Larval mesenchyme cell specification in the primitive echinoid occurs independently of the double-negative gate

Understanding how developmental mechanisms evolved is a predominant topic in the research of evolutionary biology, and echinoids (sea urchins) are an attractive model for this field. The embryonic development of indirectly developing regular sea urchins, such as Strongylocentrotus purpuratus, has been extensively examined, and knowledge of this gene regulatory network is accumulating (McClay, 2011). Additionally, various types of developmental modifications – e.g. putative ancestral developers and various derived modified developers, with developers being a species and/or group showing subjected developmental mode – are now recognized and are considered as potential subjects for evolutionary development studies (Schroeder, 1981; Wray and Raff, 1991). Based on both paleontological and molecular studies, a steadfast phylogenetic relationship that includes these various species has been hypothesized, and it provides a stable phylogenetic basis for evolutionary development studies (Smith, 1984; Kroh and Smith, 2010). Among echinoids, the cidaroid (a ‘primitive’ group of echinoids) has been considered as a crucial taxon to estimate the early evolutionary history (Schroeder, 1981; Wray and McClay, 1988). Because many characteristics of adult cidaroid morphology are considered to be ancestral, cidaroid development is also expected to preserve and reflect the ancestral state (Schroeder, 1981; Smith, 1984).

Although cidaroids have a pluteus larval stage that is similar to that of euechinoids (a ‘derived’ group of echinoids), their embryonic development clearly differs from that of euechinoids, particularly in relation to the process of mesoderm formation (Schroeder, 1981) (Fig. 1). Two types of mesenchyme cells are common in both euechinoids and cidaroids – skeletogenic cells and nonskeletogenic cells (Mortensen, 1921, 1938). The ontogeny of these two cell types in cidaroids, however, differs from that in euechinoids. In euechinoids, skeletogenic cells are derived from four micromeres at the vegetal pole of the 16-cell-stage embryo (Fig. 1E). The micromeres give rise to primary mesenchyme cells (PMCs), which ingress into the blastocoel before archenteron invagination and then differentiate into skeletogenic mesenchyme cells (Fig. 1F,H). During gastrulation, another type of mesenchyme cell emerges from the tip of the archenteron and differentiates into nonskeletogenic mesenchyme cells, such as pigment cells (Okazaki, 1975) (Fig. 1G,H). By contrast, in cidaroids, no typical micromere quartet is formed at the 16-cell stage (Schroeder, 1981; Yamazaki et al., 2012). Although the fourth cleavage in the vegetal hemisphere is unequal, the size of the blastomeres at the vegetal pole varies (Fig. 1A). The cidaroid embryos also have no typical PMCs (Fig. 1B). Both the skeletogenic and nonskeletogenic mesenchyme cells ingress after the initiation of archenteron invagination (Fig. 1C,D; supplementary material Fig. S2A2) (Schroeder, 1981; Emlet, 1988; Yamazaki et al., 2012).

The aforementioned differences in embryonic development strongly suggest that distinct molecular mechanisms are adopted for mesenchyme specification in cidaroids and euechinoids. In euechinoids, skeletogenic cell specification is regulated by the double-negative gate (DNG), a regulatory sub-circuit that consists of two sequential repressions by micro1 (also known as pmar1) and hesC (hairy/Enhancer of split C) (Revilla-i-Domingo et al., 2007). Micro1, a transcription repressor, is exclusively activated in the micromere quartet, and it blocks the transcription of hesC, another transcription repressor, in the micromeres. In the remaining part of the embryo, hesC is expressed and the product of this gene represses the transcription of downstream mesenchyme specification genes, including alx1 (aristaless-homeobox 1), tbr (T-brain), ets1 (E26 transformation-specific 1) and delta. The silencing of hesC by micro1 in the micromere lineages leads to specific activation of these mesenchyme specification genes (Kurokawa et al., 1999; Fuchikami et al., 2002; Kitamura et al., 2002; Oliveri et al., 2002; Sweet et al., 2002; Ettensohn et al., 2003; Röttinger et al., 2004; Revilla-i-Domingo et al., 2007) (see Fig. 6B). The repression of transcription of hesC by micro1 is the core mechanism of DNG. Following DNG activation, the transcription of cyp1 (cyclophilin 1) – which encodes a peptidylprolyl cis-trans isomerase and gcm (glial cells missing) – starts in skeletogenic and nonskeletogenic cells, respectively, which promotes the differentiation of each mesenchyme cell type (Fig. 1I; Amore and Davidson, 2006; Ransick and Davidson, 2006).

