Deuterostome phylogenetic relationships

Deuterostome phylogenetic relationships have been reviewed extensively elsewhere (Schaeffer, 1987; Cameron et al, 2000; Bourlat et al, 2003; Smith et al, 2004; Blair and Hedges, 2005; Zeng and Swalla, 2005; Delsuc et al, 2006) so will be briefly discussed here. Deuterostomes are monophyletic, and include two great clades: Ambulacraria, which consists of Echinodermata and Hemichordata (Figure 1I; Cameron et al, 2000; Peterson, 2004; Smith et al, 2004), and Chordata, which consists of Tunicata, Cephalochordata (lancelets), and Vertebrata (Figure 1II; Cameron et al, 2000; Zeng and Swalla, 2005). Xenoturbellida, a potentially new phylum of the Deuterostomia that has been described, is thought to be related to the Ambulacraria, but the exact placement within the deuterostomes is not yet clear (Bourlat et al, 2003).

Figure 1
figure 1

Current deuterostome phylogeny, with the three major invertebrate clades marked on the right: Echinodermata, Hemichordata and Tunicata. Vertebrates and Cephalochordata (lancelets) form a fourth clade, Chordata. Ciliated Ambulacraria larvae (I) and Tunicata tadpole larvae (II) are likely to have separate origins. Uncertainties in the Tunicata phylogeny are marked by dotted lines. Modified from Zeng and Swalla (2005).

Full size image

There is a plethora of evidence for echinoderms and hemichordates as sister groups (Peterson, 2004; Smith et al, 2004; Zeng and Swalla, 2005), so features that hemichordates share with chordates were likely to have been present in the deuterostome ancestor (Gerhart et al, 2005; Rychel et al, 2006). There is both molecular and morphological evidence that the Ambulacraria are monophyletic (Cameron et al, 2000; Peterson, 2004; Smith et al, 2004). Similarities in the larvae of echinoderms and hemichordates have been noted for years (Figure 2; Morgan, 1891; Dautov and Nezlin, 1992; Strathmann and Eernisse, 1994; Nielsen, 1997) and echinoderms and hemichordates have recently been shown to share motifs in three posterior Hox genes, Hox 11/13a, 11/13b and 11/13c, as discussed later (Peterson, 2004). This raises the interesting question of how the two phyla have such different adult body plans, when they have such similar larvae.

Figure 2
figure 2

Deuterostome larvae, showing (a) a sea star echinoderm larvae, (b) a hemichordate tornaria larva and (c) a tunicate larva, all oriented with the mouth to the left and anus to the bottom. (a) Sea star larvae have an anterior (top left), which was the original animal pole of the egg. (b) Anterior in the hemichordate tornaria larva is the apical tuft (top of photo). (a, b) Both of these larvae feed with ciliary beating and have well-developed guts and coeloms. The mouth of the sea star and hemichordate larvae are seen to the left (arrow). The posterior anus forms at the former vegetal pole (arrows at bottom). In hemichordates, the larval mouth becomes the adult mouth and the proboscis develops anterior to the mouth. The gill slits and abdomen of the worm will develop posteriorly. (c) The tunicate larva is nonfeeding and lacks a heart, blood and gut, which will develop after metamorphosis. An arrow marks the anterior, where the mouth will form after metamorphosis, but is not yet open. There is no anus at this stage.

Full size image

Conversely, within chordates, Tunicata is the only subphylum that is classified by larval, rather than adult characteristics (Zeng and Swalla, 2005). Tunicates are monophyletic (Swalla et al, 2000) and have a unique adult body plan, including the cellulose tunic, as discussed in Zeng and Swalla (2005). It is difficult to place tunicates reliably within the deuterostomes, due to the long branches found for most of their genes (Winchell et al, 2002; Blair and Hedges, 2005; Zeng and Swalla, 2005). Interestingly, new genome analyses suggest that tunicates are more related to vertebrates than cephalochordates, but these results could be an artifact of incomplete data from cephalochordates and hemichordates (Blair and Hedges, 2005; Philippe et al, 2005; Delsuc et al, 2006). Understanding the position of the tunicates within the deuterostomes will require continued phylogenetic and genomic analyses, coupled with careful studies of evolutionary and developmental processes, including analyses of gene networks (Davidson and Erwin, 2006).

