Mouse Models for Chromosomal Instability

Aneuploidy means any karyotype that is not euploid, anything that stands outside the norm. Two particular characteristics make the research of aneuploidy challenging. First, it is often hard to distinguish what is a cause and what is a consequence. Secondly, aneuploidy is often associated with a persistent defect in maintenance of genome stability. Thus, working with aneuploid, unstable cells means analyzing an ever changing creature and capturing the features that persist. In the book Aneuploidy in Health and Disease we summarize the recent advances in understanding the causes and consequences of aneuploidy and its link to human pathologies.


60
link the centromeric DNA to the mitotic spindle. The inner kinetochore plate is the DNA interacting domain and is comprised of several structural proteins such as the centromeric proteins (CENP). Some of these CENP genes (e.g. CENPA, CENPF and CENPH) frequently show altered expression in human cancer, underscoring the importance of a precisely balanced protein composition of the kinetochore (Yuen et al., 2005).

The mitotic spindle checkpoint
To ensure the accurate segregation of chromosomes, cells initiate segregation only after the correct attachment of all chromosomes to microtubules of the mitotic spindle. The spindle checkpoint (SAC) operates at the outer kinetochore plate by monitoring the attachment of the kinetochore to the microtubules. Improperly attached kinetochores recruit a family of SAC proteins (Mad1, Mad2, Bub1, BubR1, Bub3, Mps1, Rod and ZW10) (Musacchio & Salmon, 2007). Additionally, CENPE, a motor protein, is recruited to unattached kinetochores, where it is implicated in regulating microtubule attachment (Kops et al., 2005). Even a single unattached kinetochore is sufficient to trigger a robust spindle checkpoint response that prevents premature anaphase onset and thus missegregation of chromosomes (Rieder et al., 1995, Rieder et al., 1994. This robust response is achieved by CDC20 sequestering by the SAC proteins, the specificity factor required for anaphase promoting complex/cyclosome (APC/C) activity. Activation of the APC/C, an E3 ubiquitin ligase, is essential for anaphase progression, as it is responsible for the degradation of several mitotic proteins including Securin and Cyclin B1. Securin degradation leads to activation of Separase, which cleaves Cohesin that glues sister chromatids together. Cyclin B1 degradation results in the inactivation of CDK1 and thereby allows mitotic exit (Karess, 2005, Kops et al., 2005, Musacchio & Salmon, 2007. Thus, precise regulation of APC/C activity by the SAC allows for timely and accurate progression of mitosis preventing numerical CIN.

Tension versus attachment
Kinetochore-microtubule attachment by itself is not sufficient to satisfy the spindle checkpoint. To achieve proper segregation, it is also essential that one kinetochore of a pair is attached to microtubules from one spindle pole and the other kinetochore to the opposite pole (Biggins & Walczak, 2003). This correct geometry of attachment generates tension between the sister chromatids, which is another prerequisite for mitotic progression. Aurora B is implicated in sensing this tension and correcting erroneous attachments (Andrews et al., 2003). In the absence of tension, for instance when both sister chromatid kinetochores are connected to the same centrosome, Aurora B kinase activity remains uninhibited. This results in the release of improper microtubule attachments and allows reattachment of the kinetochores with microtubules emanating from the opposite pole (Andrews et al., 2003, Liu et al., 2009, Musacchio & Salmon, 2007. While details of error correction mechanisms are just emerging, drugs targeting this pathway (e.g. Aurora inhibitors) are already in clinical trials for cancer (Jackson et al., 2007).
into the two daughter cells during mitosis (see for detailed review: (Doxsey et al., 2005, Fukasawa, 2005, Nigg, 2007, Tsou & Stearns, 2006 ). Defects in these control mechanisms can lead to abnormal centrosome numbers and are thus another cause of numerical CIN. In agreement with this, most tumours exhibit abnormal centrosome numbers (Nigg, 2006, Pihan et al., 1998. It was originally hypothesized that abnormal centrosome numbers would provoke CIN through multipolar mitoses. Although careful tracking revealed that such divisions occasionally do take place, the resulting daughter cells all died in the next cell cycle. It's therefore unlikely that cells undergoing multipolar divisions contribute to a developing tumour. Instead, in a large proportion of the cells supranumeral centrosomes clustered to form one 'supercentrosome' allowing 'normal' bivalent divisions to take place. Such divisions frequently coincided with lagging chromosomes in the next mitosis, presumably the result of merotelic attachments caused by abnormal centrosomal structure (Ganem et al., 2009). Therefore, lagging chromosomes might very well be the driving force behind numerical CIN associated with supranumeral centrosomes.

