Introduction

Echinoids, or sea urchins, are constituents of the phylum Echinodermata and are comprised of two extant subclasses, the cidaroids (Cidaroidea) and the euechinoids (Euechinoidea). Fossil evidence suggests these two clades had already diverged by the middle of the Permian period at least 268 million years ago (mya) (Thompson et al. 2015). In addition to the conspicuous differences observed in the adult morphologies of these clades (Gao et al. 2015), embryological evidence also indicates numerous developmental differences between these two clades, suggesting extensive rewiring of the gene regulatory networks (GRNs) directing their development (Schroeder 1981). Two decades of research have parsed out the elaborate circuitry of the GRNs guiding early euechinoid development (Peter and Davidson 2011). Thus, with an abundance of euechinoid data in hand, an auspicious opportunity presents itself not only to enumerate the observed developmental differences between these two clades at the molecular level but also to utilize them to better understand the plasticity and role of GRNs in developmental evolution.

The most striking difference between the early development of cidaroids and euechinoids is exhibited in the skeletogenic mesenchyme. In contrast to euechinoids, which exhibit a precociously ingressing, pre-invagination skeletogenic primary mesenchyme (PMC) lineage, the skeletogenic mesenchyme of cidaroids ingresses from the tip of the archenteron well after gastrulation has begun (Wray and McClay 1988). In euechinoids, the micromere quartet gives rise to the PMC lineage and is fated from very early in development to become embryonic skeleton (Oliveri et al. 2008). Cidaroids, in juxtaposition, exhibit a variable number of micromeres in each embryo (Schroeder 1981), and yet in spite of this variance their micromeres were found to be homologous to euechinoid PMCs (Wray and McClay 1988).

The skeletogenic GRN of the purple sea urchin Strongylocentrotus purpuratus (Sp), a euechinoid, proffers a highly detailed account of the specification and differentiation of the PMC lineage (Oliveri et al. 2008). Multiple lines of experimental evidence support the hypothesis that this developmental GRN is highly conserved amongst euechinoids (Ettensohn 2013). Recently, similar investigations have been extended to cidaroids (Erkenbrack and Davidson 2015; Yamazaki et al. 2014). For example, a recent study investigated expression and function of genes at the top of the euechinoid PMC GRN in the Atlantic Basin-dwelling cidaroid Eucidaris tribuloides (Et) (Erkenbrack and Davidson 2015). Functional interrogation of these genes in Et revealed nine specific regulatory inputs that are absent and that are likely to be gain-of-function changes that occurred during the evolution of euechinoids. More specifically, the localization mechanism of skeletogenic factors to the euechinoid PMCs—the double-negative gate—was missing in this cidaroid, suggesting large-scale rewiring of this circuitry occurred in the lineage leading to extant euechinoids. Furthermore, the transcription factors ets1/2 and tbrain were not spatially restricted to the Et skeletogenic cells. This was also the case in the Indo-Pacific-dwelling cidaroid Prionocidaris baculosa (Pb), suggesting a broader role in mesodermal specification for these genes in the cidaroid clade (Yamazaki et al. 2014). In euechinoids, ets1/2 and tbrain are restricted to the PMCs and function as important early inputs into an erg–tgif–hex–alx1 skeletogenic GRN circuit (Oliveri et al. 2008). An additional function of the micromeres in Et is to exclude the skeletogenic fate in the surrounding non-skeletogenic mesoderm (Erkenbrack and Davidson 2015). Many of these findings in Et are consistent with those in Pb, suggesting that many of these observations are conserved amongst cidaroids (Yamazaki et al. 2014). These results beg the question of how manifold the changes to the downstream circuitry are between the cidaroid skeletogenic GRN and the euechinoid PMC GRN.