Early development in the deuterostomia

All deuterostomes gastrulate at the vegetal pole, thus the blastopore is formed at or near the vegetal pole, later becoming the anus (Chea et al, 2005). However, the chordate larvae of ascidians have completely different structures and functions than the larvae of echinoderms and hemichordates (Figure 2; Ettensohn et al, 2004). Many echinoderm and hemichordate species have feeding larvae that capture food by ciliary motion and can spend months feeding in the plankton (Figure 2; Dautov and Nezlin, 1992; Strathmann and Eernisse, 1994; Nielsen, 1997). On the other hand, chordate ascidian larvae are nonfeeding, and must metamorphose in order to be able to feed (Figure 2; Davidson et al, 2004). We believe that these larvae have independent evolutionary origins (Zeng and Swalla, 2005). Ascidian embryos and larvae share many genetic pathways with chordate embryos (Passamaneck and Di Gregario, 2005), while echinoderms and hemichordates share similar genetic pathways during the embryonic and larval stages (Shoguchi et al, 1999). In hemichordates and indirect-developing echinoderms, the animal–vegetal axis of the egg becomes the anterior–posterior axis of the larva, so a mouth is formed secondarily at the location where the archenteron contacts the ectoderm (Figures 2, 3 and 4; Chea et al, 2005). In contrast, in chordates, gastrulation results in the movement of large amounts of mesoderm into the archenteron, in order to form the notochord and the surrounding muscular somites, so the anterior–posterior axis lies at a right angle to the animal–vegetal axis (Chea et al, 2005).

Figure 3
figure 3

Echinoid development (a). Early blastula shows a thickening at the vegetal pole (bottom of the photo), where gastrulation will begin. (b) The archenteron invaginates at the vegetal pole (bottom of the photo), which will become the anus, and the mouth is formed to the left side in this photo, where the archenteron touches the ectoderm. This is also the site of nodal and brachyury expression in sea urchins (Duboc et al, 2004; Peterson et al, 1999b). (c) Later, the adult rudiment is formed where the ectoderm touches the larval coelom on the left side of the larva (arrow). This is opposite the second site of nodal expression in sea urchin larvae (Duboc et al, 2005).

Full size image
Figure 4
figure 4

Sea star development (a). Sea stars gastrulate similarly to sand dollars, with the vegetal pole the site of archenteron formation (bottom of the photo), although the mouth is formed more at the midpoint of the larva. (b) Later, in the bipinnaria larva, the adult rudiment is seen as new pigmentation, forming in the posterior of the larva. (c) The advanced brachiolaria larva shows the anterior–posterior larval organization, while the radial adult is formed at the posterior of the animal (arrow). After metamorphosis, the tiny new radial sea star will engulf the rest of the bilateral larva.

Full size image

Different adult body plans built from similar larvae

Hemichordate tornaria larvae are similar morphologically to the bipinnaria larvae of sea stars and the auricularia larvae of sea cucumbers (Figures 2, 3 and 4; Dautov and Nezlin, 1992; Strathmann and Eernisse, 1994; Urata and Yamaguchi, 2004). This type of larva, with a distinct gut and three coeloms has been called collectively a dipleurulid larva (Nielsen, 1997). Development of a hemichordate planktonic larvae and its complete metamorphosis into an enteropneust worm, Balanoglossus misakiensis, has recently been published (Urata and Yamaguchi, 2004). This study documents larval and adult structures throughout early development and metamorphosis in an enteropneust worm with tornaria larvae (Urata and Yamaguchi, 2004). The authors show that the hemichordate adult body plan is elaborated onto the larval body plan (Urata and Yamaguchi, 2004), as suggested by earlier studies (Morgan, 1891; Hadfield, 1975). Balanoglossus misakiensis embryos gastrulates at the vegetal pole and the blastopore at the vegetal pole becomes the anus (Urata and Yamaguchi, 2004; Chea et al, 2005). The mouth is formed secondarily to the side when the archenteron touches the ectoderm, similarly to echinoderm embryos (Figures 3 and 4). An apical tuft is formed at the anterior, where the animal pole was located in the fertilized egg (Figure 2; Hadfield, 1975). After metamorphosis, the nerves of the apical tuft degenerate and the proboscis coelom is formed, developing a proboscis at the very anterior of the animal (Figure 2; Urata and Yamaguchi, 2004). The larval mouth becomes the adult mouth of the worm, located in the collar region, while the posterior of the enteropneust worm is elaborated by growth posterior to the neck region (Urata and Yamaguchi, 2004). In summary, in enteropneust hemichordates, the adult body plan is dramatically different inmorphology from the larval body plan, but the adult retains the same anterior–posterior and oral–aboral axes.

