Emerging mechanisms regulating mitotic synchrony during animal embryogenesis
Abstract
The basic mechanisms controlling mitosis are highly conserved in animals regardless of cell types and developmental stages. However, an exceptional aspect of mitosis is seen during early animal embryogenesis in which a large fertilized egg is quickly divided into smaller blastomeres according to the reproducible spatiotemporal pattern that does not rely on the cell‐cycle arrest or growth. This mitosis, referred to as cleavage, overlaps in the timeframe with the specification of cell fate. The precise spatiotemporal regulation of cleavages is therefore essential to the creation of the appropriate cell number and to the morphology of an embryo. To achieve the reproducibility of cleavage during embryogenesis, the relative timing of mitosis between cells, which we refer to as synchrony, must be properly regulated. Studies in model organisms have begun to reveal how the synchrony of mitosis is regulated by the developmental modulation of cell‐cycle machineries. In this review, we focus on three such mechanisms: biochemical switches that achieve the synchrony of mitosis, the nucleo‐cytoplasmic ratio that provokes the asynchrony of mitosis, and the transcriptional mechanisms coupled with cell fate control that reestablish the synchrony of mitosis in each fate‐restricted compartment. Our review is an attempt to understand the temporal patterns of cleavages in animal embryos created by the combinations of these three mechanisms.
Introduction
Animal embryogenesis is initiated by the spatially and temporally coordinated mitotic cell division named cleavage that continues rapidly without cell‐cycle arrest or cell growth. These cell cycles are composed of only two phases: DNA synthesis (S) phase and mitotic (M) phase. One remarkable feature of cleavage is its synchronies of nuclear division and cytokinesis (Fig. 1A). For example, rapid and highly synchronous mitotic cell divisions are commonly observed in early embryos of many animals (reviewed by Masui & Wang 1998). The synchrony of mitosis in the whole embryo roughly persists until the maternally supplied cell cycle regulators are depleted at the mid‐blastula transition (MBT). At the MBT, the transition from the maternal to zygotic control of the genetic programs as well as the slowing and asynchronization of cleavage occur (reviewed by Langley et al. 2014). The timings of the several pre‐MBT mitoses show a graded spatial variation within each round of mitosis, which is referred to as parasynchronous or metachronous mitosis (Fig. 1B; Satoh 1977; Newport & Kirschner 1982a; Boterenbrood et al. 1983; Foe & Alberts 1983; Keller et al. 2008; Olivier et al. 2010; Mendieta‐Serrano et al. 2013). At the MBT, S phase is elongated and gap 2 (G2) phase is introduced before M phase. The introduction of gap 1 (G1) phase occurs several cell cycles later than that of G2 phase (McKnight & Miller 1977; Frederick & Andrews 1994; Iwao et al. 2005). Due to these modifications of the cell cycle, post‐MBT mitoses exhibit higher asynchrony than the pre‐MBT mitosis, but still show synchrony or parasynchrony within each cell‐fate‐restricted compartment (Foe 1989; Kane et al. 1992; Fig. 1C). A similar trend to lose the synchrony of mitosis in a stepwise fashion can also be seen during the embryogenesis of C. elegans and ascidians, both of which finish synchronous mitosis at earlier cell cycle (Deppe et al. 1978; Nishida 1986). However, the mechanisms that regulate the mitotic synchrony during embryogenesis have not been fully clarified.
The critical parameter for the mitotic synchrony is the duration of the cell cycle, which is controlled by highly conserved machineries of the cell cycle including the modulation of DNA replication. First, post‐translational regulations of the core cell cycle regulator, cyclin‐dependent kinase 1 (Cdk1) behave as biochemical switches to allow the synchrony of the cleavages, as suggested by the studies in 1‐cell stage Xenopus embryos (Chang & Ferrell 2013) and Drosophila blastoderm embryos (Deneke et al. 2016). Second, the variation of nucleo‐cytoplasmic (N/C) ratio as a result of repeated unequal cleavage provokes asynchrony of mitosis through the N/C ratio‐dependent slowing of DNA replication, as suggested by the studies of the MBT in Xenopus embryos (Wang et al. 2000; Collart et al. 2013) and the cleavages of C. elegans embryos (Brauchle et al. 2003; Arata et al. 2015). Finally, cell fate cues provided by the combination of transcription factors can re‐establish the mitotic synchrony within each cell‐fate‐restricted compartment, as suggested by the studies in the post‐MBT embryos of Ciona (Ogura & Sasakura 2016a) and Drosophila (Momen‐Roknabadi et al. 2016). Our review article is an attempt to comprehend the transitions in the synchrony of cleavages in animal embryos by the combination of the three mechanisms.