By contrast, the molecular mechanism of cidaroid development has scarcely been studied. The expression patterns of two putative endomesoderm specification genes in cidaroids – wnt8 and hox11/13b – suggest that the molecular mechanism responsible for cidaroid endomesoderm formation is similar to that of euechinoids (Yamazaki et al., 2012); however, little is known about the expression patterns and functions of mesenchyme specification genes. Considering the fact that cidaroids have neither typical micromeres nor PMCs, it is probable that their mesenchyme cells are specified by a mechanism that differs from that of euechinoids.

In the present study, we isolated six genes – hesC, alx1, tbr, ets1, gcm and cyp1 – from a cidaroid, Prionocidaris baculosa, and examined their expression patterns and functions in mesenchyme specification. Our results suggest that no euechinoid-type DNG exists in P. baculosa, although the roles of alx1, tbr, ets1 and gcm are similar between cidaroids and euechinoids.

Seven euechinoid genes that are involved in mesenchyme cell specification – micro1, hesC, alx1, tbr, ets1, gcm and cyp1 – were selected and attempts were made to isolate the gene orthologs in P. baculosa. All orthologs, except for micro1, were successfully identified using degenerate polymerase chain reaction and rapid amplification of cDNA ends (RACE) PCR. Phylogenetic analyses based on the deduced amino acid sequences strongly support the orthology for the relevant genes (supplementary material Fig. S1). Therefore, we assigned the prefix Pb (P. baculosa) and designated these assembled cDNAs as PbhesC (Accession No. in the DNA Data Bank of Japan AB853323), Pbalx1 (AB853324), Pbtbr (AB853325), Pbets1 (AB853326), Pbgcm (AB853328) and Pbcyp1 (AB853327).

The temporal and spatial expression profiles of these genes were analyzed by quantitative real-time PCR (qPCR) and whole-mount in situ hybridization (WMISH; Fig. 2 for late-blastula stage; Fig. 3 for late-gastrula stage; supplementary material Fig. S2 for other identified stages). At the late-blastula stage (16 h post-fertilization; hpf), PbhesC expression was observed in two regions – the vegetal pole (lower double-headed arrow in Fig. 2A,A″) and the presumptive ectodermal region (upper double-headed arrow in Fig. 2A). These two regions are separated by a torus domain that lacked expression of this gene (between the white arrowheads in Fig. 2A). The expression of Pbalx1, Pbtbr, Pbets1 and Pbgcm was exclusively observed in the vegetal pole region (Fig. 2B-E,B″-E″). No localized expression of Pbcyp1 was detected at the late-blastula stage (Fig. 2F). Thus, at this stage, all genes, except for Pbcyp1, were expressed together in the vegetal pole region. However, the widths of the expression domains were not identical (Fig. 2A′-E′). The proportions of the vegetal expression region of PbhesC, Pbalx1, Pbtbr, Pbets1 and Pbgcm to the perimeter of the embryos (on average) were 12%, 8%, 11%, 15% and 11%, respectively. As summarized in Fig. 2G, the expression regions of Pbalx1 and Pbets1 were the narrowest and the broadest, respectively, whereas the sizes of the expression regions of PbhesC, Pbtbr and Pbgcm were intermediate to these.

