Regulators specifying cell fate activate cell cycle regulator genes to determine cell numbers in ascidian larval tissues

In animal development, cell cycle control is important for proper differentiation of tissues and organs. In many tissues, terminal differentiation is coupled with withdrawal from the cell cycle (G0 arrest). If cell cycle withdrawal is interrupted, it may affect normal development and could lead to cancer or other diseases (Buttitta and Edgar, 2007; Sun and Buttitta, 2017). For example, a myogenic basic helix-loop-helix transcription factor, Myog, induces cell cycle exit in differentiated myocytes in mammalian embryos (Liu et al., 2012). In nematode embryos, mutations in genes encoding cyclin-dependent kinase inhibitors induce hyperplasia in multiple somatic tissues, suggesting that these proteins normally stop cells from dividing (Fukuyama et al., 2003). Similarly, in Drosophila embryos, dacapo, encoding a cyclin-dependent kinase inhibitor, is necessary to stop epidermal cell proliferation (de Nooij et al., 1996; Lane et al., 1996). The retinoblastoma protein (Rb) is another important regulator of cell cycle withdrawal (Wikenheiser-Brokamp, 2006). In mice, deficiency of Rb in epidermis leads to hyperplasia (Ruiz et al., 2004). It is also important to maintain cell cycle progression for different cell populations. In mice, Mycn is required to maintain proliferating neural cell progenitors, and a mutation of this gene results in precocious differentiation of neural cells (Knoepfler et al., 2002).

In ascidian embryos, cell lineages are invariant among individuals, and numbers of cell divisions are strictly controlled (Conklin, 1905; Nishida, 1987). Notochord cells become post-mitotic after nine rounds of cell division, and every larva contains exactly 40 notochord cells. Similarly, in Ciona, 36 muscle cells are differentiated in the larval tail, and the anterior 28 cells are derived from the posterior vegetal quadrant of each embryo. Among these cells, 16 cells become post-mitotic after nine rounds of cell division, and 12 cells become postmitotic after eight rounds of cell division. Epidermal cells stop dividing after 11 rounds of cell division (Ogura et al., 2011; Pasini et al., 2006; Yamada and Nishida, 1999), although these cells resume cycling around the time when a larva develops competence for metamorphosis (Nakayama et al., 2005). By contrast, mesenchymal cells and endodermal cells, which produce adult tissues after metamorphosis, may continue to divide after hatching (Nakayama et al., 2005; Yamada and Nishida, 1999). In this way, these tissues in ascidian larvae regulate the cell cycle differently, which suggests that developmental programmes for these tissues are involved in determining when cells stop dividing.

In the notochord lineage of embryos of Halocynthia roretzi, an ascidian belonging to a different order than that of Ciona, the number of cell divisions is controlled by a mechanism that is under the control of Brachyury (T) (Fujikawa et al., 2011), a key transcription factor for notochord differentiation (Chiba et al., 2009; Satou et al., 2001a; Yasuo and Satoh, 1994). Similarly, the number of cell divisions in muscle cells is controlled by Tbx6-r (Kuwajima et al., 2014), a key transcription factor for muscle differentiation (Yagi et al., 2005; Yu et al., 2019). Because Cdkn1.b (also known as CKI-b), a gene encoding a Cdk inhibitor, is expressed in notochord and muscle cells, and because its expression in the notochord is under control of Brachyury, Cdkn1.b may be involved in regulation of cell cycle in these tissues (Kuwajima et al., 2014). Indeed, injection of Cdkn1.b mRNA into fertilized eggs halts the cell cycle around the 8- or 16-cell stage (Kuwajima et al., 2014), although its function in differentiation of notochord and muscle has not been evaluated.

By contrast, mesenchymal and endodermal cells likely continue to divide even after hatching (Nakayama et al., 2005), and produce a large number of small cells; however, it is not known how these tissues maintain cell cycle progression. It is also uncertain whether Twist-r.a/b and Lhx3/4, which encode key transcription factor genes for fate specification of mesenchyme and endoderm (Imai et al., 2003; Satou et al., 2001b), control cell numbers in these tissues. In the present study, we addressed these issues and found that Cdkn1.b and Myc serve essential functions in cell cycle control in ascidian embryos.

Notochord cells stop dividing after nine rounds of cell division, and epidermal cells stop after 11 rounds of cell division (Nishida, 1987; Ogura et al., 2011; Pasini et al., 2006; Yamada and Nishida, 1999). The primary lineage of muscle cells (the anterior 28 muscle cells) stops dividing after eight or nine rounds of cell division depending on their lineages (Nishida, 1987). To examine whether zygotic gene expression is involved in regulation of cell cycle counts, we isolated cells of presumptive notochord and muscle at the 64-cell stage, and presumptive epidermal cells at the 32-cell stage using a fine glass needle, because developmental fate of these lineages is restricted to notochord, muscle and epidermis at these stages (Nishida, 1987; Nishida and Satoh, 1985). Because development of Ciona savignyi embryos is stable, especially in case of isolation of blastomeres, we used C. savignyi embryos in the present study unless otherwise noted.