In contrast, indirect-developing echinoderms exhibit a radical metamorphosis, where the axes of the adult body plan do not necessarily correspond to the larval body axes (Figures 3 and 4). For example, in echinoids, the touching of invaginating ectoderm and coelom will mark where the adult rudiment forms inside of the pluteus larva (Figure 3). The same pattern is seen in sea stars as shown in Figure 4. The adult sea star, with fivefold symmetry, emerges from a bilateral larva. The larval anterior was the original animal pole of the fertilized egg, and the posterior is the anus, where the blastopore was formed (Figure 4). Smith et al (2004) discuss how it is likely that the echinoderm ancestor was a bilateral adult and that the prevalence of pentaradial symmetry in extant forms involved a mass extinction of many clades of echinoderms.

In summary, if the Ambulacraria were classified on larval morphology, they would be considered one phylum. Instead, they are divided into separate phyla, echinoderms and hemichordates, based on adult morphology. The echinoderms have a derived body plan, many groups show pentaradial adult symmetry, and all extant echinoderms lack gill slits (Smith et al, 2004). It has been noted that hemichordates share gill slits with the chordates, so the deuterostome ancestor was likely a benthic worm-like creature with gill slits (Cameron et al, 2000; Bourlat et al, 2003; Peterson, 2004; Gerhart et al, 2005; Zeng and Swalla, 2005; Rychel et al, 2006). When one examines chordate features, hemichordates share at least three of these morphological characters: the gill slits, an endostyle and a postanal tail (Gerhart et al, 2005; Rychel et al, 2006). The simplest interpretation of these results is the chordate ancestor was worm-like, with an endostyle, a postanal tail and gill slits. The key chordate innovation, then, was the evolution of a notochord and a dorsal neural ectoderm with the resulting loss of an ectodermal nerve net in the adult (Swalla, 2006).

Coloniality is a fast track to new body plans

In both hemichordates and tunicates, very different body plans evolved by switching from a solitary to a colonial life history and perhaps even vice versa, from colonial to a solitary mode (Cameron et al, 2000; Zeng and Swalla, 2005; Zeng et al, 2006). Hemichordate phylogenies have shown that the colonial pterobranchs may be more closely related to the Harrimaniids, which have direct developing larvae, than to the Ptychoderids (Figure 1; Halanych, 1995; Cameron et al, 2000). Tunicate phylogenies show that the evolution of coloniality has occurred several times independently in tunicates (Swalla et al, 2000) and within ascidians (Wada et al, 1992; Zeng et al, 2006; Yokobori et al, 2006).

There are several consequences of evolving a colonial lifestyle from a solitary one, which affect adult axes and mode of reproduction (Figure 5; Davidson et al, 2004). First, there is usually a miniaturization of the adult when comparing closely related solitary and colonial species (Kardong, 2006; Zeng et al, 2006). This is also normally accompanied by changes in polarity, as colonial species form a superstructure of individuals (Zeng et al, 2006). Finally, in colonial species, asexual reproduction goes on continuously by budding to form new individuals (Figure 5; Kardong, 2006). In contrast, solitary species reproduce solely by sexual reproduction (Figure 5; Davidson et al, 2004). Colonial species also brood their larvae, and spawn competent larvae that are ready to settle (Davidson et al, 2004; Zeng et al, 2006). These life history and body plan changes, miniaturization, budding, and brooding, are seen in the evolution of coloniality in both hemichordates (Cameron et al, 2000) and tunicates (Davidson et al, 2004).

Figure 5
figure 5

Coloniality in tunicates – a schematic diagram of the development of colonial ascidians. On the left is shown embryonic development into a tadpole larva, going through metamorphosis. On the right is a colonial ascidian in the process of budding. An exact replica of the adult can be formed either by a tadpole or by budding in colonial species. Used with permission from Kardong (2006).

