Abstract
Chromosome instability is a commonly observed feature in ovarian carcinoma.Mitotic checkpoint controls are thought to be essential for accurate chromosomal segregation, and MAD2 is a key component of this checkpoint. In this study, we investigated the competence of the mitotic checkpoint and its relationship to the expression of MAD2 protein in seven ovarian cancer cell lines. We found that a significant number (43%, three of seven cell lines) of the tested ovarian cancer cells failed to arrest in the G2-M phase of the cell cycle in response to microtubule disruption. This loss of mitotic checkpoint control was associated with reduced expression of the MAD2 protein. To additionally understand the significance of the MAD2 to mitotic checkpoint control, we established an inducible expression system in which MAD2 was induced by the addition of ponasterone A. Notably, the induced expression of MAD2 in two checkpoint-defective ovarian cancer cell lines led to the restoration of mitotic checkpoint response to spindle-disrupting agents. Taken together, our findings suggest that the steady-state amount of MAD2 inside cells may represent a molecular switch for mitotic checkpoint control. This provides a novel insight into the molecular basis of CIN in ovarian carcinoma and has implications for effective use of checkpoint-targeting drugs.
INTRODUCTION
Accurate chromosomal segregation is essential for cell survival and genomic stability. The fact that the majority of human cancer cells exhibit gains or losses of chromosomes suggests that CIN 3 may contribute to tumorigenesis (1) . The mitotic checkpoint, also known as the spindle assembly checkpoint, detects errors occurred in the spindle structure or in the alignment of the chromosomes on the spindle, and delays chromosome segregation and anaphase onset until the defects are corrected. Disruption of the spindle with microtubule toxins such as nocodazole and colcemid arrests cells in mitosis, and this arrest depends on a functional mitotic checkpoint (2 , 3) . Two major groups of mitotic checkpoint genes, budding uninhibited by benomyl (BUB)1–3 and MAD1–3, have been identified in budding yeast (4 , 5) . Mammalian homologues of the yeast mitotic checkpoint proteins have also been characterized (6, 7, 8, 9) .
Kinetochores, which are linked to both chromosomes and microtubules, play an important role in generating the mitotic checkpoint signal. The function of kinetochores is to ensure that chromosomes are not segregated until every one of them is aligned and attached to the spindle (2 , 10) . The majority of proteins associated with mitotic checkpoint function have been shown to localize to kinetochores unattached to the microtubules (6 , 7 , 11 , 12) . It has been proposed that the mitotic checkpoint proteins, especially MAD2, may be crucial for generating the “wait” signal to prevent the onset of anaphase after microtubule disruption (13, 14, 15) . Several lines of evidence support MAD2 as a key component of the mitotic checkpoint. Microinjection of anti-MAD2 antibodies into HeLa cells abolished nocodazole- induced mitotic arrest and caused premature mitosis (6 , 16) . MAD2 has also been shown to interact with other mitotic checkpoint proteins including MAD1 and MAD3 (8 , 17) . In addition, MAD2 directly interacts with CDC20 and inhibits the anaphase-promoting complex (18, 19, 20, 21, 22, 23) . Furthermore, chromosome missegregation was observed in MAD2 knockout (MAD2−/−) mice (24) , and deletion of one MAD2 allele resulted in a defective mitotic checkpoint in both human colon cancer cells and murine primary embryonic fibroblasts (25) .
Although the importance of MAD2 in mitotic checkpoint control has been established in yeast and mammalian cells, the significance of MAD2 and the mitotic checkpoint to CIN in human cells and their association with human tumorigenesis are incompletely understood. Human MAD1 has been identified as a cellular target of the human T-cell leukemia virus type 1 oncoprotein Tax (8) . Mitotic checkpoint defects have also been found frequently in human (6) , colon (50%; Ref. 26 ), lung (44%; Ref. 27 ), and NPC (40%) cells (28) . Although MAD2 gene mutations are very rare in human bladder (29) , breast (30 , 31) , digestive tract (32) , liver (29) , and lung (27) cancers, aberrantly reduced expression of MAD2 protein has been correlated with defective mitotic checkpoint in breast cancer (6) and NPC (28) cells. Furthermore, mice with heterozygous deletion of MAD2 developed lung tumors at high rates after long latencies (25) , suggesting that MAD2 haplo-insufficiency might contribute to CIN and tumorigenesis.