Biochemical switches for mitotic synchrony
Simultaneous occurrence of cytokinesis across a cell at the first cleavage is probably a prerequisite for the synchrony of both nuclear division and cytokinesis in the subsequent cleavages. Therefore, we begin with the first cleavage in Xenopus that serves as a simple model of biochemical switches for mitotic synchrony.
Cytokinesis is kept nearly simultaneous in the Xenopus first cleavage
It is well established that M‐phase in a eukaryotic cell is initiated by the accumulation of B‐type cyclin that activates cyclin‐dependent kinase 1 (Cdk1) (Morgan 2007). In human cells, the active Cdk1 and Cyclin B initially appears on centrosomes in prophase, and then it spreads toward the whole cytoplasm (De Souza et al. 2000; Jackman et al. 2003). Since the size of human cells is in the order of 10‐μm radius, the simultaneous onset of M‐phase across the entire cell could be explained by the diffusion of active Cdk1 from the centrosome; however, 1‐cell stage Xenopus embryo is too large (600‐μm radius) for the simultaneous onset of M‐phase to be accounted for solely by the diffusion of activated Cdk1 (Chang & Ferrell 2013).
The key phenomenon to fill this discrepancy is the surface contraction waves (SCWs) that rapidly propagate from the animal to the vegetal pole ahead of cleavage furrow at the cell cortex of 1‐cell stage amphibian embryos (Hara 1971; Hara et al. 1980). In Xenopus, these waves represent the activation and inactivation waves of Cdk1‐CyclinB complex, which are initiated at the localized source around the animal pole and propagate toward the vegetal pole (Rankin & Kirschner 1997; Pérez‐Mongiovi et al. 1998). In the model proposed by the experiments in Xenopus egg extracts, Cdk1‐CyclinB complex is held inactive through the inhibitory phosphorylation by Wee1 kinase, which is relieved by Cdc25 phosphatase at M‐phase entry; once Cdk1‐CyclinB is activated, it then activates Cdc25 phosphatase and inactivates Wee1 through their phosphorylation (Morgan 2007). Due to the interlinked feedback loops for the activation of Cdk1 (Fig. 3A), the dependency of Cdk1 activity to the concentration of its activator CyclinB shows either on or off response with no intermediate state (Fig. 3E). Such response of Cdk1, referred to as ‘bistable’ threshold response (Pomerening et al. 2003), generates discontinuous transitions between on/off states, just like a toggle switch (Fig. 3F). Bistability is closely related to the ‘excitability’ by which a neuronal cell rapidly transmits an electrical signal down the axon as a trigger wave (Gelens et al. 2013). The kinematic analysis of the first SCW (SCWa) was consistent with the model in which the activation of Cdk1‐CyclinB complex propagates as a trigger wave from the vicinity of the centrosome (Chang & Ferrell 2013; Fig. 2A′). Thus, the excitability shared by the Cdk1 activation switch likely allows the rapid propagation of SCWa.
The SCWa is closely followed by the second SCW (SCWb; Fig. 2A″). In SCWb, the Cdk1‐CyclinB complex inactivates itself by activating the ubiquitin‐protein ligase anaphase‐promoting complex (APC) to degrade CyclinB (Fig. 3B). The positive feedback loop coupled with the delayed negative feedback generates the ultrasensitive threshold response of CyclinB degradation rate to the activity of Cdk1 (Yang & Ferrell 2013). The relationship between the stimulus (Cdk1 activity) and the response (the CyclinB degradation rate) shows a sigmoidal curve, which is characteristic of an ultrasensitive threshold response (Fig. 3E). Such a response generates continuous transitions between the high and low activities, just like a rheostat switch (Fig. 3F). The increase in the Cdk1 activity in SCWa would cause the ultrasensitive activation of APC and would lead to the rapid inactivation of Cdk1 during SCWb (Chang & Ferrell 2013). Thus, SCWs are the manifestation of the rapid activation and inactivation of Cdk1‐CyclinB.