In examined euechinoids, erg, hex, and tgif form a recursively wired subcircuit downstream of ets1/2 and tbrain that plays a role in stabilizing the regulatory state of the skeletogenic lineage (Oliveri et al. 2008). There is strong evidence that in the echinoderm clade this subcircuit serves as a highly conserved, recursively wired GRN kernel that can be deployed at alternative embryonic addresses to lock down and stabilize regulatory states (McCauley et al. 2010; Davidson and Erwin 2006). Importantly, it is clear that there has been significant GRN rewiring in the lineage leading to euechinoids, in which much of this circuitry has been restricted to the skeletogenic mesoderm as opposed to the broader endomesodermal roles seen in the sea star Patiria miniata (Pm) and mesodermal roles in the sea cucumber Parastichopus parvimensis (Pp) (McCauley et al. 2010; McCauley et al. 2012). Comparative analyses in sea stars and euechinoid sea urchins indicate that the initial onset of this circuit differs between these clades. In euechinoids, ets1 and tbrain set off the erg–hex–tgif regulatory cascade (Oliveri et al. 2008). However, in asteroids, tbrain is the major driver of this subcircuit, suggesting that ets1 is a derived input unique to euechinoids (McCauley et al. 2010). We hypothesized that in Et this conserved circuit would play roles mainly in mesoderm specification, as the drivers of this kernel—ets1/2 and tbrain—are expressed throughout the mesoderm in cidaroids, as opposed to their more restricted roles in the PMCs in euechinoids. Thus, we predicted that in cidaroid sea urchins the spatial expression pattern of this subcircuit would mirror more closely that of asteroids and holothuroids than that of euechinoids. To address these queries, we cloned the genes from the conserved erg–hex–tgif subcircuit and investigated their spatiotemporal expression in Et. We also present here expression dynamics of the skeletogenic GRN gene tel, as similar data in other echinoderms is lacking.

Materials and methods

Cloning and phylogenetic analyses

RNA was extracted from embryos following in vitro fertilization of eggs obtained from adult E. tribuloides collected by Gulf Specimens Marine Lab (Panacea, FL) or Reeftopia (Sugarloaf Key, FL). Complementary DNA (cDNA) was prepared using the SMART RACE cDNA Amplification Kit. Full-length sequences of erg, hex, tgif, and tel were obtained through a combination of cloning and existing transcriptome data available in Echinobase (echinobase.org). DNA binding domains were predicted using ExPASy: Bioinformatics Resource Portal (expasy.org). Multiple sequence alignments were performed using ClustalX 2.1. Phylogenetic reconstruction was carried out using maximum likelihood methods with bootstrap confidence intervals determined by using 1000 replicates. The output was viewed using FigTree 1.4.0.

The following sequences were used to construct the phylogenetic trees: SpErg (SPU_018483), LvErg (retrieved by BLAST in Echinobase), PmErg (GU_251975), CiErg (NM_001078474), SkErg (XM_006822711), BfErg (XM_002613065), Xl (AJ_224125), and GgErg (X_77159); SpHex (SPU_027215), LvHex (retrieved by BLAST in Echinobase), PmHex (GU_251972), CiHex (NM_001078262), SkHex (SQ_431047), BfHex (EU_296398), XlHex (NM_001085590), and GgHex (NM_205252); SpTel (SPU_028479), LvTel (retrieved by BLAST in Echinobase), BfTel (XP_002608583), XlTel (NM_001124423), and GgTel (NM_001199273); and SpTgif (SPU_018126), LvTgif (retrieved by BLAST in Echinobase), PmTgif (GU_251973), CiTgif (XP_002124000), SkTgif (NM_001164980), BfTgif (NP_001071803), XlTgif (NP_001080420), and GgTgif (NM_205379).

Whole-mount in situ hybridization

Probes were synthesized from cDNA using the DIG RNA Labeling Kit (SP6/T7) (Roche). Primers used for whole-mount in situ hybridization (WMISH) probe amplification were designed from full-length coding sequences (Table 1). Embryos were fixed in PFA-MAB (4 % PFA, 32.5 % MFSW, 32.5 mM maleic acid (pH 7), 162.5 mM NaCl) solution on ice and left overnight at 4 °C. Embryos were transferred into hybridization buffer (50 % formamide, 5× Denhardt’s, 5× SSC, 1 mg/ml yeast tRNA, 100 mM NaCl, 0.1 % Tween-20, 50 μg/ml heparin) by series (10 %, 25 %, 50 %, 75 %, 100 %). Probe concentrations were 0.5–1.0 ng/μl. Probe hybridization and post-hybridization stringency washes were carried out at 63 °C. Antibody concentration was 0.25 μg/ml. Detailed procedures are described in Erkenbrack and Davidson (2015).

Table 1 Sequences of WMISH primer sets used in this study
Full size table

Real-time quantitative PCR

Eggs from two different females were used as the starting point for two timecourse cultures. From these, 100 embryos were counted at hourly intervals, gently centrifuged, and lysed with Buffer RLT. To allow quantification of messenger RNA (mRNA) transcripts, each timepoint was spiked with ∼1000 transcripts of synthetic Xeno RNA (Cells-to-Ct Control Kit, Life Technologies) and harvested according to the manufacturer’s protocol (RNeasy, Qiagen). Approximately one embryo per reaction was assayed in triplicate by quantitative polymerase chain reaction (qPCR) (SYBR Green, Life Technologies). qPCR primers used in this study are listed in Table 2.

