Spindle Checkpoint Function Is Required for Mitotic Catastrophe Induced by DNA-Damaging Agents

Spindle checkpoint function is required for mitotic catastrophe induced by DNA-damaging agents

Masayuki Nitta1,2,3, Osamu Kobayashi1,2,3, Shinobu Honda1, Toru Hirota1, Shinji Kuninaka1, Tomotoshi Marumoto1, Yukitaka Ushio2 and Hideyuki Saya*,1

1Department of Tumor Genetics and Biology, Graduate School of Medical Sciences, University of Kumamoto, 1-1-1 Honjo, Kumamoto 860-8556, Japan; 2Department of Neurosurgery, Graduate School of Medical Sciences, University of Kumamoto, Kumamoto, Japan

Mitotic catastrophe is an important mechanism for the ments such as chemotherapy and radiation. Most cancer induction of cell death in cancer cells by antineoplastic cells are deficient in G1 checkpoint function and agents that damage DNA. This process is facilitated by therefore fail to arrest in G1 phase on exposure to geno- defects in the G1 and G2 checkpoints of the cell cycle that toxic agents. Instead, they accumulate temporarily in are apparent in most cancer cells and which allow the cells G2 phase. However, given that the G2 checkpoint is to enter mitosis with DNA damage. We have now also partially impaired in cancer cells, they are unable characterized the dynamics of mitotic catastrophe induced to maintain G2 arrest and eventually die as they enter by DNA-damaging agents in p53-deficient cancer cells. mitosis (Fingert et al., 1986; Chan et al., 1999). This Cells that entered mitosis with DNA damage transiently process is known as mitotic catastrophe or mitotic death arrested at metaphase for more than 10 h without (Chan et al., 1999; Roninson et al., 2001). segregation of chromosomes and subsequently died Downregulation or inhibition of G2 checkpoint directly from metaphase. In those metaphase arrested proteins, including p53, p21, and 14-3-3, induces G2 precatastrophic cells, anaphase-promoting complex ap- checkpoint abrogation and subsequent catastrophic peared to be inactivated and BubR1 was persistently death in response to DNA damage (Bunz et al., 1998; localized at kinetochores, suggesting that spindle check- Chan et al., 1999, 2000). A new strategy has therefore point is activated after the DNA damage. Furthermore, been developed to sensitize cancer cells to the effects suppression of spindle checkpoint function by BubR1 or of genotoxic agents with the use of small molecules that Mad2RNA interference in the DNA damaged cells led inhibit the induction of G2 arrest by DNA damage to escape from catastrophic death and to subsequent (Sausville et al., 2001). The induction of mitotic abnormal mitosis. Dysfunction of the spindle checkpoint catastrophe is thus an important goal of cancer in p53-deficient cancer cells is thus likely a critical factor therapies using DNA-damaging agents. However, mito- in resistance to DNA-damaging therapeutic agents. tic catastrophe has been characterized mostly by Oncogene (2004) 23, 6548–6558. doi:10.1038/sj.onc.1207873 morphological observations, with the underlying mole- Published online 28 June 2004 cular mechanism remaining largely unknown in human cells. Keywords: mitotic catastrophe; spindle checkpoint; The spindle checkpoint inhibits the activity of the DNA damage; p53; time lapse; BubR1 anaphase-promoting complex (APC) and blocks the transition from metaphase to anaphase until all chromosomes are aligned at the metaphase plate and achieve bipolar attachment to the mitotic spindle, thus Introduction ensuring accurate sister chromatid segregation during cell division. Impairment of spindle checkpoint function Cell cycle checkpoints prevent transition from one phase results in premature mitosis and missegregation of of the cell cycle to the next until all processes of the chromosomes, leading to aneuploidy (Michel et al., present phase are completed. Defects in the checkpoint 2001). The signal activating spindle checkpoint is functions result in gene mutations and chromosome supposed to be generated from kinetochores that either damages, which contribute to the development and fail to bind spindle microtubules or are not experiencing progression of tumors. However, loss of checkpoint the tension that normally results from bipolar attach- function in some cancer cells is considered to be ment to the spindle. Several proteins that localize to associated with their sensitivity to antineoplastic treat- kinetochores, including Mad1, Mad2, BubR1 (Mad3), Bub1, Bub3, and Mps1, contribute to generation of the spindle checkpoint signal. Recent observations have *Correspondence: H Saya; E-mail: [email protected] 3These authors contributed equally to this work shown that cells having spindle checkpoint abnormal- Received 6 February 2004; revised 26 April 2004; accepted 26 April 2004; ities are insensitive to Taxol, a spindle poison (O’Con- published online 28 June 2004 nor et al., 2002; Anand et al., 2003). However, in case of Mitotic catastrophe in cancer cells M Nitta et al 6549 DNA-damaging agents, it is largely unknown whether phosphorylation of histone H2AX on serine-139 (Ro- dysfunction of the spindle checkpoint correlates with the gakou et al., 1998), which then results in the formation ineffectiveness of these drugs aimed at inducing mitotic of g-H2AX complexes that can be detected by the catastrophe. antibodies (Rogakou et al., 1999). Given that DBSs We have now characterized the dynamics of mitotic are introduced in cells even in