Evolutionary crossroads in developmental biology: sea urchins

Echinoderms and chordates are deuterostomes (see Glossary, Box 1), although echinoderms branched from the chordate lineage prior to the Cambrian period, more than 500 million years ago (Fig. 1A). Nevertheless, this shared relationship places echinoderms in an important position for developmental and evolutionary studies. The term echinoderm means ‘spiny skin’, and the Echinodermata phylum includes sea stars (Asteroidea), sea urchins and sand dollars (Echinoidea), brittle stars (Ophiuroidea), sea cucumbers (Holothuroidea), and sea lilies (Crinoidea) (Fig. 1B). Because their development is relatively simple and rapid, and because gene function in these organisms is easy to perturb and manipulate, echinoderms, especially sea urchins and sea stars, have long been used for testing hypotheses about how early development works.

Specification (see Glossary, Box 1) in echinoderms begins early in development. Zygotic gene transcription begins shortly after fertilization and zygotic mRNA quickly becomes responsible for the diverse specification activities. This is in contrast to the extended delay in the appearance of zygotic mRNA seen in Xenopus and Drosophila. In sea urchin embryos, by the time gastrulation (see Glossary, Box 1) is reached, each cell of the embryo is differentially specified to one of at least 14 early cell fates. Gastrulation then involves a series of complex and dynamic movements that ultimately lead to the formation of the free-floating larval form. In the final stage of development, metamorphosis (see Glossary, Box 1) occurs and the larva transforms into the juvenile sea urchin. Each of these stages in sea urchin development has been studied in great detail, and, most importantly, many of these details have provided us with insights into the conservation of these events across deuterostomes.

This primer provides an overview of the reproduction, biology and development of the echinoderms, focusing on the sea urchin, as used in studies of evolutionary developmental biology. Importantly, this article highlights how discoveries in the sea urchin have provided us with insights into some of the key mechanisms used to build a deuterostome organism.

The echinoderm phylum dates at least to the Cambrian period, about 500-540 million years ago, when crinoids, animals similar to today's sea lilies, appear in the fossil record (Shu et al., 2001). From the Cambrian period onwards, fossil deposits of echinoderms track a good evolutionary history of this ancient and exclusively marine phylum of animals. Echinoderms were included with deuterostomes by early systematists, based on the shared embryonic character of the mouth as the second invagination during development. Genomic sequence data has since verified the deuterostome clade as the correct phylogenetic grouping of these animals (Swalla and Smith, 2008). This poses a basic question: to what extent is the genetic basis of development and morphogenesis conserved among deuterostomes? Research on echinoderms is of great value for addressing this question, as it offers an opportunity to discover remnants of the common template upon which the deuterostome style of development is built.

Most developmental research on echinoderms is restricted to a small number of model sea urchin and sea star species (Fig. 2). On the west coast of the USA, Strongylocentrotus purpuratus is studied at the molecular, cellular and genomic levels, and its genome was the first of this phylum to be sequenced (Sodergren et al., 2006). On the eastern coast of the USA, Lytechinus variagatus is commonly studied and valued for the transparency of its eggs. In Japan, Hemicentrotus pulcherrimus is the most commonly studied species. Around the Mediterranean, Paracentrotus lividus is the sea urchin species of choice for most of the European research community, and was the echinoderm species used by late 19th and early 20th century embryologists for the discovery of many principles of early development (see Box 2). Several sea star species are also studied by developmental and evo-devo labs around the world, including Asterias amurensis and Asterina pectinifera from Japan, Asterias forbesi and Asterina miniata (now called Patiria miniata) in the USA, and Astropecten aranciacus in the Mediterranean.

