from G₂ to M phase in the absence of IR (Fig. 1,B, top panels), consistent with a role for Chk1 in control of G₂ to M progression in both the presence and absence of IR. Furthermore, CEP-3891 efficiently abrogated the delayed IR-induced G₂ checkpoint, which was assayed at 20 hours after 6 Gy (Fig. 1,C). When nocodazole was added to the culture medium, the CEP-3891–treated and CEP-3891 + 6 Gy-treated cells accumulated at mitosis, demonstrating that CEP-3891–treated cells were cycling over the time course of this experiment (data not shown). Consistent with our previous report (3), CEP-3891 also abrogated the IR-induced S-phase checkpoint and slightly increased the rate of DNA synthesis in the absence of IR (Fig. 1 D).

To further investigate how CEP-3891 abrogates the S and G₂ checkpoints, we assayed the activities of major cell cycle-regulating kinases in a time course after CEP-3891 and IR treatment (Fig. 2). A transient down-regulation of cyclin A- and B-associated kinase activities was found in IR-treated cells (Fig. 2,A and B, 6 Gy), consistent with the induction of the S and G₂ phase checkpoints (Fig. 1) and previous reports (see refs. 3, 4, and 23). Addition of CEP-3891 in the absence of IR caused a marked transient increase of these kinase activities (Fig. 2,A and B, CEP), consistent with the effects of CEP-3891 on cell cycle progression seen in Fig. 1,B and D. Similar effects were seen for cyclin E- associated kinase activity (data not shown). When U2-OS cells were treated with CEP-3891 in addition to IR, the IR-induced down-regulation of cyclin A- and B-associated kinase activities was abrogated (Fig. 2,A and B, CEP + 6 Gy), consistent with the abrogation of the S and G₂ checkpoints (Fig. 1).

Having established that addition of CEP-3891 would impair both the S and G₂ checkpoints, we explored the effects of CEP-3891 on IR-induced cell death (Fig. 3). The predominant type of IR-induced cell death of most solid tumors is mitosis-linked death, which is manifested by the appearance of nuclear fragmentation (or micronucleus formation after low radiation doses), occurring as a consequence of passage of IR-treated cells through mitosis (23, 24, 25, 26). We first tested whether our IR-treated U2-OS cells underwent nuclear fragmentation (Fig. 3,A). As evident from the results, U2-OS cells irradiated with 6 Gy underwent massive nuclear fragmentation after exit from the G₂ checkpoint (Fig. 3,A, 48 and 72 hours). The majority of the cells with nuclear fragmentation had DNA content close to G₁-phase cells (Fig. 3,A, DNA histograms), indicating that although the nucleus had split into several subnuclear fragments, the total amount of DNA of each daughter cell was close to normal G₁ phase cells. A similar pattern of IR-induced nuclear fragmentation has been reported in HeLa cells (23), and in U2-OS cells irradiated with 10 Gy (30). When the IR-induced S and G₂ checkpoints were abrogated by addition of CEP-3891, massive nuclear fragmentation was seen already at 24 hours after IR (Fig. 3,B). Quantification of the number of cells with nuclear fragmentation in a time course showed that treatment with CEP-3891 in addition to IR accelerated the time for onset of nuclear fragmentation compared with treatment with IR alone (Fig. 3 C). Acceleration of mitotic cell death by CEP-3891 appeared to be a more general response rather than a specific phenomenon limited to U2-OS cells because other cell types, including normal human fibroblasts, also showed a significant fraction of cells with nuclear fragmentation at 24 hours after treatment with CEP-3891 and IR (data not shown).