First, we isolated one presumptive notochord cell (A7.3/A7.7) from each of 11 normal 64-cell embryos. When isolated cells were incubated in normal sea water until unperturbed control embryos hatched, they divided three times to become eight cells in most cases (Fig. 1A). Because each of these cells produces eight notochord cells in normal embryos, this observation indicated that the number of cell cycle rounds in these isolated cells was regulated tissue-autonomously, as previously indicated in Halocynthia embryos (Fujikawa et al., 2011). When we isolated presumptive notochord cells from unperturbed 64-cell embryos and isolated cells were incubated in sea water containing actinomycin D, which is an inhibitor of transcription (Crowther and Whittaker, 1984; Davidson and Swalla, 2001; Green et al., 2002; Krasovec et al., 2021; Meedel, 1983; Meedel and Whittaker, 1979; Miyaoku et al., 2018; Nishida and Kumano, 1997; Nishikata et al., 1987; Satoh and Ikegami, 1981; Shirae-Kurabayashi et al., 2006; Terakado, 1973; Whittaker, 1973, 1977), each of these cells became 13.4 cells on average (Fig. 1A). Thus, inhibitors of transcription increased the number of rounds of the cell cycle of isolated presumptive notochord cells, indicating that genes zygotically expressed after the 64-cell stage were required to stop cell division in the notochord lineage.

Similarly, we isolated two different lineages of presumptive muscle cells from normal 64-cell embryos. First, when B7.4 cells were isolated and incubated in normal sea water, isolated cells divided three times to become 7.9 cells on average (Fig. 1B). This number was close to the expected number (8=2³), because B7.4 cells divide three times before stopping division in normal embryos (Nishida, 1987). Therefore, as in notochord and muscle cells in Halocynthia embryos (Fujikawa et al., 2011), this experiment showed that the number of cell cycle rounds of isolated blastomeres was tissue- or lineage-autonomously determined without requiring interactions with other cell lineages. By contrast, when we isolated this lineage of cells from unperturbed 64-cell embryos and incubated isolated cells in sea water containing actinomycin D, each of them became 12.5 cells on average (Fig. 1B). Thus, genes zygotically expressed after the 64-cell stage are also required to stop cell division in the muscle lineage.

Similarly, a previous study reported that B7.8 presumptive muscle cells divide two additional times before stopping division in normal embryos (Nishida, 1987), and we found that isolated B7.8 cells usually divide twice in normal sea water (Fig. 1C). However, isolated B7.8 cells incubated in sea water containing actinomycin D became 6.6 cells on average (Fig. 1C). Therefore, zygotic gene expression after the 64-cell stage is also necessary to stop the cell cycle in this muscle lineage.

Finally, we isolated presumptive epidermal cells (a6.6 and b6.8) from unperturbed 32-cell embryos. These cells divide six more times (11 times in total from 1-cell embryos) in normal embryos (Ogura et al., 2011; Pasini et al., 2006; Yamada and Nishida, 1999). When isolated a6.6 and b6.8 cells were incubated in normal sea water, they became 61.5 and 60.1 cells on average. These numbers were close to the expected value (64=2⁶) (Fig. 1D,E). By contrast, a6.6 and b6.8 cells incubated in sea water containing actinomycin D became 89.3 and 83.0 cells (Fig. 1D,E). Therefore, cell cycle rounds of epidermal cells were also under control of zygotic gene expression after the 32-cell stage. Note that cell numbers did not always increase with actinomycin D treatment, because numbers of mesenchymal and endodermal cells were reduced with this treatment (see Fig. 6A,B).

The above experiments indicated that a gene or genes zygotically activated after fate specification are required to stop cell division properly in notochord, muscle and epidermis. A gene encoding a Cdk inhibitor was a strong candidate, because involvement of a Cdk inhibitor in regulation of cell division of notochord and muscle cells has been indicated in another ascidian (Kuwajima et al., 2014). Indeed, in situ hybridization showed that an orthologue (Cdkn1.b) was expressed zygotically, as shown in Fig. 2. Although no clear signal for zygotic expression was observed before the gastrula stage, the first evident in situ hybridization signal was detected in primary muscle lineage cells at the middle-gastrula stage. Soon after, presumptive notochord cells expressed Cdkn1.b. Signals were also detected in a few neural cells at the early neurula stage, and many neural cells expressed Cdkn1.b at the middle- and late-neurula stages. At the late-neurula stage, signals in epidermal cells also became evident transiently. At early and middle tailbud stages, signals in notochord, muscle and neural cells were evident, whereas signals in epidermal cells became undetectable.

Next, we injected a morpholino antisense-oligonucleotide (MO) into unfertilized eggs, which was designed to inhibit translation of Cdkn1.b mRNA. After insemination, injected eggs were incubated until the 32- or 64-cell stage. Then, to count accurately how many times cells divided, we isolated presumptive notochord cells (A7.3 or A7.7), presumptive muscle cells (B7.4 and B7.8) and presumptive epidermal cells (a6.6 and b6.8), and incubated them until uninjected control embryos hatched. As expected, isolated cells from Cdkn1.b MO-injected embryos produced a larger number of cells than isolated cells from embryos injected with a control MO against Escherichia coli lacZ, and the differences were statistically significant in all cases (Fig. 3A-E). These numbers were scarcely changed when we incubated for an additional 6 h. These results strongly supported the previously presented hypothesis that Cdkn1.b is responsible for stopping cell division in notochord and muscle (Kuwajima et al., 2014). In addition, the present study showed that Cdkn1.b is also transiently expressed in epidermal cells in order for epidermal cells to stop dividing.

