Potentiation of Cytotoxicity of Topoisomerase I Poison by Concurrent and Sequential Treatment with the Checkpoint Inhibitor UCN-01 Involves Disparate Mechanisms Resulting in Either p53-Independent Clonogenic Suppression or p53-Dependent Mitotic Catastrophe

INTRODUCTION

Proliferating cells respond to genotoxic stress by triggering a series of signaling events known as cell cycle checkpoints, which function to delay cell cycle progression, thereby facilitating repair and preventing propagation of damaged genetic materials (1) . In a broader sense, checkpoints can be viewed as an integral component of a more global DNA damage response, which relays genotoxic stimuli to other processes such as activation of DNA repair, and induction of apoptosis or cellular senescence (2) . Checkpoints are present in all of the phases of the cell cycle and are believed to be essential for maintaining genomic integrity and suppressing tumorigenesis (1) .

Chk1 and chk2 are two evolutionarily conserved but structurally unrelated serine/threonine kinases that share overlapping function in controlling both the S and G₂-M checkpoints (2) . Current evidence suggests that in response to UV irradiation, DNA replicative block, and to a lesser extent ionizing radiation, chk1 is activated by its upstream phosphoinositide-3-kinase-related kinase ATR via phosphorylation on serines 317 and 345 (3, 4, 5, 6) . Conversely, chk2 appears to be a major effector of ATM, another phosphoinositide-3-kinase-like kinase, which phosphorylates chk2 at its amino-terminal SQ/TQ cluster domain after ionizing radiation-induced double-strand DNA breaks (7) . Both chk kinases halt cell cycle progression via phosphorylation and inhibition of the dual specificity cdc25 phosphatases, cdc25A and cdc25C. In response to ionizing radiation, cdc25A, which is required for S-phase progression, is rapidly degraded via the ubiquitin–proteasome pathway after phosphorylation on serine 124 by chk1/chk2 (6 , 8 , 9) . After DNA damage, both chk1 and chk2 can phosphorylate cdc25C on serine 216 and inhibit mitotic entry (10) . In addition to the ATM/ATR→ chk1/2→ cdc25 pathways, it has been shown that both ATM and chk2 can phosphorylate and stabilize the tumor suppressor p53 (11, 12, 13) . Because p53 transcriptionally activates its downstream targets p21 and 14-3-3σ, which are important for maintaining the G₂ arrest after DNA damage (14 , 15) , the ATM-chk2-p53 pathway is expected to be critical for regulating the G₂ checkpoint as well. However, two recent reports have shown that the p53 response to DNA damage was apparently intact in cell lines where chk2 was depleted by either gene knockout technique (16) or small interfering RNA (17) , therefore, questioning the role of chk2 in activating p53.

Mounting evidence has indicated that dysregulation of checkpoint gene products is common in both sporadic and hereditary human cancers, e.g., inactivation of p53 and chk2 (18 , 19) and overexpression of cdc25A (20 , 21) . Because most conventional agents used for cancer treatment impart damage to the genome and, therefore, also activate cell cycle checkpoints, pharmacological disruption of checkpoints will possibly create “synthetic lethality” (22) selectively in tumors with intrinsic checkpoint defects after genotoxic challenge. This concept of enhancing the cytotoxicity of DNA-damaging agents by checkpoint inhibition was first exemplified by caffeine (23) , an inhibitor of both ATM and ATR (24) , and more recently shown with UCN-01. UCN-01, at nontoxic concentrations, abrogates both the S and G₂-M checkpoints and potentiates the cytotoxic effects of a wide spectrum of DNA-damaging agents, including ionizing radiation (25) , cisplatin (26) , gemcitabine (27) , temozolomide (28) , and camptothecin (29) . Current model suggests that UCN-01 inhibits chk1, and possibly chk2, and abrogates the S and G₂-M checkpoints by targeting the cdc25A-cdk2/cyclin E and cdc25C-cdc2/cyclin B1 pathways, respectively (6 , 30 , 31) . There has been conflicting data as to whether UCN-01 selectively enhances the cytotoxicity of DNA-damaging agents in cells with nonfunctional p53. Several reports have suggested an enhanced effect of UCN-01 combination in cells with defective p53 (29 , 32) , whereas others show no association between p53 status and sensitivity to combined treatment (28 , 31) . We now report that part of the discrepancy probably relates to the different treatment schedules used, the particular checkpoint events examined, and assays used to evaluate cytotoxicity.

MATERIALS AND METHODS

Chemicals.

UCN-01 was provided by Drs. Edward Sausville and Robert Schultz (National Cancer Institute, Bethesda, MD). SN-38 was a generous gift from Dr. J. Patrick McGovren (Pharmacia Upjohn Inc., Kalamazoo, MI). All of the drugs were dissolved in dimethyl sulfoxide and stored in aliquots at −20°C. [¹⁴C]-thymidine (58 mCi/mmol) and [methyl-³H]thymidine (79 mCi/mmol) were purchased from Perkin-Elmer (Boston, MA).

Cell Culture.