Although the time for onset of nuclear fragmentation after IR correlated with the time of progression through mitosis (Fig. 3; ref. 23), and abnormal mitotic figures were present (data not shown; ref. 23), no direct and dynamic visualization of the process of formation of cells with nuclear fragmentation has thus far been reported. We generated a human model system for a constitutive expression of green fluorescent protein (GFP)-tagged histone H2B (31) to directly visualize chromosomal dynamics in the absence and presence of CEP-3891 by time-lapse fluorescent video microscopy (Fig. 4). These experiments revealed that U2-OS-H2B-GFP cells underwent normal chromosome segregation in the absence of IR or CEP-3891 treatment (Fig. 4,A; Supplementary Fig. 1). In response to IR (6 Gy), mitotic cells appeared after exit from a prolonged G₂ checkpoint arrest. The results of videorecording of arbitrary selected mitotic cells that appeared at 40 to 50 hours after IR clearly demonstrated that nuclear fragmentation was a result of defective chromosome segregation in mitosis (Fig. 4,B; Supplementary Fig. 2). When CEP-3891 was added together with IR, cells were entering mitosis prematurely without induction of the S and G₂ checkpoints, and mitotic cells were present at all times within the first 24 hours after IR. Similar to the delayed process of nuclear fragmentation after exit from G₂ arrest in the IR-treated cells (Fig. 4,B), the more rapid nuclear fragmentation occurring when the checkpoints were abrogated by CEP-3891 also reflected defective chromosome segregation in mitosis (Fig. 4,C; Supplementary Fig. 3).

Nuclear fragmentation is often associated with apoptotic cell death (32). To explore whether the mitotic induction of nuclear fragmentation was related to apoptosis, we measured the extent of caspase activity in U2-OS cells at 12, 24, and 48 hours after IR and CEP-3891 treatment (Fig. 5,A). Although massive nuclear fragmentation was found in U2-OS cells treated with IR + CEP-3891 at 24 hours, no significant increase in caspase activity was found at 12 or 24 hours (Fig. 5,A), indicating that nuclear fragmentation was not a result of classical apoptosis. However, a small increase in caspase activity was found at 48 hours after treatment with CEP-3891 and IR, suggesting that cells died at later times at least partly via apoptosis (Fig. 5 A).

Furthermore, nuclear fragmented cells at 24 hours after treatment with CEP-3891 and IR did not show the characteristic nuclear condensation (Fig. 5,B), which would have been expected to accompany apoptosis (32). The majority of CEP-3891 and IR-treated cells also did not stain with the membrane-impermeable dye Sytox green (Fig. 5 B; data not shown); therefore, these cells were not likely to be at late stages of apoptosis. Taken together, these results show that the massive nuclear fragmentation after inhibition of Chk1 and IR is not a result of apoptosis.

Finally, we explored the effect of CEP-3891 on the extent of overall cell killing after IR as measured in clonogenic survival assays (Fig. 6,A). The results showed that CEP-3891 sensitized U2-OS cells to the cytotoxic effects of IR (Fig. 6 A), to an extent nearly comparable with the sensitization obtained by 2 mmol/L caffeine, an established radiosensitizer (15).

The results of this study suggest that inhibition of Chk1 abrogates the IR-induced S and G₂ checkpoints, followed by premature mitotic entry, defective chromosome segregation, and nuclear fragmentation (Fig. 6,B). In the absence of CEP-3891, IR-treated cells arrest in G₂ for several hours before they eventually enter mitosis. After a relatively high dose of IR, such as 6 Gy, most of the cells that eventually enter mitosis contain so much chromosome damage that defective chromosome segregation and nuclear fragmentation occur, followed by cell death. Most cells that arrest in G₂ after 6 Gy therefore contain irreparable lesions; nevertheless, the cells are eventually released from the checkpoint and progress through mitosis. Therefore, a process of adaptation to DNA damage likely occurs at the G₂ checkpoint (Fig. 6 B). Adaptation to DNA damage has also been described in yeast (33) and Xenopus egg extracts (34).

Consistent with this concept, time-lapse studies have shown that both cells that survived and cells that died in clonogenic assays after IR (6 Gy), would transiently delay their cell cycle progression in the first cell cycle after IR (35). However, the cells that ended up as nonclonogenic (dead cells), on average, showed longer cell cycle delays compared with the clonogenic survivors (35). Thus, there may be two different processes that promote exit from the G₂ checkpoint: one that allows exit when the DNA damage is repaired, and another that eventually allows exit of cells with irreparable damage due to adaptation. The molecular mechanisms of these processes are unknown; however, dephosphorylation of the activating phosphorylations on Chk1 may be involved (36). It was also suggested that overaccumulation of cyclin B during checkpoint arrest due to a prolonged time in G₂ phase may contribute to cause exit from the G₂ checkpoint (23).