Adults appear five-sided or pentaradial, although this apparent symmetry is actually superimposed on a true bilaterian underlying body plan (Sprinkle, 1992). Adults have a thin epithelium covering an internal calcium carbonate skeleton. This skeleton consists of many plates that collectively form five ambulacra (see Glossary, Box 1), which extend from the mouth to the anus and are penetrated by many holes through which tube feet extend. The tube feet are part of a water-based vascular system that connects externally via a single opening next to the anus, called the madreporite. Side branches of this system connect to the tube feet, enabling them to extend/withdraw by hydraulic inflation/deflation. Muscles lining the tube feet provide the adult with a mode of movement, and objects are grasped via suckers at the end of the tube feet. The mouth of an adult usually faces downwards towards the substrate and is opened by a unique pentaradial arrangement of muscles and teeth used for biting and scraping food. This arrangement is known as ‘Aristotle’s lantern', after the great scientist and philosopher who described this structure and how it works. Food is digested in a five-part gut that branches from the mouth and then converges again at the anus on the top of the animal. The echinoderm nervous system consists of a diffuse network of neurons with ganglia but there is no central nervous system.

In most species the sexes are separate, and indirect development produces a free-swimming larva that then undergoes metamorphosis, at which point the juvenile emerges with the characteristic fivefold symmetry and grows into the adult. About one-fifth of all sea urchin species follow a direct development strategy. Heliocidaris erythrogramma, for example, gastrulates and reaches the juvenile stage 3 days after fertilization (Wray and Raff, 1989). Such direct developing species generally have large eggs (~300 μm in diameter), allowing them to bypass the larval feeding stage (Parks et al., 1988; Wray and Raff, 1989). Direct developing strategies arose relatively recently in several groups, and because of this it is likely that molecular remnants of the transition from indirect to direct development will be found in the not too distant future. The lifespan of species varies from a few years to more than 200 years (Ebert, 2008). Most echinoderms also have the capacity to develop asexually. Sea

stars, for example, can regenerate a complete animal from an isolated arm or half. Furthermore, sea urchin larvae have recently been found to ‘clone’ themselves (Eaves and Palmer, 2003) as another form of asexual reproduction.

Sea star development is similar to sea urchin development, although the bipinnaria (Fig. 2H) and brachiolaria larvae are not the same shape as the pluteus larva of the sea urchin (Fig. 2E-G). In addition, there is no internal skeleton initially. Furthermore, some sea stars species brood their embryos until metamorphosis.

The gravid season for echinoderms varies greatly between species: some species are gravid year round and other species are gravid for very short periods of time; others spawn with a lunar periodicity. Eggs and sperm of sea urchins are spawned into the seawater and fertilization occurs externally. Development to the larval stage is simple relative to other deuterostomes. Fertilization triggers activation of a high rate of metabolism (Epel, 1975), and the zygote begins a stereotypic radial pattern of cleavage (Fig. 3). The fourth and fifth cleavages are unequal in the vegetal half of the embryo, giving rise to the following groups of micromeres, macromeres and mesomeres (see Glossary, Box 1) arranged in tiers, beginning at the vegetal pole: four small micromeres (fated to the coelomic pouches, see Glossary, Box 1), four large micromeres (the future skeletogenic cells), four macromeres (future endoderm, see Glossary, Box 1, and nonskeletogenic mesoderm) and eight mesomeres (future ectoderm, see Glossary, Box 1, and neural ectoderm) (Fig. 3A-C). By the sixth cleavage, the embryo hollows, forming a blastocoel cavity in the center, and is now called a blastula. At mid-blastula stage, cilia appear and the embryo hatches from its fertilization envelope and enters the plankton community (Fig. 3D). Gastrulation then occurs in two temporally distinct phases. First, the vegetal half of the embryo flattens and the primary mesenchyme cells (the skeletogenic cells) ingress at the center of the vegetal plate (Fig. 3A). A short time after ingression, invagination of the archenteron (see Glossary, Box 1) begins. Non-skeletogenic mesoderm (NSM) cells in the center of the vegetal plate initiate invagination by changing shape to produce an inward bending sheet of cells. The archenteron then elongates to extend across the blastocoel by recruitment of endoderm cells into the archenteron, by elongation via convergent extension movements of cells and by extension of thin filopodia by the NSM cells to help pull the archenteron toward the animal pole (Fig. 3A,E). The archenteron reaches a target site near the animal pole and, shortly thereafter, the oral ectodermal epithelium invaginates to form the stomodeum (see Glossary, Box 1), which then fuses with the foregut epithelium.