To confirm the specificity of the MO, we designed another MO against Cdkn1.b. We isolated presumptive notochord (A7.3/A7.7) and muscle (B7.4) cells from embryos injected with the second Cdkn1.b MO, and found that injection of this MO similarly produced a larger number of cells than the control MO (Fig. S1).

Next, to confirm that knockdown of Cdkn1.b promoted extra rounds of cell cycle, not only in isolated cells but also in intact embryos, we injected the Cdkn1.b MO into eggs, and labelled presumptive notochord (A7.3/A7.7), muscle (B7.4 and B7.8) and epidermal (a6.6 and b6.8) cells with DiI, and incubated them until uninjected control embryos hatched. As shown in Fig. S2, the numbers of labelled cells were significantly increased in Cdkn1.b morphant embryos.

Finally, we injected Cdkn1.b mRNA into eggs. A previous study reported that cell division stopped at the 8-cell or 16-cell stage in Halocynthia embryos in response to overexpression of Cdkn1.b (Kuwajima et al., 2014). However, in C. savignyi embryos, cells of embryos injected with Cdkn1.b mRNA divided normally until the 64-cell stage. Therefore, we isolated the presumptive notochord (A7.3/A7.7) and muscle (B7.4 and B7.8) cells at the 64-cell stage, and epidermal cells (a6.6 and b6.8) at the 32-cell stage, and incubated them until uninjected control embryos hatched. As expected, isolated cells from Cdkn1.b mRNA-injected embryos produced a smaller number of cells than isolated cells from embryos injected with a control lacZ mRNA (Fig. S3A-E). Therefore, overexpression of Cdkn1.b delayed or stopped cell cycle in the notochord, muscle and epidermal lineages.

Because Brachyury directly or indirectly controls many notochord-specific genes (Takahashi et al., 1999), we examined whether expression of Cdkn1.b is under control of Brachyury in notochord. Cdkn1.b expression was indeed lost in presumptive notochord cells of embryos injected with a Brachyury MO, whereas it was unchanged in embryos injected with the control MO (Fig. 4A,B). This observation is consistent with what has been shown in Halocynthia embryos (Kuwajima et al., 2014). Next, we injected an MO against Zic-r.b (also known as ZicL). Because knockdown of Zic-r.b abolishes Brachyury expression (Imai et al., 2002), it was expected that knockdown of Zic-r.b would also suppress Cdkn1.b. Indeed, Cdkn1.b expression was lost in presumptive notochord cells of these embryos (Fig. 4C). These data indicated that this gene is under control of Zic-r.b and Brachyury. However, the data included in a previous study indicate that overexpression of Brachyury does not induce Cdkn1.b (Reeves et al., 2017), which suggests that Brachyury alone may not be sufficient to activate Cdkn1.b in the notochord lineage.

Zic-r.b is also involved in specification of muscle fate by activating Tbx6-r.b and Mrf in cooperation with a paralogous gene, Zic-r.a (also known as Macho-1) (Nishida and Sawada, 2001; Yu et al., 2019). Although knockdown of Zic-r.a or Zic-r.b does not completely abolish expression of muscle-specific genes, simultaneous knockdown of Zic-r.a and Zic-r.b abolishes it completely (Imai et al., 2002; Satou et al., 2002). Similarly, knockdown of Zic-r.b alone did not have an apparent effect on Cdkn1.b expression in muscle cells (Fig. 4C), but double knockdown of Zic-r.a and Zic-r.b abolished Cdkn1.b expression in muscle cells (Fig. 4D). Thus, Cdkn1.b is controlled by the same developmental programmes that direct fate specification in notochord and muscle.

Finally, to demonstrate the importance of cell cycle control in proper formation of notochord, we injected the Cdkn1.b MO into eggs. The resulting larvae had short and thick tails, compared with normal larvae (Fig. 5A,B). Because major morphogenetic events of notochord formation start after notochord cells become post-mitotic (Jiang and Smith, 2007), it is likely that inhibition of Cdkn1.b prevented notochord morphogenesis. Indeed, at 0 h post-hatch (hph), there were many small notochord cells, and normal morphogenesis was apparently disturbed (Fig. 5C,D). At 6 hph, whereas a cell-free tubular lumen was conspicuous in the notochord of normal larvae (Fig. 5C′), such a lumen was not observed in Cdkn1.b morphants (Fig. 5D′).