Another question is what kind of DNA lesion evokes the “signal” that triggers the prolonged G₂ checkpoint arrest. Unrejoined double-strand breaks (DSBs) and/or changes in chromatin structure are thought to trigger the rapid activation of ATM in response to DNA damage (37). Unrejoined DSBs likely contribute to the ATM-dependent S and G₂ checkpoints in the first few hours after IR. However, the majority of DSBs are rejoined within several hours after IR, and the prolonged G₂ checkpoint is independent of ATM (19, 38, 39). Although the signal for the prolonged G₂ checkpoint still could reflect the small proportion of remaining unrejoined DSBs, another possibility is that additional lesions are involved. IR-induced DSBs are converted into large-scale chromosomal damage [such as dicentrics, rings, and bridges (40)], which possibly could also be sensed by the cell’s checkpoint machinery. In such a scenario, exit from G₂ checkpoint arrest could depend on complete rejoining of DSBs in addition to adaptation to large-scale chromosome damage.

We found that checkpoint abrogation by CEP-3891 increases cell cytotoxicity as measured in clonogenic survival assays (Fig. 6 A). Thus, CEP-3891 not only accelerated IR-induced nuclear fragmentation but also caused an increased total number of dead cells. This result is consistent with a concept in which Chk1-mediated checkpoint arrest promotes repair of DNA damage before cells enter mitosis, in which defective chromosome segregation takes place if the damage was not properly repaired. It remains to be investigated whether Chk1 plays a direct role in regulating repair pathways after IR or whether Chk1 simply promotes repair by allowing more time before entry into mitosis.

We did not observe much cell death over the first 24 hours after CEP-3891 and IR treatment; however, caspase activity was slightly increased at 48 hours, indicating that cells died at later times at least partly via apoptosis (Fig. 5 A). Future studies can address whether other death mechanisms such as necrosis or senescence are also involved. Previous extensive time-lapse studies have demonstrated that rat embryo cells transfected with c-myc underwent apoptosis after irradiated cells had divided once or several times (27, 41). This process was associated with formation of micronuclei in mitosis that preceded cell death (27), which may be analogous to our study, in which nuclear fragmentation in mitosis precedes cell death.

Chk1 is thought to regulate the S and G₂ checkpoints via phosphorylation of the human phosphatases Cdc25A, Cdc25B, and Cdc25C (4, 8). Although CEP-3891 efficiently abrogated the G₂ checkpoint, the level of Ser²¹⁶ phosphorylation of Cdc25C was not altered by CEP-3891 treatment (data not shown). This was in contrast to the effects seen with UCN-01, which effectively reduced Ser²¹⁶ phosphorylation under analogous experimental conditions (data not shown; ref. 17). Therefore the other phosphatases (Cdc25A and Cdc25B) may be more important than Cdc25C in control of the G₂ checkpoint. This finding is consistent with previous reports in which cells lacking Cdc25C showed a normal G₂ checkpoint response (11) and with the involvement of Cdc25A in the G₂ checkpoint (4, 10).

It has become increasingly clear that one of the general features of human cancer cells is defective DNA damage checkpoint control. It is therefore attractive to exploit new strategies to interfere with checkpoints in an attempt to find new cancer-specific treatments. Chk1 is one of several potential targets for checkpoint interference. Although Chk1 inhibition by UCN-01 showed promise as a cancer-selective treatment, particularly for cancer cells lacking p53 (42, 43), other studies failed to detect such specificity (44). It was proposed that more specific inhibition of Chk1 than that obtained with UCN-01 would likely be more effective (14). CEP-3891 was developed to make such a specific Chk1 inhibitor. The results of our study, using CEP-3891 to abrogate the IR-induced checkpoints, contribute to a better understanding of the cellular consequences of specific Chk1 inhibition. Additional studies of the effects of specific Chk1 inhibitors such as CEP-3891 on a broad panel of human cancer and normal cells, as well as in vivo studies in mice, will be needed to evaluate the true potential of Chk1 inhibition as a strategy for cancer treatment.

We thank Cephalon Inc. for providing the CEP-3891 Chk1 inhibitor and Nicole Fehrenbacher for helpful technical advice.