an ordinary cultured catastrophe induced by DNA damage in p53-deficient condition (Haber, 1999), the g-H2AX complexes were cancer cells. Most cells entering mitosis with DNA detected in a small population of cultured cells. damage arrested at metaphase and subsequently under- However, treatment with aphidicolin induced a marked went cell death. Furthermore, metaphase arrest prior increase in the number of g-H2AX-positive spots to the catastrophe was clearly shown to result from the apparent in the nuclei of both p53 þ / þ and p53À/À activation of the spindle checkpoint, and inhibition of HCT116 cells in G2 phase (Figure 1d, left and right checkpoint function using RNA interference allowed panels). In addition, multiple g-H2AX-positive spots cancer cells to escape mitotic catastrophe. Our findings were detected on the chromosomes of p53À/À cells that demonstrate that DNA damage can activate spindle subsequently entered mitosis (Figure 1d, right panel). checkpoint in cells with impaired G2 checkpoint Exposure to ionizing radiation at G2 phase, which function in a way where persistent kinetochore localiza- also induces DNA damage, showed effects similar to tion of BubR1 seems to have a critical role. The spindle those of aphidicolin treatment in p53 þ / þ and p53À/À checkpoint function is thus required for the induction of HCT116 cells. The cells were synchronized in G0 phase catastrophe in cancer cells treated with DNA-damaging by serum deprivation, returned to serum-containing antineoplastic agents. medium and, after 12 h, exposed to 20Gy of X- radiation. The p53À/À cells entered mitosis after a transient G2 delay and underwent cell death (see Supplemental Figure 1). Results Aphidicolin induced the expression of both p53 and p21 proteins in p53 þ / þ HCT116 cells, but no p53 and Loss of G2 checkpoint function results in mitotic cell death À/À after DNA damage very little p21 proteins in p53 cells (Figure 1e and f). Consistent with previous observations (Bunz et al., Cancer cells with impaired G2 checkpoint exhibit an 1998, 1999), these results suggest that the failure to increased sensitivity to DNA-damaging agents, to which induce sufficient expression of p21 in the p53-deficient they respond by undergoing mitotic death (Bunz et al., cells results in G2 checkpoint abrogation. To confirm 1999; Borel et al., 2002). We established a system to that the death of p53À/À HCT116 cells treated with investigate mitotic death in response to DNA damage. aphidicolin occurs during mitosis, we blocked the G2–M To avoid the effect of G1 checkpoint, we used transition in these cells by exposure to butyrolactone I, aphidicolin to induce DNA damage. This drug, at which specifically inhibits the activity of cyclin-depen- concentrations between 0.3 and 1.0 mg/ml, inhibits DNA dent kinase 1 (Cdk1), 24 h after the onset of aphidicolin polymerases a, d,ande, and thereby induces a treatment. Butyrolactone I (50 mM) induced G2 arrest prolonged stall in DNA replication that prevents its and significantly suppressed cell death (see Supplemen- completion (Borel et al., 2002). The resulting gaps in the tal Figure 2). These results thus indicate that cells having double-stranded DNA are recognized as DNA strand DNA damage with impaired G2 checkpoint undergo cell breaks (Toledo et al., 2000) and activate the DNA death after they enter mitosis. þ / þ damage-induced G2 checkpoint. We treated p53 and À/À p53 HCT116 human colon cancer cells with aphidi- DNA damage induces metaphase arrest and subsequent colin (0.5 mg/ml) and examined cell cycle distribution by death in p53-deficient cells flow cytometry. Both p53 þ / þ and p53À/À cells transiently arrested at S phase and subsequently entered G2 phase We next analysed the dynamics of mitotic catastrophe þ / þ (Figure 1a). The p53 cells arrested in G2 phase and triggered by aphidicolin-induced DNA damage. To did not enter mitosis, whereas the p53À/À cells were facilitate visualization of the kinetics of chromosomes À/À unable to maintain G2 arrest and entered mitosis (Figure in living cells, we used HeLa and p53 HCT116 cells 1a and b). Furthermore, less than 15% of p53 þ / þ cells that stably express green fluorescent protein (GFP)- treated with aphidicolin for 96 h underwent cell death tagged histone H2B. Time-lapse fluorescence and (Figure 1c). In contrast, B50% of p53À/À cells under- differential interference contrast (DIC) microscopy went cell death after entry into mitosis. Similar to p53À/À using GFP-HH2B-expressing HeLa cells revealed that HCT116 cells, HeLa cells, in which the p53 is cells undergo mitotic catastrophe in three phases functionally inactivated by human papilloma virus-E6, (Table 1, Figure 2a, Supplemental video 1). In the first entered mitosis and subsequently underwent cell death phase, the cells entered mitosis after G2 checkpoint after treatment with aphidicolin (Figure 1b and c). abrogation and their chromosomes became aligned To confirm that aphidicolin treatment induces DNA at the metaphase plate. The duration from nu- damage, we examined cells for the presence of the g- clear envelope breakdown to metaphase alignment H2AX complex by immunofluorescence analysis using was slightly longer in aphidicolin-treated cells antibodies against phosphorylated histone H2AX. (40.0715.2 min) than in control cells (29.176.0min). Double-strand breaks (DSBs) in DNA induce the rapid The second phase was characterized by the failure