Such morphogenetic defects were observed even before lumen formation. In early tailbud embryos injected with the control MO, 40 notochord cells were intercalated and aligned in a line (Fig. S4A). By contrast, in Cdkn1.b MO-injected embryos notochord cells were not aligned, although the number of notochord cells was the same as in normal embryos at this stage (Fig. S4B). In this experiment, we injected the Cdkn1.b MO into a pair of anterior vegetal blastomeres (A4.1 cell pair) of 8-cell embryos. Because this cell pair produces 32 notochord cells, but does not produce primary lineage muscle cells or epidermal cells, we can assess the effects of Cdkn1.b MO injection on notochord more clearly.

We also examined expression of two marker genes, Talin (Satou et al., 2001c; Yamada et al., 2003) and Tropomyosin-like 1 (Di Gregorio and Levine, 1999), by in situ hybridization (Fig. 5E-H). As shown in Fig. 5F,H, these markers were expressed in Cdkn1.b morphant embryos, indicating that knockdown of Cdkn1.b did not affect expression of Talin or Tropomyosin-like 1 in notochord cells. Similarly, we confirmed that muscle markers and epidermal markers were expressed in Cdkn1.b morphant embryos (Fig. S5).

A previous study showed that mesenchymal and endodermal cells in Ciona robusta (also called Ciona intestinalis type A) continue to divide after hatching (Nakayama et al., 2005). We confirmed that these cells continue to divide after hatching in C. savignyi by labelling one of three presumptive mesenchyme cells (B7.7, B8.5 and A7.6) and five presumptive endoderm cells (A7.1, A7.2, A7.5, B7.1 and B7.2) with DiI. Numbers of labelled cells in larvae at 6 hph were greater than those at hatching (0 hph) (Fig. S6A,B).

For further confirmation, we transferred normal larvae into sea water containing 5-bromo-2'-deoxyuridine (BrdU) for 30 min before fixation. At 0 and 6 hph, incorporation of BrdU in mesenchymal and endodermal cells was confirmed with a specific antibody (Fig. S6C), as in the case of C. robusta (C. intestinalis type A) (Nakayama et al., 2005).

Next, we examined whether zygotic transcription was necessary for mesenchymal and endodermal cells to continue to divide. For this purpose, we similarly labelled B7.7 (presumptive mesenchyme) and A7.2 (presumptive endoderm) of unperturbed embryos with DiI between the 68- and 112-cell stages, and then incubated them in normal sea water (control) or sea water containing actinomycin D. When control embryos hatched, we counted the labelled cells. We observed 109.0 labelled mesenchymal cells and 62.4 labelled endodermal cells in control embryos on average (Fig. 6A,B), and 22.6 and 24.3 labelled cells, respectively, in embryos incubated in sea water containing actinomycin D (Fig. 6A,B). Thus, inhibitors of transcription decreased the number of rounds of cell division or delayed the cell cycle of the mesenchymal and endodermal lineages, indicating that zygotically expressed genes are required to promote cell cycle progression in these cell lineages.

A previous study showed that a proto-oncogene, Myc, is zygotically expressed in mesenchyme and endoderm in C. robusta (C. intestinalis type A) (Imai et al., 2004), and we confirmed that Myc was zygotically expressed the same way in C. savignyi (Fig. S7). On the basis of this expression pattern, we tested the possibility that Myc is involved in cell cycle regulation of mesenchymal and endodermal cells. For this purpose, we injected an MO against Myc into eggs, and labelled a presumptive mesenchymal or endodermal cell with DiI between the 68- and 112-cell stages. When the control lacZ MO was injected and B7.7 (presumptive mesenchymal cell) was labelled with DiI, 107.4 and 117.4 labelled cells were detected at 0 hph and 6 hph, respectively (Fig. 6C). By contrast, when the Myc MO was injected, 27.1 and 28.4 labelled cells were observed at 0 hph and 6 hph (Fig. 6C). This suggested that control larvae divided approximately two additional times compared with Myc-morphant embryos. We obtained similar results in the two remaining mesenchymal lineages (B8.5 and A7.6 lineages; Fig. S8A). Therefore, the above observation indicated that Myc is necessary for mesenchyme cells to continue to divide.

Similarly, when the control MO was injected and A7.2 (presumptive endodermal cell) was labelled with DiI, 60.5 and 102.5 cells were detected at 0 hph and 6 hph (Fig. 6D). By contrast, when the Myc MO was injected, 26.4 and 30.0 cells were detected at 0 hph and 6 hph (Fig. 6D). We obtained similar results in the other four endodermal lineages (A7.1, A7.5, B7.1 and B7.2 lineages; Fig. S8B). Therefore, as in the case of mesenchymal cells, our data indicated that Myc is necessary for endodermal cells to continue to divide.