Table 2 Primer sequences used in this study for quantitative PCR
Full size table

Results and discussion

Cloning and phylogenetic analysis of skeletogenic genes

Homologs of erg, hex, tgif, and tel were isolated from Et. Erg and Tel belong to the ETS (E-twenty six) family of related transcription factors, which is characterized by a DNA binding domain known as the ETS domain with a “winged” helix–turn–helix motif (Seidel and Graves 2002). The pointed (PNT) domain occurs rarely in subfamilies and regulates phosphorylation by serving as a docking site for kinases such as Erk2 (Seidel and Graves 2002). In Sp, there are 11 members of the ETS family with at least one ortholog for each of the subfamilies that have been identified in vertebrates (with the exception of one mammal-specific subfamily) (Rizzo et al. 2006). Genes in this family interact with a diverse array of co-regulatory proteins and serve as both transcriptional activators and repressors (Mavrothalassitis and Ghysdael 2000). In Et, the full-length coding sequence of erg is 1785 base pairs (bp) and encodes a 594-amino acid (aa) protein that includes a PNT domain (161–246) and an ETS domain (425–505) . The full-length coding sequence of tel is 2121 bp and encodes a 706-aa protein that includes a PNT domain (42–126) and an ETS domain (473–554). The predicted aa sequences for Erg and Tel show 69.67 and 55.06 % similarity to their homologs in Sp, respectively.

Hex and TG-interacting factor (Tgif) belong to the homeobox (HOX) family of related transcription factors, characterized by the presence of a DNA binding homeodomain with a helix–turn–helix motif. There are 96 members belonging to various subfamilies in Sp (Howard-Ashby et al. 2006). A member of the NK subfamily, Hex (PRH) is known to act both as an activator and a repressor of transcription (Soufi and Jayaraman 2008), while Tgif is a member of the three-amino acid loop extension (TALE) subfamily, serving mostly to repress transcription, although it can also function as an activator. In Et, the full-length coding sequence of hex is 885 bp and encodes a 294-aa protein that includes a HOX (151–211) domain. The full-length coding sequence of tgif is 1083 bp and encodes a 360-aa protein that contains a HOX domain (44–107). The predicted aa sequences show 64.98 and 70.01 % similarity to their homologs in Sp, respectively.

To confirm the identity of each Et sequence, we carried out phylogenetic analyses in which the cephalochordate Branchiostoma floridae was always the outgroup. In each case, our analyses resulted in a phylogenetic tree that confirmed the orthology of our sequences and their placement as anciently diverged with respect to euechinoids within a large clade consisting of ambulacrians (i.e., echinoderms and the hemichordate Saccoglossus kowalevskii) (Fig. 1).

Fig. 1
figure 1

Phylogenetic analyses of Erg, Tgif, Hex, and Tel amino acid sequences from selected deuterostomes to confirm homology with predicted amino acid sequences from the cidaroid Eucidaris tribuloides. a Erg, b Tgif, c Hex, d Tel. N–J trees were constructed in FigTree 1.4.0 following multiple sequence alignments in ClustalX 2.1. Bootstrap values are indicated at the nodes, and scale bars beneath the trees represent the average number of substitutions per site as a measure of evolutionary distance. Species abbreviations: Et Eucidaris tribuloides, Sp Strongylocentrotus purpuratus, Lv Lytechinus variegatus, Pm Patiria miniata, Sk Saccoglossus kowalevskii, Ci Ciona intestinalis, Xl Xenopus laevis, Gg Gallus gallus, Bf Branchiostoma floridae (outgroup)