Then, how the SCWs contribute to the simultaneous occurrence of cytokinesis itself? During cytokinesis, F‐actin assembly at the equatorial membrane is activated by Rho GTPase, which is activated by the recruitment of a Rho guanine nucleotide exchange factor Ect2 to the equatorial membrane (reviewed in Mishima 2016). In the current model, Ect2 is cytoplasmic before anaphase because the membrane recruitment is inhibited by Cdk1‐CyclinB; during anaphase, the APC‐mediated inactivation of Cdk1‐CyclinB complex causes the recruitment of Ect2 to the equatorial membrane (Su et al. 2011). In embryos of Xenopus and starfish, Rho signaling activates F‐actin assembly and is antagonized by the F‐actin assembly in a negative feedback (Bement et al. 2015; Fig. 3C). In principle, the feedback loops between Rho and F‐action act as a bistable switch that can produce the excitability of Rho activity that allows rapid F‐action assembly and cytokinesis (Bement et al. 2015). In summary, the above studies suggest that the bistable or ultrasensitive properties of the biochemical switches are important to achieve the simultaneous cytokinesis even in the cell with large cytoplasm. For the in‐depth explanation of the bistable or ultrasensitive threshold response, we guide the readers to Ferrell and Ha 2014.
S phase synchronization in the Drosophila syncytial cell cycles
In certain types of cells, nuclear division cycles proceed without cytokinesis, resulting in a multinucleate cell called a ‘syncytium.’ A well‐studied example is the blastoderm of Drosophila melanogaster (Fig. 2B). Thirteen cycles of nuclear divisions occur in the cytoplasm of syncytium Drosophila embryo, which is as large as 500 and 150 μm in their major and minor axes, respectively. The cell cycle 13 is the MBT in Drosophila and is accompanied by the specialized cytokinesis called ‘cellularization’ that subdivides 5000 nuclei into multiple cells at the same time (Foe & Alberts 1983). The nuclear division cycles exhibit minutes‐long parasynchronous waves that sweep from both the anterior and posterior poles toward the middle of the embryo (Fig. 3B), the speed of which gradually slows from the mitosis 10–13 (Foe & Alberts 1983; Idema et al. 2013). Deneke et al. (2016) visualized the propagating wave of Cdk1 activity during the nuclear division cycles in live embryos (Fig. 2B′,B″). They found that, in contrast to the SCWs in Xenopus, a slow rise of Cdk1 activity at the end of S phase but not in M phase was crucial for the synchronization of nuclear division in the Drosophila blastoderm.
For the regulation of Cdk1 activity during S phase, Chk1, the main effector of the conserved DNA replication checkpoint is important. Chk1 is activated by a sensor kinase for un‐replicated DNA named ATR (Morgan 2007). In the pre‐MBT cell cycle, the ATR/Chk1 pathway does not halt cell division, but it does slow down cell‐cycle progression by prolonging the S phase (Sibon et al. 1997, 1999). Chk1 inactivates Cdk1 through both the activation of Wee1 and the inactivation of Cdc25 (Morgan 2007; Fig. 3D). The requirement of Chk1 for the slowdown of Cdk1 activation rate during S‐phase was revealed by the fluorescence resonant energy transfer (FRET) biosensor for Cdk1‐CyclinB activity; Chk1‐dependent slowdown of Cdk1 activation rate explains the slowing of the mitotic wave during mitosis 10–13 (Deneke et al. 2016). Thus, the additional negative regulations to the Cdk1 bistable switch provided by Chk1 is responsible for the parasynchronous mitoses of the Drosophila blastoderm. In their model, as an embryo approaches the MBT, the Chk1 activity becomes higher due to the increase of the nucleo‐cytoplasmic ratio (the details are discussed in the next chapter). The propagation of Cdk1 activity is then slowed down (Fig. 2B″) because a longer time is required to overcome the inhibition of Cdk1 by Chk1. In both 1‐cell stage Xenopus embryo and Drosophila blastoderm, the biochemical switches depending on Cdk1 regulatory systems are the basis for rapidly propagating and synchronous onset of mitosis in a single cytoplasm.
The nucleo‐cytoplasmic (N/C) ratio and the asynchrony of mitosis
Titration of replication factors at the MBT
Since early embryos undergo cell divisions without the increase of cell volume (cell growth), the N/C ratio (used in the sense of DNA‐to‐cytoplasmic ratio in this review) becomes greatly increased as a consequence of the repeated cell divisions. In Xenopus, the first 12 mitoses are rapid and generally synchronous, and the 13th mitosis is slower and asynchronous accompanied by a parasynchronous mitotic wave (Satoh 1977; Newport & Kirschner 1982a; Boterenbrood et al. 1983). These changes in cell behaviors are delayed by one cell cycle in haploid embryos, whereas they occur one cell cycle earlier in tetraploid embryos (Newport & Kirschner 1982a). Therefore, a progressive increase in the N/C ratio has been hypothesized to titrate some cytoplasmic material that can bind DNA in a concentration dependent manner: when the N/C value surpassed the threshold value, it triggers the switch from the rapid and synchronous to the slow and asynchronous mitosis (Newport & Kirschner 1982b; Dasso & Newport 1990). The elongation of the cell cycle is required for the onset of transcription (Kimelman et al. 1987). A similar titration model has been proposed for zebrafish (Kane & Kimmel 1993). However, it is only recent years that the titration model was directly examined in Xenopus through an investigation of the role of DNA replication factors in the MBT (Collart et al. 2013).