Mouse models for whole chromosome instability
To test whether numerical CIN is an early event in tumourigenesis and thereby infer whether it can be causative of cancer, several proteins playing a role in accurate segregation were disrupted in mice. The results of these studies are discussed below and collectively underscore the importance of the spindle checkpoint in the different aspects of tumourigenesis.

Loss of centromeric (CENP) genes frequently causes embryonic lethality
CENPA, B and C reside at the inner plate of the kinetochore and bridge between the centromeric DNA and the kinetochore-docking microtubules emanating from the centrosomes. Several of these genes have been knocked out in mice to determine their role in the prevention of aneuploidy. Whereas homozygous CENPB loss resulted in no overt phenotypes other than slightly lower testis and body weights (Hudson et al., 1998, Kapoor et al., 1998, Perez-Castro et al., 1998, disruption of CENPA or CENPC led to embryonic lethality in early stages of embryogenesis (E3.5-E6.5). The early embryos show micronuclei, a lowered mitotic index (indicative of mitotic delay), and enlarged nuclei (suggestive of tetraploid cells). Heterozygous mice develop normally and are fertile, although animals were not tested for cancer predisposition using carcinogens or in combination with other predisposing mutations (Howman et al., 2000. Animal longevity and spontaneous cancer predisposition are currently being tested [Andy Choo, personal communication].
The chromosomal passenger proteins Incenp, Survivin and Aurora B are part of the chromosomal passenger complex. This complex is essential for forming a bipolar spindle by measuring tension during metaphase and is involved in later events in cytokinesis as well (Lens et al., 2006). Knockouts for Survivin or Incenp revealed that ablation of either protein causes embryonic lethality before day E8.5, similar as observed for CENP proteins and SAC protein knockouts (Cutts et al., 1999, Uren et al., 2000.

Does reduced expression of CIN protective genes result in spontaneous tumour formation?
Several cancers show decreased expression of spindle checkpoint proteins, indicating that a partially compromised checkpoint can be the underlying cause for tumour aneuploidy (Weaver & Cleveland, 2006). As homozygous inactivation of SAC components aborts embryonic development well before mid-gestation, heterozygous mice (Mad1, Mad2, BubR1, Bub1, Bub3, Rae1, Cdc20, Cdh1 and Cenp-E) were used to study the in vivo consequences of a partially defective spindle checkpoint. In all cases, heterozygotes are born at a Mendelian ratio and none show clear developmental defects, except for a mild hematopoietic defect in BubR1 heterozygotes (Babu et al., 2003, Baker et al., 2006, Garcia-Higuera et al., 2008, Iwanaga et al., 2007, Jeganathan et al., 2007, Kalitsis et al., 2005, Li et al., 2009, Michel et al., 2001, Q. Wang et al., 2004, Weaver et al., 2007. However, despite normal development, Mad2, Mad1 and CENPE heterozygotes develop cancers in a substantial number of the animals (20-30%), albeit relatively late (at 18-20 months of age). Lung tumours are predominant, and CENPE heterozygotes develop hematopoietic malignancies as well (Iwanaga et al., 2007, Michel et al., 2001, Weaver et al., 2007, indicating that Mad1, Mad2 and CENPE function as haplo-insufficient tumour suppressors in the mouse by suppressing CIN. Heterozygousity for the SAC-downstream-targets Cdh1 or Cdc20 predisposes for late tumours as well: up to 50% of the Cdc20 hypomorphic mice develop tumours, mostly in the hematopoietic compartment, whereas Chd1 heterozygotes mainly develop solid tumours in the mammary gland, lungs, liver and testes (Garcia-Higuera et al., 2008, Li et al., 2009. Heterozygous BubR1, Bub1, Bub3 and Rae1, and even Bub3; Rae1 double heterozygotes do not develop any malignancies, suggesting that in those models aneuploidy is not sufficient for tumours to arise (Baker et al., 2004, Baker et al., 2006, Jeganathan et al., 2007, Q. Wang et al., 2004. Finally, similar to SAC knockout heterozygotes, Survivin +/-and Incenp +/-mice are indistinguishable from their wildtype littermates. Whether these animals are cancer-predisposed is currently not known [Andy Choo, personal communication]. In another approach, a hypomorphic allele for Bub1 was created, reducing protein levels down to 20% of wildtype levels. These mice develop several malignancies within 18-20 months of age, mainly lymphomas, lung and liver tumours (Jeganathan et al., 2007). Using the same approach, a hypomorphic allele for BubR1 was engineered (reducing BubR1 protein levels by 90%). Surprisingly, although these mice develop massive aneuploidy in several tissues, they are not cancer-predisposed, suggesting that aneuploidy does not cause cancer per se (Baker et al., 2004). and muscle atrophy. The median lifespan of these animals is six months and none of them age over 15 months (Baker et al., 2004), which might obscure a late-in-life cancer phenotype. Bub3; Rae1 double heterozygotes develop a similar phenotype, albeit less severe as these animals survive up to 27 months (Jeganathan et al., 2007). Interestingly, BubR1 levels seem to decline in aging wildtype tissue, which suggests a role for this gene and possibly aneuploidy in natural aging as well (Baker et al., 2004). However, why the premature aging phenotype affects these two knockouts and not other aneuploidy-related mouse models is currently unknown.
Phenotypes of the various mouse models are summarized in Table 1. Taken together, the results show that SAC defects have a profound effect on development and cancer predisposition. Additionally, mouse models targeting the SAC have revealed a potential role for aneuploidy in aging.