In the late-gastrula stage (36 hpf), all genes were expressed in both mesenchyme cells (black arrowheads in Fig. 3A-F) and the archenteron (brackets in Fig. 3A-F); however, the loci of expression were not identical. Two types of mesenchyme cells can be identified in the late gastrula – mesenchyme cells with skeletogenic cell-specific antigen (P4) expression (skeletogenic mesenchyme cells) (Shimizu et al., 1988) and those that lack P4 expression (nonskeletogenic mesenchyme cells). By simultaneously using WMISH and immunohistochemistry with P4 antibody, the gene expression patterns of the two mesenchyme cell types were revealed. PbhesC was expressed in both mesenchyme cell types (Fig. 3A′-A‴), whereas the expression of Pbalx1 was observed in almost all skeletogenic cells but not in nonskeletogenic cells (Fig. 3B′-B‴). The expression of Pbtbr was observed in all skeletogenic cells and in some nonskeletogenic cells (Fig. 3C′-C‴); the expression of Pbets1 and Pbcyp1 was observed in almost all mesenchyme cells of both types (Fig. 3D′-D‴ and F′-F‴); and the expression of Pbgcm was observed in only a small portion of nonskeletogenic cells (Fig. 3E′-E‴). The expression patterns of these genes were classified into five patterns that are summarized in Fig. 3G.

The expression domains in the archenteron also differed among the genes (brackets in Fig. 3A-F). PbhesC expression was observed in the middle and lower region of the archenteron (Fig. 3A), whereas the rest of the genes were expressed in the tip of the archenteron or upper region (Fig. 3B-F). In addition, PbhesC and Pbgcm were expressed in parts of the ectodermal region (arrows in Fig. 3A,E).

The overlapping expression of PbhesC and mesenchyme specification genes at the vegetal pole of the blastula (Fig. 2G) suggests the absence of DNG in P. baculosa mesenchyme specification. To verify this possibility, we performed a perturbation experiment on PbhesC using morpholino antisense oligonucleotides (MOs) and examined the effects on the expression of mesenchyme specification genes at the mid-blastula stage (12 hpf).

The perturbation of PbhesC did not affect the spatial expression patterns of all genes examined. There were no significant differences in the widths of gene expression domains between the control embryos that had been injected with an MO designed against hesC of another echinoid species, Scaphechinus mirabilis (SmhesC; hereafter, SmhesC-MO) and the experimental embryos (PbhesC-MO-injected embryos; hereafter, PbhesC-MO embryos) (Fig. 4Aa-Ea,Ab-Eb for images of embryos; Fig. 4Ac-Ed for the size distribution of the gene expression domains). This suggests that PbhesC does not participate in the spatial regulation of Pbalx1, Pbtbr, Pbets1 and Pbgcm. The only visible difference between the control and PbhesC-MO embryos was the density of the WMISH signal for PbhesC (Fig. 4Ea,Eb).

The high intensity of the PbhesC WMISH signal in PbhesC-MO embryos suggested a repressive role for PbhesC on the transcription of PbhesC itself. To verify this possibility, qPCR analysis was conducted. The amount of transcribed PbhesC mRNA was significantly increased in the PbhesC-MO embryo (Fig. 4F), suggesting that PbhesC plays a repressive role in PbhesC transcription. The result of the increased hesC transcription in hesC-MO-injected embryos was also obtained when similar experiments were performed for two euechinoids, S. mirabilis and Hemicentrotuspulcherrimus (supplementary material Fig. S3B,C). These results strongly suggest that the autorepression of hesC is a conserved function in echinoids. The specificity of the MOs was experimentally supported – no obvious change in the amount of hesC transcripts in all three species was observed when the hesC-MO that had been designed against a different species was introduced (supplementary material Fig. S3). These results indicate the efficacy and specificity of the PbhesC-MO in P. baculosa embryos, even though this MO did not alter the spatial expression patterns of alx1, tbr and ets1.

qPCR analyses also revealed the modest repressive potential of PbhesC on Pbalx1 and Pbtbr. In PbhesC-MO embryos, an increase in the amount of Pbalx1 transcript was detected in two batches (separate experiments using different eggs), and an increase in Pbtbr transcripts was detected in one of the two batches that were examined. No significant change was observed for the other three genes (Pbcyp1, Pbets1 and Pbgcm; Fig. 4F).

