One of the most remarkable larval types among spiralians, and invertebrates in general, is the planktotrophic pilidium. The pilidium is found in a single clade of nemerteans, called the Pilidiophora, and appears to be an innovation of this group. All other nemerteans have either planktotrophic or lecithotrophic juvenile-like planuliform larvae or have direct development. The invention of the pilidium larva is associated with the formation of an extensive blastocoel that supports the delicate larval frame and elaborate ciliary band. Perhaps the most striking characteristic of the pilidium is the way the juvenile worm develops inside the larva from a series of isolated rudiments, called the imaginal discs. The paired cephalic discs, cerebral organ discs, and trunk discs originate as invaginations of larval epidermis and subsequently grow and fuse around the larval gut to form the juvenile. The fully formed juvenile ruptures the larval body and, more often than not, devours the larva during catastrophic metamorphosis. This review is an attempt to examine the pilidium in the context of recent data on development of non-pilidiophoran nemerteans, and speculate about the evolution of pilidial larval development. The author emphasizes the difference between the planuliform larvae of Palaeonemerteans and Hoplonemerteans, and suggest a new name for the hoplonemertean larvae—the decidula.
Nemerteans are good spiralians
The nemerteans, commonly referred to as ribbon worms, constitute a separate phylum of marine invertebrates with ∼1275 formally recognized species (Kajihara et al. 2008). The vast majority of nemerteans are benthic marine predators, which attack their prey using a unique eversible proboscis. The proboscis is housed dorsal to the gut in a fluid-filled cavity, called the rhynchocoel. Other characteristics of nemerteans include a brain that encircles the rhynchocoel, instead of the foregut, as in canonical protostome invertebrates, and the two main lateral blood vessels (compared to the dorsal and ventral). Historically grouped with flatworms as acoelomates, nemerteans are now widely considered to be closely related to coelomate protostomes, such as annelids and mollusks (Turbeville 2002; Jenner 2004). The coelomate affinities of nemerteans were initially suggested by the characters of microscopic anatomy, e.g., the presence of continuous endothelial lining in the rhynchocoel and the two main lateral blood vessels (which are thus homologized with the coelomic cavities), gliointerstitial system (Turbeville and Ruppert 1985; Turbeville 1991), and schizocoelous formation of nemertean blood vessels (Turbeville 1986). Additional support comes from many independent molecular and total-evidence phylogenetic studies (Turbeville et al. 1992; Winnepenninckx et al. 1995; Giribet et al. 2000; Peterson and Eernissee 2001).
Recent phylogenies based on mitochondrial gene arrangement and EST data sets firmly place nemerteans in the Lophotrochozoa (arguably synonymous with Spiralia) and suggest close ties to coelomate taxa, although the nemertean sister group is yet to be identified (Turbeville and Smith 2007; Dunn et al. 2008; Bourlat et al. 2008, Chen et al. 2009, Podsiadlowski et al. 2009). Nemerteans display stereotypical equal spiral cleavage (Delsman 1915; Hammarsten 1918; Maslakova et al. 2004a) and a large degree of conservation of classical spiralian cell fates, including ectomesoderm derived from 3a and 3b cells, and endomesoderm derived from the 4d cell (Henry and Martindale 1998).
Aside from the spiral cleavage itself, many spiralians posses a trochophore larva, in the sense of the word used by Rouse (1999), and broadly defined as a larva with a preoral ring of large ciliated cells, called the prototroch. The prototroch is primarily used for larval locomotion, and sometimes for feeding. Critically, the prototroch arises from particular embryonic progenitor cells in a spiralian embryo, called the trochoblasts. Many spiralians have variously modified larvae that deviate from the generalized trochophore body plan, e.g., molluscan veligers, the sipunculid pelagosphaera, uniformly ciliated larvae, endolarvae, or the mitraria larva of annelids. However, many of these still go through a brief trochophore-like stage. At the very least, other representatives of these phyla possess trochophore larvae. Until recently, however, no nemertean was known to possess a trochophore larva, and it was not clear how to relate nemertean larvae to the classical spiralian trochophores. Descriptive and cell-lineage studies in the palaeonemertean Carinoma, however, revealed clear vestiges of the trochophore (Maslakova et al 2004a, 2004b). This means that the origin of the pilidium larva should be interpreted in the context of classical spiralian developmental program, including the stereotypic behavior of the trochoblast lineage.
Larval diversity in nemerteans
Nemertean larvae can be classified by shape and development into pilidium larvae (Fig. 1A) and planuliform larvae (Fig. 1B). The two correspond to the traditionally recognized categories of “indirect” (pilidial) and “direct” (non-pilidial) development in nemerteans (Henry and Martindale 1997; Norenburg and Stricker 2002). The terms “direct” and “indirect” are somewhat misleading because many of the “direct-developing” nemerteans nevertheless have long-lived feeding planktonic juvenile-like larvae. “Direct” in this case refers to a developmental trajectory without a distinct larval body plan, or an overt metamorphosis, rather then to the lack of a swimming or feeding planktonic stage in the life cycle.