Human HCT116 colonic carcinoma cell line and its p53-null variant were kindly provided by Dr. Bert Vogelstein (Johns Hopkins University, Baltimore, MD). Cultures were maintained at 37°C in the presence of 5% CO₂ in RPMI 1640 supplemented with 10% heat-inactivated fetal bovine serum, 100 units/ml penicillin, and 100 μg/ml streptomycin. For time-lapse experiments, cells were cultured in the same medium containing 20 mm HEPES without phenol red. Cell synchronization using double thymidine block was done by treating cells with 2 mm thymidine for 16 hours, washed, released into regular medium for 8 hours, and again blocked with 2 mmol/L thymidine for an additional 16 hours. All of the cultures were tested consistently as Mycoplasma free.

Generation of HCT116 Transfectant Cell Lines.

Parental and p53−/− HCT116 cells were transfected with a green fluorescent protein (GFP)-histone 2B expression plasmid (PharMingen, San Diego, CA) using Fugene (Roche, Indianapolis, IN). Stable transfectant clones were selected in the presence of 2 μg/ml blastocidin (Invitrogen, Carlsbad, CA). Parental HCT116 cells transfected with a GFP-tubulin expression plasmid (Clontech, San Jose, CA) were selected in the presence of 600 μg/ml G418 (EMD Bioscience, San Diego, CA).

Clonogenicity Assays.

Log-phase cells were plated, in triplicate, onto 100-mm dishes at 1000–5000 cells/dish and were allowed to attach for 24 hours before treatment. At the end of treatment, cells were cultured in drug-free medium for 10–14 days. The resulting colonies were scored after staining with 0.01% crystal violet. The plating efficiencies for untreated control were ∼35 and 25% for parental and p53−/− HCT116 cells, respectively, which are typical for these cell lines (33) .

Measurement of Apoptosis and Micronucleation by Quantitative Fluorescent Microscopy.

After drug treatment, both adherent and floating cells were harvested and fixed in 3% paraformaldehyde. In experiments involving sequential therapy, floating cells were collected after initial treatment and were added back to the plate for subsequent treatment. The nuclear morphology of cells was examined under fluorescence after staining with 4′,6-diamidino-2-phenylindole. Cells were scored as apoptotic based on the presence of condensed fragmented chromatin. Micronucleated cells were defined as cells containing multiple (≥3) small decondensed interphase nuclei. A minimum of 400 cells were counted for each sample. The statistical significance of the experimental results was determined by the two-sided t test.

Biparameter Flow Cytometry.

Cell cycle analysis was done as described previously (34) . Briefly, both adherent and nonadherent cells were harvested and fixed in 70% EtOH. Cells were first labeled with the MPM-2 monoclonal antibody, which recognizes specific phosphorylated epitopes during mitosis (35) and subsequently with fluorescein isothiocyanate-conjugated antimouse secondary antibody. Cells were then treated with RNase (5 μg/ml) and stained with propidium iodide (50 μg/ml). Samples were analyzed on a FACScan (Becton Dickinson, Franklin Lakes, NJ) for cell cycle distribution and mitotic index (percentage of MPM-2-positive cells) using the CellQuest software.

S-Phase Checkpoint Assays.

Log-phase cells were plated in triplicate and prelabeled with 0.010 mCi/ml [¹⁴C]-thymidine for 24 hours. After treatment with SN-38 in the presence or absence of UCN-01 for 1.5 hours, cells were pulse-labeled with 2 mCi/ml of [methyl-³H]thymidine for 15 minutes at 37°C. After washing twice with ice-cold PBS, cells were scraped into 1 ml of 10% trichloroacetic acid. Samples were then mixed and collected by centrifugation. The precipitates were dissolved in 0.5 ml of 0.5 n NaOH overnight at room temperature. After neutralization with HCl, radioactivity was determined by liquid scintillation counting. DNA synthesis rate was calculated as [(3H:14C)_{treated samples}/(3H:14C)_{untreated controls}] × 100%, after correction for channel crossover. The counting efficiency for [³H] and ¹⁴C after correcting for channel crossover was 44 and 75%, respectively.

Immunoblots.

Cell lysates were prepared by lysis of both floating and adherent cells in radioimmunoprecipitation assay buffer. Twenty-five to 50 μg of protein were fractionated by SDS-PAGE and transferred onto immobilion membranes (Millipore, Billerica, MA). Equal protein loading was confirmed by Amido black staining (Bio-Rad, Hercules, CA). After blocking with 5% nonfat milk, membranes were probed with mouse monoclonal p53, p21 (Santa Cruz Biotechnology, Santa Cruz, CA), and γ-H2AX (Upstate Biotechnology, Waltham, MA). Bound primary antibodies were detected with horseradish peroxidase-conjugated secondary antibodies (ICN/Jackson ImmunoResearch, West Grove, PA) and visualized by enhanced chemiluminescence reagent (Amersham Pharmacia, Piscataway, NJ).

Time-Lapse Fluorescent Microscopy.