Full size image

Characterization of spatiotemporal expression of Et-erg

Previous studies in Sp showed that erg is initially ubiquitous and subsequently restricted to the PMCs by 21 h post fertilization (hpf). Upon ingression of the PMCs, Sp-erg is activated in the oral non-skeletogenic mesenchyme (NSM), where it functions in blastocoelar cell fate (Materna et al. 2013; Solek et al. 2013). Pm-erg is initially expressed at blastula stage in mesoderm precursors at the center of the vegetal pole (McCauley et al. 2010), and is localized during gastrulation to the bulb of the archenteron with some expression in ingressing mesenchymal cells. Similar expression is also observed in the holothurian Pp (McCauley et al. 2012). In Et, our data indicate that, as in euechinoids, erg is not maternally expressed. Both qPCR and WMISH data indicate that zygotic expression in Et begins specifically in a few cells at the center of the vegetal pole 4–5 hpf prior to gastrulation (Fig. 2(a, e, e′)). This localized expression is similar to that for Sp-erg, which begins to be expressed in the PMCs around 15 hpf (Rizzo et al. 2006). In Et, this restricted initial expression pattern is reminiscent of mesodermal genes at the top of the euechinoid PMC GRN that begin their expression in the micromere descendants, including that of alx1, ets1/2, and tbrain, which begin to be expressed at 4, 8, and 6 hpf prior to erg, respectively (Erkenbrack and Davidson 2015). Et-erg expression is seen in the entire mesoderm by 24 hpf; later, expression is observed in skeletogenic mesenchyme ingressing into the blastocoel (Fig. 2(f, f′)). Moreover, staining persists at 60 hpf not only at the tip of the archenteron and in all mesenchyme but also in three to five cells that have migrated to the ventral lateral clusters and which likely constitute the skeletogenic mesenchyme (Fig. 2(g′)). These data are consistent with the hypothesis that Et-erg potentially is a broad mesodermal regulator that is downstream of the first wave of mesodermal GRN circuitry (Erkenbrack and Davidson 2015). Though the initial input of erg is not known, it is highly probable there is an intermediate regulator between the initial mesodermal regulators ets1/2 and tbrain, since these genes are already expressed in many more cells than the number of cells in which erg begins to be expressed. Minimally, these data further substantiate the claim that erg played an ancestral role in mesoderm specification in eleutherozoans, and possibly all echinoderms, as it is expressed in this tissue in all taxa where spatial expression data is available and thus is a crucial cog in GRN circuitry of mesodermal specification (Table 3).

Fig. 2
figure 2

Spatiotemporal expression of the erg–hex–tgif subcircuit and the euechinoid skeletogenic gene tel in the early development of the cidaroid Eucidaris tribuloides. a–d mRNA expression sampled approximately every 3 h over the course of the first 30 h post fertilization (hpf) of E. tribuloides development as revealed by quantitative PCR. a qPCR profile of erg showing it is first detected between 15 and 17 hpf. b qPCR profile of tgif showing it is maternal and is first detected between 8 and 14 hpf. c qPCR profile of hex showing it is first detected between 10 and 15 hpf. d qPCR profile of tel showing it is maternal and is first detected between 10 and 14 hpf. e–m Whole-mount in situ hybridization (WMISH) of erg, tgif, hex, and tel at late blastula, early gastrula, and mid-gastrula. Gene of interest, view of the embryo, and hours post fertilization (h) are indicated on each image. e–g, e′–g′ WMISH of erg at four timepoints in development. e, e′ At 18 hpf, erg is expressed in a few cells at the vegetal pole. f, f′ At 26 hpf, erg is expressed in all invaginating cells (mesoderm) and is also expressed in skeletogenic mesenchyme (arrow). g At 40 hpf, erg is expressed throughout the mesoderm and ingressing mesenchyme. g′ At 60 hpf, erg expression is present at the tip of the archenteron and in migrating mesenchyme, including mesenchymal cells that have migrated to the ventral lateral cluster (skeletogenic cells, arrows). h–j, h′–j′ Expression of tgif at four timepoints in development. h, h′ At 22 hpf, tgif is expressed in a few cells at the vegetal pole. i, i′ At 24 hpf, tgif is now seen throughout the mesoderm. j By 28 hpf, tgif is expressed throughout the mesoderm, including skeletogenic cells (arrow) and also in the endoderm. j′ At 60 hpf, tgif is expressed at the tip of the archenteron, in the mid- and hindgut endoderm, and also in cells that have migrated to the ventral lateral clusters (arrow). Tgif is not, however, expressed in most migrating mesenchyme. k, k′ At 28 hpf, hex is expressed throughout the mesodermal domain and is also present in ingressing skeletogenic mesenchyme (arrow). l, l′, m, m′ tel expression at two different developmental timepoints. l, l′ At 22 hpf, tel is expressed in isolated mesodermal cells. m, m′ At 28 hpf, tel expression is seen in the majority of the mesodermal domain but is absent from ingressing skeletogenic cells (arrow). h hours post fertilization, LV lateral view, VV vegetal view, AV apical view