Full size image

There are no known extant deuterostome species that can switch from a solitary to colonial lifestyle. In extant deuterostomes, one sees closely related species that are colonial or solitary, and an intermediate social phenotype that may be an evolutionary step in becoming colonial (Zeng et al, 2006). However, in some aplousobranch colonial ascidians, reproduction continues asexually unless the possibility of outcrossing is detected through nonself sperm in the water (Bishop et al, 2000). As basal metazoans (cnidarians) have some solitary species, some colonial species and some species that are capable of switching life histories, it is possible that the deuterostome ancestor was capable of either a colonial lifestyle with asexual reproduction or a solitary lifestyle with sexual reproduction or a combination of both life histories (Davidson et al, 2004). In the evolution of echinoderms and chordates, especially in the cephalochordate/vertebrate lineage, the colonial lifestyle and asexual reproduction capacities were lost, after which all species became solitary and exclusively sexual.

In addition, it appears that echinoderm larvae have the ability to clone themselves when they are in the plankton for long periods of time (Eaves and Palmer, 2003; Knott et al, 2003). Under controlled laboratory conditions, the clones have been shown to be able to metamorphose and form normal juveniles (Eaves and Palmer, 2003; Knott et al, 2003). It is not yet known whether hemichordate tornaria larvae also have the ability to clone, but it would be an interesting finding. Solitary tunicate larvae and adults have not been reported to be able to clone, but colonial tunicates can form a new individual from single epidermal ampullae (Rabinowitz and Rinkevich, 2003). This capacity for cloning and regeneration in nonvertebrate deuterostomes has not received the attention and interest that it deserves. It is likely that the genetic programs involved in larval cloning and colonial budding will be similar to the programs necessary for regeneration and stem cell renewal (Laird et al, 2005). These processes will likely share genetic pathways across the deuterostomes, including vertebrates and humans.

Hox developmental gene expression in deuterostomes

One of the key innovations during the evolution of vertebrates was the duplication of developmental genes, including the clusters of homeobox gene transcription factors that are involved in anterior–posterior patterning in both protostomes and deuterostomes (Figure 6; Carroll, 1995). While ascidians (Di Gregorio et al, 1995; Spagnuolo et al, 2003) and lancelets (amphioxus) (Holland et al, 1994; Wada et al, 1999) have only a single Hox cluster, tetrapods have four clusters (Wada et al, 1999) and teleost fish have eight clusters (Figure 6; Amores et al, 1998; Meyer and Malaga-Trillo, 1999). In vertebrate embryos (Holland et al, 1994; Carroll, 1995), lancelet (Holland et al, 1994; Wada et al, 1999) and ascidian larvae (Di Gregorio et al, 1995; Spagnuolo et al, 2003) expression of the Hox genes proceeds in a collinear temporally and spatially defined pattern from anterior to posterior (Figure 6). In all of the invertebrate deuterostomes there is a single Hox cluster. In echinoderm and hemichordate Hox clusters (Martinez et al, 1999; Long et al, 2003), the posterior Hox genes share motifs, suggesting that they diverged independently from the posterior Hox genes in the chordates (Peterson, 2004; Figure 6). In hemichordates, the expression of the Hox genes is in an anterior to posterior manner, with the anterior genes being expressed in the proboscis, and the posterior genes expressed in the postanal tail (Lowe et al, 2003; Figure 6). The expression of Hox genes in sea urchins initially did not appear to proceed in a colinear manner during embryonic development (Popodi et al, 1996; Arenas-Mena et al, 1998; Martinez et al, 1999). However, recent studies in a direct developing sea urchin suggest that the oral–aboral axis of echinoid echinoderms is similar to the anterior–posterior axis of hemichordates and chordates (Morris and Byrne, 2005). Furthermore, most of the Hox genes in the cluster are expressed only during the development of the adult, and so far only two show any early larval expression (Arenas-Mena et al, 1998; Martinez et al, 1999; Morris and Byrne, 2005). For Hox cluster members 6–11/13, expression is detected during late larval development in nested domains of the posterior coeloms; cluster member six is expressed in the anterior part and 11/13 in the posterior part with the intervening genes exhibiting overlapping domains (Martinez et al, 1999). These results suggest that the larvae of echinoderms and hemichordates may have evolved secondarily in the Ambulacraria clade of the deuterostomes. It is interesting that Hox 1, the most anterior Hox gene, is expressed just after the first gill slit forms in hemichordates and in vertebrates, suggesting that the positioning of the gill slits along the anterior–posterior axis is homologous (Lowe et al, 2003; Gerhart et al, 2005; Rychel et al, 2006; Figure 6).