Replication factors are categorized into two types: the one required for the pre‐replicative complex (pre‐RC) assembly (licensing) and the other required for the initiation of DNA replication (origin firing). Collart et al. 2013 examined the abundance of the replication factors before and after the MBT. Four initiation factors, Cut5, RecQ4, Treslin, and Drf1, are greatly reduced at the MBT in both protein and mRNA level. When the four factors were overexpressed, the rapid and synchronous mitosis continued beyond the 12th mitosis, suggesting the delay of the MBT. The phenotype was rescued by the partial depletion of the pre‐RC component Cdc6, suggesting that the effect of the overexpressed factors is mediated by the increased rate of DNA replication (Collart et al. 2013). Thus, the four initiation factors are qualified as what were postulated as limiting cytoplasmic materials in the titration model. The increase of N/C ratio can cause dose‐dependent activation of the conserved Chk1/ATR pathway (Shimuta et al. 2002; Conn et al. 2004). The slowing of DNA replication by the titration of replication factors is partly responsible for the activation of the Chk1/ATR pathway (Collart et al. 2013). Thus, titration of replication factors can account for the change of cell cycle duration at MBT. This mechanism, when it is coupled with the variability of N/C ratio, can be a trigger of mitotic asynchrony, as we explain below.
Inverse correlation between the cell size and the cell‐cycle duration
In isolated animal blastomeres of early Xenopus embryos, Wang et al. 2000 revealed the inverse correlation between the cell‐cycle duration and the cell radius (as a surrogate for the N/C ratio on the assumption of a constant amount of DNA). The inverse correlation was significant when the cell size surpassed the threshold cell radius at the MBT (Fig. 4A). Since early cleavages in Xenopus embryos generate considerable inequality of cell size between the daughter blastomeres even within the animal blastomeres, this correlation explains the onset of asynchronous mitosis at the MBT through the increase in the dependency of the cell‐cycle duration on the N/C ratio. A subsequent study revealed that the cell‐to‐cell variation of cell‐cycle duration at the MBT primarily attributable to the variation of S phase length (Iwao et al. 2005). It is likely that the variation of S phase length is originated from the reduced origin firing or the activation of the Chk1/ATR pathway.
Similar inverse correlation between cell volume and cell cycle duration is found in intact embryos of zebrafish (Kane & Kimmel 1993) and C. elegans (Arata et al. 2015). In early embryo of C. elegans, the cleavages exhibit highly invariant stem cell‐like divisions that segregate founder lineages (AB, EMS, C, and D lineages) from the germline lineage (P1–P4) at each asymmetric division (Fig. 4B). In these divisions, the stem cell lineage (germline) at the posterior end is always smaller than the founder lineages, thereby generating the progressively smaller blastomeres in the posterior side (Deppe et al. 1978). As it would be expected from shorter cell‐cycle duration in larger cells, anterior blastomeres divide earlier than the posterior ones, generating an apparent wave of mitosis from the anterior to the posterior of the embryo (Deppe et al. 1978). Arata et al. 2015 quantified the dependency of cell‐cycle duration on the cell volume in each lineage, and they revealed that the strength of the dependency differed among the lineages (Fig. 4C). The P and C lineages showed the strongest dependency on cell volume while the AB and MS lineages show weaker dependency and the D and E lineages show no apparent dependency. Since the cell‐cycle duration in the cleavage stage embryos of C. elegans is controlled by the length of the S phase (Edgar & McGhee 1988), there should be mechanisms that link the duration of the S phase and cell volume for achieving size‐correlated cell‐cycle durations. Involvements of other factors such as the unequal cell division in the P and C lineages (Arata et al. 2015) and the introduction of G2 phase in the E lineage (Edgar & McGhee 1988) remains to be elucidated. Moreover, C. elegans embryos exhibit the dependency of cell‐cycle duration on the N/C ratio from the beginning of embryogenesis in stark contrast to Xenopus embryos. This may correspond to the absence of a clear MBT in this animal (Schauer & Wood 1990). Alternatively, the smaller cell size (versus nuclear size) in C. elegans blastomeres compared to Xenopus blastomeres may explain the earlier onset of cell size‐dependent cycle duration.