We could not obtain another Myc MO that could sufficiently knock down Myc; therefore, we made a construct that drove a dominant-negative version of Myc (dnMyc) under the upstream regulatory region of Twist-r.a. Because C. robusta (C. intestinalis type A) eggs are more suitable for electroporation, we used C. robusta eggs for this experiment. When we introduced a construct that contained a lacZ reporter and the same Twist-r.a upstream region into fertilized eggs, resultant larvae expressed the reporter in mesenchymal cells. In larvae co-introduced with constructs encoding dnMyc and lacZ, we found larger cells than those in normal larvae (Fig. S9A,B). To quantify larvae with large cells, we picked several cells that appeared largest in each of five control and eight experimental larvae into which only dnMyc was introduced, and measured areas of confocal images of these cells. As shown in Fig. S9C, embryos that expressed dnMyc contained larger cells than control embryos. We also counted the number of cells expressing β-galactosidase in embryos into which lacZ and dnMyc were co-introduced. The number of β-galactosidase-positive cells was markedly smaller in larvae with lacZ and dnMyc constructs than in larvae with the lacZ construct only (Fig. S9D). These results indicated that the dominant-negative form of Myc suppressed cell division, and were consistent with results obtained using the Myc MO in C. savignyi larvae; therefore, it is likely that the Myc MO successfully inhibited translation of Myc mRNA.

Twist-r.a/b and Lhx3/4 are key transcription factor genes for fate specification of mesenchyme and endoderm, respectively (Imai et al., 2003; Satou et al., 2001b). Myc expression was lost in mesenchymal cells of embryos injected with a Twist-r.a/b MO and in endodermal cells of embryos injected with an Lhx3/4 MO (Fig. 7). By contrast, expression of two mesenchymal marker genes [Hm13 (previously called Psl3) and Tram1/2] (Tokuoka et al., 2004) and two endodermal marker genes (Thr and Gata.a) (Imai et al., 2004, 2006) were not changed in Myc MO-injected embryos (Fig. S10). These data indicated that Myc is under control of Twist-r.a/b and Lhx3/4 in mesenchymal cells and endodermal cells, respectively, and that Myc is not involved in specification of mesenchymal fate or endodermal fate.

Because Myc is a transcription factor, it may regulate cell cycle regulator genes. To test this hypothesis, we quantified levels of mRNAs encoding Ccna (Cyclin A), Ccnb (Cyclin B), Ccnd (Cyclin D), Ccne (Cyclin E), Cdk1, Cdk2/3 and Cdk4/6 by reverse-transcription followed by quantitative PCR (RT-qPCR). As shown in Fig. 8, expression levels of Ccnb, Ccne, Cdk1, Cdk2/3 and Cdk4/6 were obviously reduced in Myc MO-injected embryos, whereas Ccna and Ccnd expression levels were not significantly altered.

Previous studies have shown that zygotic factors are required to control cell cycling in ascidian early and late embryos (Dumollard et al., 2013; Kuwajima et al., 2014; McDougall et al., 2012). The present study showed that numbers of cell cycle rounds in normal embryos are controlled by zygotically expressed Cdkn1.b and Myc, depending on cell lineages. In the A-line notochord lineage, cells stopped dividing after precisely nine rounds of cell division under control of Cdkn1.b. Similarly, in the primary muscle lineage, cells stopped dividing after eight (B7.8 lineage) or nine (B7.4 lineage) rounds of cell division, under control of Cdkn1.b.

As in the case of Halocynthia Cdkn1.b, Ciona Cdkn1.b begins to be expressed in notochord and muscle lineages within the cell cycle once or twice before the final cycle. Kuwajima et al. suggested the possibility that a certain duration, but not a certain number of cell divisions, is necessary for a sufficient amount of Cdkn1.b to accumulate (Kuwajima et al., 2014). Our data are consistent with this hypothesis. Expression of Cdkn1.b in epidermal cells is restricted to a short time during the neurula stage. This might be related to the fact that epidermal cells resume cell cycling several hours after hatching (Nakayama et al., 2005).

Numbers of notochord, muscle and epidermal cells in embryos injected with the Cdkn1.b MO were slightly greater in intact embryos than in isolated partial embryos (compare Fig. 3 with Fig. S2). Although this may be due to damage unexpectedly caused by isolation, it is possible that these cells may receive unknown signalling molecules that promote cell division through cell–cell interactions.

Mesenchymal and endodermal cells do not have apparent functions in larvae, and they produce adult mesodermal and endodermal tissues after metamorphosis (Hirano and Nishida, 1997, 2000; Tokuoka et al., 2005). The number of mesenchymal cells is estimated to be around 900 at 0 hph (Monroy, 1979). As we showed in the present study, numbers of the three lines of mesenchymal cells became 1.1-1.6 times (1.3 times on average) larger at 6 hph than at 0 hph (Fig. S6); therefore, there may be ∼1200 cells at 6 hph. Because mesenchymal cells are derived from five pairs of cells (ten cells) in the eighth generation (Nishida, 1987), these presumptive cells are estimated to divide six or seven more times (13-14 times in total from 1-cell embryos).

Similarly, the number of endodermal cells is estimated to be around 500 at 0 hph (Monroy, 1979). We observed that numbers of the five lineages of endodermal cells became 1.1-1.9 times (1.6 times on average) greater at 6 hph than at 0 hph; therefore, there may be ∼800 cells at 6 hph. Endodermal cells are derived from five pairs of cells in the seventh generation. Hence, it is likely that the five pairs of presumptive endodermal cells (ten cells) divide six or seven additional times (12-13 times in total). Thus, the number of cell cycles of mesenchymal and endodermal cells greatly exceed the number of cell cycles of the notochord and muscle lineages of cells. The present study indicated that Myc expression is required for mesenchymal and endodermal cells to divide so many times.