Oncogene Mitotic catastrophe in cancer cells M Nitta et al 6550 of sister chromatid separation and metaphase arrest ‘the precatastrophic phase’. The precatastrophic phase (140.4733.9 min). In the third phase, the chromosomes persisted for 3–10h (380.0 7150.4 min), after which the at the metaphase plate began to be stretched toward the cells collapsed and died. The same pattern of changes spindle poles, with some of them moving to the poles was observed in GFP-HH2B-expressing p53À/À HCT116 without complete segregation; we designated this phase cells after exposure to aphidicolin (data not shown) as

Oncogene Mitotic catastrophe in cancer cells M Nitta et al 6551 Table 1 Duration of the three phases in mitotic catastrophe induced by aphidicolin treatment in HeLa Cells Mitotic catastrophea Normal mitosisb

NEBD metaphase (phase I) 40.0715.2 min 29.176.0min Metaphase arrest (phase II) 140.4733.9 min Metaphase – mitotic exit 17.972.5 min Precatastrophic phase (phase III) 380.07150.4 min Total time 560.47148.7 min 47.075.7 min

Data are means7s.e.m. of values from 30cells. aHeLa/GFP-HH2B cells were treated with aphidicolin and the mitotic process was monitored from 48 h after the onset of drug treatment by time-lapse fluorescence and DIC video microscopy. bCells were synchronized in early S phase by thymidine treatment and monitored from 6 h after release well as in HeLa cells treated with 20Gy of X-radiation with aphidicolin or X-irradiation induced uncondensed in G2 phase (Figure 2b, Supplemental video 2), suggest- and entangled mitotic figures (Figure 4a). Multiple spots ing that the process of mitotic catastrophe observed in of BubR1 immunoreactivity were apparent at kineto- our experiments is not specific to a particular DNA- chores, whereas Mad2 was not detected at kinetochores damaging agent or cell line. in almost all these arrested cells (Figure 4a and b). As positive controls, we examined BubR1 and Mad2 DNA damage induces metaphase arrest by spindle localization in cells treated with monastrol (100 mM checkpoint activation for 24 h) or ICRF-193 (4 mM for 48 h), which induce metaphase arrest by targeting the motor protein Eg-5 We next attempted to explore the origin of the mitotic and topoisomerase II, respectively. Consistent with arrest of cells at precatastrophic phase. Arrest at previous observations (Kapoor et al., 2000; Mikhailov metaphase is usually caused by the activation of the et al., 2002), both BubR1 and Mad2 were located at spindle checkpoint and by the inhibition of proteolysis kinetochores in most cells arrested at metaphase by promoted by the APC. To investigate whether the APC treatment with either of these agents (Figure 4a). These is inactive in precatastrophic metaphase cells, we results thus suggested that the prolongation of meta- examined the expression of cyclin B and securin, the phase induced by DNA damage is attributable to degradation of both of which is mediated by the APC– the activation of BubR1-related spindle checkpoint CDC20complex at the metaphase–anaphase transition. function. We also examined the expression of cyclin A, which is To confirm that spindle checkpoint activation is degraded in prometaphase before metaphase–anaphase required for DNA damage-induced metaphase arrest, transition (den Elzen and Pines, 2001). Immunofluores- we treated HeLa cells with Mad2 or BubR1 RNA cence analysis revealed that both cyclin B and securin, interference. Transfection of the cells with small inter- but not cyclin A, were abundant in the cytoplasm of fering RNAs (siRNAs) specific for Mad2 or BubR1 precatastrophic HeLa and p53À/À HCT116 cells that had mRNAs resulted in time-dependent reduction of the been treated with aphidicolin for 72 h (Figure 3a). corresponding proteins (Figure 4c). Immunohistochem- Immunoblot analysis also showed that, whereas cyclin A istry showed that RNAi effectively wiped out both had been degraded in precatastrophic HeLa cells, both Mad2 and BubR1 stainings on kinetochores of prome- cyclin B and securin accumulated to levels similar to taphase cells (Figure 4d). As described previously (Li those apparent in cells treated with nocodazole (0.1 mg/ and Benezra, 1996), the reduction of Mad2 resulted in ml), which prevents APC activation by triggering the the abrogation of nocodazole-induced prometaphase spindle checkpoint (Figure 3b). These observations arrest (Figure 4e and f); the cells entered into mitosis, indicate that the APC is not activated in the precatas- then quickly exited without cytokinesis and became trophic phase of mitotic death. tetraploid G1 cells. In contrast, cells transfected with a To determine whether the metaphase arrest and control siRNAs were arrested at prometaphase after failure to activate the APC apparent after DNA damage exposure to nocodazole for 24 h. Reduction of BubR1, result from the activation of the spindle checkpoint, like that of Mad2, also abrogated nocodazole-induced we analysed the subcellular distribution of BubR1 and prometaphase arrest (Figure 4f). Next, we transfected Mad2 proteins in metaphase-arrested HeLa cells after cells 24 h after exposure to aphidicolin with Mad2 or exposure to aphidicolin or X-irradiation. Treatment BubR1 siRNAs and monitored the mitotic process by