Cell fate cooperates with the cell size for the mitotic asynchrony between cell lineages
The two‐cell stage blastomeres of C. elegans have unequal volume (AB > P1). As expected from the smaller volume, P1 cells divide 2 min later than AB cells. In fact, 40% of the asynchronous time between the AB and P1 blastomeres is attributable to the preferential activation of ATR/Chk1 pathway in the P1 blastomere, which is thought to result from the unequal volume of AB and P1 (Encalada et al. 2000; Brauchle et al. 2003; Fig. 4D). However, a microsurgical experiment suggested that the cell‐cycle asynchrony in the two‐cell stage C. elegans embryo cannot be explained by the unequal volume alone (Schierenberg & Wood 1985). The remaining 60% of the asynchrony is attributable to several mechanisms acting downstream of the PAR‐3/PAR‐4 proteins that play key roles in the specification of cell fates at the two‐cell stage (Fig. 4D). First, PDZ‐domain protein PAR‐3 localized in AB blastomere causes the enrichment of Polo‐like kinase PLK‐1, which positively regulates Cdk1‐CyclinB activity to cause an earlier mitotic entry in the AB blastomere (Budirahardja & Gönczy 2008; Rivers et al. 2008). Secondly, regulation of DNA replication factors by the serine‐threonine kinases PAR‐4 and PAR‐1 work together to decrease the origin firing in P1 blastomere, thereby delaying the progression of S phase (Benkemoun et al. 2014). The effect of PAR‐4/PAR‐1 is mediated by a specific isoform of Cyclin B that has a non‐canonical role in promoting S‐phase progression (Michael 2016). These mechanisms provide lineage‐specific regulations of cell‐cycle duration independently of the cell size. Thus, the first cleavage of C. elegans embryo illustrates the cooperation of the cell size and cell fate cues in the asynchronization of mitosis between different cell lineages.
Synchronous mitosis despite the remarkable inequality of cell volume in ascidian embryos
Unequal cleavage accompanying the segregation of essential maternal mRNAs
Early embryos of the chordate ascidians are another well‐studied models of asymmetric and unequal cell divisions (reviewed in Kumano & Nishida 2007). The cleavage patterns of the ascidian embryo show significant inequality along the animal–vegetal (A‐V) axis, which is perpendicular to the future anterior‐posterior (A‐P) axis. The ascidian egg is preloaded with the maternal mRNAs, some of which are localized to the region named the posterior vegetal cytoplasm (PVC) (Fig. 5A). The proteins encoded by these postplasmic/PEM RNAs are crucial for patterning along the A‐P axis (Nishida 1994; Yoshida et al. 1996; Kumano & Nishida 2009), cell differentiation (Sasakura et al. 1998; Nishida & Sawada 2001; Nakamura et al. 2005; Prodon et al. 2007), germ cell specification (Shirae‐Kurabayashi et al. 2006), spindle attraction for the asymmetric cell divisions (Negishi et al. 2007), the localization of mRNAs (Nakamura et al. 2005), and morphogenesis (Nishida 1996). The centrosome attracting body (CAB) is the structure that attracts the nuclei to the posterior pole by a microtubule‐dependent mechanism (Hibino et al. 1998; Nishikata et al. 1999). By the eight‐cell stage, postplasmic/PEM RNAs are concentrated into a very restricted region surrounding the CAB (Fig. 5A), and this process is dependent of cytoskeletal elements such as the microtubule and F‐actin (Sasakura et al. 2000). The subsequent unequal divisions are directed by the CAB to concentrate the postplasmic/PEM RNAs into the posterior‐most B7.6 blastomeres fated to germline precursors (Nishikata et al. 1999). According to the quantitative measurement of blastomere volumes by a computational image analysis (Tassy et al. 2006), the successive unequal cleavages result in the increasing inequality of cell volume across the embryo (Fig. 5B). This is particularly evident between the germline precursors and their sister blastomeres, in which the volume inequality calculated according to the formula in Figure 5C is 17% at the eight‐cell stage, 67% at the 16‐cell stage and 71% at the 32‐cell stage (Fig. 5C). Thus, remarkably unequal cleavages accompany the segregation of certain maternal mRNAs in ascidian embryos.