The long-lived transparent planktotrophic pilidium larva is found only in the monophyletic clade Pilidiophora (Fig. 2), which comprises the Heteronemertea and the formerly palaeonemertean family Hubrechtidae (Thollesson and Norenburg 2003), with a total of ∼450 species (Kajihara et al. 2008). The pilidium stands out because of its characteristic morphology, and the unusual development of the juvenile via imaginal discs (Salensky 1912; Dawydoff 1940; S. A. Maslakova and L. S. Hiebert, manuscript in preparation). Its development culminates in dramatic metamorphosis, in which the juvenile emerges from the larva by rupturing the larval epidermis, and, in most cases, devours the larval body (Cantell 1966a, 1969; Maslakova and Hiebert, manuscript in preparation). Pilidial development is so unlike any other, that it is difficult to relate it to the development of other spiralians.
Planuliform larvae, found in the non-pilidiophoran nemerteans, the monophyletic Hoplonemertea, and the basal paraphyletic Palaeonemertea, are uniformly ciliated, non-transparent, and resemble cnidarian planulae, hence the name. In contrast to the pilidium, the juvenile body is largely prefigured by the larva. Development of a planuliform larva is gradual, and does not involve a conspicuous metamorphosis, i.e., a dramatic change in body plan between the larval and the juvenile stages. Although all nemertean planuliform larvae look superficially similar to each other, and have been traditionally lumped together as “direct developers”, it is clearly a heterogeneous group (Maslakova and von Döhren 2009).
The most obvious, but frequently overlooked, distinction between the palaeonemertean and the hoplonemertean planuliform larvae lies in their ecology. Most of the studied palaeonemertean planuliform larvae are planktotrophic (Smith 1935; Iwata 1960; Jägersten 1972; Norenburg and Stricker 2002; Maslakova et al. 2004a, 2004b; S. A. Maslakova, personal observations). Relatively small eggs (∼100 μm) are free-spawned and, within a few days, develop into simple planula-like larvae with a conspicuous mouth (derived directly from the blastopore), which leads into a blind gut with a distinct lumen. These larvae apparently require food to grow and develop juvenile structures such as the proboscis, the cerebral ganglia, and the lateral nerve cords. Unlike the pilidium larvae, which are equipped with a ciliated band with which they can capture unicellular algae, palaeonemertean larvae do not seem to be able to collect microscopic particles. Laboratory observations show that they are capable of swallowing larger particles, including isolated sea urchin blastomeres and bivalve veligers (S. A. Maslakova, personal observations), dinoflagellate-size microplankton (Norenburg and Stricker 2002), and even pilidium larvae (Jägersten 1972). However, no one to this day has identified the preferred food for any given species, or succeeded in rearing a palaeonemertean larva to the point of developing juvenile structures. Advanced stages, which are clearly feeding and growing, are found in plankton samples (Jägersten 1972; Norenburg and Stricker 2002; S. A. Maslakova, personal observations).
In contrast to palaeonemerteans, the hoplonemerteans have lecithotrophic (non-feeding) development (Norenburg and Stricker 2002). In many studied hoplonemertean species that free-spawn small eggs (∼100–200 μm) the larva develops the proboscis and the juvenile nervous system without feeding. This is correlated with a rather complex morphogenesis of the hoplonemertean digestive system. The blastopore closes, and the mouth and foregut develop later via a secondary invagination, which eventually fuses with the midgut (Friedrich 1979; Maslakova and von Döhren 2009). Thus, unlike the palaeonemertean larvae, the hoplonemertean larvae are simply not equipped to feed. In some species, settlement is observed soon after the juvenile structures, which include the proboscis, the brain, and the complete gut develop (Maslakova and von Döhren 2009). In others, larvae likely remain in the plankton as feeding juveniles (S. A. Maslakova, personal observations). Many hoplonemertean species with larger eggs (200–500 μm) deposit egg cocoons, and the development is entirely encapsulated, i.e., a complete juvenile emerges from the egg envelopes (Hickman 1963; Norenburg 1986, Maslakova and Malakhov 1999; Maslakova and von Döhren 2009). A few hoplonemertean species bear live young (Coe 1904; Norenburg 1986; Maslakova and Norenburg 2008).
Several pilidiophoran species are known to exhibit encapsulated development (e.g., see Schmidt 1964), or possess an ovoid non-feeding uniformly ciliated planktonic larva that superficially resembles planuliform larvae of hoplonemerteans and palaeonemerteans (Iwata 1958; Fig. 3D). Furthermore, Schwartz and Norenburg (2005) described a transverse equatorial ciliary band in an otherwise planula-like larva of the pilidiophoran Micrura rubramaculosa. Nevertheless, in each case the juvenile develops via imaginal discs, indicating that these larvae and encapsulated pilidiophoran embryos (called Desor’s and Schmidt’s larvae) represent secondary modifications of pilidial development.