Clonal HCT116 transfectant cells expressing GFP-histone 2B or GFP-tubulin fusion protein were grown on 40-mm round coverslips and allowed to attach for 48 hours before assembly onto a FCS2 close-system flow-observation chamber (Bioptechs, Butler, PA). The chamber was mounted onto the stage of a Zeiss Axiovert 200 M inverted microscope with the chamber and coverslip temperature maintained at 37°C. For experiments involving sequential treatment, cells were treated with the first agent inside a regular 5% CO2 incubator and were loaded onto the flow chamber for time-lapse study during subsequent treatment. Drug containing media was introduced into the chamber via a peristaltic pump. Epifluorescent and phase contrast images were acquired with a ×40 objective lens, using a cooled CCD camera every 5 minutes, with exposure time limited to 4 to 5 seconds/image (MicroMax-1300YHS, Roper Scientific, Tucson, AZ). For each time point, images were taken with five to six different focal planes along the z-axis 2 to 3 μm apart. Imaging data were analyzed using MetaMorph 3.0.

Cyclin B1/cdc 2 Kinase Assays.

Cyclin B1-associated kinase assays were done as described previously (34) . Briefly, cyclin B1 was immunoprecipitated from 200 μg of cell lysate protein with 1 μg of anticyclin B1 (GN1) antibody (Santa Cruz Biotechnology) and immobilized protein A (Upstate Biotechnology). Precipitates were washed three times with lysis buffer and twice with kinase buffer (34) . Reactions were carried out in 30 μl kinase buffer containing 5 μCi of [γ-³²P]ATP, 15 μmol/L ATP, and 1 μg histone H1 (Boehringer Mannheim, Indianapolis, IN) at 30°C for 20 minutes. Products were resolved by 14% SDS-PAGE, transferred onto Immobilon-P membrane, and visualized by autoradiography. The activity levels on membranes were additionally quantified by a phosphoimager (Bio-Rad).

Senescence-Associated β-Galactosidase Expression.

Staining for senescence-associated β-galactosidase activity was done using the senescence detection kit according to the manufacturer’s instructions (Cell Signaling Technology, Beverly, MA). The percentage of senescent cells was determined by counting at least 400 cells/sample and scoring for the incidence of blue-stained cells (senescence-associated β-galactosidase positive) under a bright-field microscope at ×200 magnification.

RESULTS

Induction of Apoptosis in HCT116 Colonic Carcinoma Cells by SN-38 and UCN-01 Combination Is Sequence Dependent.

We have shown previously that treatment of p53 wild-type HCT116 colon cancer cells with 20 nm SN-38, the active metabolite of the topoisomerase I poison CPT-11, resulted in a transient delay in S-phase progression followed by a sustained G₂ arrest, even on removal of drug. These G₂-arrested cells did not undergo apoptosis based on both morphologic and biochemical criteria (36) ; they remained metabolically active and showed a senescence-like phenotype—flattened morphology, increased granularity, and expression of senescence-associated β-galactosidase activity (Fig. 1A) ⇓ . We next examined the treatment effect of 20 nm SN-38 in combination with 200 nm UCN-01 in these cells given in four separate schedules: (1) individually for 24 hours (SN or UCN-01 or washed and followed by no drug for 24 hours (SN→ no drug or UCN-01→ no drug); (2) concurrently for 24 hours followed by no drug (SN+UCN-01→ no drug); (3) sequentially with SN-38 for 24 hours, washed, and followed by UCN-01 (SN→ UCN-01); or (4) the same drugs given in the reverse sequence (UCN-01→ SN; Fig. 1B ⇓ ). Sequential treatment with SN-38 followed by UCN-01 resulted in 11.6 ± 0.8% apoptosis, which is significantly higher than the other schedules that caused only an additive effect of the two agents (P < 0.05 by t test; Fig. 1B ⇓ ). The lack of apoptosis observed with SN+UCN-01→ no drug could not be because of loss of apoptotic cells during the drug-free period, because a comparable level of apoptosis (<4%) was seen when cells were harvested and examined immediately after drug treatment (data not shown).

Examination of the nuclear morphology of cells treated sequentially with SN→ UCN-01 at serial time points revealed that at 8 hours after UCN-01 addition, cells entered an aberrant mitosis (mit) characterized by condensed but disorganized chromatin and congression of some, but not all, chromosomes on a metaphase plate. At 16 hours after UCN-01 treatment, the loss of the abnormal mitotic figures was accompanied by the appearance of frank apoptotic cells (Apo) as well as interphase cells with lobulated micronuclei (MN), a morphologic hallmark of mitotic catastrophe (ref. 37 ; Fig. 1C ⇓ ). Conversely, cells treated with the two agents concurrently or in the reverse sequence showed minimal apoptosis and no evidence of micronucleation (data not shown).

Sequential Treatment with SN-38 followed by UCN-01 Abolishes the Maintenance of the G₂-M Checkpoint, whereas Concurrent Treatment Abrogates the S-Phase but not the G₂-M Checkpoint.