Ovarian cancer is a leading cause of mortality among gynecological cancers. Notably, chromosomal aberrations were frequently found in ovarian cancer (33, 34, 35, 36) . However, it is not known whether the mitotic checkpoint is functional in ovarian cancer cells. Neither is it clear whether and how defects in this checkpoint might cause CIN in ovarian cancer. In the present study, we investigated the competence of the mitotic checkpoint in seven human ovarian cancer cell lines. We showed that defects in the mitotic checkpoint are rather common in ovarian cancer cells. To shed light on the mechanisms underlying the defects, we examined the expression of mitotic checkpoint proteins and demonstrated an association between MAD2 expression and checkpoint response. The significance of MAD2 to the mitotic checkpoint control was additionally studied in an inducible MAD2-expression system. The stable introduction of a MAD2-expressing plasmid into two ovarian cancer cell lines with low basal levels of MAD2 resulted in the restoration of the checkpoint response to microtubule-disrupting agents. Our findings implicate that reduction of MAD2 expression may represent a critical event in the development of CIN in ovarian cancer.
MATERIALS AND METHODS
Cell Lines and Cell Culture Conditions.
Seven ovarian cancer cell lines, DOV13, SKOV3, OVCA3 (obtained from ATCC), Ovca420, Ovca429, Ovca432, and Ovca433 (37) were maintained in RPMI 1640 (Life Technologies, Inc.) supplemented with 2 mm l-glutamine and 5% (v/v) FCS at 37°C. The cultures were grown for a maximum of 10 passages before retrieving fresh cells from frozen stock. Ovca420, Ovca432, and Ovca433 cell lines were established from freshly isolated ascites or tumor explants from patients with late-stage ovarian adenocarcinomas with distinct characteristics (37) . Several independent groups have also demonstrated the differential expression patterns of genes and proteins, such as BRCA1, BRCA2, HER-2, and epidermal growth factor receptor in these cell lines (37 , 38) . Moreover, the independent origins of these lines were additionally verified by PCR analysis of polymorphic alleles. Using four sets of primers for four microsatellite markers on chromosomes 3 (D3S162), 5 (D5S82), 17 (D17S855), and X (DXS981), we found that each of the cell lines exhibited a distinct pattern of allelic polymorphism (data not shown).
Mitotic Index.
Cells were grown on Chamber slides and treated with nocodazole or Colcemid. The cells were then fixed in cold methanol/acetone (1:1) for 5 min and stained with 4′,6-diamidino-2-phenylindole. The presence of condensed nuclear DNA was considered to indicate cells undergoing mitosis. To measure the mitotic index (percentage of viable cells arrested in mitosis), at least 500 cells were counted for each experiment using fluorescence microcopy, and the data points represent the average results from three independent experiments.
BrdUrd Staining.
