Heterozygous Deletion of Mitotic Arrest–Deficient Protein 1 (MAD1) Increases the Incidence of Tumors in Mice

Cancer is a disease of damaged genes. Currently, the precise mechanisms for cellular transformation remain incompletely elucidated. Abnormal chromosome numbers from losses or gains in whole chromosomes (i.e., aneuploidy) are commonly seen in cancer cells. Eukaryotic cells have evolved a spindle assembly checkpoint (SAC) to monitor the fidelity of chromosomal segregation in mitosis (1, 2) to guard against aneuploidy. In lower eukaryotes (i.e., yeasts), the SAC is composed of the mitotic arrest–deficient (MAD) proteins (MAD1, MAD2, and MAD3), the budding uninhibited by benzimidazole (BUB) proteins (BUB1, BUB2, and BUB3), and the monopolar spindle 1 protein (3). In higher eukaryotes, SAC components also include BUBR1 (a vertebrate variant of yeast MAD3 and BUB1), the ROD-ZW10-Zwilch complex, and the microtubule motor centromere protein E (4). A link between SAC function and neoplasia is suggested, although not proven, by findings that expression of many checkpoint proteins is aberrant in cancers (5–11).

Several knockout (KO) mouse models (Mad2, BubR1, or Bub3) have been constructed to investigate the physiologic functions of SAC proteins. An underlying goal, among others, of these KO studies is to clarify whether a weakened SAC creates a proclivity for tumor development in vivo. Regrettably, the full biological import of these KO studies has been unrealized because mice homozygous null for BubR1, Mad2, or Bub3 are embryonic lethal (12–14). Nonetheless, analyses of heterozygous null animals have been instructive, albeit slightly confounding. Here, an increased incidence of constitutive tumors was revealed in Mad2^+/− mice but not in either BubR1^+/− or Bub3^+/− mice (12–16). On the other hand, a higher tumor incidence can be induced in BubR1^+/− compared with wild-type (wt) control mice when animals are treated with carcinogens (12). Whereas other interpretations are possible, these data suggest potentially two qualitatively discrete paths within the SAC, in which attenuation of one, but not the other, suffices to trigger constitutive tumors. Alternatively, MAD2 and BUBR1 may have different quantitative effects on the SAC because BubR1^+/−ApcMin^+/− compound mice exhibit greater genetic instability than BubR1^+/− mice and do develop constitutive tumors (17).

To further clarify tumorigenic differences, if any, between a Mad2^+/− genotype and its BubR1^+/− or Bub3^+/− counterpart, we constructed a KO mouse model for the MAD2 partner protein, MAD1. Here, we report that Mad1^−/− like Mad2^−/− (13) confers embryonic lethality, and that Mad1^+/− mice are similar to Mad2^+/− mice in developing increased constitutive tumors.

Construction of Mad1 KO mice. The Mad1 KO vector (Mad1 KO) was constructed by cloning exon 10 of the mouse Mad1 gene into vector pGEM-7 (Promega, Madison, WI). Neo was positioned downstream of exon 10. Introns 9 (1.4 kb) and 10 (6.1 kb) were placed before the 5′ end of exon 10 and after the 3′ end of neo, respectively (Fig. 1A). The HSV-TK gene was introduced at the 3′ end for negative selection. In addition, three loxP (locus of X-over of bacteriophage P1) sites were created between the 5′ arm and exon 10, exon 10 and neo, and neo and the 3′ arm, respectively (Fig. 1A). Mad1 KO mice were generated by the Mouse Genome Engineering Facility at the University of Nebraska. In brief, Mad1 KO was introduced into embryonic stem (ES) cells by electroporation and doubly selected using G418 and ganciclovir. Surviving clones were confirmed by PCR. Heterozygous ES cells were injected into C57BL/6J blastocysts. Mosaic founder animals were screened for germ-line transmission of the KO genotype by breeding to C57BL/6J mice. F₁ mice heterozygous for the KO allele were mated to a Cre-expressing transgenic mouse [BALB/c-TgN(CMV-Cre)#Cgn; The Jackson Laboratory, Bar Harbor, ME], resulting in deletion of exon 10 of Mad1 (Fig. 1A) in some of the offsprings. Loss of exon 10 was verified by PCR using the following primers: 5′-cggacgaggtatttgcacgtgcagctctattttagg-3′ and 5′-gcatgggtgagctcagtcacactgg-3′.