We next investigated the effects of UCN-01 on cell cycle checkpoint response induced by SN-38, using biparameter flow cytometry. Treatment of HCT116 cells with 200 nm UCN-01 alone did not result in any appreciably changes in cell cycle distribution (Fig. 2A) ⇓ ⇓ . Cells treated with 20 nm SN-38 for 24 hours underwent a G₂ arrest (SN_24h), as judged by the accumulation of cells with 4N DNA content and loss of the mitosis-specific marker MPM-2 (Fig. 2A) ⇓ ⇓ . This G₂ arrest sustained for at least 24 hours after drug removal with cells maintained in drug-free medium (SN→ no drug₂₄). On SN-38 removal, sequential treatment with 200 nm UCN-01 abrogated the G₂ arrest and resulted in mitotic entry as indicated by an increase in MPM-2 labeling (SN→UCN-01). The entry into mitosis usually peaked between 6 and 9 hours of subsequent UCN-01 (SN→ UCN-01_6h and SN→UCN-01_9h), which was in agreement with our examination of cellular morphology (Fig. 1C) ⇓ . Thus, consistent with previous reports, UCN-01 abrogated the G₂ checkpoint induced by topoisomerase I poison, causing cells to enter mitosis (29 , 38) . However, the current experimental design only allowed examination of the inhibitory effect of UCN-01 on maintenance of the G₂ checkpoint specifically, without revealing its effect on the S-phase checkpoint, because >90% of the cells had completed S phase and arrested in G₂ after 24 hours of SN-38 treatment before UCN-01 addition.

Therefore, we next elected to examine the cell cycle effects of SN-38 and UCN-01 combination in these cells within the first 24 hours of drug exposure. Treatment of asynchronous HCT116 cells with 20 nm SN-38 alone resulted in a transient accumulation of S-phase cells (SN_12h) before development of a sustained G₂ arrest (SN_24h; Fig. 2B ⇓ ⇓ ). Concurrent treatment with UCN-01 inhibited the transient accumulation of S-phase cells induced by SN-38 (most evident with SN+U_6h and SN+U_12h; Fig. 2B ⇓ ⇓ ). However, cells treated with concomitant SN-38 and UCN-01 ultimately remained arrested in G₂ with marginal increase in mitotic index, suggesting that UCN-01 was ineffective in abolishing the G₂ checkpoint when given in this treatment schedule (Fig. 2B) ⇓ ⇓ . We have treated the cells concurrently with both drugs for 24 hours followed by drug-free medium for an additional 12 hours or continuously with both drugs up to 36 hours and found that the G₂ arrest appeared to be sustained (Fig. 2B ⇓ ⇓ ; data not shown).

The differential ability of UCN-01 to inhibit the G₂ checkpoint between concurrent and sequential treatment with SN-38 was additionally investigated. To allow a more legitimate comparison between treatment effects of the two schedules on the G₂ checkpoint, HCT116 cells were first synchronized at the G₁-S boundary by double thymidine block, cell cycle progression was then monitored after release from the block. Cells released into drug-free medium (no drug) entered mitosis at ∼9 hours after release with a peak mitotic index of 16% (Fig. 2C) ⇓ ⇓ . Cells treated with UCN-01 underwent mitosis sooner than control with mitoses seen as early as 6 hours after release (Fig. 2C) ⇓ ⇓ . Cells released into SN-38-containing medium underwent a G₂ arrest without mitotic entry as expected (Fig. 2C) ⇓ ⇓ . Concomitant treatment with UCN-01 did not substantially abrogate the G₂ arrest induced by SN-38, as mitotic index only increased to 3% at 12 hours after release. Conversely, sequential exposure of cells to SN-38 for 9 hours followed by UCN-01 resulted in a marked checkpoint abrogation and mitotic entry when compared with SN-38 treatment followed by no drug (12% and <1%, respectively; Fig. 2C ⇓ ⇓ ).

To corroborate our results on cell cycle analysis, we determined the activity of the promitotic cyclin B1/cdc2 kinase at different time points after release of synchronized cells. Cyclin B1-associated kinase activity derived from cells released into drug-free medium peaked at 9 hours (no drug_9h) after release from the block (Fig. 2D) ⇓ ⇓ . Treatment with UCN-01 resulted in a more rapid increase in mitotic kinase activity mirroring the earlier mitotic entry seen with this treatment, whereas SN-38 prevented activation of the mitotic kinase (Fig. 2D) ⇓ ⇓ . Concomitant treatment with SN-38 and UCN-01 resulted only in a modest increase in cyclin B1/cdc2 kinase activity, whereas sequential treatment with SN-38 for 9 hours followed by UCN-01 for 6 hours (SN_9h→ U_6h) caused a much higher activation of the mitotic kinase (Fig. 2D) ⇓ ⇓ . Taken together, these results suggest that SN-38 activates a replication checkpoint, causing a delay in S-phase progression (see below) as well as a subsequent G₂ arrest. Concurrent exposure to UCN-01 abrogates the S phase but not the G₂ checkpoint, whereas sequential treatment with UCN-01 abolishes maintenance of the G₂ checkpoint. One caveat with cell synchronization using double thymidine block is that it may result in activation of the S-phase checkpoint because of replicative stress induced by the high dose thymidine. Indeed, we observed an initial activation of chk1 by phosphorylation, which was transient and disappeared within 3 hours after release of cells in drug-free medium (data not shown). However, this S-phase checkpoint activation is unlikely to affect our conclusion regarding the lower extent of G₂-M abrogation associated with concurrent treatment, because the mitotic events being examined occurred at a much later time (8 to 10 hours) after release from the block, and cells had already progressed through S into G₂-M phase by then.