Figure 1 Abrogation of the G2 checkpoint and aphidicolin-induced mitotic death in p53-deficient cancer cells. (a) Cell cycle distribution of p53 þ / þ and p53À/À HCT116 cells after treatment with aphidicolin. Cells in mitosis were detected with antibodies to phosphorylated histone H3 (P-H3) as shown in the dot plots. (b) Time course of the mitotic index of p53 þ / þ HCT116 cells (filled circles), p53À/À HCT116 cells (open circles), and HeLa cells (filled squares) treated with aphidicolin. (c) Time course of the induction of cell death in p53 þ / þ HCT116 cells, p53À/À HCT116 cells, and HeLa cells treated with aphidicolin. The percentage of dead cells with a sub-G1 DNA content was determined by flow cytometry. (d) Immunofluorescence analysis of the formation of g-H2AX complexes (green) in p53 þ / þ and p53À/À HCT116 cells treated with aphidicolin for 48 h. Nuclei (red) were stained with PI. (e) Immunofluorescence analysis of the expression of p53 and p21 proteins in p53 þ / þ and p53À/À HCT116 cells treated with aphidicolin for 48 h. (f) Immunoblot analysis of p53 and p21 in p53 þ / þ and p53À/À HCT116 cells treated with aphidicolin. All data are representative of at least three independent experiments; those in (b and c) are means7s.e.m. of values from three such experiments

Oncogene Mitotic catastrophe in cancer cells M Nitta et al 6552

Figure 2 Dynamics of mitotic catastrophe induced by DNA damage in HeLa/GFP-HH2B cells. The process of mitosis was observed by time-lapse ﬂuorescence video microscopy. (a) Cells were treated with aphidicolin. Time is shown in hours:minutes, with entry into mitosis characterized by chromatin condensation at 00 : 00. (b) Cells were exposed to X-radiation (20Gy) 6 h after release from synchronization in early S phase by thymidine block

light microscopy and flow cytometry. Cells depleted BubR1-deficient cells did not (Figure 4h). Given that Mad2 or BubR1 escaped from metaphase arrest and Mad2 and BubR1 are thought to act cooperatively in a rapidly divided (Figure 4e and f). Similarly, Mad2- or single pathway to inhibit the APC (Shannon et al., BubR1-deficient cells in which DNA damage was 2002), our results demonstrate that metaphase arrest induced by ionizing radiation failed to undergo mitotic induced by DNA damage is mediated by spindle arrest (Figure 4e and f). Flow cytometry revealed that checkpoint activation. the Mad2-deficient cells did not accumulate in M phase after aphidicolin treatment but rather divided, entered Spindle checkpoint function is essential for DNA damage- the next G phase and proceeded to S phase (Figure 4g). 1 induced mitotic catastrophe Whereas the mitotic index of cells transfected with the control siRNA increased in a time-dependent Finally, we examined whether spindle checkpoint manner after aphidicolin treatment, that of Mad2- or function is required for the induction of mitotic

Oncogene Mitotic catastrophe in cancer cells M Nitta et al 6553 video 3). BubR1-deficient cells exposed to aphidicolin progressed to metaphase but were unable to maintain metaphase arrest and divided rapidly (Figure 5a, Supplemental video 4). Moreover, reduction of Mad2 or BubR1 resulted in marked suppression of cell death induced by aphidicolin (Figure 5b). In particular, focusing on cells that had entered mitosis by time-lapse video microscopy, only 6.3 or 14.3% of those deficient in Mad2 or BubR1, respectively, underwent mitotic death, whereas 48.6% of those transfected with the control siRNA did so (Figure 5c). These data suggest that suppression of spindle checkpoint function reduces the sensitivity of cancer cells to DNA-damaging agents. Furthermore, the aphidicolin-treated Mad2- or BubR1- deficient cells underwent cell division with severe DNA damage and manifested an irregular nuclear morphology (Figure 5d). Therefore, spindle checkpoint function is essential not only for induction of mitotic catastrophe but also for prevention of the marked morphological changes characteristic of cancer cells that have under- gone DNA damage.