Cells were grown on Chamber slides for 24 h and then incubated with culture medium containing BrdUrd (10 μm) for 2 h at 37°C. The BrdUrd incorporation rate was examined by ABC method using a Vectastain ABC kit (Vector Laboratories, Inc., Burlingame, CA). Briefly, after washing with PBS, the cells were fixed in cold methanol/acetone (1:1) for 5 min and then incubated with methanol containing 2% H2O2 for 10 min at room temperature. After washing three times with PBS, the cells were incubated in blocking solution (provided in the ABC kit) for 30 min at 37°C. The blocking solution was drained and primary Ab (1:20; mouse anti-BrdUrd Formalin Grade; Roche Diagnostics, Mannheim, Germany) was added onto the monolayers, and the cells were incubated at 37°C for 1 h. After washing three times with PBS, the cells were incubated with secondary Ab (biotinylated antimouse Ab, provided in the ABC kit) for 30 min at 37°C, and visible brown color in the positive cells was developed according to the protocol provided by the manufacturers. Then the slides were counterstained with hematoxylin, dehydrated, and then mounted. The positive cells were identified as the presence of brown color in the cell nucleus. At least 500 cells were counted in each experiment, and the percentage of BrdUrd-positive cells was calculated and compared with untreated controls. Each experiment was repeated at least three times, and the SE of the means was used as error bars.
Cell Cycle Analysis.
Flow cytometric analysis was performed on an EPICS profile analyzer and analyzed using the ModFit LT2.0 software (Coulter) as described previously (28) .
Western Blotting Analysis.
Detailed experimental procedures were described previously. Briefly, cells were harvested in lysis buffer (28) and protein concentrations were determined. Approximately 30 μg of protein was separated on a 15% SDS-polyacrylamide gel, transferred to nitrocellulose, and incubated with antibodies against MAD1 (polyclonal, 1: 500) and MAD2 (monoclonal, 1:500; Ref. 8 ), or actin (1:200; Roche). The relative amounts of each protein were quantitated as ratios to actin. Results represent the average of three independent experiments.
Transfection.
A 1.5-kb fragment containing the full-length human MAD2 cDNA (Ref. 8 ; GenBank U) was cloned into the EcoRI and XbaI sites of an expression vector pIND(SP1) driven by the ecdysone-inducible promoter (Invitrogen, Carlsbad, CA). The resulting construct pIND(SP1)-MAD2 (conferring neomycin resistance) was then cotransfected with pVgRXR vector (conferring Zeocin resistance), which permits formation of a functional heterodimeric ecdysone receptor in mammalian cells. Using the calcium-phosphate method, the stable pIND(SP1)-MAD2 and pVgRXR transfectants were generated by selecting in both neomycin (150 μg/ml) and Zeocin (250 μg/ml). Briefly, cells were plated at ∼50% confluency, and the cell culture medium was changed to 5% DMEM containing 5% fetal bovine serum 1 h before transfection. DNA suspension (1.1 ml) was made [69 μl of 2 m CaCl2, 550 μl of 2 × HBSS, 5 μg of each plasmid DNA, and appropriate amount of H2O; HBSS [560 mm NaCl, 20 mm KCl, 3.0 mm Na2HPO4, 24 mm dextrose, and 100 mm HEPES (pH 7.02)] and added to the cell culture flask over night. Selective drugs were added 24 h later, and individual transfectants were visible ∼10 days after drug selection. Clones were then isolated and tested for MAD2 expression. To induce MAD2 expression in the transfectants, 5 μm of ponasterone A (Invitrogen) was added for 20 h. Two transfectants of Ovca432 and Ovca433 cell lines were generated and studied in detail.
Confocal Immunofluorescence Microscopy.
The cells were grown on coverslips and treated with nocodazole (50 nm) for 24 h. Confocal immunofluorescence microscopy was performed on a Zeiss Axiophot system as described previously (8) . Cells were then fixed in 4% paraformaldehyde, permeabilized with methanol, and stained with anti-α-tubulin Ab (clone B-5-1-2; Sigma Chemical Co.).
RESULTS
Competence of Mitotic Checkpoint in Ovarian Cancer Cells.