To examine whether PbhesC plays a part in the differentiation of mesenchyme cells, PbhesC-MO embryos were reared to the late-gastrula stage, and the expression of the skeletogenic cell-specific P4 antigen and Pbcyp1 (a pan-mesenchyme cell marker) was examined. No significant changes were observed in the number of either skeletogenic cells or total mesenchyme cells. The average numbers of P4-positive cells and Pbcyp1-expressing mesenchyme cells were approximately 20 and 55, respectively, in both control and PbhesC-MO embryos (Fig. 4G-L).

Mesenchyme cell differentiation in euechinoids is regulated by various transcription factors – including alx1, tbr and ets1 (Kurokawa et al., 1999; Fuchikami et al., 2002; Ettensohn et al., 2003; Röttinger et al., 2004). The expression patterns of the P. baculosa orthologs resembled those in euechinoids (Figs 2 and 3), suggesting that these genes have similar roles to their counterparts in euechinoids. To examine the roles of these genes in the mesenchyme cell specification of P. baculosa, we performed functional perturbation experiments and examined the effects on mesenchyme cell differentiation. The functions of alx1 and tbr were perturbed by gene-specific MOs. To estimate the ets1 function, the MEK inhibitor U0126 was used because it blocks the mitogen-activated protein kinase (MAPK) pathway that is required for ets1 activation in euechinoids (Röttinger et al., 2004; Ettensohn et al., 2007).

First, we examined the effects on the differentiation of mesenchyme cells at the gastrula stage (36 hpf). The expression of cyp1 was used as a pan-mesenchyme cell marker. The numbers of cyp1-positive mesenchyme cells in the Pbalx1-MO (Fig. 5B) and Pbtbr-MO embryos (Fig. 5C) were not significantly different from that in the control (control-MO embryo) (Fig. 5A) – the mean numbers were approximately 60 (Fig. 5E). By contrast, almost no migrating mesenchyme cells were observed in the U0126-treated embryos, and the number of cyp1-positive mesenchyme cells was strikingly decreased (approximately one cell per embryo upon treatment with U0126, and about 45 cells per embryo upon treatment with the control reagent dimethylsulfoxide) (Fig. 5D,E).

The differentiation of skeletogenic cells (Fig. 5F-I) and some of the nonskeletogenic mesenchyme cells (Pbgcm-positive cells) (Fig. 5K-N) at the late-gastrula stage was examined under perturbation. In the Pbalx1-MO and U0126-treated embryos, P4-positive skeletogenic cells were rarely formed (approximately one cell per embryo upon treatment with Pbalx1-MO, and less than one cell per embryo in those treated with U0126) (Fig. 5G,I,J). No significant difference in the number of P4-positive cells was observed between Pbtbr-MO and control-MO embryos (about 20 cells per embryo upon treatment with Pbtbr-MO, and 22 cells per embryo in control-MO embryos) (Fig. 5F,H,J).

The differentiation of Pbgcm-positive mesenchyme cells was strongly affected in Pbtbr-MO or U0126-treated embryos. In Pbtbr-MO embryos, the number of Pbgcm-positive mesenchyme cells (pink arrowheads) was approximately 60% or less than that in the control embryos (Fig. 5M,O). In U0126-treated embryos, there were almost no gcm-expressing mesenchyme cells; however, Pbgcm expression was observed in the tip of the archenteron and the bilateral ectodermal regions (Fig. 5N,O). Control and Pbalx1-MO embryos had a mean of 14-16 gcm-positive mesenchyme cells (pink arrowheads; Fig. 5K,L,O).

The larval skeletons and red-colored pigment cells are visible characteristics of mesenchyme differentiation in the pluteus larvae. To evaluate the effect on skeleton formation, the total length of postoral rods (TLPR, blue double-arrows in Fig. 5P-S) was measured. In the control pluteus larva (control-MO or dimethylsulfoxide-treated embryos), the TLPR was approximately 300-400 µm (Fig. 5P,T). The TLPR values in Pbalx1-MO, Pbtbr-MO and U0126-treated larvae were below 100 µm (Fig. 5Q-T). Pigment cell differentiation was severely affected by the perturbation of Pbtbr and Pbets1. Control pluteus larvae had 31-35 pigment cells on average (Fig. 5P,U). The number of pigment cells in Pbtbr-MO larvae was 19 on average, which is approximately 60% of the number of those in the control, whereas U0126-treated larvae had even fewer pigment cells (approximately five cells per embryo) (Fig. 5R,S,U). By contrast, in Pbalx1-MO larvae, the number of pigment cells was almost the same as that in the control (Fig. 5Q,U).