Discussion

We have shown that mitotic catastrophe requires not only a defective G2 checkpoint but also intact spindle checkpoint function. Dysfunction of the spindle checkpoint in cancer cells may thus confer resistance to DNA- damaging antineoplastic drugs. Our analysis of the dynamics of mitotic catastrophe induced by DNA damage in cancer cells with impaired G2 checkpoint function revealed that metaphase arrest is an important event in this process. Recent studies with budding yeast have shown that DNA damage induces metaphase arrest through activation of the spindle checkpoint pathway (Garber and Rine, 2002; Maringele and Lydall, 2002). Phosphorylation of Pds1 by Chk1 has also been shown to be essential for metaphase arrest after DNA damage in budding yeast (Wang et al., 2001). In some human cancer cells, the DNA damage-induced metaphase arrest has also been observed previously Figure 3 Failure of APC activation in p53-deficient cells (Erenpreisa et al., 2000). However, the mechanism of in the precatastrophic phase of mitotic death induced by this effect has been unclear. Three lines of evidence DNA damage. (a) Immunofluorescence analysis of the presented here suggest that the prolonged metaphase expression and subcellular localization of cyclin A, cyclin B, and securin in HeLa and p53À/À HCT116 cells treated arrest induced by DNA damage in p53-deficient cancer with aphidicolin for 72 h. (b) Immunoblot analysis of the cells is induced by the activation of the spindle expression of cyclin A, cyclin B, and securin in HeLa cells checkpoint. First, cyclin B and securin, both of which treated with nocodazole for 24 h or with aphidicolin for 24 are essential targets of the APC, accumulated in the or 72 h. The mitotic fractions of cells treated with nocodazole or aphidicolin (72 h) were collected by a arrested cells, suggesting that the APC was not ‘mitotic shake-off’ method activated. Second, the spindle checkpoint kinase BubR1 was detected at kinetochores during the metaphase arrest. And third, reduction of either BubR1 or Mad2 catastrophe in p53-deficient cancer cells by DNA by RNA interference abolished the metaphase arrest, damage. We monitored the effects of Mad2 or BubR1 resulting in abnormal mitosis. Given that BubR1, but reduction on mitotic catastrophe in HeLa/GFP-HH2B not Mad2, was localized at the kinetochores of cells by time-lapse video microscopy. Cells deficient in precatastrophic cells and that BubR1 siRNAs sufficient Mad2 by RNA interference underwent aberrant cell to wipe out this molecule on kinetochores abrogated division from prometaphase, without metaphase arrest, metaphase arrest, existence of BubR1 on kinetochores after exposure to aphidicolin (Figure 5a, Supplemental seems to be critical for the activation of the spindle

Oncogene Mitotic catastrophe in cancer cells M Nitta et al 6554 checkpoint. Since Mad2 was not detected at kineto- kinetochores, we speculate that some of sister chroma- chores of precatastrophic chromosomes, the microtu- tids may not be attached in an amphitelic manner in the bules–kinetochore attachment seems to be completed. precatastrophic cells. This hypothesis is supported by Having known that the loss of interkinetochore tension our time-lapse movies data that cells at precatastrophic inhibits the release of BubR1 from the attached phase showed the dynamic movement of chromosomes

Figure 4 Activation of spindle checkpoint in precatastrophic HeLa cells. (a) Immunofluorescence analysis of BubR1 and Mad2 localization in control (left) and in cells exposed to various drugs or X-radiation (right). Cells were treated with aphidicolin for 48 h or were irradiated (20Gy) 6 h after release from thymidine block and examined 18 h later. (b) The percentage of cells that exhibited Mad2 or BubR1 immunoreactivity at kinetochores. Data were obtained from 20cells randomly selected for each mitotic phase. ( c) Reduction of Mad2 or BubR1 by specific siRNAs. The Mad2 and BubR1 immunoreactive bands are shown for cells transfected with the corresponding specific siRNA. (d) Cells were transfected with each siRNAs for 24 h and were fixed and stained with corresponding antibodies. (e) Protocols for siRNA transfection and treatment of cells with nocodazole, aphidicolin, or X-radiation. (f) Phase-contrast microscopy of cells subjected to the protocols shown in (e). (g) Cells were transfected with control or Mad2 siRNAs between 24 and 48 h of an incubation with aphidicolin for 72 h. Cell cycle distribution was analysed by flow cytometry. (h) Time course of the mitotic index of cells transfected with control (filled squares), Mad2 (filled circles), or BubR1 (open circles) siRNAs during incubation with aphidicolin as in (g)

Oncogene Mitotic catastrophe in cancer cells M Nitta et al 6555

Figure 4 (Continued) and some fragments of chromosomes were actively Mad2 and BubR1 have been proposed to act stretched and dislocated from the metaphase plate to independently as sensors for the spindle checkpoint, in one of the poles. Since these dynamic movements of which Mad2 detecting for kinetochore microtubule chromosomes during precatastrophic phase were im- attachment and BubR1 detecting for interkinetochore mediately eliminated by the depolymerization of micro- tension, but to function cooperatively in a single tubules by nocodazole treatment (see Supplemental pathway to inhibit APC activity, resulting in metaphase video 5), each chromatid is subjected to poleward arrest (Shannon et al., 2002). Therefore, the BubR1- pulling force generated by spindle microtubules. dependent mitotic arrest after DNA damage was