A typical pilidium larva looks like a deerstalker cap (the kind worn by Sherlock Holmes)—with ear flaps pulled down (Figs. 1A, 3A and 5A). The external shape of pilidium varies by species (Dawydoff 1940; Cantell 1969; Norenburg and Stricker 2002; Lacalli 2005; Fig. 3), but almost uniformly these are transparent larvae with a blind gut and an unusually extensive blastocoel supporting a delicate larval muscle frame (Fig. 5A). The pilidium is equipped with a long-apical tuft, which can be partially retracted into the dome by a special muscle. The mouth is located inside the cap, between the ear flaps. The surface of the dome is uniformly covered by short cilia. A band of longer cilia spans the margins of larval lappets (Figs. 1A, 3A and B) and produces the feeding current. The pilidium spends weeks to months in the plankton feeding on microscopic particles.
Analysis of lineage contributions to the larval body of the pilidiophoran Cerebratulus lacteus (Henry and Martindale 1998) revealed both conserved and novel features. Among the classical spiralian lineage traits is the dual source of mesoderm. Pilidial larval muscles are derived from the 3a and 3b cells, and the rest of the mesoderm (and a portion of the gut) is derived from the 4d cell, the spiralian mesentoblast. Among the unusual characteristics is the relatively large contribution of the first quartet micromeres (1q) to the larval body—the entire external surface of the pilidium is produced by the first-quartet micromeres.
Because the pilidium is derived within nemerteans, and larvae of the basal members of the phylum are uniformly ciliated (i.e., lack distinct-ciliated bands), the ciliated band in the pilidium larva is not homologous to the prototroch of the trochophore larvae of other spiralians as a ring of differentially ciliated cells employed in locomotion and feeding. Unlike the classical spiralian prototroch which is composed of 24–40 relatively large cleavage-arrested cells (Damen and Dictus 1994) derived as a rule from the first and second quartet micromeres (Henry et al. 2007), the pilidial ciliated band consists of numerous, perhaps hundreds, of very small ciliated cells and includes a contribution of the first-quartet, second-quartet, and also the third-quartet micromeres 3c and 3d (Henry and Martindale 1998). The sheer number and size of cells in the pilidial ciliated band suggests that even though they are partially derived from the spiralian trochoblast lineage, they must escape the cleavage arrest and continue to divide to form the elaborated and ever expanding ciliated band of the pilidium larva. Although this behavior of the trochoblast lineage is likely a derived condition for nemerteans (see below), it is certainly not unique. Prototrochs of several other kinds of spiralian larvae seem to have converged upon the same solution to the problem of making an extensive larval ciliated band, e.g., many gastropod and bivalve veligers, the mitraria larva of the polychaete Owenia (Smart and von Dassow 2009), and the endolarva of the polychaete Polygordius (Woltereck 1902).
The most remarkable thing about the pilidium is the development of the juvenile inside the larva from a series of isolated rudiments, called imaginal discs. The larval body plan is carved out according to the spiralian cell lineage, but formation of the juvenile is largely disconnected from larval development. This type of development is unique among spiralians. Although, of course, the pilidial imaginal discs are not homologous to the identically-named rudiments in the development of holometabolous insects such as Drosophila, it is not inconceivable that similar developmental processes and mechanistic solutions are utilized in both groups to address analogous challenges.
A typical pilidium has three pairs of imaginal discs, which develop as invaginations of larval epidermis (Fig. 3A). The cephalic imaginal discs, which give rise to the head of the juvenile, appear first. The second pair is the trunk discs, which form the major portion of the juvenile body. The third pair is the cerebral organ discs, which give rise to a pair of chemosensory organs (unique to nemerteans) in the head of the juvenile (Salensky 1912; S. A. Maslakova and L. S. Hiebert, manuscript in preparation). The cephalic discs in the pilidium larva are derived from the first quartet micromeres (Henry and Martindale 1998). The formation of the other imaginal discs is nutrition-dependent and takes a much longer time (weeks) than the injected fluorescent dextrans can be detected. Until more persistent lineage markers are developed for nemerteans, one can only guess which lineages give rise to the other imaginal discs based on their position in the larva. Both the trunk discs and the cerebral organ discs invaginate below the ciliated band (Salensky 1912; Maslakova and Hiebert, manuscript in preparation), which suggests that they are derived from the second or third micromere quartets.