Defects in mitotic checkpoint are thought to contribute to CIN and tumorigenesis (14 , 26) . Deregulation of the mitotic checkpoint in various cancers has been documented (6 , 26, 27, 28) . However, there is no information on the mitotic checkpoint control and its association with CIN in ovarian cancer. As a first step toward understanding the mechanisms of CIN, we sought to assess the competence of mitotic checkpoint control in seven ovarian cancer cell lines using nocodazole and Colcemid, which inhibit spindle assembly by disrupting microtubules. On addition of nocodazole or Colcemid, three of the seven (43%) ovarian cancer cell lines, Ovca420, Ovca432, and Ovca433, did not show a significant increase in the percentage of cells arrested at mitosis (Fig. 1, A and B ⇓ , dotted lines), as measured by the number of cells with condensed chromosomes, a characteristic of mitotic block (Fig. 1C) ⇓ . In addition, there were no significant differences in the patterns of nuclear staining between nocodazole-treated cells and the control, because both of them showed evidence of nuclear division (Fig. 1C ⇓ , cells in anaphase or telophase are indicated by arrows, compare panel 6 to panel 5). These data suggest that nocodazole treatment did not alter mitosis and cell cycle progression in these three cell lines. In contrast, four of the cell lines (Ovca429, Ovca3, Skov3, and Dov13) responded to these two agents with an increase in mitotic cells reaching a peak at ∼24 h after exposure (Fig. 1, A and B ⇓ , solid lines). This response is similar to the reaction of HeLa cells, which have been shown to be mitotic-checkpoint competent (6 , 27 , 28) . Flow cytometric analysis of Ovca420, Ovca432, and Ovca433 cell lines showed a lack of G2-M phase arrest after treatment with nocodazole or Colcemid for 18 h (Fig. 2A) ⇓ in contrast to Ovca429, Ovca3, Skov3, Dov13, and HeLa cells, which showed a clear G2-M phase block (Fig. 2A) ⇓ . However, a small percentage of cells with >4 n DNA content was observed in Ovca420, Ovca432, and Ovca433 cell lines at this time point, and the percentage of >4 n cells increased with the extended incubation time (data not shown). These results indicate that three of seven (43%) ovarian cancer cell lines failed to arrest at mitosis in response to microtubule disruption.
Mitotic indices of seven ovarian cancer cell lines. The cells were treated with Colcemid (0.1 μg/ml; A) or nocodazole (50 nm; B) for the indicated times and analyzed at 6 h intervals. ····, highlight cells with defective mitotic checkpoint control. HeLa cell line was used as a control with a normal mitotic checkpoint. C, representative results of fluorescence microscopic examination of HeLa (panels 1 and 2), Skov3 (panels 3 and 4), and Ovca433 (panels 5 and 6) before (panels 1, 3, and 5), and after (panels 2, 4, and 6) treated with nocodazole (50 nm) for 18 h. Accumulation of cells with condensed DNA indicates a normal spindle checkpoint in HeLa and Skov3 (panels 1–4) after exposure to nocodazole. Photos were taken under ×400 magnification. Arrows in Ovca433 cells indicate anaphase or telophase cells undergoing nuclear division.
Cell cycle distribution (A) and BrdUrd incorporation (B) in ovarian cancer cell lines before and after Colcemid (0.1 μg/ml) or nocodazole (50 nm) treatment for 18 h. A, representative results showing the accumulation of G2-M phase cells in HeLa and Skov3 cell lines after drug exposure in contrast to the lack of this accumulation in Ovca433 cells. B, significant reduction of BrdUrd incorporation was observed in HeLa, Ovca429, Ovca3, Skov3, and Dov13 but not in Ovca420, Ovca432, and Ovca433 cell lines compared with the untreated control after exposure to nocodazole (50 nm; P < 0.001). Each data point represents the average of three independent experiments; bars, ±SE.
To eliminate the possibility that the Ovca420, Ovca432, and Ovca433 cells failed to accumulate in mitosis simply because they were not actively cycling during drug treatment, we measured the fraction of cells in S phase by staining with BrdUrd. As shown in Fig. 2B ⇓ , after nocodazole treatment, Ovca420, Oca432, and Ovca433 cell lines had a similar S phase index to the untreated controls, suggesting that these cells did proceed through mitosis and enter the next S phase in the presence of nocodazole. In contrast, the Ovca429, Ovca3, Skov3, Dov13, and HeLa cells showed >80% reduction in BrdUrd-positive cells compared with the untreated control, reflecting a failure to reinitiate DNA synthesis because of the mitotic block.