To form the pluteus larva, the ectoderm changes shape, driven in part by growth of the underlying skeleton (Fig. 3E). The skeletal rods elongate in a species-specific pattern, whereas the gut subdivides into the foregut, midgut and hindgut, with muscular sphincters forming at the compartment junctions (Fig. 3F). Lateral to the foregut, coelomic pouches, progeny of the non-skeletogenic mesoderm and small micromeres, bulge to either side. The pluteus larva differentiates to produce a number of cells necessary for the skeleton, for neural transmission and for feeding (Fig. 4). The pluteus locomotes and feeds on plankton using coordinated beating of cilia growing in the ciliary band (Fig. 4C). As the larva grows, the rudiment of the adult grows from the left coelomic pouch (Fig. 4B). At metamorphosis, the larva settles and the mature rudiment transforms into the juvenile sea urchin.

By the sixth cleavage, zygotic transcription is active in all cells and specification of the germ layers is well under way. To depict the sequence of specification, models of gene regulatory networks (GRNs, see Glossary, Box 1) in progeny of the macromeres, mesomeres and micromeres are used to show the relationship of

Many approaches enable mechanistic studies of the development of echinoderms, in particular that of sea urchins [discussed in detail by Ettensohn et al. (Ettensohn et al., 2004)]. Adult sea urchins are collected or obtained from commercial shippers when the animals are gravid. Eggs and sperm are obtained from different species either by shaking adults vigorously, by injection of 0.5 M KCl, by isolating gonads from the adults, by electrical stimulation or by neurotransmitter stimulation to cause gamete release. To fertilize eggs, concentrated sperm is diluted in seawater and a few drops of this suspension are added to the eggs. The embryos grow well in artificial seawater, making them appropriate for both laboratory and classroom use, and amenable to many experimental techniques.

Millions of sea urchin eggs can be fertilized at the same time and will develop synchronously, providing ample material for biochemical studies. There are many ways to isolate and purify individual cell types of the embryo for biochemical purification and/or molecular and cell biological analysis. Embryos can be dissociated by removal of calcium (McClay, 2004), and specific cells can be isolated using gradient centrifugation (Hynes and Gross, 1970), lectin adhesive affinity, which selectively isolates skeletogenic cells (Ettensohn and McClay, 1987), or by fluorescent cell sorting following expression of fluorescent reporters that are expressed in specific cell types (Arnone et al., 2004). Large quantities of RNA, DNA, proteins or other cellular components for study can be isolated from such cultures at any stage (Foltz et al., 2004).

Injections can be used to perturb embryos at the molecular level, or to introduce various molecular reagents into eggs or individual cells (Cheers and Ettensohn, 2004). Thousands of eggs, if necessary, can be injected in one sitting for a wide range of molecular studies. For example, the injection of morpholinos (see Glossary, Box 1) can be used to block translation of specific mRNAs (or of splice junctions in RNA processing) and, thus, to assess protein function. The expression of dominant-negative molecules, the ectopic expression of molecules, the overexpression of a molecule, or the inclusion of inhibitors that block specific signals or pathways can similarly be used to assess gene function. The advantage to the investigator is that endpoint assays involving in situ hybridization, antibody staining or PCR allow one to gain an impression of gene function during development. These approaches allow the investigator to assess whether a gene function is necessary and whether it is sufficient. This, in turn, progresses to epistatic studies that aim to determine which protein depends on which and to establish the regulatory relationships between genes. For lineage-tracing studies, individual cells can be injected up to the 16-cell stage with constructs or fluorescent markers. Mosaic approaches allow the investigator to study how the progeny of a single inserted founder cell develop relative to a control founder cell, starting at that same position in the embryo. Immunostaining or in situ analyses at the end of the study then provide spatial information about the expression of a specific protein or mRNA relative to controls. Time-lapse microscopy enables one to follow the trajectory and behavior of fluorescently labeled cells in whole mounts, with the small size and transparency of the embryo allowing it to be visualized in its entirety.

The ability to inject DNA into eggs also provides an easy method for the production of transgenic embryos. If the construct contains an appropriate enhancer and promoter, it will be activated and transcribed in the correct cells and at the correct time in development. Such constructs usually are introduced along with a reporter (Arnone et al., 2004) that allows the visualization of expression of that molecule. This approach has many uses, one of the most common being cis-regulatory analysis (see below). BACs (bacterial artificial chromosomes) that are modified by insertion of a reporter to replace the coding sequence of the gene offer another transgenic approach.