Myc is a well-known oncogene that controls cell cycling. One function of Myc is to regulate cell cycle-related genes (Bretones et al., 2015). We examined this possibility in the present study. In our experiment, we used whole embryos that included tissues in which Myc was not expressed. Because these tissues might mask effects of Myc knockdown, it is possible that Ccna and Ccnd, levels of which were not changed in the present study, are also controlled by Myc in mesenchymal and endodermal cells. By contrast, despite the possible masking effect, Ccnb, Ccne, Cdk1, Cdk2/3 and Cdk4/6 were clearly downregulated by injection of the Myc MO. Thus, Myc controls cell cycle-related genes directly or indirectly, and ascidians use this conserved mechanism to continue the cell cycle in embryonic/larval mesenchyme and endoderm. Myc can also regulate the cell cycle by hyperactivating Cyclin/Cdk complexes, inhibiting transcription of CDKN1A, encoding a Cdk inhibitor, and activating genes that encode proteins involved in DNA replication (Bretones et al., 2015). It is possible that these mechanisms also act in ascidian mesenchyme and endoderm. We injected the Myc MO into eggs to evaluate its effects in whole embryos. Therefore, we cannot rule out the possibility that mesenchyme and endoderm interact with each other to control cell cycle rounds under control of Myc. However, because cell cycle rounds of these two tissues are determined without interaction among tissues in Halocynthia embryos (Fujikawa et al., 2011), it is likely that cell cycle rounds of mesenchyme and endoderm are also controlled tissue-autonomously in Ciona embryos.

In the in situ hybridization experiment, we detected maternal Cdkn1.b mRNA, but this maternal mRNA was hardly seen in 112-cell embryos. Injection of the Cdkn1.b MO did not apparently change the morphology of embryos before the 112-cell stage. Therefore, it is highly likely that the phenotypes observed in the present study were due to knockdown of zygotic Cdkn1.b mRNA. Likewise, we detected maternal Myc mRNA, and this maternal mRNA was hardly detected at the gastrula stage. Injection of the Myc MO did not apparently change the morphology of embryos before the gastrula stage. Therefore, it is highly likely that the observed phenotypes are due to knockdown of zygotic Myc mRNA.

Inhibition of progression of the cell cycle by Cdk inhibitors and maintenance of cell proliferation by Myc are well-known mechanisms for cell cycle control in animal development (de Nooij et al., 1996; Fukuyama et al., 2003; Knoepfler et al., 2002; Lane et al., 1996; Liu et al., 2012). The present study showed that both of these mechanisms are used in different tissues of ascidian larvae, and identified their upstream regulators. Cdkn1.b is under control of Brachyury (and its upstream regulator Zic-r.b) in the notochord lineage, and it is under control of Zic-r.a and Zic-r.b in the muscle lineage, whereas Myc is controlled by Twist-r.a/b in the mesenchymal lineage and by Lhx3/4 in the endodermal lineage. These upstream regulators are transcription factors important for specifying cell fate (Corbo et al., 1997; Imai et al., 2002, 2003; Nishida and Sawada, 2001; Reeves et al., 2017; Satou et al., 2001b; Takahashi et al., 1999; Yasuo and Satoh, 1993, 1994; Yu et al., 2019). In muscle cells, Tbx6-r.b is activated under control of Zic-r.a and Zic-r.b (Yagi et al., 2004, 2005; Yu et al., 2019), and involvement of Tbx6-r.b in cell cycle regulation in muscle cells has been suggested in Halocynthia embryos (Kuwajima et al., 2014). Therefore, it is possible that Cdkn1.b is under the control of Tbx6-r.b in Ciona embryos. In this way, the present study showed that the same transcription factors that specify cell fate also regulate cell cycle regulator genes, directly or indirectly. However, our data do not necessarily indicate that tissue specification mechanisms and cell cycle control mechanisms are decoupled. In a related species, Ciona robusta (Ciona intestinalis type A), Dlx.b and Tfap2-r.b contribute to epidermal fate specification (Imai et al., 2017). Although currently we cannot efficiently knock down or knock out these genes in C. savignyi, it is possible that Cdkn1.b is regulated by either or both of these factors. Thus, cell fate specification and cell cycle control are tightly linked at the point of these key transcription factors in ascidian embryos.