Tumorigenicity assay in nude mice. Mouse embryonic fibroblasts (MEF; 5 × 10⁶) were resuspended into 200 μL PBS containing 5% fetal bovine serum and injected s.c. into the posterior neck of 6-week-old athymic nude mice (Harlan, Inc., Indianapolis, IN). Tumor growth was monitored every 3 days.

Immunofluorescence and confocal microscopy. MEFs were fixed in 4% paraformaldehyde for 30 min and permeablized with 0.1% Triton X-100 in PBS for 5 min at room temperature. To prevent nonspecific binding, cells were equilibrated in 1% bovine serum albumin in PBS for 30 min. Antibodies against α-tubulin (Sigma-Aldrich, St. Louis, MO) or anti–phosphorylated histone H3 on Ser¹⁰ (Cell Signaling Technology, Beverly, MA) were added and incubated for 1 h at room temperature. Fluorescent-conjugated secondary antibodies (Molecular Probes, Carlsbad, CA) were used for detection. Nuclei were stained with 4′,6-diamidino-2-phenylindole (DAPI; Molecular Probes). Cells on the coverslips were mounted on glass slides with antifade reagent (Molecular Probes). Slides were monitored using a Leica (Wetzlar, Germany) TCS-NP/SP confocal microscope.

Fluorescent in situ hybridization. Fluorescent in situ hybridization (FISH) was done using chromosome-specific Cy3-labeled probe for chromosome 2 (Cambio Ltd., Cambridge, United Kingdom) according to the manufacturer's protocol. Briefly, cells were fixed in 3:1 (v/v) methanol/acetic acid and dropped onto slides. Slides were denatured with 70% formamide at 65°C for 1.5 min. After 16 h of hybridization, the slides were washed with 50% formamide, 1× SSC, and 4× SSC/0.05% Tween 20 successively. Finally, slides were mounted onto coverslips with antifade reagent.

Construction of a Mad1 KO mouse. MAD1 is a protein-binding partner of MAD2. RNA interference– or dominant-negative mutant protein–mediated knockdown of MAD1 in somatic cells has been shown to produce MAD2 mislocation from kinetochores and failed SAC function (18). To investigate in vivo MAD1 function, we embarked on generating a Mad1 KO mouse (see Materials and Methods). Mad1 has 17 coding exons; we chose to target exon 10 from the coding sequence (Fig. 1B) because removal of this exon abolishes the ability of MAD1 to bind MAD2 and creates loss of SAC function (19). Independent founder mice with exon 10 of Mad1 flanked by loxP sequences were created, and germ-line transmission from founder to F₁ offsprings was achieved. To delete exon 10 from Mad1, we mated F₁ animals to a strain of transgenic mouse from The Jackson Laboratory [strain BALB/c-TgN(CMV-Cre)#Cgn] that overexpress a CMV promoter-driven Cre expression vector (Fig. 1A). Cre is a site-specific DNA recombinase of bacteriophage P1 that recognizes a 34-bp site on the P1 genome (i.e., loxP) and efficiently catalyzes reciprocal conservative DNA recombination between pairs of loxP sites (20). Cre-mediated removal of loxP-flanked exon 10 in F₂ mice (Fig. 1B) was confirmed by PCR. We checked by direct sequencing (data not shown) that wt mouse DNA generated the expected 312-bp PCR fragment, whereas the deleted Mad1 genotype produced the expected 260-bp fragment (Fig. 1C). Western blotting of Mad1^+/− mouse embryo fibroblasts showed a one-third to one-half expression level of MAD1 protein with no clear detection of truncated polypeptide (data not shown). This suggests that cell endogenous exon 10–deleted MAD1 protein may be unstable.

We paired Mad1^+/− mice to breed for Mad1 homozygous null mouse. To this end, 15 deliberate Mad1^+/− × Mad1^+/− pairings were made, which produced a total of 319 pups (Table 1A). Genotyping of the pups revealed that 209 were Mad1^+/− and 110 mice were wt (i.e., Mad1^+/+); no Mad1^−/− mouse was born. The Mad1^+/− to wt ratio conformed to Mendelian inheritance and suggested embryonic lethality as the explanation for the absence of live Mad1^−/− births. Subsequent matings have generated another 1,000+ offsprings; no live Mad1^−/− mouse has ever been detected.