Real-Time Visualization of Mitotic Catastrophe by Time-Lapse Fluorescent Microscopy.

Our results have shown that sequential treatment with SN-38 followed by UCN-01 resulted in the greatest potentiation of apoptosis and formation of micronucleated cells, which appeared to be preceded by an aberrant mitosis (Fig. 1C) ⇓ . However, the exact temporal-spatial relationship among abnormal mitosis, apoptosis, and micronucleation was not clear. To determine the mechanism giving rise to these peculiar cellular changes, HCT116 cells were stably transfected with a vector that constitutively expressed a GFP-histone 2B chimera. Incorporation of the fusion protein into nucleosomes allowed real-time visualization of cellular events that involved dynamic chromatin changes, such as mitosis and apoptosis, using time-lapse fluorescent microscopy (39) . Using this technique, we are able to follow the fate of individual cells treated sequentially with SN-38 and UCN-01. As shown in Fig. 3, A and B ⇓ , on the addition of UCN-01, ∼25% of those cells arrested in G₂ by SN-38 underwent a highly abnormal mitosis characterized by disarrayed condensed chromosomes and incomplete alignment of all of the chromosomes on a metaphase plate (Fig. 3, A and B ⇓ , mit). The percentage of cells that underwent abnormal mitosis, as determined by this technique, was consistent with the results estimated by flow cytometry (Fig. 2A) ⇓ ⇓ . During mitosis, the chromatin mass appeared to be stretched (Fig. 3A ⇓ , panel 2), but not necessarily toward the opposite poles, and the chromatin became fragmented into lobulated (Fig. 3A ⇓ , panel 3) masses, which subsequently underwent decondensation on mitotic exit (Fig. 3A ⇓ , panel 5; movie 1, supplementary data). The majority of these cells did not undergo cytokinesis and appeared as interphase cells with micronuclei (Fig. 3A ⇓ , panel 5, mn; movie 1, supplementary data). As shown in Fig. 3B ⇓ , some of these micronucleated cells (Fig. 3B ⇓ , panel 3, mn) underwent apoptosis, as evidenced by recondensation of fragmented chromatin (Fig. 3B ⇓ , panel 5, apo; movie 2, supplementary data). Apoptotic cells became apparent at ∼12 hours after UCN-01 treatment, and ∼15% of cells developed apoptosis during the 24-hour period of time-lapse experiment. All of the apoptotic cells were derived from cells that had undergone G₂ checkpoint abrogation and entered mitosis. Thus, mitotic catastrophe appeared to be a prerequisite for subsequent apoptosis induced by sequential treatment with SN-38 and UCN-01. The UCN-01 by itself did not cause micronucleation, and treatment with SN-38 followed by drug-free medium resulted in interphase cells arrested in G₂ without mitosis or apoptosis (data not shown). Additionally, consistent with the data from cell cycle analysis and cyclin B1/cdc 2 kinase assays (Fig. 2) ⇓ ⇓ , concurrent treatment with SN-38 and UCN-01 did not result in any appreciable level of aberrant mitosis or subsequent apoptosis/micronucleation as observed with sequential treatment (data not shown).

The improper segregation of chromosomes during mitosis in cells treated sequential with SN-38 and UCN-01 could be because of a defect in bipolar spindle formation. To investigate this possibility, HCT116 cells were stably transfected with GFP-α-tubulin-expressing plasmid and examined by time-lapse microscopy after sequential treatment. Surprisingly, bipolar spindles were present in the majority of cells that entered mitosis (Fig. 3C ⇓ , panel 2, bi; movie 3, supplementary data), with tripolar spindles sporadically detected in a few mitoses (data not shown), yet these cells failed to undergo appropriate segregation of chromosomes and developed micronucleation (Fig. 3C ⇓ , panel 5, mn; movie 3, supplementary data). Thus, mitotic abnormalities of cells after sequential treatment with SN-38 and UCN-01 did not seem to arise from a defect in the assembly of bipolar spindles.

Checkpoint-Defective p53-Null Cells Are Prone to Undergoing Mitotic Catastrophe After Sequential Treatment with SN-38 and UCN-01 Compared with p53-Wild-Type Parental HCT116 Cells.