Adult specimens of Ciona savignyi were obtained from the International Coastal Research Center (the University of Tokyo), the Onagawa Field Center (Tohoku University), the Research Center for Marine Biology (Tohoku University), the Maizuru Fisheries Research Station (Kyoto University), Honmoku fishing harbour (Kanagawa, Japan), and the University of California, Santa Barbara (CA, USA). Identifiers in Ensembl (Howe et al., 2021) for genes examined in this study are ENSCSAVG00000005700 for Cdkn1.b, ENSCSAVG00000000363 for Myc, ENSCSAVG00000002247 for Brachyury, ENSCSAVG00000001907 for Zic-r.a, ENSCSAVG00000002442/2450/2454/2459/2465 for Zic-r.b (multiple copies exist), ENSCSAVG00000011698 for Talin, ENSCSAVG00000009518 for Tropomyosin-like 1, ENSCSAVG00000008180 for Ma1, ENSCSAVG00000003085 for Epi1, ENSCSAVG00000009525 for Cesa, ENSCSAVG00000008611/8613 for Twist-r.a/b, ENSCSAVG00000000590 for Lhx3/4, ENSCSAVG00000007855 for Hm13 (Psl3), ENSCSAVG00000009046 for Tram1/2, ENSCSAVG00000006056 for Thr, ENSCSAVG00000008350 for Gata.a, ENSCSAVG00000010339 for Ccna (Cyclin A), ENSCSAVG00000003185 for Ccnb (Cyclin B), ENSCSAVG00000003746 for Ccnd (Cyclin D), ENSCSAVG00000005373 for Ccne (Cyclin E), ENSCSAVG00000001660 for Cdk1, ENSCSAVG00000009914 for Cdk2/3 and ENSCSAVG00000005088 for Cdk4/6.

Adult specimens of Ciona robusta (also called Ciona intestinalis type A) were obtained from the National Bio-Resource Project for Ciona. Identifiers for genes examined in this study are KY21.Chr1.384 for Myc and KY21.Chr5.361 for Twist-r.a (gene models were based on annotation of the latest genome assembly of this species; Satou et al., 2019, 2022).

Nucleotide sequences of MOs (custom-made by Gene Tools) were as follows: Cdkn1.b, 5′-CATTGTACGAAGGCGGTGGAACCAT-3′ and 5′-AAGGCGGTGGAACCATTTTGATTAC-3′; Myc, 5′-TGTTTATCGGTGTGCTGAGCTTCAT-3′; Brachyury, 5′-AACACGATTCTAAAGTGCTGGTCAT-3′; Twist-r.a/b, 5′-CTTGATTGTACTCTAGTGATGTCAT-3′; Lhx3/4, 5′-AAAGCGGGCTTGACTCGATACACAT-3′; Zic-r.a, 5′-AACCAAGCGTGCCAACGAAAGCCAT-3′; Zic-r.b, 5′-GTACATGATATTGATTGCTGTCTAA-3′. These MOs were designed to block translation. MOs for Brachyury, Twist-r.a/b, Zic-r.a and Zic-r.b have been described previously (Imai et al., 2002, 2003; Satou et al., 2001a, 2002). Specificity of MOs for Cdkn1.b and Myc was examined and is described in the Results section. In larvae injected with the Lhx3/4 MO, alkaline phosphatase activity was rarely detected by histochemical staining (Fig. S11). This phenotype was similar to that obtained in a previous study in which a different MO against Lhx3/4 was used (Satou et al., 2001b), suggesting that the Lhx3/4 MO used in the present study acted specifically. We also used an MO against E. coli lacZ as a negative control (5′-TACGCTTCTTCTTTGGAGCAGTCAT-3′). These MOs were microinjected into unfertilized eggs unless otherwise specified. Concentrations of the MOs were 1.5 mM for Cdkn1.b, 1.0 mM for Brachyury and Zic-r.b, and 0.5 mM for all others.

Synthetic capped mRNAs for Cdkn1.b and lacZ were synthesized from Cdkn1.b or lacZ cDNA cloned into the pBluescript RN3 vector (Lemaire et al., 1995) using the mMESSAGE mMACHINE T3 Transcription Kit (Thermo Fisher Scientific). The concentration of the mRNA was 1.5 μg/μl.

A mutant C. robusta Myc encoding a dominant-negative form of Myc lacking Myc box II (Haque et al., 2016) was fused to the upstream region of Twist-r.a, and E. coli lacZ was similarly fused to the upstream region of Twist-r.a. These constructs were introduced into C. robusta fertilized eggs by electroporation. To measure the area of a cell, we used a confocal slice in which a given cell looked largest among all slices, and approximated each cell using polygons. All image analysis was performed in ImageJ (https://imagej.nih.gov/ij/).

All experiments were performed at least twice using different batches of embryos. They gave essentially the same results. Total numbers of embryos examined are shown with images in figures.

Blastomeres were isolated with a fine glass needle under a binocular microscope, and isolated blastomeres were incubated using agar-coated dishes without further isolation or dissociation. DiI (CellTracker CM-DiI, Thermo Fisher Scientific) was dissolved in soybean oil at a concentration of 1 mg/ml. We labelled blastomeres with the DiI solution under a binocular microscope. To count cell numbers, isolated partial embryos, embryos and larvae were fixed for 30 min in 4% paraformaldehyde solution and mounted with Vectashield mounting medium with 4′,6-diamidino-2-phenylindole (DAPI) (Vector Laboratories, H-1200). Numbers of nuclei were counted under a fluorescence microscope or a confocal laser-scanning microscope. Actinomycin D was dissolved in DMSO and added to sea water at a final concentration of 200 μg/ml.

