Since the discovery of Bub1 mutations in colorectal tumor cell lines several years ago (8
), considerable interest has focused on the role of spindle checkpoint impairment in human cancer. Several subsequent studies found that mutations in known spindle checkpoint genes are rare in a variety of tumors or tumor cell lines, although checkpoint impairment by epigenetic mechanisms was observed in some cases (7
). These studies cumulatively call into question the role of spindle checkpoint impairment in cancer.
We examined the consequences of overexpressing a dominant-negative Bub1 fragment in mouse thymocytes in vivo to determine if spindle checkpoint impairment contributes to lymphomagenesis. Surprisingly, we did not detect any evidence that the BubDN protein impaired the spindle checkpoint, as transgenic thymocytes cultured in vitro arrested normally in response to nocodazole treatment. The only phenotypes observed in Tg-LBDN mice were (i) a modest reduction in thymocyte numbers with increased age and (ii) impaired in vitro proliferation of thymocytes from aged animals. Tg-LBDN+ mice did not show significantly increased tumor susceptibility. Heterozygous or homozygous mutation of the p53 gene also failed to uncover significant tumorigenic effects of the transgene that might have been suppressed by p53-dependent checkpoints. Together, our data indicate that thymic overexpression of Bub1DN has no significant effect on thymic development or tumorigenesis.
The Bub1DN fragment used in the present work has been used extensively in cultured cells, inducing significant spindle checkpoint impairment (2
). For example, HeLa cells expressing Bub1DN displayed accelerated mitotic passage under normal growth conditions and decreased mitotic arrest and apoptosis after nocodazole treatment (38
). Bub1DN expression reversed growth arrest and transformed Brca2-deficient mouse embryonic fibroblasts, rendering them resistant to nocodazole-induced mitotic arrest (26
). There are several possible explanations for why expression of Bub1DN did not elicit the same spindle checkpoint effects in thymocytes. First, it is possible that cell lines used in previous in vitro studies have accumulated changes in additional factors, “relaxing” the mitotic checkpoints and making them abnormally sensitive to Bub1 inhibition. Thus, the dominant-negative Bub1 fragment may not be sufficient for checkpoint deregulation in primary cells. However, the same Bub1DN protein was recently shown to cause accelerated passage through meiosis I and to disrupt nocodazole-induced arrest in meiosis II when expressed in mouse oocytes (39
). These results suggest that Bub1DN can have similar effects in primary cells cultured in vitro, although effects on meiosis and mitosis could be distinct. Thymocytes could also be resistant to the dominant-negative effects via differences in spindle checkpoint regulation or in expression levels of spindle checkpoint components. The lack of effect could also be due to technical challenges associated with the animal studies described here. For example, it is possible that Bub1DN expression levels achieved in Tg-LBDN mice are insufficient to completely override endogenous Bub1 function in thymocytes. However, we have shown that the Bub1DN fragment is expressed at significantly higher levels than endogenous Bub1. It has been noted that this fragment displays dominant-negative activity in cell culture only when the cells are challenged with spindle inhibitors (2
). Nevertheless, that observation differs from other reports in which the same fragment caused shortened mitosis in untreated HeLa cells (16
). Such differences could reflect distinct spindle checkpoint regulation in distinct cell types, different levels of Bub1DN expression, or other unknown, variables between studies. It is noteworthy that an amino-terminal fragment of yeast Bub1 supports partial spindle checkpoint activity in vivo, independently of the C-terminal kinase domain. Nevertheless, overexpression of full-length wild-type or N-terminal Bub1 fragments in yeast caused chromosome segregation errors, supporting a dominant-negative effect due to overexpression (43
). Thus, while the evidence supporting dominant-negative activities of Bub1 N-terminal fragments is compelling in yeast and cell culture systems, the behavior of these fragments in distinct cell types in vivo will need to be carefully evaluated to determine if dominant activities are induced.
The equivalent survival of p53 heterozygous mice with or without the LBDN transgene supports the conclusion that the Bub1DN fragment does not induce significant genomic instability in vivo. p53 heterozygous mice have a shortened life span and high rates of thymic lymphomagenesis in response to irradiation or chemical mutagenesis, both of which elicit increased genomic instability (22
). The majority of irradiation-induced tumors in p53 heterozygous mice display loss of the wild-type p53 allele (p53LOH), suggesting that p53 loss plays a key role in tumor development. If thymic expression of the Bub1DN protein in Tg-LBDN mice induced significant genomic instability, the frequency of p53LOH and subsequent tumorigenesis should be greater in Tg-LBDN+ p53+/−
animals than in controls. The absence of such cooperativity in our experimental cohorts supports the conclusion that the Bub1DN protein does not elicit significant genomic instability in vivo.
