Contribution of mutant mice to the study of in vivo cell cycle regulation during early development

Read Also

Plasticity and functional redundancy

For the last 15 years or so, gene targeting experiments have challenged the canonical view of cell cycle regulation, generating a tremendous amount of data. One of the most striking conclusions raised by these studies is that most key cell cycle regulators are largely dispensable during development (reviewed in [78,79,85,86]). To date, regarding Cyclin/Cdk complexes, early developmental failure was observed after disruption of only Cyclin A2 [87,88], Cyclin B1 [89] and Cdk1 [90]. Consistent with the fact that several members of a given type of cyclin or Cdk are present in the mouse genome, more severe phenotypes were observed in compound mutants. For example, while mice defective for a single member of the cyclin D family (D1, D2 or D3) are viable [91-93], combined invalidation of the three CyclinD genes leads to an embryonic lethality around E16.5 [94]. Similarly, double knock-out of Cyclin E1 and E2 leads to a lethality towards E16.5 whereas the simple knock-outs are viable [95,96]. Lastly, the double knock-out of Cdk4 and Cdk6 is embryonic lethal around E14.5-E16.5 whereas simple knock-outs are viable [97,98]. Strikingly, progression to late developmental stage of the aforementioned double and triple mutant embryos demonstrates that removal of all these ‘‘key’’ regulators of proliferation provides a surprisingly minimal barrier to cell proliferation in the early mouse embryo. It also suggests that the embryonic cell cycle has a high degree of plasticity more sophisticated than simple redundancy (figure 2). Hence, in some circumstances functional compensatory mechanisms may occur between proteins acting on different aspects of the cell cycle regulation. An evidence for such mechanism was recently provided by the analysis of embryonic fibroblasts deficient for Cdk2. Indeed, in the absence of CDK2, CDK1 normally regulating mitotic progression is able to bind Cyclin E and to drive cells into the S phase [99]. Other examples of functional compensation will certainly arise from future studies on genetically modified mice. In these studies, an important question will be to determine whether such compensatory mechanisms are essentially activated following the disequilibrium induced by gene disruption or whether they represent accessory pathways normally used in wild-type context.

Control of endoreplication

Another striking fact uncovered by gene knock-out studies is that many cell cycle regulators happen to have tissue-specific functions. Even in the case of a combined inactivation of several genes leading to an embryonic lethality, one finds defects restricted to some embryonic structures. Thus, the inactivation of cyclins D1, D2 and D3 induces problems in the formation of the heart at the origin of the lethality of embryos towards E16.5 [94]. Contrarily to the situation found in Drosophila [100] and C. elegans [101] where cyclin E is absolutely required for normal development, the E-type cyclins appear to be dispensable for the development of the embryo proper in the mouse. However, the disruption of cyclin E in mice brought some important informations on the control of endoreplicative cell cycle in vivo. Indeed, lethality of embryos deficients for both cyclin E1 and E2 is essentially due to abnormal placental development consecutive to failure of endoreplicative cycles of trophoblast giant cells (TGC) [95,96]. This shows that E-type cyclins are key players in the control of endoreplication. Regulation of cyclin E levels is therefore expected to be critical for TGC endoreplication. Interestingly, Skp2 deficient mice display elevated cyclin E levels and polyploidy in several tissues of postnatal animals [102]. Abnormally high levels of cyclin E and early embryonic lethality have been observed in mice deficient for other components of ubiquitin and ubiquitin-like modification pathways as for example Cul1 [103] or Cul3 [104], two members of the SCF complexes, Csn2 [105], a subunit of Cop9 signalosome and Uba3 [106], the catalytic subunit of NEDD8-activating enzyme. Importantly, in Cul3 or Uba3 deficient embryos, constitutive expression of cyclin E in trophoblast cells has been shown to correlate with a block in endoreplication. Gene disruption studies also provide evidences that other pathways regulate endoreplicative cycle progression in vivo. Indeed, inactivation of Mat1, coding for a subunit of the trimeric Cdk7-CylinH-Mat1 kinase, results in peri-implantation lethality [107]. Mat1 deficient TGC are rapidly arrested in the cell cycle progression, although they underwent several cycles of endoreplication. Finally, genetic ablation of Geminin, an inhibitor of pre-replication complex assembly, causes premature endoreplication and trophoblast cell differentiation of inner cells [108]. In wild-type blastocysts, Geminin’s down regulation in trophoblast cells correlates with active endoreplication. Altogether these observations suggest that Geminin is involved in suppression of endoreplication and trophoblast differentiation.