Cell fate control of mitotic asynchrony at the MBT
In ascidians, earliest zygotic gene expression is detected at the eight‐cell stage in the animal blastomeres (Miya & Nishida 2003; Rothbächer et al. 2007), which is followed by the wider zygotic transcription at the 16‐cell stage (Imai et al. 2004). The next mitosis from the 16‐cell stage (5th mitosis) occurs about 5 min earlier in the eight vegetal blastomeres, creating a brief 24‐cell stage (Dumollard et al. 2013). The asynchronous timing of the 5th mitosis coincides with the endomesoderm specification by the nuclear accumulation of maternal β‐catenin (Imai et al. 2000). From these observations, Dumollard et al. 2013 proposed that the 16‐cell stage is the MBT in ascidians. They also showed that the asynchrony between the animal and vegetal blastomeres was attributable to the activation of β‐catenin that facilitates the S‐phase progression in the vegetal blastomeres. Indeed, knockdown of β‐catenin retards the mitosis of the vegetal blastomeres, whereas the stabilization of β‐catenin in the animal blastomeres causes their precocious mitosis (Dumollard et al. 2013). Importantly, both perturbations resulted in the synchronization of mitosis in the animal and vegetal blastomeres, thereby eliminating the 24‐cell stage. Thus, the cell fate cue provided by β‐catenin but not the difference of cell size provokes the asynchrony of mitosis at the MBT in ascidian. Moreover, the role of N/C ratio in the control of mitosis in later embryogenesis was investigated by creating various types of egg fragments by cutting the unfertilized egg or removing the cytoplasm/nucleus from the unfertilized eggs (Reverberi & Ortolani 1962; Ishida & Satoh 1998; Yamada & Nishida 1999). Surprisingly, these partial embryos develop into normal larvae and the timings of mitoses at least up to the 110‐cell stage were not affected. These observations suggest a minor role played by the N/C ratio in the control of mitotic timing in ascidian embryos, making ascidian embryos as straightforward models to investigate the role of cell fate cues in the control of mitotic synchrony.
Cell‐fate coupled transcriptional mechanisms of mitotic synchrony
Mitotic domain: the cell lineage‐based compartment of mitotic timing
The cleavage patterns in the early embryos of C. elegans and ascidians suggest that the regulation of the cell‐cycle duration by the N/C ratio can be modulated or overwhelmed by the cell fate‐coupled cell‐cycle regulations. In an early C. elegans embryo, each cell lineage has a unique mitotic timing that is synchronous within each lineage. It was therefore proposed that each lineage has its own unique clock (Deppe et al. 1978). Similarly, it was reported that the mitotic timing in the ascidian gastrula obeys cell‐lineage specific patterns (Nishida 1986). After the studies in Drosophila and zebrafish, such a compartment of the mitotic timing coupled with the cell lineage was termed as ‘mitotic domain’ (Foe 1989; Kane et al. 1992). How the cell fate cues can provide information about the synchrony of mitosis in each mitotic domain is the topic of the remainder of this review.
The transcriptional regulation by a combination of transcription factors is a widespread strategy in cell fate decisions (Spitz & Furlong 2012). Similarly, a mechanism for achieving diverse mitotic timings from a limited number of transcription factors is the combinations of these factors in the regulation of genes encoding key cell‐cycle regulators. In ascidians, the fates of most blastomeres are restricted to single tissues at the 110‐cell stage (Nishida 1987). The choice of cell fate is conducted by a specific set of transcription factors for each lineage (Imai et al. 2004). For example, Brachyury (Bra) is a key transcription factor responsible for the specification of the notochord (Yasuo & Satoh 1998). Bra regulates the number of cell divisions in the notochord lineage through the upregulation of Cdk inhibitor; however, Bra does not significantly affect the cell‐cycle duration (Fujikawa et al. 2011). Another transcription factors are thus likely to act together with Bra for the cell‐cycle duration that is specific to the notochord lineage. As seen in fate decisions of ascidian blastomeres (Imai et al. 2004), key cell‐cycle regulators in ascidian embryos would also be regulated by the combinations of transcription factors to achieve the lineage‐specific cell cycle duration. Indeed, our recent study revealed that the cell‐cycle duration of epidermal cells is regulated by the combination of the transcription factors GATA and AP‐2 (Ogura & Sasakura 2016a), both of which are essential for the specification of epidermis in ascidians (Rothbächer et al. 2007; Sasakura et al. 2016; Imai et al. 2017).
Mitotic domains in the epidermis of ascidians during neurulation
Epidermal cells occupy most of the animal hemisphere of ascidian cleavage‐stage embryos (Fig. 5B). Seven rounds of cleavage generate 50 fate‐restricted epidermal precursors at the 110‐cell stage (Nishida 1987). Following the 110‐cell stage, three rounds of epidermal mitosis occur during the gastrula stage. The intervals between the 7th and 8th, 8th and 9th, and 9th and 10th mitoses are less than 1 h. The 11th mitosis then occurs during the neurula stage after an approximately 1.5‐h interval from the 10th mitosis (Ogura et al. 2011). The 11 rounds of mitosis in total give rise to about 800 epidermal cells after the neurula stage (Nishida 1987; Pasini et al. 2006).