These effects caused by alx1-MO, tbr-MO and U0126 in P. baculosa were considerably similar to those observed previously in euechinoids (Fuchikami et al., 2002; Ettensohn et al., 2003; Röttinger et al., 2004). Based on these results, we suggest that alx1 is necessary for skeletogenic cell differentiation, that tbr is necessary for skeleton formation and pigment cell differentiation and that the MAPK pathway (and possibly ets1) is necessary for the general mesenchyme cell differentiation process and/or migration processes in P. baculosa. The reduction of Pbgcm-expressing mesenchyme cells and pigment cells in Pbtbr-MO embryos suggests that there is a role for Pbgcm in pigment cell differentiation.

The present study revealed various differences in the larval mesenchyme specification processes between the cidaroid P. baculosa and euechinoids. The most prominent question is as to whether DNG is involved in mesenchyme specification in cidaroids. In euechinoids, the complementary expression pattern of hesC with other component genes (micro1, alx1, tbr and ets1) is an indication of the existence of the DNG (Fig. 6B) (Revilla-i-Domingo et al., 2007). In P. baculosa, the overlapping expression patterns of PbhesC and mesenchyme specification genes in the vegetal pole region strongly suggest the absence of a euechinoid-type DNG that involves hesC (Fig. 2 and Fig. 6A), which is also supported by the PbhesC perturbation experiments (Fig. 4). Smith and Davidson (Smith and Davidson, 2009) have recently proposed the hypothesis that the DNG, consisting of hesC and blimp1, is an ancestral system in echinoids and that the micro1 initiation system might have developed more recently in the modern euechinoid lineage. However, the hesC-mediated DNG system, regardless of whether the repressor of hesC is micro1 or blimp1, seems to be absent in the mesenchyme specification mechanism of cidaroids. The change from blimp1-mediated DNG to micro1-mediated DNG, if any, might have occurred in the euechinoid lineage after divergence from the cidaroid ancestor.

It might be postulated that P. baculosa contains a DNG framework that consists of an unknown repressor, instead of hesC, which controls the expression of alx1, tbr and ets1; however, this cannot be the case because the borders of the expression regions for alx1, tbr and ets1 did not correspond with each other, even at the earliest stages, following the onset of their region-specific expression (Fig. 4A-C), suggesting that the spatial expression of these three genes is regulated through distinct factors rather than a single repressor.

Another difference is the timing of mesenchyme cell ingression (supplementary material Fig. S2A2). Unlike in the case of euechinoids, in all indirectly developing cidaroids that have been studied to date, mesenchyme cell ingression starts after the initiation of archenteron invagination (Schroeder, 1981; Emlet, 1988; present study). However, the order of the timing of the ingression of mesenchyme cell types can be similar to that in euechinoids – i.e. skeletogenic cells ingress early during development and pigment cells ingress later. In P. baculosa, the first ingressing mesenchyme cells at the mid-gastrula stage express both the skeletogenic cell-specific P4 antigen (supplementary material Fig. S2A2) and Pbalx1 mRNA (supplementary material Fig. S2C2), suggesting that these cells are precursors of skeletogenic cells. A previous study on another cidaroid, Eucidaris tribuloides, has also demonstrated that the skeletogenic cells ingress at an earlier stage (Wray and McClay, 1988). The ingression of nonskeletogenic mesenchyme cells follows the ingression of skeletogenic cells. Our study shows that Pbgcm-expressing mesenchyme cells, which are possibly pigment cell precursors, ingress long after the appearance of skeletogenic cells (Fig. 3E; supplementary material Fig. S2F2). The order of the ingression of skeletogenic cells, which is followed by that of pigment cells, seems to have been established in the echinoid ancestor.