Oncogene Mitotic catastrophe in cancer cells M Nitta et al 6556

Figure 5 Suppression of DNA damage-induced mitotic catastrophe by reduction of Mad2 or BubR1 in HeLa cells. (a) Dynamics of mitotic progression in HeLa/GFP-HH2B cells transfected with Mad2 or BubR1 siRNAs during aphidicolin treatment. The cells were monitored from 48 h after the onset of aphidicolin treatment. (b) HeLa cells were transfected with control (filled squares), Mad2 (filled circles), or BubR1 (open circles) siRNAs 24 h after the onset of exposure to aphidicolin for 72 h. The percentage of cell death was determined from a sub-G1 population on flow cytometry. (c) The percentage of aphidicolin-treated, siRNA-transfected HeLa/GFP- HH2B cells that underwent catastrophic death among cells that had entered mitosis was determined from time-lapse records similar to those shown in (a). Data were obtained from 30cells randomly selected in a representative experiment. ( d) HeLa cells transfected with control, Mad2, or BubR1 siRNAs were fixed 48 h after the onset of aphidicolin treatment and stained with PI and antibodies to g-H2AX

Oncogene Mitotic catastrophe in cancer cells M Nitta et al 6557 skipped not only by BubR1 depletion but also by Mad2 Materials and methods depletion (Figure 5a). Consistent with previous observations (Bunz et al., Cell culture and chemicals 1998, 1999), we have shown that p53 plays a critical role HeLa cells were maintained in DMEM-F12 containing 10% þ / þ À/À in maintenance of G2 checkpoint and in prevention of fetal calf serum (FCS). HCT116 (p53 and p53 ) cells were mitotic death in response to DNA damage occurring kindly provided by Dr B Vogelstein and maintained in between S and G2 phases. Given that p53 or its related McCoy-5A containing 10% FCS. Nocodazole (Sigma, St pathway is inactivated in most human cancers, abroga- Louis, MI, USA), aphidicolin (Wako, Osaka, Japan), butyrolactone I (Wako), monastrol (Calbiochem, San Diego, CA, tion of the G2 checkpoint and consequent mitotic catastrophe likely underlie the killing of cancer cells by USA), and ICRF-193 (Zenoaq, Fukushima, Japan) were maintained in DMSO. therapeutic agents. However, whereas some studies have shown that p53 inactivation results in an enhanced Flow cytometry sensitivity of cancer cells to DNA-damaging agents (Fan et al., 1995; Servomaa et al., 1996; Bradford et al., 2003), The cells were collected and fixed in ice-cold 70% ethanol. The others have found that loss of p53 increases cellular cells were treated with rat monoclonal antibodies to phos- resistance to such agents (Brachman et al., 1993; Fan phorylated histone H3 (kindly provided by M Inagaki) (Goto et al., 1994), suggesting that additional factors modify et al., 1999) followed by incubation with FITC-conjugated anti-rat IgG. After further incubation for 30min at 37 1C with the drug sensitivity of p53-deficient cancer cells(Vogt RNase A (100 mg/ml), the cells were stained with propidium et al., 2002). Our results now indicate that spindle iodide (PI) (25 mg/ml) and analysed by FACScan (Becton checkpoint function is a crucial factor for the induction Dickinson, Franklin Lakes, NJ, USA). of mitotic death in p53-deficient cells treated with DNA- damaging agents. Cancer cells with both p53 mutations Time-lapse video microscopy and a defective spindle checkpoint are therefore not able to undergo metaphase arrest in response to DNA- Cells stably expressing GFP-tagged histone H2B (Marumoto et al., 2003) were grown in 35-mm Delta-T dishes (Bioptechs damaging agents; instead, they undergo abnormal Inc., Butler, PA, USA) containing Leibovitz’s CO2-indepen- mitosis and subsequent replication of their damaged dent medium (Gibco). Time-lapse fluorescence and DIC video genomes as shown in Figure 4g. Although a large microscopy was performed with an IX70microscope (Olym- fraction of those damaged cells eventually became a pus, Tokyo, Japan); images (100-ms exposure) were acquired senescence-like phenotype and/or underwent cell death every 5 or 10min with a Sensys-charge-coupled device (CCD) in long-term culture, approximately 5–10% of cells camera controlled by Metamorph imaging software (Universal formed colonies and rapidly grew with an aberrant Imaging, Downingtown, PA, USA). nuclear morphology and a high frequency of micro- nuclei (data not shown). RNA interference Recently, several reports have demonstrated a posi- The sequences of the siRNAs specific for human Mad2 or tive relationship between spindle checkpoint impairment BubR1 were 50-ACCUUUACUCGAGUGCAGA-30 (corre- and resistance to antimicrotubule agents. (Anand et al., sponding to nucleotides 124–142 relative to the start codon) 2003; Masuda et al., 2003; Sudo et al., 2004). However, and 50-CAAUACUCUUCAGCAGCAG-30 (nucleotides 293– the contribution of spindle checkpoint in inducing cell 311), respectively. A double-stranded RNA targeting luciferase death after treatments of DNA-damaging agents has not (GL-2: 50-CGUACGCGGAAUACUUCGAdTdT-30) was been sufficiently studied previously. Masuda et al. (2003) used as a control. Annealing of the component strands of have recently shown that there is no significant each siRNA and transfection were performed as described (Elbashir et al., 2001). difference in the susceptibility to cis-platin between nine spindle checkpoint-proficient and four spindle Immunofluorescence analysis checkpoint-impaired human cancer cells. However, as we have shown in this study, not only the status of Cells grown in 35-mm Petri dishes were fixed for 20min at spindle checkpoint but also that of p53 is important for room temperature with 4% paraformaldehyde in phosphate- determining the sensitivity of cancer cells to DNA- buffered saline (PBS) and then permeabilized with 0.2% Triton damaging agents. Spindle checkpoint function is im- X-100 in PBS. The cells were incubated first with primary 1 paired in many types of cancer cells (Li and Benezra, antibodies overnight at 4 C and then for 45 min at room temperature with FITC-conjugated secondary antibodies 1996; Takahashi et al., 1999; Saeki et al., 2002; (Biosource). The stained cells were examined with a confocal Wang et al., 2002; Lens et al., 2003) even if there microscope (FV300, Olympus). have been found few mutation of candidate genes (Cahill et al., 1998; Ru et al., 2002). It is possible Antibodies that spindle checkpoint function is affected by protein level. In fact, haploinsufficiency of Mad2 gene Rabbit antiserum against Mad2 was generated by injecting induces downregulation of Mad2 protein and loss of rabbits with recombinant GST-hsMad2 fusion protein. Mouse monoclonal antibodies to cyclin A, cyclin B1, and a-tubulin spindle checkpoint (Michel et al., 2001). Assessment of (B512) were from Transduction Laboratories. Mouse mono- spindle checkpoint function should thus provide im- clonal antibodies to p53 (DO-1) and rabbit polyclonal portant information for clinical decision-making with antibodies to p21 (C19) were from Santa Cruz Biotechnology regard to the most appropriate treatment for advanced (Santa Cruz, CA, USA). Rabbit polyclonal antibodies to tumors. securin were from Zymed (South San Francisco, CA, USA).