To exclude the possibility that the failure of mitotic arrest in Ovca420, 432, and 433 cell lines was simply because the microtubules became resistant to nocodazole compared with the other four cell lines, we studied the changes in microtubule structure in the presence and absence of nocodazole. The seven cell lines were immunostained for α-tubulin, and the patterns were compared with those in HeLa and CNE3 cells, which have been shown previously to be competent and defective, respectively, for mitotic checkpoint control (6 , 27 , 28) . In all of the untreated cells, α-tubulin was observed in microtubules arranged in long and slender fibers, which spread to the entire cytoplasm. The microtubule organizing centers or centrosomes were also evident in some cells (Fig. 3 ⇓ , panels 1 and 3). In contrast, the microtubule arrays in cells treated with nocodazole were largely disrupted, and α-tubulin was found in the cell periphery (Fig. 3 ⇓ , panel 4) as well as in the remaining microtubule organizing centers (Fig. 3 ⇓ , panels 2 and 4). Notably, microtubule disruption occurred in all of the cell lines. Generally the severity of microtubule disruption was not significantly different between cell lines that showed a mitotic arrest in response to nocodazole (i.e., HeLa, Ovca429, Ovca3, Skov3, and Dov13) and the lines that failed to arrest (i.e., Ovca420, Ovca432, and Ovca433). For instance, nocodazole induced mitotic arrest in Skov3 cells (Figs. 1 ⇓ and 2 ⇓ ; Fig. 3 ⇓ , panel 2), whereas Ovca432 cells treated with nocodazole kept cycling (Figs. 1 ⇓ and 2 ⇓ ; Fig. 3 ⇓ , panel 4). However, the microtubules in Ovca432 cells (Fig. 3 ⇓ , panel 4) appeared to be even more severely disrupted than in the Skov3 cells (Fig. 3 ⇓ , panel 2). These results suggest that microtubules in cells defective for mitotic arrest are not resistant to nocodazole.
Nocodazole-induced microtubule disruption in ovarian cancer cell lines. Representative results are shown for Skov3 and Ovca432 cells. Cells were mock-treated (panels 1 and 3) and treated with 50 nm nocodazole for 24 h (panels 2 and 3). Photos were taken under ×126 magnification. Arrow indicates a cell arrested at metaphase. Similar results were obtained for the other five ovarian cancer cell lines.
Collectively these results demonstrate that in the presence of microtubule-interfering agents, three of seven (43%) ovarian cancer cell lines progress through mitosis despite microtubule depolymerization and keep cycling into S phase, indicating a loss of mitotic checkpoint.
Correlation of Defective Mitotic Checkpoint with Reduced Expression of MAD2.
The loss of mitotic checkpoint control in some colorectal cancer cells has been shown previously to be associated with the mutational inactivation of mitotic checkpoint protein BUB1 (26) . However, mutations of relevant checkpoint genes are extremely rare in various cancers tested (27 , 29, 30, 31, 32) . Instead, aberrantly reduced expression of the MAD2 protein has been correlated with defective mitotic checkpoint control in breast and NPC (6 , 28) . To additionally study the mechanisms underlying the defects in the mitotic checkpoint response in ovarian cancer cells, we asked whether the expression level of checkpoint proteins might be significantly different. As a first step, we examined the expression of two key components of the mitotic checkpoint, MAD1 and MAD2 (6 , 8) , by Western blot analysis (Fig. 4) ⇓ . We found that the expression of MAD2 protein was significantly decreased (70–80% reduction) in Ovca420, Ovca432, and Ovca433 cell lines, which also exhibited a defective mitotic checkpoint response (Fig. 4) ⇓ . The other four cell lines with competent mitotic checkpoint expressed MAD2 protein to ∼60% or more of the steady-state amount in HeLa cells. However, there was no significant correlation between MAD1 expression and mitotic checkpoint competence in ovarian cancer cells as reported previously for NPC cells (28) . These results suggest that decreased MAD2 expression may contribute to the defective mitotic checkpoint control in Ovca420, Ovca432, and Ovca433 cells.