The existence of DNG is not the only feature that distinguishes euechinoids from cidaroids on a molecular level. The present results suggest that the array of genes involved in skeletogenic cell differentiation had been also modified during echinoid evolution. In the euechinoid S. purpuratus, cyp1 is specifically expressed in the skeletogenic mesenchyme region (Amore and Davidson, 2006). Alternatively, in P. baculosa, almost all mesenchyme cells expressed cyp1 (Fig. 3F), suggesting that cyp1 is involved in the differentiation of all mesenchyme cell types. The PMC-restricted expression pattern of cyp1 in S. purpuratus strongly suggests that an evolutionary modification occurred of cyp1 function in the euechinoid lineage.

For more than three decades, it has been deemed that the developmental mechanism for mesenchyme specification was modified in modern euechinoids during echinoid evolution (Schroeder, 1981). However, it is still unknown how a molecular mechanism effected this change. In the present study, we hypothesize that the introduction of the DNG is one of the crucial modifications in the euechinoid ancestor. Based on the study of asteroid embryos, McCauley and colleagues (McCauley et al., 2010) have suggested that the DNG is a novelty of euechinoids. The result of the present study supports this view of McCauley et al. and provides experimental evidence to suggest that there is an absence of the DNG in the ancestral mechanism of echinoids.

Our study shows that there are two types of molecular mechanism that specify the mesenchyme in echinoids – i.e. the mechanism without DNG, which is observed in P. baculosa, and that with DNG, which is observed in euechinoids. These findings could indicate one of the two following evolutionary scenarios – first, P. baculosa retained its ancestral mechanism and DNG evolved in the lineage of euechinoids; second, the echinoid ancestor had already obtained DNG and cidaroids lost DNG following divergence from the euechinoid lineage. We consider the former possibility to be more likely because it has also been suggested that another group of echinoderms – the asteroid group – does not use DNG. In the asteroid embryos, hesC, alx1, tbr and ets1 are co-expressed at the vegetal pole of the blastula (Fig. 7) and perturbation of hesC does not affect the spatial expression of tbr and ets1 (McCauley et al., 2010, 2012). The similarity between the asteroid and cidaroid leads us to conclude that the mesenchyme specification mechanism without DNG is the ancestral trait of echinoids.

We further propose that DNG was established in the common ancestor of euechinoids (asterisk in the phylogenetic tree of Fig. 7). We have previously shown that the DNG-characteristic gene expression pattern is also observed in the euechinoid S. mirabilis, the ancestor of which had been separated from the camarodonta (the group including S. purpuratus) approximately 200 million years ago (Yamazaki et al., 2010) (Fig. 7), indicating that the euechinoid ancestor had obtained DNG.

This study on cidaroids might also provide the key towards understanding the evolutionary history of echinoderms. Our data suggest that the functions of alx1, tbr, ets1 and gcm had been established before the diversification of echinoids, because the roles of these genes are quite similar between P. baculosa and euechinoids (Fig. 5) (Kurokawa et al., 1999; Fuchikami et al., 2002; Ettensohn et al., 2003; Röttinger et al., 2004; Ransick and Davidson, 2006). Tbr might have been recruited for the specification of both skeletogenic and pigment cells, whereas gcm might have been recruited for the specification of pigment cells in the common ancestor of echinoids (see circle in the phylogenetic tree in Fig. 7). It is likely that, in echinoderm embryos, tbr and gcm lack these functions. The asteroids do not form larval skeletons and pigment cells, whereas holothuroid embryos have no pigment cells (Ben-Tabou de-Leon and Davidson, 2009; Takata and Kominami, 2011; McCauley et al., 2012). In addition, holothuroid tbr is not responsible for skeletogenic cell specification (McCauley et al., 2012). Our data also support previous notions about the origins of alx1 and ets1 functions – that these were obtained before diversification of the echinoid from holothuroids, a closely related echinoderm class (square and triangle in the phylogenetic tree in Fig. 7) (McCauley et al., 2012; Koga et al., 2010).