Aneuploidy as a collaborating factor in tumorigenesis
Even though mouse models clearly indicate that numerical CIN can contribute to tumour initiation, the actual tumour phenotypes of these mouse models are relatively weak, with tumours arising sporadically and relatively late in life. Why is this effect so weak despite the high incidence of aneuploidy in human cancer? One explanation is that the current mouse models rely on partial inactivation of the spindle checkpoint, since complete inactivation is incompatible with embryonic development. Another possible explanation for the weak phenotypes is that aneuploidy requires predisposing mutations in order to be tumorigenic and therefore plays a more important role in tumour progression. To test the latter hypothesis, several of the described models were exposed to carcinogenic drugs or combined with other mouse models carrying predisposing mutations in established tumour suppressor genes.
Indeed, tumour phenotypes are exacerbated by carcinogenic insults. For instance, whereas Bub1, Bub3, Rae1 and Bub3; Rae1 double heterozygous mice are not clearly cancer predisposed, DMBA treatment (a known carcinogen) provokes more tumours in all four knockouts than in wildtype mice (Baker et al., 2006, Jeganathan et al., 2007. This is also true for DMBA-treated BubR1 hypomorphs (Baker et al., 2004). Furthermore, BubR1 heterozygotes are more susceptible to the formation of intestinal tumours in response to treatment with the colon carcinogen azoxymethane (Dai et al., 2004). Finally, 40% of Mad1 heterozygous mice treated with Vinicristine (a microtubule-depolymerizing agent) develop tumours whereas control-treated mice do not develop any (Iwanaga et al., 2007). Similarly, numerical CIN synergizes with predisposing mutations. For instance, decreased levels of Bub1 accelerate tumorigenesis in a p53 +/-or Apc +/Min background. In this setting, numerical CIN acts by facilitating LOH of the remaining wildtype allele of the tumour suppressor gene (Baker et al., 2009).