Increased tumor incidence in Mad1^+/− mice. It has been proposed that weakening of the SAC without a total loss in checkpoint function is relevant to the development of aneuploidy (4). Compared with Mad1^+/+, we reasoned that Mad1^+/− represents a weakened SAC. To ask if a MAD1-attenuated SAC influences tumor development in mice, we monitored age- and sex- matched Mad1^+/+ and Mad1^+/− cohorts (128 and 121 animals, respectively) for >18 months. Although Mad1^−/− is incompatible with live birth (Table 1A), Mad1^+/− mice are born normally. Further, the life spans of Mad1^+/+ and Mad1^+/− mice are similar as evidenced by 90% (114 of 128) of Mad1^+/− and 93% (112 of 121) of wt mice being alive at 18 months. This observation is similar to published survival curves comparing wt with Mad2^+/−, BubR1^+/−, or Bub3^+/− mice (12, 16, 21).

At 18 months of age, both wt and Mad1^+/− mice were sacrificed and detailed necropsy and histopathology were done. Necropsies of wt and Mad1^+/− mice revealed a statistically significant (P < 0.05, χ²) difference in tumor incidence (Table 1B). Lymphoma is a common background tumor in aging mice (22). The incidence of lymphoma was statistically indifferent between wt and Mad1^+/− mice and was excluded from our comparisons. Previously, 18-month-old Mad2^+/− mice were reported to have increased papillary lung adenocarcinomas (16). Interestingly, mice in our Mad1^+/− cohort developed neoplasia in a wide range of tissues (hepatocellular carcinoma, rhabdomyosarcoma, osteosarcoma, hemangiosarcoma, and uterine sarcoma) in addition to the lung (Table 1B). Spontaneous rhabdomyosarcoma and hemangiosarcoma are exceedingly rare in 18-month-old mice with BALB/c, C57BL/6, or B6:129 backgrounds (23); hence, the presence of these tumors in our Mad1^+/− mice suggests specificity for Mad1 haploinsufficiency. The Mad1^+/− rhabdomyosarcoma and hemangiosarcoma showed typical morphologies of aggressively proliferating mitotic cells (Fig. 2A , arrows).

Vincristine treatment increased tumor development in Mad1^+/− mice. Vincristine is a microtubule-depolymerizing drug used frequently in cancer chemotherapy. Depolymerized microtubules attach poorly to kinetochores, and this inefficiency is expected to elicit SAC-mediated mitotic arrest (24). Cells with weakened SAC may fail to arrest in mitosis despite microtubule depolymerization; a failure to arrest in mitosis followed with progression through cytokinesis could produce aneuploid progeny cells. To ask if the increased tumors in Mad1^+/− mice could be explained by a weakened SAC, we deliberately increased chromosomal missegregations by dosing animals with vincristine. We hypothesized that such provocation would induce a higher frequency of aneuploidy in Mad1^+/− versus Mad1^+/+ cells, and induction of aneuploid cells might then engender more tumors in Mad1^+/− versus Mad1^+/+ mice.

We took 6-month-old wt and Mad1^+/− mice, in groups of 12, and treated them i.p. with vincristine once every 7 days for three injections. As a control, we also treated 12 Mad1^+/− mice with etoposide, a topoisomerase II inhibitor, which does not affect microtubule physiology (Table 1C). The mice were followed for 12 months after the conclusion of treatment and sacrificed to examine for tumors (Table 1C). We found that none of the 12 wt mice treated with vincristine (Mad1^+/+;Vin) nor any of the 12 Mad1^+/− mice treated with etoposide (Mad1^+/−;Eto) developed tumors. By contrast, 5 of 12 (42%) Mad1^+/− mice treated with vincristine (Mad1^+/−;Vin) did have tumors. These included three cases of premalignant bronchioloalveolar adenoma (Fig. 2B), one case of hepatocellular carcinoma, and one case of Harderian gland adenoma. The significantly higher induced tumor incidence in vincristine-treated Mad1^+/− than vincristine-treated wt mice or etoposide-treated Mad1^+/− mice suggests that a weakened Mad1^+/− SAC is less competent at arresting cells with deliberately provoked chromosomal missegregations. Aneuploid progeny cells that arise from uncensored missegregations could seed tumorigenesis in Mad1^+/− mice.

Genetic instability in dually haploinsufficient Mad1^+/−Mad2^+/− MEFs. Above, we inferred that the increased tumor presentation in 18-month-old Mad1^+/− mice may be due to a weakened SAC. To verify this inference, we isolated fresh Mad1^+/− MEFs and compared them with wt MEFs for stringency of mitotic arrest in response to treatment with nocodazole (Fig. 3A). Cells arrested in mitosis were identified by immunostaining for phosphorylated Ser¹⁰ on histone H3 (Fig. 3A). We counted 1,000 cells per group and found that 34% of nocodazole-treated wt and 10% of nocodazole-treated Mad1^+/− MEFs were in mitosis. These numbers support a weaker SAC-mediated arrest in Mad1^+/− versus wt cells.