In cis-regulatory analyses, a reporter gene that expresses a fluorescent or colorimetric marker and replaces the protein-coding sequence of the gene is constructed. This gene construct is then injected into eggs and the spatial and temporal expression of the reporter is followed (Arnone et al., 2004). If the reporter

Genomic information enabling cis-regulatory analysis for several species can be found on a frequently updated website (see SpBase at http://www.spbase.org/SpBase/). These databases provide increasingly valuable resource information for cis-regulatory analysis, and for evo-devo studies that examine how genomes compare and how they evolved.

Because of its accessibility, its rapid synchronous development, its amenability to experimental manipulation and its optical transparency, the sea urchin embryo continues to be a valuable model organism for studying the mechanisms that pattern embryonic tissues and for constructing the GRNs that control this early development. With the addition of molecular and genomic information, the sea urchin is a leading model for determining specification mechanisms that establish the major tissue territories in the embryo. Furthermore, the recent advances in GRN construction have allowed us both to understand how these specification events occur and to track how these complex networks have evolved. Some of the key findings enabled by research with this model are discussed below.

Three laboratories independently discovered that the Wnt pathway is crucial for the activation of endomesoderm development in sea urchins (Emily-Fenouil et al., 1998; Wikramanayake et al., 1998; Logan et al., 1999). β-Catenin, a key component of the Wnt signaling pathway, was observed to enter the nucleus of micromeres at fourth cleavage and later entered the nucleus of macromeres and their progeny (Logan et al., 1999). As a transcriptional co-activator, β-catenin was found to be essential for activation of the endomesoderm GRN and further analyses have established that β-catenin activates pmar1 (paired-class micromere anti-repressor, a transcription factor, see Fig. 5) to launch the micromere GRN (Oliveri et al., 2002; Oliveri et al., 2003) and that it activates Wnt8 and homeobox (Hox) 11/13b to initiate the endoderm GRN (Peter and Davidson, 2010).

Elimination of Notch signaling was observed to prevent specification of mesoderm (Sherwood and McClay, 1997; Sherwood and McClay, 1999). It was soon discovered that Delta, a Notch ligand that is produced first by micromeres and later by the non-skeletogenic mesoderm, activates mesoderm specification in two intervals of signaling (Sweet et al., 2002; Croce and McClay, 2010). The pattern of Delta expression is controlled dynamically, allowing progressive expression of Delta in cells more distant from the vegetal pole (Smith and Davidson, 2008).

The Nodal pathway, along with BMP (bone morphogenetic protein) signaling, Lefty and Chordin, was discovered to organize oral-aboral (also called ventral-dorsal) territories of ectoderm (Duboc et al., 2004; Duboc et al., 2005; Duboc et al., 2008; Bradham et al., 2009). An apparent initial asymmetry involving subtle redox differences (Coffman and Denegre, 2007), plus a possibly related asymmetric inactivation of p38 mitogen-activated protein kinase (Bradham and McClay, 2006), are somehow necessary for the asymmetric expression of Nodal on the oral ectoderm. Nodal expression then controls activation of both the oral GRN state and activation of BMP2/4, which is important for aboral ectoderm specification (Angerer et al., 2000; Duboc et al., 2004). Nodal later provides a signaling input that is important for specifying left-right asymmetries (Duboc et al., 2005), although, surprisingly, Nodal operates on the right side (compared with the situation in vertebrates where it operates on the left). In addition, Hedgehog signaling contributes to left-right patterning in the sea urchin (Walton et al., 2009), a property that is also shared with left-right specification in vertebrates (Levin et al., 1995).