Aneuploidy as a tumour suppressor?
Although decreased levels of Bub1 have a profound effect on tumour initiation in p53-of Apc-mutated backgrounds, this synergy is not observed in Rb or Pten heterozygous mice. Paradoxically, Bub1 hypomorphism acts tumour suppressive in Pten +/-mice, with Pten +/-; Bub1 hypomorphs developing fewer hyperplastic prostate lesions than Pten +/-mice expressing wildtype Bub1 levels (Baker et al., 2009). Similarly CENPE heterozygousity delays tumourigenesis in a p19 Arf -deficient background. Whereas normally p19 Arf-/-animals develop lymphomas and sarcomas with an average latency of 6-7 months (Kamijo et al., 1997), tumour latency is delayed to 12 months in CENPE +/-mice. Furthermore, despite a predisposition for hematopoietic malignancies, the number of spontaneous liver tumours in CENPE +/-mice is reduced 3-fold and liver tumour sizes are remarkably smaller. Finally, CENPE heterozygousity also delays DMBA-induced tumours (Weaver et al., 2007). These observations suggest that aneuploidy can delay tumourigenesis as well and that the response to aneuploidy might very well be tissue specific.

Overexpression of genes with a role in chromosome segregation
Diminished Mad2 expression, mutations and even gene loss do occur in cancer, but are infrequent (Percy et al., 2000, X. Wang et al., 2002, X. Wang et al., 2000. More often Mad2 expression is elevated (Alizadeh et al., 2000, Hernando et al., 2004, van 't Veer et al., 2002. This upregulation is likely the result of increased E2F activity in cancers, as the RB (retinoblastoma) -E2F pathway is altered in more than 80% of human tumours (Malumbres & Barbacid, 2001) and the E2F transcription factor family is an important modulator of Mad2 transcription (Hernando et al., 2004). Furthermore, deregulation of the RB pathway has been associated with CIN (Lentini et al., 2006, Lentini et al., 2002, Schaeffer et al., 2004, and elevated Mad2 levels can provide an explanation for this. Indeed, reducing Mad2 protein levels to wildtype levels in Rb deficient cells reduces the number of aneuploid cells. Conversely, overexpression of either E2F or Mad2 in wildtype cells increases aneuploidy (Hernando et al., 2004). These results strongly suggest that unscheduled activation of E2F contributes to aneuploidy through elevated Mad2 expression. To test if this is true in vivo, an inducible Mad2 overexpression mouse model was engineered. These mice display a wide spectrum of tumours (lung adenomas, lymphomas, hepatocellular carcinomas and fibrosarcomas) with a similar latency (~20 months), but higher frequency (~50% at 20 months) than Mad2 heterozygous mice (Sotillo et al., 2007). These results seem somewhat counterintuitive: why would increasing levels of checkpoint protein lead to cancer? Perhaps, delayed degradation of both Securin and Cyclin B1 during mitosis causes defects in mitotic exit and thus explains the observed chromosomal aberrations and subsequent tumourigenesis. Even though this data indicates a clear role for Mad2 overexpression in tumorigenesis, several other mechanisms might contribute to the CIN phenotype of cells with aberrant E2F activity. For instance, Rb inactivation in mouse embryonic fibroblasts also reduces cohesion between sister chromatids, leading to premature chromosome segregation and CIN (Coschi et al., 2010, Manning et al., 2010, van Harn et al., 2010. Furthermore, inducible overexpression of Hec1, another Rb-interacting protein residing at the kinetochore, provokes aneuploid cancers 15-18 months following induction (Diaz-Rodriguez et al., 2008) similar to Mad2 overexpressing mice. Therefore, even though overexpression of Mad2 has a clear cancer phenotype, it is unlikely that Mad2 is the only downstream target of the RB pathway involved in the maintenance of a euploid genome.