We next crossed Mad1^+/− and Mad2^+/− mice (16) to generate doubly heterozygous Mad1^+/−Mad2^+/− mice. We reasoned that cells doubly heterozygous null for Mad1 and Mad2 might have an even more attenuated SAC than either Mad1^+/− or Mad2^+/− cells (25). Mad2^+/− and Mad1^+/−Mad2^+/− MEFs were also isolated and evaluated for SAC arrest after nocodazole treatment. Nine percent of Mad2^+/− and only 6% of Mad1^+/−Mad2^+/− MEFs were found in M phase (Fig. 3A and B), verifying a further weakened SAC in the doubly heterozygous cells. In our experiments, many Mad1^+/−, Mad2^+/−, and Mad1^+/−Mad2^+/− cells that completed mitosis and cytokinesis contained micronuclei; the finding of micronuclei is consistent with increased missegregation of chromosomes in these cells (26).

The above experiments measured SAC function after drug provoked microtubule depolymerization. However, constitutive tumors developed in drug-untreated Mad1^+/− mice at a rate higher than in wt mice (Table 1B). To assess and compare ambient SAC function, we carefully examined mitoses of drug-untreated wt, Mad1^+/−, Mad2^+/−, and Mad1^+/−Mad2^+/− MEFs. Interestingly, abnormal mitoses were frequent in the latter three cell types. Figure 3C shows several examples of chromosomal aberrations (lagging chromosomes, DNA bridging between the separated chromatids, and multipolar spindles) during various points of mitosis in Mad1^+/−, Mad2^+/−, and Mad1^+/−Mad2^+/− MEFs. Statistically, 11% of Mad1^+/−, 14% of Mad2^+/−, and 26% of Mad1^+/−Mad2^+/− MEFs had mitotic abnormalities (Fig. 3D). These numbers indicated a rank order of Mad1^+/−Mad2^+/− > Mad2^+/− ≥ Mad1^+/− in constitutive genetic instability.

We independently checked chromosomal instability of wt, Mad1^+/−, Mad2^+/−, and Mad1^+/−Mad2^+/− MEFs by FISH using a mouse chromosome 2–specific probe (Fig. 4A). Here, we used losses and gains in chromosome 2 as measures of propensity for aneuploidy. Consistent with results in Fig. 3, Mad1+Mad2 doubly haploinsufficient MEFs were more prone to gain or lose chromosome 2 than their singly haploinsufficient counterparts (Fig. 4B). Thus, 14.5% of Mad1^+/−Mad2^+/− MEFs were aneuploid for chromosome 2 compared with 6.0% of Mad1^+/−, 7.2% of Mad2^+/−, and 1.4% of wt MEFs.

Doubly haploinsufficient Mad1^+/−Mad2^+/− MEFs are more tumorigenic in nude mice. Collectively, the above data suggest a gradation of spontaneous genetic instability with Mad1^+/−Mad2^+/− > Mad2^+/− ≥ Mad1^+/− > wt. Our comparisons of constitutive in vivo tumor development in cohorts of aging Mad1^+/−, Mad2^+/−, and Mad1^+/−Mad2^+/− mice are ongoing, and data will not be available for >1 year. However, at this stage, we can assess the relative efficiencies of the various MEFs as cell explants (∼2 weeks after isolation from embryos) for tumorigenesis. Hence, we injected equal numbers of each cell type into athymic nude mice (Fig. 5A) and assessed tumor growth. Tumors appeared within 4 weeks in 100% (eight of eight) of mice injected with Mad1^+/−Mad2^+/− cells. By comparison, eight of eight mice injected with Mad2^+/− MEFs also developed tumors within 4 weeks; however, these tumors averaged 40% smaller in size (243 ± 142 mm³ for Mad2^+/− MEFs compared with 399 ± 141 mm³ for Mad1^+/−Mad2^+/− MEFs, measured at 6 weeks postexplant; Fig. 5B). On the other hand, only 37.5% (three of eight) of animals injected with Mad1^+/− MEFs developed tumors, and these tumors grew more slowly and did not become visually apparent or palpable until weeks 6 to 10. None of the eight mice injected with wt MEFs developed tumor (Fig. 5A and B). Histologically, all tumors were fibrosarcomas with pleomorphic spindle cells (Fig. 5C), consistent with the expected cell type from the injected cells. The rapidity, size, and frequency of tumor development correlated with the relative degrees of ambient genetic instability in MEFs subscribing to an ordering of Mad1^+/−Mad2^+/− > Mad2^+/− ≥ Mad1^+/− > wt.