Western blot analysis of MAD1 and MAD2 expression in seven ovarian cancer cell lines. Protein (30 μg) was analyzed, and the relative amounts of each protein were quantified as ratios to actin and then compared with HeLa cells. Results represent the average of three independent experiments; bars, ± SE.
Effect of MAD2 Expression on Mitotic Checkpoint Control in Ovca432 and Ovca433 Cells.
The correlation of defective mitotic checkpoint with reduced expression of MAD2 prompted us to additionally investigate the roles of MAD2 in the checkpoint control. To this end, we sought to reintroduce MAD2 into the checkpoint-defective ovarian cancer cell lines. One route to express MAD2 in mammalian cells is under the control of a constitutive promoter. However, we were concerned that constitutive and stable overexpression of MAD2 might cause irreversible changes in cell physiology and cell cycle control. Thus, we constructed the human MAD2 expression system using an inducible expression system in which MAD2 can be induced transiently by a diffusible small molecule.
The ecdysone-inducible expression system is based on molting induction in the fruit fly and has been adapted for inducible expression in mammalian cells (39) . This system uses heterodimeric nuclear receptors induced by a synthetic analogue of ecdysone (ponasterone A) to activate the expression of the gene of interest. The aberrant reduction of MAD2 expression in checkpoint-defective cells (Ref. 6 , 28 and data shown above) raised the concept that the steady-state amount of MAD2 inside the cells might determine the competence of the mitotic checkpoint. Thus, the MAD2 expression plasmid driven by an ecdysone-inducible promoter [pIND(SP1)-MAD2] and the plasmid expressing the heterodimeric ecdysone receptor (pVgRXR) were cotransfected into Ovca432 and Ovca433 cell lines. Individual stable transfectants were isolated by selecting in culture medium containing both Zeocin (250 μg/ml) and G418 (150 μg/ml). Exogenous MAD2 expression was induced by exposure to 5 μm ponasterone A for 20 h. As shown in Fig. 5 ⇓ , ponasterone A treatment resulted in a 2–4-fold increase in MAD2 levels in two Ovca432-MAD2 clones (Clone 1 and Clone 2) and two Ovca433-MAD2 clones (Clone 1 and Clone 2) compared with the untreated controls. The MAD2 expression levels in all of the transfectants were ≥50% compared with HeLa cells.
Forced expression of MAD2 restores mitotic checkpoint response in Ovca432 and Ovca433 cells. Western blot analysis of MAD2 protein was performed in cells cotransfected with pIND(SP1)-MAD2 and pVgRXR plasmids. Increased MAD2 expression was observed in cells treated with ponasterone A (5 μm) for 20 h (+P) compared with the untreated control (−P). The relative MAD2 protein levels before and after ponasterone A treatment were quantified as ratios to actin and compared with the level in HeLa cells. Results represent the average of three independent experiments; bars, ± SE.