Full size image
Table 3 Spatial expression patterns of erg, hex, tgif, and tel in echinoderms. Data for the cidaroid lineage are from this study. All other data are from previously published work, for which the references are cited
Full size table

Characterization of spatiotemporal expression of Et-tgif

In Sp, tgif is a maternally deposited factor, the zygotic expression of which begins at 16 hpf, where it is initially expressed in both the PMCs and NSM (Howard-Ashby et al. 2006). After gastrulation begins, Sp-tgif is seen in NSM and midgut endoderm. In Pm, tgif expression is first observed broadly in the endomesoderm and by mid-gastrula is observed in the endoderm, whereas in Pp, tgif is expressed first in mesodermal cells and later is seen in endoderm and non-ingressed mesoderm at the tip of the archenteron (McCauley et al. 2010; McCauley et al. 2012). Our qPCR data indicate that Et-tgif is maternally expressed (Fig. 2(b)). Zygotic expression begins by at least 14 hpf, indicating this is the first gene to be zygotically expressed in the erg–hex–tgif subcircuit. Up to gastrulation Et-tgif is restricted to a subset of cells in the mesoderm, after which point it comes to be expressed throughout the mesoderm by early gastrula and then subsequently in mesoderm and endoderm by mid-gastrula (Fig. 2(h, i, j)). By late gastrula, very faint expression can be seen in migrating mesenchyme cells, weak expression is seen in mesodermal cells at the tip of the archenteron, and strong expression occurs in the mid- and hindgut—while expression is conspicuously absent from the foregut (Fig. 2(j′)). These data indicate that, in all eleutherozoans examined, tgif has an early, spatially restricted role in mesodermal specification and, later, a distinct role in endoderm specification. However, only in echinozoans is tgif restricted to a mesodermal linage early in development. Given that the role of tgif in endodermal specification appears conserved in all eleutherozoans, these data suggest that following the echinozoan–asterozoan divergence (481 mya; Jell 2014), this endodermal activity was activated later in embryonic development in echinozoans and is a derived feature of this clade.

Characterization of spatiotemporal expression of Et-hex

Zygotic expression of Sp-hex, which is not maternal in Sp, begins prior to hatching in the early blastula and is observed in the PMCs by 20 hpf (Poutska et al. 2007). Similarly to Sp-erg, Sp-hex expression is also activated in the oral NSM upon ingression of the PMCs (Materna et al. 2013). By late gastrula stage, expression is observed in PMCs and SMCs but not endoderm. Pm-hex is initially expressed at blastula stage throughout the presumptive endomesoderm at the vegetal pole. During gastrulation, Pm-hex continues to be weakly expressed in both the endoderm and the mesoderm with the exception of some cells at the tip of the archenteron (McCauley et al. 2010). Our data suggest that, like in Sp, hex is not maternally expressed in Et (Fig. 2(c)). Zygotic expression of Et-hex begins concurrently with Et-erg after hatching in the late blastula by 16 hpf (Fig. 2(c)). By 28 hpf, spatial expression occurs broadly in mesodermal cells and also occurs in cells ingressing into the blastocoel, very likely overlapping in its expression with Et-erg (Fig. 2(k, k′)). The spatiotemporal expression patterns of Et-hex are indicative of the possibility that it shares inputs with erg. Unlike in sea stars, however, these data indicate that Et-hex is not expressed in the endoderm up to the time that the skeletogenic mesenchyme ingresses at early gastrula. Indeed, this is also the case in euechinoids (Poutska et al. 2007). These data suggest that in early embryonic development of echinoids, hex is strictly mesodermal, whereas in asteroids it functions in both endoderm and mesoderm. Additionally, even though the spatial expression pattern of hex is not known in Pp, comparative analyses of spatiotemporal expression patterns from three taxa predict that the expression pattern of hex in this holothuroid will mirror that of erg and tgif in Pp.