Several reports have suggested that UCN-01 selectively potentiates the cytotoxicity of DNA-damaging agents in cells that lack functional p53. The general interpretation is that p53-deficient cells are unable to arrest in G₁ after genotoxic stress and, therefore, rely more on the G₂ checkpoint for survival. However, it has been shown that p53 is also critical for the maintenance of the G₂ DNA-damage checkpoint (14) . Thus, we were interested in determining the consequence of p53 loss in affecting the sensitivity to combined treatment with SN-38 and UCN-01. As shown is Fig. 4A ⇓ , isogenic HCT116 p53−/− cells were somewhat more prone to undergoing apoptosis and micronucleation than parental cells after treatment with SN-38 alone, consistent with an intrinsic G₂-M defect in these cells (14) . However, sequential treatment with SN-38 followed by UCN-01 markedly enhanced the induction of apoptosis and micronucleation in p53-null cells (14 and 2%, respectively, in parental, and 21 and 23%, respectively, in p53-null cells at 24 hours after UCN-01; Fig. 4A ⇓ ).

The ability of UCN-01 to abrogate the G₂ checkpoint induced by SN-38 was compared between the two cell lines, using biparameter flow cytometry (Fig. 4B) ⇓ . Treatment with SN-38 alone for 24 hours resulted in a loss of mitosis in both cell lines. However, p53-null cells showed a gradual increase in mitosis after SN-38 removal, whereas parental remained arrested in G₂, indicating that maintenance of the G₂ checkpoint was compromised when p53 was absent (Fig. 4B) ⇓ . This intrinsic checkpoint deficiency was markedly enhanced by UCN-01 as 43 ± 3% (n = 3) of the G₂-arrested p53-null cells were at mitosis 9 hours after UCN-01 addition compared with a peak mitotic index of only 11 ± 4% (n = 3) at 9 hours in parental HCT116 cells (Fig. 4B) ⇓ . Time-lapse studies confirmed that the higher MPM-2 labeling in p53-null cells was because of an increase in aberrant mitotic entry but not to an increase in mitosis transit time (data not shown). Therefore, we conclude that p53-null HCT116 cells are selectively more susceptible to undergoing apoptosis and micronucleation induced by sequential SN-38 and UCN-01 treatment than parental cells by virtue of an intrinsically defective G₂ checkpoint as a result of p53 loss, which is additionally compromised by pharmacological disruption of the chk1-mediated pathway by UCN-01. Although it has been reported that premature mitosis after SN-38 treatment could be induced by a lower dose of UCN-01 (<10 nm) in cells with mutant p53 (38) , we did not observe any significant increase in G₂ checkpoint abrogation when UCN-01 concentration was <50 nm even in HCT116 p53−/− cells (data not shown).

Concurrent but not Sequential Treatment with SN-38 and UCN-01 Results in an Increased Loss of Clonogenicity Independent of p53 Status.

To corroborate our observation on mitotic cell death associated with G₂ checkpoint abrogation, we performed colony formation assays on parental and p53-null HCT116 cells treated with SN-38 and UCN-01. Surprisingly, whereas sequential treatment with SN-38 followed by UCN-01 resulted in marked enhancement of apoptosis via mitotic catastrophe, particularly in p53-null cells (Fig. 4) ⇓ , this was not accompanied by an increased loss of clonal growth over that caused by SN-38 alone (Fig. 5) ⇓ . On the other hand, concurrent exposure with SN-38 and UCN-01, a treatment schedule that did not cause apoptosis, resulted in potentiation of inhibition of colony formation in both parental and p53-null cells (Fig. 5) ⇓ . Because concurrent treatment did not lead to G₂ checkpoint abrogation and premature mitosis (Fig. 2) ⇓ ⇓ , the increased loss of clonogenic survival resulted from this treatment schedule must involve a different and yet undefined mechanism. We have attempted to quantify the percentage of senescence cells induced by treatment with 5 nm SN-38 alone or with the combination followed by no drug in parental HCT116 cells but did not find appreciable difference between the two [35.3 ± 4.0% versus 34 ± 5.2% (n = 3), respectively; P > 0.05]. However, we could not exclude the possibility that our inability to detect difference in senescence was related to the qualitative rather than quantitative nature of the assay.

Concurrent Treatment with SN-38 and UCN-01 Results in Abrogation of DNA Replication Checkpoint and an Enhanced DNA Damage Response.

To probe for the biochemical basis of the increased inhibition of clonogenicity with concurrent SN-38 and UCN-01 treatment when apoptosis was not observed, expression of key DNA-damage regulators were studied. The histone variant H2AX is rapidly phosphorylated at serine 139 (referred as γ-H2AX) in response to double-strand breakage induced by DNA-damaging agents and can therefore be used as a surrogate for double-strand breakage induced by topoisomerase I poison (40) . As expected, 200 nm UCN-01 alone did not cause any increase in S139 phosphorylation (Fig. 6A) ⇓ . Concurrent treatment with SN-38 and UCN-01 resulted in a much higher level of γ-H2AX than single-agent SN-38, indicating an increased number of double-strand breakage accumulation caused by the combination (Fig. 6A) ⇓ . This enhancement of DNA damage was independent of p53 status (Fig. 6A) ⇓ . Conversely, sequential treatment with SN-38 followed by UCN-01 did not cause any increase in H2AX S139 phosphorylation in both cell lines up to 8 hours after UCN-01 addition, indicating that aberrant mitotic entry per se did not result in additional increase in double-strand breakage (Fig. 6A) ⇓ . The increase in γ-H2AX seen in p53-null cells treated sequentially by SN-38 and UCN-01 was presumably because of apoptosis that was induced after mitotic catastrophe.