Other mouse models for numerical CIN
Not all aneuploidy is the result of spindle checkpoint failure; other mitotic defects can also perturb chromosome segregation and thus cause aneuploidy. Lzts1 was discovered as a gene that is frequently lost in breast, lung, gastric, esophageal, prostate, and bladder cancers. Lzts1 impairs Cyclin B1 activation late in mitosis, resulting in lowered Cyclin B1-Cdk1 activity in mitosis and premature mitotic exit, similar to Mad2 overexpression. Indeed, Lzts1-deficient mice develop a wide spectrum of tumours within 19 months of age with a high penetrance. Lzts1 loss also facilitates carcinogen-induced tumourigenesis, as NMBA treatment resulted in a tumour incidence of 100% in both Lzts1-deficient and heterozygous animals, whereas only 15% of the treated control animals develop tumours. Therefore, premature mitotic exit through decreased Cyclin B1 levels might be an important driver of (numerical) CIN and cancer (Vecchione et al., 2007).
Entry into mitosis is guarded by a prophase checkpoint that can be activated by chromosomal damage or disruption of microtubules (Mikhailov et al., 2002, Pines & Rieder, 2001. Chfr is a potential E3 ubiquitin ligase that functions early in prophase (Matsusaka & Pines, 2004), presumably by controlling expression levels of mitotic proteins such as Aurora A. Unlike SAC components, homozygous deletion of Chfr in the mouse is dispensable for normal development. However, half of the Chfr-deficient animals develop tumours (lymphomas, lung, liver and intestinal tumours) within 20 months of age and furthermore the knockouts show accelerated tumour development upon carcinogenic insults (DMBA). (Yu et al., 2005). Both phenotypes (Lzts1, Chfr) are very similar to those of the cancerpredisposed SAC-compromised heterozygotes, emphasizing the importance of accurately regulated mitotic kinase activity (Cyclin B1-Cdk1, Aurora A) in preserving normal chromosomal numbers and suppression of cancer.
Micronuclei formation is another indication of chromosomal instability. An elegant genetic screen for genes that facilitate micronucleus formation in erythrocytes identified a point mutation in Mcm4 (Chaos3). Mcm4 is part of the Mcm2-7 replication licensing complex and therefore a crucial part of the DNA replication machinery. Mcm4 mutant mice showed decreased expression of several Mcm proteins and primary cells from this mouse also showed increased sensitivity for the replication inhibitor aphidicolin in line with a defect in DNA replication (Shima et al., 2007). However, as this mutation was identified as a mutation that provokes micronuclei formation, this model suggests that defects in DNA replication can cause numerical chromosomal abnormalities as well.
The Adenomatous polyposis coli (APC) gene is a tumour suppressor gene that is often mutated in human colorectal cancers (note that this gene is unrelated to the anaphase promoting complex APC/C). In addition to APC's well-known role in regulating/dampening β-catenin's transcriptional activity and cell proliferation, multiple lines of evidence indicate that mutated APC contributes to chromosome missegregation and numerical CIN. For instance, cell culture studies have shown that APC localizes to kinetochores and microtubule ends. Perturbing APC function by mutation, overexpression or depletion causes aneuploidy (Draviam et al., 2006, Green et al., 2005, Tighe et al., 2004. In addition, mice expressing truncated APC mutant protein display multiple intestinal tumours with aberrant mitosis and polyploidy, indicating a strong link between the loss of APC's mitotic function and tumourigenesis (Caldwell et al., 2007, Dikovskaya et al., 2007, Oshima et al., 1995, Su et al., 1992. Interestingly, this phenotype can be substantially aggravated by concomitant heterozygous BubR1 loss (Rao et al., 2005). Finally, transcriptome analysis of APC-mutated cells revealed that the mutation resulted in the increase of BubR1 and Mad2 expression, which is also observed in adenomas and colorectal carcinomas (Dikovskaya et al., 2007). Together, these findings illustrate an important link between deregulation of APC and the proteins that control mitosis in the progression from early adenomas to aggressive carcinomas.