Oncogene Mitotic catastrophe in cancer cells M Nitta et al 6558 Mouse monoclonal antibodies to BubR1 were gently kindly Chase Cancer Center) for the anti-BubR1 monoclonal provided by Dr T Yen. antibodies; Dr WM Bonner for the anti-g-H2AX antibodies; Drs GM Wahl and T Kanda for the GFP-tagged histone H2B expression plasmid; Dr K Brocklehurst for editorial Abbreviations assistance; Y Fukushima for help in preparing the manuscript; APC, anaphase-promoting complex; GFP, green ﬂuorescent members of the Saya lab for valuable suggestions; and protein; DIC, differential interference contrast; CCD, charge- members of the Gene Technology Center at Kumamoto coupled device. University for their contributions to the technical assistance. This work was supported by the Research for the Future Acknowledgements program of the Japan Society for the promotion of Science We thank Dr AB Pardee for helpful discussions and critical and by a grant for Cancer Research from the Ministry of reading; Dr B Vogelstein (Johns Hopkins University) for Education, Culture, Sports, Science, and Technology of Japan providing p53 þ / þ and p53À/À HCT116 cells; Dr T Yen (Fox (to HS).

References

Anand S, Penrhyn-Lowe S and Venkitaraman AR. (2003). Li Y and Benezra R. (1996). Science, 274, 246–248. Cancer Cell, 3, 51–62. Maringele L and Lydall D. (2002). Genes Dev., 16, 1919–1933. Borel F, Lacroix FB and Margolis RL. (2002). J. Cell Sci., 115, Marumoto T, Honda S, Hara T, Nitta M, Hirota T, Kohmura 2829–2838. E and Saya H. (2003). J. Biol. Chem., 278, 51786–51795. Brachman DG, Beckett M, Graves D, Haraf D, Vokes E and Masuda A, Maeno K, Nakagawa T, Saito H and Takahashi T. Weichselbaum RR. (1993). Cancer Res., 53, 3667–3669. (2003). Am. J. Pathol., 163, 1109–1116. Bradford CR, Zhu S, Ogawa H, Ogawa T, Ubell M, Narayan Michel LS, Liberal V, Chatterjee A, Kirchwegger R, Pasche B, A, Johnson G, Wolf GT, Fisher SG and Carey TE. (2003). Gerald W, Dobles M, Sorger PK, Murty VV and Benezra R. Head Neck, 25, 654–661. (2001). Nature, 409, 355–359. Bunz F, Dutriaux A, Lengauer C, Waldman T, Zhou S, Brown Mikhailov A, Cole RW and Rieder CL. (2002). Curr. Biol., 12, JP, Sedivy JM, Kinzler KW and Vogelstein B. (1998). 1797–1806. Science, 282, 1497–1501. O’Connor DS, Wall NR, Porter AC and Altieri DC. (2002). Bunz F, Hwang PM, Torrance C, Waldman T, Zhang Y, Cancer Cell, 2, 43–54. Dillehay L, Williams J, Lengauer C, Kinzler KW and Rogakou EP, Boon C, Redon C and Bonner WM. (1999). Vogelstein B. (1999). J. Clin. Invest., 104, 263–269. J. Cell Biol., 146, 905–916. Cahill DP, Lengauer C, Yu J, Riggins GJ, Willson JK, Rogakou EP, Pilch DR, Orr AH, Ivanova VS and Bonner Markowitz SD, Kinzler KW and Vogelstein