Complete loss of SAC proteins, MAD1, MAD2, BUBR1, or BUB3, is embryonic lethal in mice (present study; refs. 12, 14, 16), suggesting essentiality of these factors for mouse development. Although the reasons for embryonic lethality are unknown, SAC normally functions to guard against missegregation of chromosomes during somatic mitosis. Observations that cancer cells are frequently aberrant for expression of SAC proteins (4, 5) have raised a suggestion that loss in or weakening of SAC may be causal for carcinogenesis. Thus far, in vivo verification of this hypothesis based on mice heterozygous null for Mad2, BubR1, or Bub3 have yielded inconsistent results. Mad2^+/− mice show higher than wt constitutive tumor development (16), whereas BubR1^+/− and Bub3^+/− mice do not (12, 14, 21). Nevertheless, one should be cautious in making cross-comparisons because it remains unknown how MAD2, BUBR1, and BUB3 may differ in additional non-SAC functions and whether noncheckpoint differences might influence in vivo tumor development. For example, BubR1 mice have reduced fertility and a premature aging phenotype (15), which makes difficult age-matched comparisons of tumor development.

Here, our Mad1^+/− results are consistent with previous results from Mad2^+/− animals (16) and add credence to the notion that weakening of the SAC contributes to increased constitutive tumorigenesis. Empirically, although chromosome instability (CIN) is a feature of most human cancers, conclusive evidence correlating the degree of genetic instability with quantitative tumor development remains lacking. In an attempt to explore this correlation, we investigated the ex vivo genetic stability of Mad1^+/−, Mad2^+/−, and Mad1^+/−Mad2^+/− MEFs. Our analyses within the narrowly restricted MAD pathway, using several independent measures of genetic instability (Figs. 3–5), indicate that the degree of CIN is that Mad1^+/−Mad2^+/− > Mad2^+/− ≥ Mad1^+/− > wt. Interestingly, when the same MEFs were explanted into nude mice, the in vivo rapidity, frequency, and size of tumors also ranked Mad1^+/−Mad2^+/− > Mad2^+/− ≥ Mad1^+/− > wt (Fig. 5), consistent with the ex vivo quantification of chromosomal instability. Pending replication of results using larger numbers of mice, our findings are consistent with the hypothesis that the degree of weakness in SAC correlates with the number of aneuploid cells, which emerge and then develop into tumors. We caution that this correlation in mice may not extrapolate directly to humans because there are differences between the natural stringencies of mouse and human SAC (27, 28), and there are well-described mechanistic divergences between mice and men, which render the former far more cancer prone (29). Nonetheless, the Mad1^+/−, Mad2^+/−, and Mad1^+/−Mad2^+/− mice represent comparative tools restricted to a specific checkpoint pathway useful for investigating how SAC weakness translates into tumorigenesis. We note that others have also proposed a relationship between chromosomal instability, aneuploidy, and tumorigenesis (30, 31).

Finally, constitutive tumors developed in Mad1^+/− mice in multiple organs. Curiously, Mad1^+/− mice treated with vincristine showed a heightened propensity to develop lung tumors, similar to the tumor tissue type observed in Mad2^+/− mice (16). To date, up to 40% of human lung cancer cells have been found to carry mitotic checkpoint defects, including in MAD1 (32–34). It is intriguing that from our current results (Fig. 2B; Table 1B), mouse lung tissue seems to be relatively more sensitive to deregulation in the MAD components of the SAC. Future studies are needed to shed light on whether there are MAD-SAC functions specific to lung tissue development and pulmonary cell division.

Grant support: National Institute of Allergy and Infectious Diseases (NIAID)/NIH intramural funds and a NIAID contract (SoBran, Inc., Dayton, OH).

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.

We thank Drs. Mathew Starost and Michael Eckhaus (Division of Veterinary Resources, NIH, Bethesda, MD) and Dr. Torgny Fredrickson (Contract Management Branch, NIAID, NIH) for necropsy and histopathology analyses, Cindy Erexson and Larry Faucette for excellent technical assistance, and Dr. Dong-yan Jin and members of the Jeang laboratory for critical readings of the manuscript.