Figure 6
figure 6

Expression of Hox genes in deuterostomes – the Hox gene cluster is duplicated in vertebrates. There are eight Hox gene clusters in teleost fishes, showing an additional duplication from the four Hox gene clusters found in the tetrapod vertebrates. In contrast, the invertebrate deuterostomes each have a single cluster. Ascidians lack some of the middle Hox genes, and the cluster is broken up onto two chromosomes. Echinoderms and hemichordates share an independent duplication of the posterior genes, called Hox 11/13a, Hox 11/13b and Hox 11/13c. Hemichordates show anterior to posterior expression in the ectoderm, which will produce a nerve net later in development. Echinoderms show adult expression in the nerve ring with the oral side corresponding to anterior in chordates and hemichordates.

Full size image

Nodal gene expression and left–right asymmetry

Nodal is a member of the TGFβ superfamily of signaling molecules found in all phyla of deuterostomes, but nodal has not yet been reported in the Ecdysozoa or Lophotrochozoa (Chea et al, 2005). Nodal and the entire nodal signaling cascade of genes is expressed early during gastrulation during the formation of mesoderm in vertebrates and lancelets as reviewed in Chea et al (2005). Later, nodal is expressed on the left side of chordate embryos and is necessary for left–right asymmetry (Chea et al, 2005). In light of these results, the expression and function of nodal in echinoderms and hemichordates is very interesting.

Nodal expression has been reported in sea urchins at the site where the mouth forms during gastrulation, and has been shown to pattern the oral–aboral axes in the developing embryo and pluteus larvae (Duboc et al, 2004; Flowers et al, 2004; Figures 2, 3 and 4). Later, nodal is also expressed opposite the site of the adult rudiment formation in larval urchins, which normally forms on the left side of the larvae (Duboc et al, 2005; Figures 2, 3 and 4). It is intriguing that nodal serves multiple functions during deuterostome development, a theme that is fundamental to the evolution of developmental processes (Davidson and Erwin, 2006). Nodal expression is found in the developing mouth in the direct-developing hemichordate, Saccoglossus kowalevskii, similar to the early echinoderm expression (Christopher J. Lowe, personal communication). Later expression of nodal in hemichordates might be interesting and informative as to how the hemichordate left–right axes correspond to chordate axes (Chea et al, 2005; Gerhart et al, 2005).

Brachyury gene expression

A novel tissue, the notochord, unites the chordates as a monophyletic group (Figures 1 and 7). Notochord is necessary as a developmental signaling tissue during neural tube formation and also remains as a structural tail tissue in ascidian larvae and cephalochordates (Smith and Schoenwolf, 1989). One transcription factor known to be necessary for notochord development in chordate embryos is the T-box transcription factor brachyury T (Holland et al, 1995). Brachyury T is expressed in developing notochord cells in vertebrates (Wilkinson et al, 1990), lancelets (Holland et al, 1995), ascidians (Yasuo and Satoh, 1993, 1994) and larvaceans (Bassham and Postlethwait, 2000) (Figure 7). In lancelet and vertebrate embryos, brachyury T is expressed early in presumptive mesoderm, and later in the notochord and posterior mesoderm (Holland et al, 1995; Figure 7). However, in ascidians, brachyury T expression was seen only in the notochord, at the time of cell fate restriction (Yasuo and Satoh, 1993, 1994). As brachyury T is expressed exclusively in the ascidian notochord (Yasuo and Satoh, 1993, 1994), it was believed that the expression of brachyury T in echinoderms (Shoguchi et al, 1999; Peterson et al, 1999b) and hemichordates (Tagawa et al, 1998; Peterson et al, 1999a) might allow clues from which tissues the notochord evolved (Figure 7). However, the results of brachyury T gene expression underscore the differences in morphology between nonfeeding ascidian tadpole larvae and feeding larvae of indirect-developing hemichordates and echinoderms (Figure 2).