A small group of cells at the animal pole organizes into a unique territory of the embryo called the apical plate. This territory is defined during specification of the ectoderm by progressive restriction of FoxQ2 (forkhead Q2) and Six3 (sine oculis-related homeobox 3 homolog) expression (Yaguchi et al., 2008; Wei et al., 2009). Some of the cells within this territory initiate specification of the first neural cells of the embryo (Burke et al., 2006a; Burke et al., 2006b; Yaguchi et al., 2006; Yaguchi et al., 2007; Yaguchi et al., 2008; Wei et al., 2009). The area of neural cell specification is restricted initially to a small region of the animal pole domain by Nodal signaling from the oral side and by BMP signaling from the aboral side (Nakajima et al., 2004; Yaguchi et al., 2007; Bradham et al., 2009), although the details of these regulatory mechanisms are unclear. Neurogenesis also occurs later in a subset of cells in the ciliary band (Saudemont et al., 2010), establishing an elaborate neural network in the pluteus larva (Fig. 4D).

Small micromeres appear at the vegetal pole as products of the unequal fourth and fifth cleavages. These cells divide only once more and then the majority of the eight small micromeres are incorporated into the left coelomic pouch, which becomes the rudiment of the adult. When small micromeres were removed and embryos grown to adulthood, few, if any, gametes were present in the gonad, suggesting that small micromeres contribute to formation of the germ cell precursors (Yajima and Wessel, 2011). Small micromeres also express many of the markers that characterize germ cell precursors in other animals (Juliano et al., 2010; Gustafson et al., 2011), further supporting their role as germline precursors.

Cells of the sea urchin embryo have the capacity to transfate (see Glossary, Box 1) or to switch from one cell type to another (Ettensohn and McClay, 1988; Ettensohn et al., 2007). Interestingly, transfating was discovered to take shortcuts: rather than reverting back to the top of a GRN and repeating the specification sequence, the switch is somehow accomplished by moving to the middle of the network of the alternate fate; the cells then switch their GRN state to the newly adopted cell type (Ettensohn et al., 2007). Transfating is a regulative capacity of many sea urchin embryonic cells, originally discovered more than a century ago (Driesch, 1892). This capacity has been described as a ‘failsafe’ mechanism, at least in micromeres, in which one set of experiments shows reactivation of the micromere network after blocking upstream initiation of specification (Smith and Davidson, 2009). The full range of this regulative development (see Glossary, Box 1) is not understood but must use both signaling and altered expression of transcription factors to detect and then replace missing or incorrectly specified cells.

After a decade of transcription factor and signal discovery, a landmark GRN graphic model was produced based on perturbation analyses of transcription factors and signals used in the early sea urchin embryo (Davidson et al., 2002). Many laboratories added additional nodes and refined existing connections (Angerer et al., 2000; Angerer et al., 2001; Ettensohn et al., 2003; Wikramanayake et al., 2004; Smith et al., 2007; Oliveri et al., 2008; Smith and Davidson, 2008). As the GRN models advanced, cis-regulatory analyses of transcription factors in the network were added to authenticate, correct and expand the relationship between signals and transcription factors that control endomesoderm specification (Yuh et al., 2001; Minokawa et al., 2005; Ransick and Davidson, 2006; Nam et al., 2007; Range et al., 2007; Wahl et al., 2009; Ben-Tabou de-Leon and Davidson, 2010). The skeletogenic cell GRN is shown in Fig. 5, and a website for the current gene regulatory network model and the data that underpin it can be found at http://sugp.caltech.edu/endomes/.

Knowledge of the gene regulatory network that operates in early sea urchin embryo development also has provided a systems-level glimpse of the mechanisms that are used to control subcircuits within the network. Many regulatory mechanisms reported in other systems are included as control devices in the sea urchin GRN circuits. These include the ‘double-negative gate’ (Revilla-i-Domingo et al., 2007), a mechanism by which a repressor represses an early repressor, thereby allowing activation of early genes in the network. Other devices include ‘lockdown’ or ‘feed-forward’ loops, which are mechanisms in which one or more transcription factors in a series feed back to activate another in that series, creating a positive feed-forward regulatory mechanism and often simultaneously providing a lockdown mechanism that tends to prevent network reversal (Davidson and Levine, 2008; Oliveri et al., 2008). A number of self-repression regulatory elements have been discovered whereby a gene first activates a pathway and then, as its expression increases, begins to repress itself (Davidson, 2009). Additional mechanisms have been found to control more complex dynamics (Smith and Davidson, 2008).