Initiation and conserved wiring of erg–hex–tgif subcircuit in Et

Taken together, the spatial expressions of Et-erg, Et-hex, and Et-tgif are remarkable in their congruence, all broadly expressed in—and restricted to—the mesoderm at least until skeletogenic mesenchyme ingresses at 26 hpf. Furthermore, our data specifically show that erg, hex, and tgif are all expressed in the first mesenchyme to ingress in Et (Fig. 2(f, f′, j, k, k′)). These observations lend strong support to the supposition that micromere descendants in Et are homologous to the PMC lineage of euechinoids (Wray and McClay 1988). Additionally, these data suggest that many aspects of the downstream euechinoid PMC GRN circuitry were already in place at the divergence of the two extant echinoid clades. It is also important to note for erg and tgif that their initial activation in Et is restricted to a few cells at the pole of the vegetal plate. While we do not present the necessary experimental evidence to claim the erg–hex–tgif subcircuit initiates solely in alx1-positive cells, our data are very suggestive that its activation begins in micromere descendants and subsequently expands to the surrounding NSM rather than vice versa. However, this being the case, it can be said with certainty that the erg–hex–tgif subcircuit is running in alx1-positive cells at the time of skeletogenic mesenchyme ingression (28 hpf). Thus, by 22 hpf, Et-tgif and Et-erg have come on in cells that are very likely the micromere descendants. Given that ets1/2 and tbrain are known to be upstream of erg and tgif in Sp, their activation by ets1/2 and tbrain in Et must be addressed in future studies. However, the initial activation and spatial restriction of the erg–hex–tgif subcircuit to a few cells at the tip of the vegetal pole occur long after ets1/2 and tbrain have been initiated in the mesoderm, suggesting there are other factors at play in their activation. With that said, as soon as erg and tgif begin to run in the micromere descendants, all of the inputs required to initiate the erg–hex–tgif subcircuit are in place: ets1/2 and tbrain are running in the mesoderm; erg would feed into both hex and tgif; and hex would feed back into both erg and tgif. Lastly, it is worth noting that the repressive function of erg on tbrain, described in Pm (McCauley et al. 2010), must be missing in cidaroids, as erg and tbrain are expressed in overlapping cells at least up to 40 hpf.

Characterization of spatiotemporal expression of Et-tel

While tel is not part of the erg–hex–tgif subcircuit, it was included in the study due to its role immediately downstream of the initial activators of the Sp PMC GRN. In Sp, tel is a maternal factor that begins to be zygotically expressed by 15 hpf in the PMCs, where it remains until at least mesenchyme blastula stage (Rizzo et al. 2006). In Pm, there is no evidence of a tel homolog. We found that tel is maternally expressed in Et and that zygotic expression begins much like erg, hex, and tgif, in a few cells at the center of the vegetal pole; thereafter, it expands to be broadly mesodermal (Fig. 2(d, l, m)). Interestingly, we found that ingressing skeletogenic mesenchyme cells of Et were not positive for Et-tel even though it is clearly visible in the surrounding NSM at 30 hpf (Fig. 2(m, m′)). In Sp, tel is an input into differentiation genes in the PMC GRN (Oliveri et al. 2008). In Et, our data indicate that, while Et-tel is expressed early on in skeletogenic cells, by the time the skeletogenic cells ingress into the blastocoel, tel expression is absent from those cells. These data suggest that the restricted, PMC-specific expression of tel is a euechinoid novelty and likely the result of another euechinoid co-option event.

Evolution of the erg–hex–tgif subcircuit in echinoderms

Comparative analysis of three or more taxa in a monophyletic clade is the gold standard for making claims regarding developmental character state evolution. This is because with only two taxa, it is impossible to establish polarity of characters and thus determine which character states are ancestral and which are derived. Our results buttress and broaden already published data on spatiotemporal expression patterns from three major taxa of echinoderms—asteroids, holothuroids, and echinoids—members of which last shared a common ancestor over 481 mya (Jell 2014). By comparing data from multiple taxa, we can minimally enumerate 12 statements about the embryos of the ancestors of these modern species (Table 4). These statements are incredibly significant, as they afford predictions regarding both timing and classification of evolutionary events that must have occurred in the various evolutionary lineages that led to extant taxa. Data from asteroids and holothuroids allow for establishment of character polarity with regard to the spatiotemporal deployment of the erg–hex–tgif subcircuit in echinoids. That the deployment of the erg–hex–tgif subcircuit appears to be limited to the mesoderm in modern cidaroids, as it is in holothuroids (McCauley et al. 2012), indicates that this subcircuit was likely deployed in the mesoderm of the last common ancestor of cidaroids and euechinoids, e.g., in the embryos of the taxon Archaeocidaris, which lived over 268 mya (Thompson et al. 2015). Furthermore, this rigorously demonstrates that the PMC and NSM restricted expression of the erg–hex–tgif subcircuit in euechinoids is a derived character state and must have arisen in stem-group or early crown-group euechinoids since the euechinoid–cidaroid divergence. Although these data indicate that the broader mesodermal utilization of the erg–hex–tgif subcircuit in cidaroids is basal with respect to the more specific deployment in euechinoids, this is but one set of characters, and phylogenetic analyses indicate that neither the cidaroids nor euechinoids are more ancestral than the other (Thompson et al. 2015).