The 11th mitosis is remarkably different from the previous mitoses with respect to the duration and synchrony of mitosis (Nishida 1986; Ogura et al. 2011). First, as mentioned above, the interphase duration of the 11th cell cycle is approx. 1.5‐fold longer than those of the 8th–10th cell cycles (Ogura et al. 2011). The elongation of the 11th interphase is caused by the down‐regulation of cdc25 at the neurula stage (Ogura et al. 2011). Secondly, the 11th epidermal mitosis exhibits parasynchronous mitotic wave, whereas the 9th and 10th mitoses retain remarkable synchrony (Ogura et al. 2011; Ogura & Sasakura 2016a). At the 9th and 10th mitoses, slight asynchrony for approximately 10 min is seen between the anterior and posterior epidermal regions (Ogura et al. 2011), and a similar extent of asynchrony is already recognizable at the 76‐ to 110‐cell stage (Nishida 1986).
At the 11th mitosis, the epidermis is subdivided into four mitotic domains, MD1 (ventral epidermis), MD2, MD3b and MD3a (Fig. 6A) with different mitotic timings. The names MD1–3 correspond to the timing of the initiation of mitosis; the ventral mitotic domain MD1 starts mitosis earliest. The mitosis of each mitotic domain is generally initiated from the posterior side (Fig. 6B). Within each mitotic domain, epidermal cells divide parasynchronously with an apparent spatiotemporal pattern; the mitotic wave spreads from the posterior to the anterior side in each mitotic domain. A more complex pattern is observed in the posterior lateral mitotic domain MD3b, in which the mitotic waves are initiated bidirectionally from the anterior and posterior sides. The parasynchronous mitotic timing in the four mitotic domains forms the mitotic wave that takes over 60 min to propagate from the posterior end to the anterior end of the epidermis (Ogura et al. 2011). The switching from the rapid and synchronous to the slow and parasynchronous mitosis in the ascidian epidermis is reminiscent of the changes at the MBT in Xenopus, but the underlying mechanism is based on a cell fate‐coupled transcriptional mechanism, as we describe in the following section (Ogura & Sasakura 2016a).
Role of the graded transcription of cdc25 for mitotic synchrony
A simple way to investigate the synchrony of the cell cycle is to measure the duration of each cell‐cycle phase using live imaging. By this approach, we recently found that the length of the S phase but not that of the G2 phase is responsible for the parasynchronous pattern of the 11th epidermal mitosis (Ogura & Sasakura 2016a). The S‐phase length was progressively longer on the anterior side of the epidermis. The mechanism creating the asymmetry of the S‐phase length along the A‐P axis remains to be elucidated. The most significant point of that study is that progressively longer S‐phase lengths are seen in the previous 10th cell cycle, which is generally synchronous. At first glance, the parasynchronous S phase in the 10th cell cycle contradicts the synchronous mitosis. This apparent contradiction is resolved by the asynchronous G2 phase; the length of the 10th G2 phase is progressively shorter on the anterior side of the epidermis compared to the posterior side. Therefore, the S and G2 phases in the 10th cell cycle exhibit patterns that are complementary to each other (Fig. 6C). Because the S phase is already patterned to be asynchronous at the 10th epidermal cell cycle but is compensated for by the asynchronous G2 phase, the transition to parasynchronous 11th mitosis can be achieved simply by losing the asymmetric pattern of the G2 phase and this reflects the parasynchrony of S phase on the timing of mitosis.
The asymmetry of the G2‐phase length along the A‐P axis coincides with the expression pattern of the G2/M regulator cdc25 that is progressively stronger on the anterior side of the epidermis (Fig. 6D). Analyses of the cis‐regulatory elements of cdc25 for epidermal expression revealed the combinatorial regulation by the key patterning genes (GATA and AP‐2) of the epidermis (Rothbächer et al. 2007; Ogura & Sasakura 2016a; Sasakura et al. 2016; Imai et al. 2017). The gene expressions of these transcriptional activators were also stronger on the anterior side of the epidermis, suggesting that GATA and AP‐2 activate cdc25 transcription in a concentration‐dependent manner. As mentioned above, cdc25 is downregulated at the onset of neurulation, which also causes the loss of the asymmetry of the length of the G2 phase along the A‐P axis. The downregulation of cdc25 may be attributable to the reduction of GATA and AP‐2 that precedes the onset of neurulation. Thus, the synchronous to asynchronous switch in the ascidian epidermis is achieved by the elaborate transcriptional regulation of cdc25 that compensates for the asynchrony of the S phase by the regulation of the G2 phase. A similar compensatory cell‐cycle regulation termed ‘cell‐cycle compensation’ was reported in Drosophila (Reis & Edgar 2004), but the mechanism in ascidian epidermis is unique in that a developmentally‐relevant mechanism is responsible for the cell‐cycle compensation, thereby allowing tissue‐specific regulation. For the historical background of cell‐cycle compensation, please refer to our reviews: Ogura & Sasakura (2016b,c).