After ponasterone A treatment, there was no significant change in the mitotic index (Fig. 6, A ⇓ , +P), indicating that the levels of MAD2 expression in these transfectants did not have any effect on mitotic checkpoint control in the absence of spindle-disrupting agents. However, when the cells were treated with both nocodazole (for 18 h) and ponasterone A (Fig. 6A ⇓ , +N+P), a significant increase in mitotic cells was observed in all four of the clones compared with the cells treated with nocodazole alone (+N). This indicates that expression of MAD2 enabled the cells to arrest at mitosis in the presence of microtubule toxin. Among the four clones, Ovca432-MAD2-C1 showed the highest MAD2 expression ratio (>3-fold increases after ponasterone A treatment) compared with the untreated controls, and the MAD2 expression levels in these cells were comparable with those in HeLa cells (Fig. 5 ⇓ ; ∼70%). As shown in Fig. 6A ⇓ , the mitotic index profile of this cell line was also similar to HeLa cells, whereas the other three clones showed partial mitotic arrest in the presence of nocodazole. This additionally supports the notion that levels of MAD2 expression may be one of the key determinants for mitotic checkpoint control in these cells. Similar results were observed in cells treated with Colcemid (data not shown).
Effect of MAD2 overexpression on mitotic checkpoint control in Ovca432 and Ovca433 cells. A, mitotic indices. B, cell cycle distribution. C, BrdUrd incorporation. +N, cells were treated with nocodazole (50 nm). +P, cells were treated with ponasterone A (50 μm). +N+P, cells were treated with nocodazole plus ponasterone A. Results represent the average of three independent experiments; bars, ±SE.
Cell cycle studies by flow cytometry also confirmed these findings, because increased number of G2-M phase cells were observed in cells treated with both nocodazole and ponasterone A (Fig. 6B ⇓ , +N+P), whereas no significant changes were observed in the cells treated with nocodazole (+N) or ponasterone A (+P) alone. The fraction of S phase cells was also decreased in the cells treated with both nocodazole or ponasterone A (Fig. 6C ⇓ , +N+P) compared with the control and the cells treated with nocodazole (+N) or ponasterone A (+P) alone. Similar results were also observed when treated with Colcemid (data not shown). These results suggest that expression of MAD2 restored mitotic checkpoint control, either fully or partially, in the presence of microtubule inhibitors in ovarian cancer cells.
DISCUSSION
Ovarian cancer is the leading cause of death in female patients with genital tract carcinomas, because most of the cases are presented at an advanced stage. Although relatively little is known about the molecular basis of this cancer, CIN has been shown to be a common feature (33, 34, 35, 36) . A recent study on 106 ovarian cancer patients using comparative genomic hybridization demonstrated that 97% of the tumor samples showed aberrant comparative genomic hybridization profile (40) . In addition, the number and extent of chromosome changes were significantly increased in high-grade tumors compared with low-grade tumors. This indicates that CIN plays an important role in tumorigenesis as well as tumor progression in ovarian cancer. In this study, using seven ovarian cancer cell lines, we identified a high frequency of mitotic checkpoint defect (43%; three of seven cell lines). Our findings suggest that a defective mitotic checkpoint may contribute to CIN commonly observed in ovarian cancer cells. In addition, we found that restoration of MAD2 expression in Ovca432 and Ovca433 cells led to mitotic arrest in response to microtubule disruption. Our data point to a critical role for MAD2 in mitotic checkpoint control in human cancer cells.
Three salient points have emerged from our study. First, we demonstrated the frequent loss of mitotic checkpoint in ovarian cancer cells. Second, we correlated the loss of checkpoint with the aberrantly reduced expression of MAD2 protein. Third, we showed that the mitotic checkpoint response might be restored if the dosage of MAD2 was increased inside the cells.