Figure 7
figure 7

Expression of Brachyury (T) in animals. Brachyury is expressed in the hindgut in flies and in the gut during larval development in polychaete worms. In echinoderms and hemichordates, brachyury is expressed in the anus during gastrulation, then later where the mouth is formed. Brachyury was co-opted into the notochord in chordates, but in larvaceans, it is expressed in notochord and later in the mouth and anus after metamorphosis.

Full size image

Brachyury T expression in sea urchins was found early in the vegetal plate at the mesenchyme blastula stage, later in secondary mesenchyme after gastrulation and was absent in sea urchin larvae (Peterson et al, 1999b). As metamorphosis began, expression was seen in the mesoderm of the left and right hydrocoels in the developing adult urchin (Figure 7; Peterson et al, 1999b). In contrast, the pattern of Brachyury T expression in sea star larvae (Shoguchi et al, 1999) was similar to expression in hemichordate larvae (Tagawa et al, 1998; Peterson et al, 1999a; Figure 7). In hemichordate embryos, expression of brachyury T was first seen in the vegetal plate early in gastrulation (Tagawa et al, 1998). High expression continues at the vegetal plate and finally is still expressed during development of the larval anus (Figure 7; Tagawa et al, 1998; Peterson et al, 1999a). Expression is also seen in the larval mouth from the time of induction by the archenteron (Peterson et al, 1999a). Later, after metamorphosis, expression is high in the adult proboscis, collar, and the posterior region of the trunk and gut (Figure 8; Peterson et al, 1999a). Therefore, the expression of brachyury T in posterior gut is seen in all deuterostomes: chordates, echinoderms and hemichordates (Figure 7). In contrast, notochord expression is restricted entirely to the chordates (Figure 7). This separation of expression patterns in tunicates is documented nicely in expression results of brachyury T in larvaceans (Bassham and Postlethwait, 2000). In larvaceans, a pelagic tunicate, expression is first seen only in the larval notochord (Figure 7; Bassham and Postlethwait, 2000). However, after metamorphosis and the development of the gut in the adult, expression is seen in the oral and anal mesoderm (Bassham and Postlethwait, 2000; Figure 7). These results suggest that if brachyury T expression were examined in ascidians after metamorphosis, then there would be expression in the posterior gut. In summary, brachyury T expression patterns suggest that deuterostomes may share a common mesodermal transcription factor for development of the mouth and anus, but no light is shed on the evolutionary origin of the notochord. A screen for genes that are downstream of brachyury T in ascidians has yielded a number of candidates that will be interesting to clone and characterize in hemichordate and echinoderm embryos (Hotta et al, 2000). Brachyury T is also expressed in the hindgut of flies (Lengyel and Iwaki, 2002) and in the mouth and anus of polychaete larvae (Arendt et al, 2001). One idea is that brachyury T is a general transcription factor for a suite of genes that can be activated to allow cell movement, or convergence and extension (Lengyel and Iwaki, 2002). The other possibility is that brachyury T was originally important for ectodermal–endodermal interactions during the formation of the gut and anus (Technau and Scholz, 2003). Later, in chordates, the continued expression of brachyury T on one side of the blastopore may have induced many more cells to ingress opposite of the endoderm, allowing the evolution of the notochord and somites (Chea et al, 2005). An experiment to test this hypothesis would be to over-express brachyury T in hemichordate larvae on one side of the blastopore and determine whether the body axes of the worm are altered as a result of induced expression.

Summary

Deuterostomes show widely divergent adult body plans in extant taxa. The evolution of these divergent body plans has occurred through different developmental scenarios. Echinoderms and hemichordates share similar larvae, but in hemichordates the adult body plan is built onto the larval body plan, whereas in echinoderms a radical metamorphosis results in dramatically different adults. Coloniality has evolved several times repeatedly in the tunicates and has also evolved in hemichordates. The evolution of coloniality changes a suite of characteristics in larval and adult features in ways not well understood. Tunicate larvae probably evolved independently and may not represent the ‘ancestral’ chordate. Hox genes specify anterior–posterior polarity in the deuterostomes, but the cluster has evolved independently in the Ambulacraria, Tunicata and Chordata (lancelets and vertebrates). Vertebrates have increased mesoderm and may have co-opted brachyury T and nodal into mesoderm for increased cell movement into the blastopore during gastrulation. Further inquiry into the evolution of developmental processes may give us a better understanding of the evolution of divergent deuterostome body plans.