Whole-mount in situ hybridization was performed as described previously (Satou et al., 1995). To delineate cell shape, embryos were stained with Alexa Fluor 488 Phalloidin (1:200; Thermo Fisher Scientific, A12379) after fixation with 4% paraformaldehyde in MOPS buffer (0.5 mM NaCl, 0.1 M MOPS, pH 7.5) or 3.7% formaldehyde.

For histochemical detection of alkaline phosphatase activity, embryos were fixed with 4% paraformaldehyde in MOPS buffer for 10 min at room temperature. Embryos were washed in AP staining buffer (100 mM NaCl, 50 mM MgCl₂, 100 mM Tris-HCl, pH 9.5) twice, and then incubated with staining buffer (2.25 mg/ml Nitro Blue Tetrazolium, 1.75 mg/ml 5-bromo-4-chloro-3-indolyl phosphate).

For histochemical detection of acetylcholinesterase (AChE), embryos were fixed with 4% paraformaldehyde in MOPS buffer for 10 min at room temperature. Fixed embryos were washed in 100 mM sodium phosphate buffer three times, and incubated in AChE staining buffer (0.5 mg/ml acetylthiocholine, 65 mM sodium phosphate, 3 mM copper sulphate, 0.5 mM potassium ferricyanide, 5 mM sodium citrate, pH 6.0).

To detect β-galactosidase, embryos were fixed with 3.7% formaldehyde, and then incubated overnight with an anti-β-galactosidase monoclonal antibody (1:1000; Promega, Z3781) in Can-Get-Signal-Immunostain Solution B (TOYOBO). The signal was visualized using Alexa Fluor 555 anti-mouse secondary antibody (1:1000; abcam, ab150106) in Can-Get-Signal-Immunostain Solution B.

To monitor DNA synthesis, we added BrdU (Nacalai Tesque, 05650) to sea water at a concentration of 100 μM. Incorporation of BrdU was monitored with a specific antibody (1:50; Merck, 11170376001) following a protocol described previously (Nakayama et al., 2005). Mesenchymal cells and endodermal cells were identified by cell size and cell location.

Total RNA was extracted from 30 larvae (0 hph) using NucleoSpin RNA XS (Macherey-Nagel, U0902A). cDNA was synthesized with an iScript cDNA synthesis kit (Bio-Rad, 1708891). Quantitative PCR was performed using the MiniOpticon Real-Time PCR Detection System (Bio-Rad) with SsoFast EvaGreen Supermix (Bio-Rad, 1725200). A one-embryo-equivalent quantity of cDNA was used for each reaction. Cycling conditions were preheating at 95°C for 30 s, and 40 cycles of 95°C for 5 s and 60°C for 10 s. We took three biological replicates. Relative expression values were calculated using the ΔΔCT method, and expression values were normalized to a housekeeping gene, Ywhaz (tyrosine 3-monooxygenase/tryptophan 5-monooxygenase activation protein, zeta polypeptide). Primers used were: Ccna, 5′-GCAGCAAGACATCACAGTTGG-3′ and 5′-TGGTTTCGGTGTGGAGTTTG-3′; Ccnb, 5′-CAACATACGCTCGCAAAATACC-3′ and 5′-AAGGCACAAGGAACCAGCA-3′; Ccnd, 5′-AAGGTGTGAAGATGATGTGTTTCC-3′ and 5′-GTTGAGTTCGTTGGATTGGTTG-3′; Ccne, 5′-GGAGGTGTGCGAGGTTTATTC-3′ and 5′-GGGTTTTATGGATGTCGGTTG-3′; Cdk1, 5′-CGGTGTGACGCAGTTGAAAG-3′ and 5′-ATACGAGGCATTTAGCGAGCA-3′; Cdk2/3, 5′-TCGCACAGAGTCCTACACAGAGA-3′ and 5′-ATACATCCGCACTGGCACAC-3′; Cdk4/6, 5′-AGCGTGAAACCCAACTAATGC-3′ and 5′-AGTCCCATCAACATCTGCCTC-3′; Ywhaz, 5′-TTTCTGGACTTGGACGCAAAC-3′ and 5′-CGGCTTTCCCTTGTCCTTATC-3′.

We thank Masatoshi Hara and Sawako Hori-Oshima for earlier work. We are grateful to the staff of the International Coastal Research Center (the University of Tokyo), the Onagawa Field Center (Tohoku University), the Research Center for Marine Biology (Tohoku University), Maizuru Fisheries Research Station (Kyoto University) and Honmoku fishing harbour (Kanagawa, Japan) for help in collecting Ciona savignyi adults. We also thank Shota Chiba, Erin Newman-Smith, and William C. Smith at University of California, Santa Barbara (CA, USA) for providing Ciona savignyi adults. We thank Manabu Yoshida (the University of Tokyo) and other members working under the National Bio-Resource Project for Ciona (MEXT, Japan) at Kyoto University and the University of Tokyo for providing Ciona robusta adults. The manuscript was edited by a technical editor, Steven D. Aird.

The peer review history is available online at https://journals.biologists.com/dev/lookup/doi/10.1242/dev.201218.reviewer-comments.pdf.