Direct association between CIN and mitotic checkpoint defect has been demonstrated in yeast cells (41) ; however, the mechanisms involving CIN in human cells are yet to be characterized. The most convincing evidence of the role of mitotic checkpoint defect in CIN in mammalian cells came from two recent studies in MAD2−/− mice (24) , and in MAD2+/− human and mouse cells (25) showing that disruption of MAD2 expression resulted in CIN. In addition, the MAD2+/− mice developed lung tumors at high rates indicating that defects in mitotic checkpoint play an important role in tumorigenesis. In this study, we provide the first evidence for the high frequency of mitotic checkpoint defect in ovarian cancer cells (Figs. 1 ⇓ and 2 ⇓ ). Because there is a strong correlation between frequency of chromosomal changes and tumor grading in ovarian carcinoma (33 , 40) , which is important in determining patient survival, our data may prove useful for diagnosis and patient selection. The front-line chemotherapeutic drug for the treatment of ovarian cancer is cisplatin, but cisplatin-resistant tumors occur at fairly high frequency. A better understanding of the molecular basis of mitotic checkpoint control in ovarian cancer cells may reveal novel strategies for more efficient use of checkpoint-targeting drugs such as Taxol. In this regard, one critical issue is the differential response of mitotic checkpoint- competent versus mitotic checkpoint-defective cells to microtubule disruption. It is noteworthy that all mammalian cells, irrespective of their competence in mitotic checkpoint control, will eventually adapt to microtubule-disrupting agents and exit mitosis through an as yet unknown mechanism (3 , 42) . However, if the dose of microtubule toxins is sufficiently low, as in this study, or if the treatment is sufficiently transient, checkpoint-competent cells arrested at mitosis may have a chance to repair their spindles and later proceed through cell cycle. By contrast, if the microtubule challenge is persistent, these cells will maintain a p53-dependent G1 arrest after adaptation and undergo apoptosis (42 , 43) . On the other hand, checkpoint-defective cells do not arrest at mitosis in the presence of low-dose microtubule toxins but lose chromosomes at a higher rate to induce apoptosis. Meanwhile, a prolonged exposure to microtubule toxins causes polyploidy in checkpoint-defective cells with no evidence of apoptosis (7) . Thus, the differential reaction to microtubule inhibitors might be exploited in selective killing of checkpoint-defective tumor cells.
MAD2 was first identified in screens for yeast mutants, which were hypersensitive to drugs that disrupt microtubules (4 , 5) . In mammalian cells, inactivation of MAD2 resulted in loss of mitotic checkpoint control and CIN (6 , 24 , 25) . More recently, it has been reported that suppression of MAD2 protein by a carcinogenic compound tetrachlorodibenzo-p-dioxin is responsible for the inactivation of mitotic checkpoint control in HeLa cells (44) . Here we have shown that decreased MAD2 expression correlated to mitotic checkpoint defect in ovarian cancer cells (Figs. 1 ⇓ 2 ⇓ 3 ⇓ 4) ⇓ and that restoration of MAD2 expression induced mitotic arrest in response to microtubule disruption (Figs. 5 ⇓ and 6 ⇓ ). Our findings support the model in which the steady-state amount of MAD2 may serve as a molecular switch for the mitotic checkpoint control in human cancer cells. This is consistent with the recent finding that haplo-insufficiency of MAD2 led to development of lung cancer in mice (25) . Our model for mitotic checkpoint control has useful implications for patient selection and therapeutic intervention in ovarian cancer. Because mutations of the mitotic checkpoint genes, including MAD2, are uncommon (26, 27, 28) , analysis of the expression levels of MAD2 protein may allow identification of mitotic checkpoint-defective tumors, thus facilitating the selection of patients for more effective chemotherapy.
Footnotes
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
↵1 Supported in part by CRCG grants from the University of Hong Kong. D-Y. J. is a Leukemia and Lymphoma Society Scholar.
↵2 To whom requests for reprints should be addressed, at Department of Anatomy, University of Hong Kong, 1st Floor, Laboratory Block, 21 Sassoon Road, Pokfulam, Hong Kong. Phone: 852-2819-9228; Fax: 852-2817-0857; Email: gswtsao{at}hkucc.hku.hk
↵3 The abbreviations used are: CIN, chromosome instability; MAD, mitotic arrest deficient; NPC, nasopharyngeal carcinoma; BrdUrd, bromodeoxyuridine; Ab, antibody.
- Received October 24, 2001.
- Accepted January 18, 2002.
- Corrected online July 19, 2018.
- ©2002 American Association for Cancer Research.
References
Volume 62, Issue 6