Lessons learnt from Mouse Embryonic Fibroblasts
MEFs isolated from several of these mouse models have been invaluable in estimating the levels of aneuploidy occurring in vivo. However, as complete ablation of spindle checkpoint genes resulted in embryonic lethality before day E8.5 and MEFs are typically isolated at E15.5-14.5, it has been difficult to generate MEFs from the homozygous knockouts. Therefore heterozygous MEFs were analyzed for the extent of aneuploidy (Table I). All heterozygous MEFs show significantly elevated numbers of aneuploid cells compared to wildtype control MEFs, ranging from 10% aneuploidy in the Mad1 +/-MEFs to 50% in the Mad2 heterozygotes and Mad2 overexpressing cells. Note that wildtype MEFs typically show less than 10% aneuploid cells (Iwanaga et al., 2007, Michel et al., 2001, Sotillo et al., 2007. In addition, BubR1 hypomorphic and Bub3; Rae1 heterozygous MEFs exhibit premature senescence and increased protein levels of senescence-associated p19 Arf , p21 Cip1 and p16 Ink4a (Baker et al., 2004, Baker et al., 2006 in agreement with the observed premature aging phenotype. Even though all SAC-compromised MEFs show aneuploidy to a greater or lesser extent, the level of aneuploidy in MEFs does not fully correlate with cancer incidence (Table I). However, aneuploidy analyses might lead to different results depending on the passage number of the MEFs. For instance, detailed analysis of Cenp-E heterozygous MEFs revealed that up to 70% of the cells are aneuploid at high passage numbers whereas only 20% are aneuploid in early passages (Weaver et al., 2007). Furthermore, culturing MEFs in vitro might not be the best representation for every cell type in vivo. Therefore, even though MEF studies are useful to estimate CIN rates in vitro, in vivo assessment of aneuploidy will be crucial to link the extent of aneuploidy to cell fate.

Aneuploidy as a primary cause for cancer?
Since the first notion of a possible link between chromosomal abnormalities and malignant transformation by Theodore Boveri (Boveri, 1902, Boveri, 1914, we have learned a great deal about the causes and consequences of chromosome missegregation. Mouse models have provided crucial insight into how the SAC and mitotic machinery protect against CIN en cancer. The mouse models discussed in this chapter are far from complete especially as structural CIN and numerical CIN appear to be intertwined (Janssen et al., 2011). Therefore, mutations in genes protecting against structural CIN (Brca2, Mcm4, p53bp1, etc) might very well lead to aneuploidy too, even though these models were not the focus of the current chapter.
Complete inactivation of genes involved in chromosome segregation appears to be uncommon in human cancer. Why then is the vast majority of human cancers aneuploid? Transcriptome analysis of numerous cancers has revealed that overexpression of spindle checkpoint genes is more frequent than reduced expression (Oncomine database (Rhodes et al., 2004)). However, in many cases SAC gene products show mutations or truncations (Cahill et al., 1998, Kim et al., 2005, Olesen et al., 2001, Percy et al., 2000, Scintu et al., 2007, Tsukasaki et al., 2001. Possibly, mutated SAC proteins can act in a dominant-negative fashion, and therefore attenuate spindle checkpoint function. Careful biochemical analysis of such mutant gene products for their effect on the SAC status in combination with new mouse models for these cancer-associated mutations should clarify this issue. Mouse models for systemic SAC inactivation unequivocally show that removing the spindle checkpoint is not tolerated during embryonic development, when cells are dividing at maximum speed, and results in massive aneuploidy and apoptosis. This agrees fully with the observation that SAC inactivation is not tolerated in cancer cell lines (Kops et al., 2004). However, p53 inactivation in Mad2 deficient MEFs (derived from Mad2 -/-; p53 -/-embryonic stem cell lines) partly rescues this lethality, as these cell lines show better survival than Mad2 -/-MEFs, even though both cell lines become highly aneuploid (Burds et al., 2005). Therefore, p53 inactivation might synergize with SAC inactivation in malignant transformation by (partly) rescuing SAC-deficiency-induced cell death. Indeed, p53 heterozygosity appears to collaborate with Bub1 hypomorphism in transformation.
Partial inactivation of spindle checkpoint genes does not interfere with normal development (Table 1) and the average life span of these mice is unaltered. While some heterozygotes succumb to tumours at 18-20 months, others age without any cancer phenotype. The Bub3; Rae1 double heterozygotes show signs of premature aging (but not cancer), but do not succumb to the consequences of this phenotype earlier than their wildtype littermates (Baker et al., 2006). All together, the current mouse models for CIN argue that aneuploidy can provoke cancer. It remains unclear why some models are more prone to cancers than other, despite the fact that heterozygous MEFs from all mouse models show substantial aneuploidy, where tested. The only effect on life span is observed in BubR1 hypomorphic mice, which die a few months earlier due to premature aging (Baker et al., 2004). Premature aging coincides with increased numbers of senescent cells in vivo and furthermore, MEFs showed increased levels of the senescence-associated p16 Ink4a , p19 Arf and p21 Cip1 proteins. Remarkably, in vivo clearance of p16 Ink4A positive (i.e. senescent) cells can delay several of the premature aging sings in BubR1 hypomorphic mice (Baker et al., 2011), suggesting that senescence and aging , in addition to cancer, are important consequences of (numerical) CIN in vivo.
Premature senescence might also explain why Cenp-E +/-mice are less susceptible to tumours in a p19 Arf negative background or when tumours were induced chemically (Weaver et al., 2007). For instance, loss of p19 Arf , an activator of p53 activity, might lead to aberrant DNA damage signalling resulting in genomic instability. Similarly, challenging animals with carcinogenic compounds will induce CIN. Together with Cenp-E heterozygosity, the cumulative aneuploidy might rise to levels that ultimately result in senescence, instead of increased proliferation and cancer. These findings emphasize once more the importance of studying the relationship between senescence, aneuploidy and cancer.