Previously uncharacterized cell cycle regulators

In some cases, gene knock-out mouse models may help to uncover novel cell cycle regulators. Recently, three genes, the function of which had not been previously ascribed to cell cycle regulation, have been shown to regulate cell cycle progression in vivo. Hence, the Cdc2P1 gene encodes two kinases originally identified as regulators of RNA transcription and processing that have been renamed CDK11 ten years ago because of their possible interaction with cyclin L. The first evidence that Cdc2P1 is indeed involved in cell cycle progression came from the observation that Cdc2P1 deficient embryos exhibit mitotic arrest followed by massive apoptosis at the blastocyst stage [109]. Further studies have shown that the CDK11 p58 small isoform, the synthesis of which occurs through an internal ribosome entry site which is specifically used during the G2/M transition, is critical for centrosome maturation, bipolar spindle formation and proper completion of cytokinesis [110,111]. The second gene is E4f, one of cellular target of E1A oncoprotein during adenoviral infection, which encodes a protein required for both transcriptional repression and activation of adenoviral genes. E4f deficient embryos die at the end of pre-implantation development and exhibit mitotic progression defects, chromosomal missegregation, and increased apoptosis [57], suggesting that, in vivo, E4F participates to the cell cycle control. Recently, it has been demonstrated that E4F directly regulates p53 [112]. It will be therefore interesting to monitor the contribution of p53 to the phenotype of E4f deficient embryos. The third gene is Ovum mutant candidate gene 1 (Omcg1), which codes for a nuclear zinc finger protein [56]. Omcg1 invalidation leads to an embryonic lethality by the end of pre-implantation development. This lethality is preceded by a dramatic reduction in the total cell number, a high mitotic index, and the presence of abnormal mitotic figures at the late blastocyst stage. Importantly, Omcg1 disruption results in the lengthening of M phase rather than in a mitotic block. This mitotic delay is associated with neither dysfunction of the spindle checkpoint nor abnormal global histone modifications. Further analyses will help to decipher the molecular mechanisms underlying the role of Omcg1 in mitotic progression.

Control of developmental transitions by cell cycle regulators

Highly dynamic modulations of the cell cycle parameters occur during embryonic development. It is clearly established that the various signalling pathways at works during embryonic development trigger variations in the cell cycle progression that are necessary for proper coordination of essential developmental processes such as proliferation, growth, patterning and differentiation. Conversely, it is reasonable to assume that cell autonomous genetic control of the cell cycle regulation may be a potent way to allow developmental transitions. However, only few examples of such mechanisms have been documented so far, most of which concern non-mammalian species. Hence, while checkpoints mainly act as gatekeepers of cell division integrity, their ability to regulate cell cycle progression has also been employed for developmental purposes. In drosophila, the maternal to zygotic transition (MZT), which occurs by the 13^th mitosis and is equivalent to the mouse ZGA, requires a functional DNA damage checkpoint. Indeed, removal of either Mei-41 (Atr ortholog), Grapes (Chk1 ortholog) or Wee1 maternal stores causes a developmental arrest at the 13^th division [113-115]. These mutant embryos fail to undergo syncitial to blastoderm transition and to initiate major zygotic gene activation. In wild-type embryos, a lengthening of the 11^th and 12^th division precedes the MZT. In defective embryos, the MZT does not take place and there is no change in the cell cycle duration before the 13^th division. Thus, it seems that maternally derived ATR, CHK1 and WEE1 collectively participate in the slowing down of cleavage speed, which in turn allow time for the initiation of the MZT and embryo cellularisation. In nematode, Atl-1 (Atr ortholog) and Chk1 are necessary for the asynchrony of division observed between AB and P1 blastomere during the second embryonic mitosis [116]. Thus, in this species also, the DNA damage/DNA replication checkpoint contributes to modulation of cell cycle duration during early development. Interestingly though, Atr [39], Chk1 [40,41] and Wee1 [117] disruption in mice lead to an early embryonic lethality. However, the critical requirement for these checkpoint genes at that period of development may be explained by the higher rate of errors that is likely to occur in rapidly dividing cells of the epiblast. To our knowledge, requirement of maternally derived ATR, CHK1 and WEE1 has not been monitored in mouse. Considering the unusual lengthening of G2 phase observed before ZGA during the second division of the mouse embryo and given the role of these checkpoint proteins in drosophila MZT, one might expect that mouse oocyte specific inactivation of these genes results in very early developmental failure.

Unsuspected links between cell cycle regulators and developmental programs have also been reported in the mouse. For example, inactivation of Xrcc2, a member of Rad51 family involved in DNA repair by homologous recombination, leads to a mid-gestation lethality between 12.5 and 18.5 dpc [118]. Surprisingly enough, several anomalies observed in mutant embryos, mostly defects in neurogenesis and somitogenesis, can be explained by a severe reduction of expression of Dll1 coding for one of the Notch receptor ligand. How the removal of a gene involved in DNA damage repair affects expression of a member of one of the key signalling pathway at works during development remains unclear. Another example has been provided by the study of hematopoiesis in Mad2 heterozygous mice [119]. Under cytokine stimultion, c-KIT physically associates with MAD2 and this interaction plays a role in regulating hematopoietic stem cells (HSC) self-renewal/differentiation balance. It has been proposed that local cytokine signalling modulates the duration of mitosis in HSC, allowing or not the correct positioning of the spindle and therefore asymmetric division [120]. Future investigations will determine whether an interaction between members of the SAC and signalling pathways is a common mechanism regulating asymmetric division.