The present study provides a new interpretation of DNG function. To date, DNG has been considered to be a starting point in the series that leads to the specification of skeletogenic cells (from the micromere to PMC to larval skeletogenic cell lineage) (Revilla-i-Domingo et al., 2007; Li and Davidson, 2009). However, the larval skeletogenesis in the absence of DNG in P. baculosa indicates that this regulatory subcircuit is not essential for larval skeletogenesis. Two previous studies in euechinoids also support this idea – first, the knockdown of micro1 prevents PMC ingression but not skeletogenesis (Smith and Davidson, 2009), and second, ectopic activation of micro1 is not required for the regulation of skeletogenic cell formation in a non-PMC region, which occurs by removing PMCs (Ettensohn et al., 2007). However, it is noteworthy that DNG and micromeres and/or PMCs seem to have been synchronously acquired in the common ancestor of euechinoids (pink box in the phylogenetic tree of Fig. 7) (Gao and Davidson, 2008; Yamazaki et al., 2012; present study). Therefore, we now assume that the acquisition of DNG in the euechinoid ancestor is associated with micromere- and/or PMC-related evolutionary events – e.g. the fine-tuning of unequal cleavage mechanisms to produce a typical micromere quartet, and/or a heterochronic shift in the onset of mesenchyme ingression to occur earlier than the onset of gastrulation.

The collection of adult P. baculosa, S. mirabilis and H. pulcherrimus, and the handling of gametes and embryos were performed according to a previously described method (Yamazaki et al., 2010, 2012). To estimate the transition of total cell numbers during the early stages of development in P. baculosa embryos, the number of nuclei per embryo were counted after staining with propidium iodide (Nacalai Tesque), as described previously (Stickland et al., 2004) (supplementary material Table S1).

A conventional degenerate PCR method was used to isolate the partial sequence of hesC, alx1, tbr, ets1, gcm and cyp1 from P. baculosa. For cloning of tbr and ets1, primer sets that have been described previously were used (Koga et al., 2010; Minemura et al., 2009). For the remaining the genes, the primers listed in supplementary material Table S2 were used. To isolate 5′ and 3′ fragments, RACE was performed using primers shown in supplementary material Table S2. The template cDNA synthesis and PCR amplification were performed as described previously (Yamazaki et al., 2012). We obtained full-length sequences for hesC, alx1 and tbr and 3′ partial sequences for ets1, gcm and cyp1.

qPCR analysis was performed as described previously (Yamazaki et al., 2012). The primer sequences used are described in supplementary material Table S2. The mitochondrial cytochrome c oxidase subunit 1 gene (mitCO1) was used for the internal reference standard. The sequence of the mitCO1 gene from S. mirabilis (SmmitCO1) has been deposited in the DNA Data Bank of Japan with the accession number AB900169. For the hesC perturbation experiments, a difference of 1.6 or more in the Ct value between perturbed and control embryos was considered significant, as described previously (Oliveri and Davidson, 2004).

WMISH was performed as described previously (Yamazaki et al., 2012). For hybridization, we used a mixture of two probes that had been derived from 5′ and 3′ RACE fragments of PbhesC, Pbalx1 or Pbtbr and the probe from 3′ RACE fragments of Pbets1, Pbgcm or Pbcyp1. When WMISH, using sense probes, was performed, staining at a level higher than the background was not observed for any genes.

The lateral view photographs of the blastulae that had been stained by using WMISH were taken and analyzed by using Illustrator CS5 (Adobe Systems). A circular template was prepared and transformed to fit the perimeter of the embryo. We then measured both the length of the arc in the region where the gene of interest was expressed and that of the outer circumference of the embryo and then calculated the percentage of the arc that was within the circumference. Embryos were classified into categories that were divided every 2%, i.e. ‘2% and more’ and ‘under 4%’, according to the percentage of the expression region. The proportion of each expression sector relative to the perimeter was measured in more than 50 embryos for each gene.