An enhanced genotoxic response observed with concurrent SN-38 and UCN-01 treatment was substantiated by studying the expression of other DNA damage markers. Thus, the kinetics of induction of the tumor suppressor p53 and its downstream effector p21 was also accelerated by concurrent, over single-agent treatment (Fig. 6B) ⇓ . Under these conditions, UCN-01 alone did not induce p53 or p21 (Fig. 6B) ⇓ . To additionally investigate the basis for the enhanced checkpoint response after concomitant treatment, DNA synthesis was measured by thymidine incorporation in HCT116 cells treated with SN-38 in the absence or presence of UCN-01. SN-38 alone reduced DNA synthesis to 48 ± 3%, and concomitant treatment with UCN-01 resulted in increase of DNA synthesis back to 75 ± 2% (Fig. 6B) ⇓ . A similar extent of checkpoint override by UCN-01 was seen also with a lower dose of SN-38 (5 nm; data not shown). These results confirm that concurrent treatment with UCN-01 abrogates the SN-38 induced S-phase checkpoint and suggest that an increase in double-strand breakage occurs when DNA replication was not optimally suppressed after genomic damage by SN-38.

DISCUSSION

Dysfunction in cell cycle checkpoint appears to be a universal phenomenon in human cancers. Combination treatment with genotoxic agents and a small molecular checkpoint inhibitor such as UCN-01 provides an opportunity for targeting tumors with intrinsic checkpoint defects. A substantial amount of literature has shown that UCN-01, at nontoxic doses, can potentiate the cytotoxicity of a variety of DNA-damaging agents. Controversy exists as to whether UCN-01 selectively sensitizes cells with mutated p53 to DNA damage (28 , 29 , 31 , 32) . We now report that depending on the sequence of drug administration and the end point chosen for assessing cytotoxicity, different conclusions can be drawn regarding the pharmacological consequence of checkpoint abrogation by UCN-01 as well as the influence of p53 status on treatment outcome. Current experimental paradigm allows examination of the effect of G₂ checkpoint abrogation specifically without a contribution from the effect of disrupting the S-phase checkpoint. Thus, treatment of HCT116 cells with SN-38 alone resulted in a senescence-like sustained G₂ arrest without apoptosis. Sequential treatment with SN-38 followed by UCN-01 abrogated the G₂ checkpoint, and cells entered an aberrant mitosis, which ultimately led to cell death. Isogenic p53-null HCT116 cells were clearly more susceptible to undergoing checkpoint abrogation and subsequent apoptosis induced by sequential treatment than parental cells, strongly suggesting that chk1 and p53-dependent signaling pathways cooperate to maintain the G₂ checkpoint. Despite an enhancement of apoptotic cell death by UCN-01 on cells pretreated with SN-38, sequential treatment did not result in increased loss of clonogenicity above that induced by SN-38 alone in both parental and p53-null HCT116 cells. How do we reconcile this disparity? It is generally assumed that the function of the G₂ checkpoint is to allow time for sublethal DNA damage to be repaired and cell cycle progression can resume thereafter; disruption of this otherwise protective G₂ arrest by a checkpoint inhibitor would sensitize cells to genotoxic agents. This paradigm does not seem to be true in our model system. Thus, treatment with SN-38 alone resulted in a senescence-like G₂ arrest even on removal of the drug. Abrogation of the G₂ checkpoint by UCN-01 converted this permanent arrest state into apoptotic cell death via mitotic catastrophe. Although both processes (senescence and mitotic catastrophe) can result in loss of clonal growth in a colony formation assay, sequential treatment with UCN-01 did not lead to increased inhibition of clonogenicity, implying that cell cycle arrest and DNA repair are not coupled at the G₂ checkpoint in these cells.