B. (1998). WM. (1998). J. Biol. Chem., 273, 5858–5868. Nature, 392, 300–303. Roninson IB, Broude EV and Chang BD. (2001). Drug Resist. Chan TA, Hermeking H, Lengauer C, Kinzler KW and Update, 4, 303–313. Vogelstein B. (1999). Nature, 401, 616–620. Ru HY, Chen RL, Lu WC and Chen JH. (2002). Oncogene, 21, Chan TA, Hwang PM, Hermeking H, Kinzler KW and 4673–4679. Vogelstein B. (2000). Genes Dev., 14, 1584–1588. Saeki A, Tamura S, Ito N, Kiso S, Matsuda Y, Yabuuchi I, den Elzen N and Pines J. (2001). J. Cell Biol., 153, 121–136. Kawata S and Matsuzawa Y. (2002). Cancer, 94, Elbashir SM, Harborth J, Lendeckel W, Yalcin A, Weber K 2047–2054. and Tuschl T. (2001). Nature, 411, 494–498. Sausville EA, Arbuck SG, Messmann R, Headlee D, Bauer Erenpreisa JE, Ivanov A, Dekena G, Vitina A, Krampe R, KS, Lush RM, Murgo A, Figg WD, Lahusen T, Jaken S, Freivalds T, Selivanova G and Roach HI. (2000). Cell Biol. Jing X, Roberge M, Fuse E, Kuwabara T and Senderowicz Int., 24, 61–70. AM. (2001). J. Clin. Oncol., 19, 2319–2333. Fan S, el-Deiry WS, Bae I, Freeman J, Jondle D, Bhatia K, Servomaa K, Kiuru A, Grenman R, Pekkola-Heino K, Fornace Jr AJ, Magrath I, Kohn KW and O’Connor PM. Pulkkinen JO and Rytomaa T. (1996). Cell Prolif., 29, (1994). Cancer Res., 54, 5824–5830. 219–230. Fan S, Smith ML, Rivet DJ, Duba II D, Zhan Q, Kohn KW, Shannon KB, Canman JC and Salmon ED. (2002). Mol. Cell. Fornace Jr AJ and O’Connor PM. (1995). Cancer Res., 55, Biol., 13, 3706–3719. 1649–1654. Sudo T, Nitta M, Saya H and Ueno NT. (2004). Cancer Res., Fingert HJ, Chang JD and Pardee AB. (1986). Cancer Res., 46, 64, 2502–2508. 2463–2467. Takahashi T, Haruki N, Nomoto S, Masuda A, Saji S and Garber PM and Rine J. (2002). Genetics, 161, 521–534. Osada H. (1999). Oncogene, 18, 4295–4300. Goto H, Tomono Y, Ajiro K, Kosako H, Fujita M, Sakurai Toledo F, Coquelle A, Svetlova E and Debatisse M. (2000). M, Okawa K, Iwamatsu A, Okigaki T, Takahashi T and Nucleic Acids Res., 28, 4805–4813. Inagaki M. (1999). J. Biol. Chem., 274, 25543–25549. Vogt U, Zaczek A, Klinke F, Granetzny A, Bielawski K and Haber JE. (1999). Trends Biochem. Sci., 24, 271–275. Falkiewicz B. (2002). J. Cancer Res. Clin. Oncol., 128, Kapoor TM, Mayer TU, Coughlin ML and Mitchison TJ. 141–147. (2000). J. Cell Biol., 150, 975–988. Wang H, Liu D, Wang Y, Qin J and Elledge SJ. (2001). Genes Lens SM, Wolthuis RM, Klompmaker R, Kauw J, Agami R, Dev., 15, 1361–1372. Brummelkamp T, Kops G and Medema RH. (2003). EMBO Wang X, Jin DY, Ng RW, Feng H, Wong YC, Cheung AL J., 22, 2934–2947. and Tsao SW. (2002). Cancer Res., 62, 1662–1668.

Supplementary Information accompanies the paper on Oncogene website (http://www.nature.com/onc).

Oncogene