Skeletogenic cell differentiation was evaluated using the P4 antibody, which recognizes the skeletogenic cell-specific glycoprotein in H. pulcherrimus (Shimizu et al., 1988). The embryos that had been fixed for WMISH analysis, or the embryos that had been stained using WMISH, were washed with MOPS buffer three times and then incubated in MOPS buffer containing the P4 antibody (diluted 1:500) for 1 h at room temperature. After being rinsed five times, the embryos were incubated with the secondary goat antibody against mouse IgGs (heavy and light chains) Hilyte Fluor 555-Labeled (diluted 1:200; AnaSpec) for 1 h at room temperature. The embryos were further washed with MOPS buffer five times and then transferred to 50% glycerol for observation. The embryos were examined by using an epifluorescence inverted microscope (Leica DM IL).

To evaluate the larval skeleton formation, the TLPR value was used. TLPR was calculated as the sum of the length of bilateral fenetrated postoral rods (blue double arrows in Fig. 5P-S). The number of pigment cells was counted, as previously described (Kominami, 1998), using a Nikon Optiphot microscope.

The translational perturbations were performed using MOs (GeneTool). Sequence-specific MOs were synthesized against PbhesC, Pbalx1, Pbtbr, SmhesC and HphesC. The MO sequences were as follows: PbhesC-MO, 5′-CACAAAGATCTGCTGATGAAATCAT-3′; Pbalx1-MO, 5′-ACGGC-ATGGAGCGGTAGAAAATCAT-3′; Pbtbr-MO, 5′-TCGCTTCAAAAT-CCTCACTCATCAT-3′; and SmhesC-MO, 5′-AGTTAGTACAAGTGG-TAGTCATCAT-3′. For the perturbation of HphesC, we used an MO that has been described previously (Ochiai et al., 2010).

The MOs were introduced into the embryos by using microinjection. After fertilization in filtered natural seawater, the jelly layer was removed by treatment with acidic seawater containing 1 mM sodium citrate. The fertilized eggs were placed on the bottom of protamine-coated dishes that had been filled with seawater containing 1 mM 3-amino 1,2,4-triazole (ATA). For H. pulcherrimus, fertilization was performed just before microinjection. MOs were diluted with 0.2 M KCl to 1 mM. For experiments on S. mirabilis, 0.25 mM MO solutions were used. The injection volume was approximately 1/64 of the quantity of the egg volume (the final concentration was approximately 15 µM when using a 1 mM MO solution). In experiments on P. baculosa, we also attempted to inject 0.25 mM MO solutions and obtained similar effects – e.g. an increase of hesC mRNA and inhibition of skeleton formation in the case of PbhesC-MO and Pbalx1-MO, respectively. By contrast, the lower concentration of Pbtbr-MO caused no effect. Microinjection into fertilized eggs was performed until 50 min, 40 min and 60 min after fertilization for P. baculosa, S. mirabilis and H. pulcherrimus, respectively.

The MEK inhibitor U0126 was used to inhibit the MAPK pathway that is essential for ets1 activation in euechinoids (Röttinger et al., 2004). Embryos were treated with 5 µM U0126 (Merck) in seawater from 2 hpf until 36 hpf. We also attempted treatment with 25 µM or 1 µM U0126. Treatment with 25 µM U0126 caused embryonic lethality or a defect in hatching, whereas 1 µM U0126-treated embryos had a small number of mesenchyme cells at the late-gastrula stage. As a control, 1 µl/ml DMSO in seawater was used. Experimental embryos were cultured in seawater containing penicillin G potassium (100 unit/ml) and streptomycin sulfate (50 µg/ml). During the culture, the seawater was changed daily. All the experiments were conducted in more than two batches.

We thank Masashi Noguchi, Setsuo Kiyomoto, Norihiko Deguchi, Toshimitsu Fukuhata and Katsuhiko Furukawa for the collection of the adult P. baculosa. We also thank Takaharu Numakunai for the P. baculosa culture system; Ryoichi Matsuda and Hideki Katow for providing the P4 antibody; and Shunsuke Yaguchi for the support in a part of gene expression analyses. qPCR analyses were performed in the core facility for research equipment at Aomori University.