The term mitotic catastrophe was originally used to describe the lethal temperature-sensitive phenotype displayed by certain mutant yeast strains that underwent mitosis with gross abnormalities of chromosome segregation (41 , 42) . It is now being referred in the literature as a form of reproductive cell death caused by an aberrant mitosis induced by either mitotic spindle poisons or premature mitotic entry after checkpoint abrogation in the face of DNA damage (this study; for review, see ref. 37 ). A morphologic hallmark of mitotic catastrophe is the formation of interphase cells with micronuclei. Using GFP-histone 2B-transfected cells and time-lapse fluorescent microscopy, we investigated the cellular processes leading to micronucleation and the fate of these cells. We found that after sequential SN-38 and UCN-01 treatment, cells that entered mitosis fail to align all of the condensed chromosomes at the metaphase plate. During mitotic exit, the chromatin became fragmented into lobulated masses, which subsequently underwent decondensation and appeared as interphase cells with micronuclei. Approximately one-third of the micronucleated cells committed to apoptosis on exit from mitosis (movie 1); the remaining cells either arrested in G₁ in a p53-dependent manner or, in the case of p53-null cells, underwent endoduplication of DNA and developed polypoidy. 1 Using live cell imaging technique, we have convincingly shown that apoptosis induced by sequential exposure to SN-38 and UCN-01 must be preceded by an aberrant mitosis. However, how aberrant mitosis and apoptosis are coupled after G₂ checkpoint abrogation remains elusive. The mechanism responsible for the observed mitotic aberrations is also unknown. Surprisingly, improper segregation of chromosomes during mitotic catastrophe was not resultant from a defect in bipolar spindle formation (movie 3). Interestingly, the mitotic abnormalities observed in our studies are reminiscent of that found in human cell lines, where key mitotic checkpoint regulators such as BUBR1, MAD2, and aurora-B have been disrupted, raising the possibility of an asynchrony of activities among mitotic regulators as the cause of aberrant mitosis after checkpoint abrogation (43, 44, 45) .

Type I topoisomerase produces transient single-strand nicks by forming reversible cleavage complexes with duplex DNA during replication, transcription, and repair. Camptothecin analogs such as SN-38 stabilizes these DNA-topoisomerase complexes and can convert single-strand DNA breaks to double-strand in cells during S-phase when advancing replication forks collide with the cleavage complexes (46 , 47) . Once double-strand breakages are generated, initiation of new replication and elongation of nascent chains are inhibited by checkpoint mechanisms. This intra-S-phase checkpoint appears to be mediated in trans by several checkpoint kinases, including ATM, ATR, chk1, and chk2 (6 , 8 , 48 , 49) . Inhibition of chk1/chk2 function by UCN-01 seems to divest the cells of this protective mechanism in response to DNA damage by SN-38 and results in additional double-strand breakages. Consistent with this is the presence of continued DNA synthesis and enhanced formation of the double-strand breakage marker γ-H2AX when cells were treated with concurrent SN-38 and UCN-01 (Fig. 6, A and C) ⇓ . Our data indicated that concurrent treatment with UCN-01 actually enhanced the checkpoint response induced by SN-38, including an up-regulation of p53 and p21 (Fig. 6B) ⇓ . This observation is in sharp contrast with the results of a recent report, which showed that UCN-01 abolished the up-regulation of p53 induced by ionizing radiation in HCT116 cells (31) . It is unclear whether such a discrepancy is related to the specific genotoxic agent used. It has been postulated that chk2, rather than chk1, is the primary checkpoint kinase responsible for p53 stabilization in response to ionizing radiation-induced DNA damage (31) . It is possible that chk1 might represent the predominant checkpoint regulator after treatment with SN-38 in these cells.

In summary, we have determined that the biochemical mechanism underlying the enhancement of cytotoxicity of SN-38 by sequential and concurrent treatment with UCN-01 in HCT116 cells is complex and disparate. Sequential exposure of cells with SN-38 followed by UCN-01 abrogates the G₂ checkpoint induced by SN-38 and results in the conversion of a senescence-like G₂ arrest into mitotic catastrophe and subsequently micronucleation as well as apoptosis. Isogenic p53-null cells, by virtue of an intrinsic G₂ checkpoint defect, are clearly more susceptible to undergoing mitotic catastrophe induced by sequential treatment. However, this did not translate into a reduction in clonogenicity as determined by a colony formation assay. In contrast, concurrent treatment with SN-38 and UCN-01 results in abrogation of the S-phase but not the G₂ checkpoint, an enhanced DNA damage response as well as a reduced clonogenicity, which is independent of p53 status. At this juncture, it is unclear from the literature which mode of reproductive cell death in tumors, namely, cellular senescence, apoptosis, and mitotic catastrophe, is the desirable pharmacological end point in cancer therapy (37 , 50) . In a primary murine lymphoma model, it has been shown that both drug-induced apoptosis and senescence contribute to chemotherapeutic outcome; the former effect is more related to tumor regression, whereas the latter to disease stability (50) . The role of mitotic catastrophe in treatment outcome has yet to be determined. Commonly used methods for assessing cytotoxicity of drug treatment, such as the clonogenicity assay, do not necessarily distinguish among the different mechanisms of reproductive death. Thus, Waldman et al. (51) have shown that checkpoint defective p21−/− HCT116 cells were more sensitive than its parental counterpart to γ-radiation in a xenograft model; this difference in sensitivity was not apparent in a colony formation assay. Of note, our results did not specifically address the issue as to whether the S- or G₂-phase checkpoint is the preferred target for combining with cytotoxic agents in cancer therapy. The fact that UCN-01 binds avidly to serum proteins in patients, resulting in prolonged half-life of the drug, might make targeting each specific checkpoint technically difficult in the clinic. The availability of newer generations of checkpoint inhibitors with lower affinity for serum proteins might circumvent this pharmacokinetic constraint (52) . Nevertheless, UCN-01 remains an invaluable pharmacological tool for testing these new concepts in cancer therapeutics.