A potential caveat in interpreting tumorigenesis data in this study concerns the use of the lck
proximal promoter for thymic expression of Bub1DN. It is possible that Tg-LBDN+
mice do not develop tumors because the transgene is not expressed in cell types relevant to the transformation process. The originating cell type of most cancers is controversial, but significant evidence suggests a stem cell origin for many cancers (31
). It is thus possible that spindle checkpoint impairment and/or chromosome missegregation in hematopoietic stem or progenitor cells would initiate tumorigenesis, while similar effects in more-differentiated thymocytes would not. Even within the thymus, lesions may need to occur in specific thymocyte subsets, most likely the proliferative subsets, to drive tumorigenesis. Thymocytes pass through several developmental stages marked by expression of specific cell surface markers and by regulated proliferation (15
). Early thymocyte subsets that lack CD4 and CD8 expression (double-negative [DN] cells) represent only 1 to 2 percent of total thymocytes but account for a significant portion of thymocyte proliferation. DN thymocytes are divided into four subtypes (DN1 to DN4) based on expression of the CD44, CD25, and CD117 surface markers. The DN3 (CD44−
) and DN4 (CD44−
) subsets make up approximately 90% of DN thymocytes and display high proliferative activity that gives rise to the large numbers of CD4+
(double-positive [DP]) thymocytes (15
). Thus, LBDN transgene expression in the proliferative DN3 and DN4 thymocyte subsets would provide the greatest opportunity for chromosome missegregation events that may facilitate tumor formation. Our data suggest that the LBDN transgene is expressed in DN thymocytes, since Bub1DN expression was maintained in cells cultured in IL-1α and IL-2. These conditions produced a predominant DN population after 1 week of culture (data not shown), consistent with published reports that IL-1α and IL-2 stimulate DN thymocyte proliferation (37
). Our results are also consistent with a study of four transgenic mouse lines in which the same lck
promoter and RNA processing signals used here drive green fluorescent protein (GFP) expression. All four lines of lck-
GFP transgenic mice initiated GFP expression by the DN3 stage, with some lines expressing GFP as early as the DN1 stage (5
). This expression pattern is also consistent with the mRNA expression of the endogenous lck
gene, which is detectable in all DN subsets (5
). We also note that several transgenes driven by the lck
proximal promoter do elicit tumors in transgenic animals, indicating that the cells targeted by this promoter can be transformed in vivo (9
). Thus, while we cannot rule out the possibility that Bub1DN would need to be expressed in stem cells to elicit tumorigenesis, the simplest explanation for our data is that the Bub1DN protein does not display sufficient dominant-negative activity to impair the spindle checkpoint in thymocytes.
The decreases in proliferation and overall thymocyte numbers in aged LBDN+ animals are puzzling given the lack of any apparent spindle checkpoint phenotype. Flow cytometry profiles of stimulated cells showed lower percentages of blasts as evidenced by forward and side scatter properties, suggesting that transgenic cells may be defective in responding to stimulation and/or cell cycle entry (data not shown). It is unclear how the LBDN transgene might cause such a defect and why it would be observed only in older mice. Additional analyses will be needed to understand the basis of this modest phenotype. Nevertheless, our data strongly indicate that the LBDN transgenic mice are not a suitable model for spindle checkpoint disruption in thymocytes.
Thus, additional mouse models will be required in order to definitively address the roles of bub1
mutation, spindle checkpoint impairment, and chromosome instability in tumorigenesis. The caveats involved in interpreting dominant-negative data such as those described here suggest the need for care in developing such models. While mice heterozygous for Mad2, BubR1, Rae1, and Bub3 display some increased propensity for tumorigenesis (3
), it remains unclear whether the observed tissue specificity reflects distinct spindle checkpoint regulation or special sensitivity in those tissues. A conditional knockout strategy represents the best initial strategy to define the role for the spindle checkpoint in specific cell types in vivo. The availability of transgenic mice expressing the cre
recombinase from a variety of promoters will allow tests in specific cell types in vivo. However, because the spindle checkpoint may play an essential role in normal cell division, it is possible that complete ablation will also be incompatible with tumorigenesis. In line with this hypothesis, a recently described hypomorphic mutation in the BubR1
gene caused aneuploidy and premature aging phenotypes without increasing tumorigenesis (4
), and reduction of Mad2 and BubR1 expression by RNA interference in human cancer cell lines was lethal (24
). It is noteworthy that spindle checkpoint mutations identified in human tumors to date are all heterozygous mutations suggested to confer dominant-negative activities (8
). Such dominant mutations may confer sufficient checkpoint impairment for tumorigenesis while allowing enough residual activity for cell survival. Thus, conditional knock-in models of tumor-derived spindle checkpoint mutants may ultimately prove to be the best way to fully investigate the role of spindle checkpoint impairment in tumorigenesis in various tissues in vivo.
In conclusion, our results indicate that thymic overexpression of an amino-terminal Bub1 fragment does not disrupt the spindle checkpoint or promote tumorigenesis in mouse thymocytes, despite evidence from yeast and cell culture systems that the Bub1 fragment used in this study induces checkpoint impairment. This suggests that distinct cell cycle control mechanisms or redundancy may limit the effects of some dominant inhibitors in vivo, suggesting the need for caution in the use of such reagents. Our data also suggest that chromosome missegregation phenotypes observed in cell culture, such as those elicited by this reagent, may not accurately predict in vivo relevance.