Table 4 Ancestral state reconstruction for embryos of ancestors of extant echinoderm clades by comparative analysis of spatial gene expression data from three or more taxa
Full size table

Lastly, these results proffer a straightforward explanation as to the evolution of the spatial control of the erg–hex–tgif conserved GRN circuitry in echinoderms (Fig. 3). An interesting hypothesis is that tbrain is the ancestral regulator of the erg–hex–tgif subcircuit in echinoderms. This is consistent with the observations that the embryonic spatial expression of tbrain grades from PMC restricted in euechinoids (Ettensohn 2013), to broadly mesodermal in cidaroids and holothuroids (Erkenbrack and Davidson 2015; McCauley et al. 2012), and to mesodermal and endodermal in asteroids (McCauley et al. 2010). That the erg–hex–tgif subcircuit also exhibits these expression patterns in the same clades suggests that all four of these genes may be recursively wired. Further, tbrain solely regulates this subcircuit in asteroids and partially regulates it in euechinoids in spite of the fact that tbrain is expressed at different development addresses in these clades. Taken together, these observations are consistent with the hypothesis that this subcircuit is downstream of tbrain in eleutherozoans. Developmentally, this suggests that, as proposed by McCauley et al. (2010), the ancestral function of the erg–hex–tgif subcircuit was to stabilize the mesodermal regulatory state.

Fig. 3
figure 3

Co-option of the erg–hex–tgif subcircuit in the echinoderm clade. a–c Gene regulatory network (GRN) diagrams of the erg–hex–tgif subcircuit showing its wiring and embryonic domain of expression during development in three echinoderm clades. GRN diagrams were constructed using the BioTapestry online software suite (biotapestry.org) a GRN diagram of the erg–hex–tgif subcircuit in asteroids based on data presented in McCauley et al. (2010). In asteroids, this subcircuit is expressed in the endomesoderm in early blastulae and tbrain is its early activator. b Hypothetical GRN diagram of the erg–hex–tgif subcircuit in cidaroids, in which it is expressed exclusively in the mesoderm in pregastrular embryos. The transparency and question mark indicate that the cis-regulatory inputs between these genes have not been verified by perturbation analysis. The early activator of this GRN is not known. c GRN diagram of the erg–hex–tgif subcircuit in euechinoids, as presented in McCauley et al. (2010). In euechinoids, this circuit is expressed exclusively in the PMCs, a population of mesodermal cells that give rise to the embryonic skeleton, and its early activators are tbrain and ets1/2. d Phylogeny depicting the geological era of divergence and co-option events of the erg–hex–tgif subcircuit in three echinoderm clades in which this subcircuit has been investigated. Divergence estimates are taken from Thompson et al. (2015) for the euechinoid–cidaroid divergence and Jell (2014) for the asterozoan–echinozoan divergence. Red bars indicate a co-option event, and the embryonic address to which the subcircuit was co-opted is also indicated

Full size image

Our data lend support to this hypothesis and also add weight to the hypothesis put forward by Gao and Davidson (2008), that whole apparatus of ancestral mesodermal GRN circuitry were loaded into the micromere embryonic address in the lineage leading to modern euechinoids following the euechinoid–cidaroid divergence—a truly remarkable case of evolutionary co-option (Fig. 3). Phenomena like those observed here are best explained by the observation that GRNs are fundamentally hierarchical and modular in nature (Davidson and Erwin 2006). The erg–hex–tgif kernel in the early embryogenesis of these echinoderms provides an extraordinary example of the modularity and clade-specific functions of GRNs in evolution and development. The correspondence of spatial expression of the erg–hex–tgif kernel to ets1/2 and tbrain in these disparate clades, sea stars, sea cucumbers, and sea urchins, suggests that, even though these organisms last shared a common ancestor over 481 mya, the regulatory embrace they find themselves locked in is so difficult to genomically disentangle, that during evolution they are deployed differentially around the embryo as a parcel.