Future directions
What are the next steps that will bring us closer to a therapy that specifically targets aneuploid cancers? We have now learnt that dramatic numerical CIN (e.g. by SAC inactivation) kills cells in vitro and interferes with development in vivo. We also now know that aneuploidy deregulates cell metabolism and restricts cell proliferation at the cellular level. Partial SAC inactivation appears to predispose for cancer in some models and to provoke a progeria/senescence phenotype in others. However, even though the vast majority of human tumours are aneuploid, few mutations in the SAC cascade have been identified, even though many human cancer cell lines display an abnormal response to anti-cancer drugs that normally activate the SAC, such as Paclitaxel.
Now that several conventional knockout mouse models have been analyzed in depth, we need to address more specific questions on the role that aneuploidy plays in cancer, for instance: -Is full SAC inactivation tolerated in adult tissues and if so, does it require additional mutations such as p53 inactivation? -What are the consequences of aneuploidy in vivo and do these resemble the consequences as found in aneuploid MEFs or yeast strains (Torres et al., 2007, Williams et al., 2008? -Which genes collaborate with expansion of aneuploid cell progeny? -What determines the cellular response to aneuploidy (proliferation/senescence)? -Can we specifically kill aneuploid cells, leaving euploid cells untouched?
To answer these questions more sophisticated mouse models are required. For instance, to answer the first question, conditional mouse models are needed. In such models the knockout allele can be deleted in tissues of choice leaving other tissues unaffected. Using this approach embryonic lethality can be circumvented in many cases. In addition, such models allow for comparing the differential responses to aneuploidy from tissue to tissue for each conditional knockout.
In addition, highly aneuploid tumour panels are required to extract the molecular consequences to aneuploidy in transformed cells. For instance, by comparing the molecular responses in different aneuploid tumour types (different knockouts, different tissues), common responses to aneuploidy can be extracted. Understanding this response is of vital importance before even starting to develop drugs that specifically kill aneuploid cells. These tumour cohorts can also be used to screen for additional mutations that occurred during tumour development. By extracting common mutations occurring in aneuploid tumours arising in different tissues and models, recurring pathways can be identified that aneuploid cells rely on to survive. With the recent advances in high throughput sequencing technology such endeavours are becoming more and more feasible.
Does this mean that we are still far away from the first therapies that target aneuploid cells?
Maybe not. A recent study has identified a few promising compounds that specifically target aneuploid cell lines making use of the deregulated metabolism of these cells (Tang et al., 2011). Such drugs can be tested in an in vivo setting using conditional mouse models for CIN.
Finally, we need a nifty model to visualize aneuploid cells and their response in the in vivo setting. One possible approach is to engineer a mouse model in which a single chromosome can be tracked, for instance a transgenic mouse strain expressing a fluorescent artificial transcription factor. By tracking aneuploidy as it arises in early tumour lesions or in aging tissues the direct consequences, but also the more long term effects of aneuploidy can be monitored in an in vivo setting. Such a model will provide unique insight into a developing tumour or otherwise affected t i s s u e , b u t , e v e n m o r e i m p o r t a n t , c a n also be used to visualize the clearance of aneuploid cell progeny in response to newly developed drugs.