The role of the cdc25 transcriptional switch in mitotic synchrony
Another case of the cell‐fate coupled synchronization of mitosis was reported in the gastrula embryos of Drosophila (Momen‐Roknabadi et al. 2016). In Drosophila, MBT occurs during cycle 14 (Yuan et al. 2016). The mitosis 14, which is subdivided into 25 mitotic domains (Foe 1989), begins in the mitotic domain named MD1 and is immediately followed by the mitosis in MD2. Both MD1 and MD2 are located in the head region of an embryo. MD1 and MD2 show reproducible parasynchronous patterns that persist for about 10 min (Momen‐Roknabadi et al. 2016; Fig. 6E,F).
In Drosophila embryos, a crucial role of cdc25 transcription for the timing of mitosis in each mitotic domain was reported (Edgar et al. 1994; Lehman et al. 1999). The correlation between the timings of the transcriptional activation of cdc25 and mitotic entry was also demonstrated by live imaging (Di Talia & Wieschaus 2012). The zygotic transcription of cdc25 is the limiting factor for the timing of Drosophila post‐MBT mitoses, due to the degradation of maternal cdc25 mRNA and protein at the MBT (Edgar & O'Farrell 1990; Sibon et al. 1997, 1999; Ferree et al. 2016). This circumstance makes it difficult to analyze the effect of transcription factors on the timing of mitosis because homozygous mutants of the activators of cdc25 transcription completely halt mitosis at the G2 phase of cycle 14. To circumvent this problem, Momen‐Roknabadi et al. (2016) conducted a genetic screen using heterozygous mutant embryos, and they identified the dosage‐sensitive regulators of MD2. To use fixed embryos to evaluate the effects of mutation on the timing of mitosis, they plotted the number of mitotic cells in each mitotic domain against the length of reference organ that continues to elongate during the period (Fig. 6G). The changes in the intercept and slope of the graph were respectively interpreted as the measures of the relative timing and the synchrony of mitosis. Through the screening for mutations that alter the intercept, they identified three candidate transcriptional activators and repressors of cdc25. However, to alter the slope, at least two heterozygous mutations of activators or repressors are required. This suggests that the synchrony of mitosis is less sensitive to the dosage of the transcriptional regulators than the relative timing of mitosis. To further address the mechanism of mitotic synchrony, it is important to measure the precise spatiotemporal transcriptional profile of cdc25 and to analyze how the profile is quantitatively achieved by the combination of transcription factors.
Conclusions
We have reviewed three regulatory mechanisms of mitotic synchrony that are seen during the early embryogenesis of animals; namely, the (i) biochemical switches, (ii) the N/C ratio and (iii) cell‐fate coupled transcriptional mechanisms. These mechanisms target different cell‐cycle phases with elaborate systems. The biochemical switch centered on Cdk1 targets M phase with the bistable or ultrasensitive threshold responses to achieve the synchrony of mitosis. The increase in the N/C ratio mainly retards the cell cycle at S phase through the titration of replication factors that causes the reduction of origin firing and the activation of DNA replication checkpoint. The variability of N/C ratio among cells is a source of mitotic asynchrony (Fig. 7A). The asynchrony often takes the form of parasynchrony that can arise either from the three mechanisms. The spatial cues that provide parasynchronous mitosis with the temporal patterns are not fully clarified. Once the parasynchrony is introduced to the cell cycle progression, cell fate cues play dual roles in the asynchronization of mitosis between cell lineages and the synchronization of mitosis within mitotic domains (Fig. 7B). Thus, the combinations of the three mechanisms reviewed here, perhaps partly explain the stepwise reduction of mitotic synchrony observed in many animal embryos. Future studies may address how these three mechanisms are combined to achieve the appropriate cell numbers and the morphologies of various animal embryos.
Acknowledgments
We thank the members of the Shimoda Marine Research Center at the University of Tsukuba for their kind cooperation during our study. This study was supported by Grants‐in‐Aid for Scientific Research from JSPS to Y.O. and Y.S. (no. 11J00127 and no. 16H04815). We apologize to the researchers whose original works could not be cited due to the limitation of space.