A comparative analysis of the endomesoderm GRN between sea urchins and sea stars showed that much of the early circuitry responsible for activation of endoderm is conserved, despite an evolutionary separation from a common ancestor more than 500 million years ago (Hinman et al., 2003; Hinman et al., 2007). More distant comparisons show that many of the genes that operate early in the sea urchin GRN are also likely to govern the activation of GRNs in hemichordates (Lowe et al., 2006; Lemons et al., 2010), and components of the early network may also be conserved in other deuterostomes. The Wnt pathway, as an example, has a role in the vegetal activation of eggs in all deuterostomes studied (Moon et al., 1997; Croce and McClay, 2006). Many other ‘developmental genes’ work together in developmental processes throughout the deuterostomes. Goosecoid, for example, works with Nodal in many organisms, including the sea urchin (Duboc et al., 2004). The processes these genes control may differ, but cassettes of genes seem to be conserved by working together. With comparative knowledge of GRNs, it becomes possible to infer mechanisms of evolutionary change (Davidson and Erwin, 2009), with the sea urchin representing a basal deuterostome. A recent analysis shows that the sea urchin skeletogenic GRN is similar between species as distant as sea urchins and sea stars (Gao and Davidson, 2008). In modern sea urchins, the progeny of the micromeres produces the larval skeleton. However, a recent study shows that non-micromere mesoderm produces juvenile and adult stage skeleton (Yajima, 2007). Those data support the notion that skeletogenesis is an ancient function of mesoderm and, in euechinoids, the skeletogenic cells co-opted the skeletogenic GRN and precociously produced the larval skeleton.

The biggest limitation in echinoderm research is the relative inability to use genetics as a research tool. The generation time (egg to egg) is too long for mutagenesis to be used practically as a tool for genetic discovery of mechanism, and handicaps the establishment of permanent transgenic lines. This limitation is of much less significance than a decade ago when the current array of molecular tools was unavailable. Another limitation is that there are no echinoderm cell lines. Therefore, processes can be studied effectively only in vivo. Fortunately, the simplicity of the embryo provides an excellent opportunity for the discovery of the mechanisms that govern many developmental processes, such as fertilization, cleavage asymmetry, epithelial-mesenchyme transitions and morphogenetic movements, all processes that cannot be imitated in cell culture.

Future research in echinoderm biology, as in every other system, will benefit from advances in technology. Additional sequenced genomes will inform cis-regulatory analyses and will improve our understanding of the function of non-coding sequences. Furthermore, comparative genomic approaches will expand our evolutionary insights. High-throughput sequencing and increased speed and sophistication of software will accelerate methods that currently are slow. The recently developed ability to perform cis-regulatory analyses on many genes at once, rather than laboriously dissecting one gene at a time (Nam et al., 2010), is an example of that rapidity, and will enable cis-regulatory analysis to be a central component of systems biology and of genomic analysis. Expanded databases will also enable comparisons between control and experimentally perturbed cells. Hopefully, tools will become available to introduce specific perturbations at any time and into at any location in the embryo. This will allow embryos, with the exception of the perturbed cell/tissue, to develop normally except for the time and place of the perturbation, thereby enhancing our ability to understand the mechanisms behind specific developmental events or the production of distinct structures during development. Furthermore, improved imaging technologies will expand our capacity to examine cellular function in this transparent embryo; with new fluorescent probes, an ability to uncage probes and novel fluorescent reporters available, we will be better able to address molecular function in the regulation of both gene expression and signaling. Such experimental advances should also provide the tools with which to dissect more thoroughly morphogenetic processes and evolutionary mechanisms.

Early specification events and the control of morphogenesis are rapidly being understood at a molecular and systems level in the sea urchin. The penetration of the complexity of early development in this relatively simple system provides important information about sea urchin development and, importantly, enables comparative studies with other deuterostomes. Many recent discoveries provide seminal insights into conserved developmental mechanisms, as well as providing examples in which processes appear to be co-opted for novel function. Much remains to be learned, however, about how specification works and how it controls morphogenesis, how morphogenetic movements are conducted at a cell biological level, how evolution acts on embryos, and, finally, how variation in embryonic development acts on evolution.

The author appreciates input from lab members and from anonymous reviewers, and images provided by a number of colleagues. The author's research is support by the NIH. Deposited in PMC for release after 12 months.