Although many textbooks suggest that all discs form as invaginations, this is incorrect. There are at least two additional juvenile rudiments, which do not develop as invaginations of larval epidermis, but instead appear to be mesenchymal in origin (S. A. Maslakova and L. S. Hiebert, manuscript in preparation). Both are unpaired and located in the plane of bilateral symmetry. The proboscis rudiment is first evident as a small cluster of mesenchymal cells between the cephalic discs and larval epidermis. It fuses with the cephalic discs, and the proboscis forms at the junction. Since the proboscis of the adult worm is clearly an epithelial derivative, while the rhynchocoel (a coelom-like sack which houses the proboscis) is likely of mesenchymal origin, one might hypothesize that the unpaired mesenchymal rudiment gives rise to the rhynchocoel, and instructs the tissue of the cephalic discs to form the proboscis proper. The other mesenchymal rudiment is the so-called dorsal disc, which appears between the gut and the larval epidermis and later fuses with the trunk discs (Salensky 1912; S. A. Maslakova and L. S. Hiebert, manuscript in preparation). The lineage of these two unpaired mesenchymal rudiments is unknown. However, it is perhaps not unreasonable to suggest that they are derived from the progeny of the spiralian mesentoblast (4d), which produces scattered populations of mesenchymal cells in the blastocoel of a young pilidium larva (Henry and Martindale 1998).
Imaginal discs fuse with each other to form the juvenile worm around the larval gut. The body plan of the planktonic pilidium larva and the benthic juvenile are dramatically different, and the transition between the two stages is rapid (minutes) and catastrophic. Because the larva and the juvenile share the gut, they are physically connected around the mouth. During metamorphosis, the juvenile ruptures the larval epidermis, and emerges from the larva while swallowing the larval tissues, until the entire pilidium is ingested (Cantell 1966a, 1969; Maslakova and Hiebert, manuscript in preparation).
The pilidium represents a striking example of spatial and temporal segregation of larval and juvenile morphogenesis. Development of the larval body is completed long before any but the first pair of juvenile rudiments appears. The formation and growth of imaginal discs is dependent on nutrient uptake (feeding) by the larva. In the absence of food, the larva can survive for many weeks, without developing juvenile structures (S. A. Maslakova, personal observations). Related to this is the apparent dissociation of the larval and juvenile body axes. In all planuliform larvae, the larval and juvenile body axes are one and the same (Fig. 4A). In a typical pilidium, in contrast, the antero-posterior (AP) axis of the juvenile is perpendicular to the AP axis of the larva (Figs. 1A and 4B). In the types of larvae called Pilidium recurvum (Dawydoff 1940; S. A. Maslakova, personal observations) and Pilidium recurvatum (Cantell 1966b), as well as in the recently described larva of M. rubramaculosa (Schwartz and Norenburg 2005) the AP axis of the juvenile coincides with that of the larva (Fig. 4C), while in Micrura akkeshiensis (Iwata 1958), the AP axes of the larva and the juvenile are reversed (Fig. 4D). This diversity suggests that the evolution of pilidial development was accompanied by the dissociation of the mechanisms responsible for patterning the larval and the juvenile body axes.
An incidental observation of abnormal development in one cohort of pilidia of M. alaskensis (S. A. Maslakova, personal observations) suggests that the individual juvenile rudiments develop largely independently from each other. While the larvae looked normal and were actively feeding, a large proportion of pilidia in this culture exhibited various abnormalities of the imaginal discs. For example, some were missing one or both of the cephalic discs, one or both of the cerebral organ discs, or the proboscis rudiment, or failed to fuse the discs properly. Remarkably, the absence of one or several rudiments did not visibly affect the development of other rudiments. This observation suggests that individual discs must be more than simply bags of proliferating cells; they are capable of extensive pattern formation and differentiation even before the whole body is assembled.
Comparative embryonic and larval development in nemerteans
Hyman (1951, p. 514) wrote that “… the chief difference between the direct and the larval [pilidial] types of development [in nemerteans] consists in the replacement in the latter of the larval ectoderm by the invaginated ectoderm…”. She was referring to the origin of the definitive epidermis from the invaginating imaginal discs, and the pilidial metamorphosis, in which the pilidial body is consumed by the emerging juvenile. However, non-pilidiophoran nemerteans may have an equivalent developmental process in which the larval epidermis is replaced by the definitive epidermis.
Several species of hoplonemerteans with planktonic larvae or encapsulated development have been reported to possess a transitory embryonic or larval epidermis, which is replaced by the definitive epidermis during development (Delsman 1915; Hammarsten 1918; Reinhardt 1941; Hickman 1963; Maslakova and Malakhov 1999)—a process possibly homologous to pilidial metamorphosis (Jägersten 1972; Maslakova and Malakhov 1999). Recent studies of hoplonemertean larval development added more species to the list, suggesting that presence of the transitory epidermis, which is shed or resorbed during development, is likely the rule for hoplonemerteans (Maslakova and von Döhren 2009; Hiebert et al. 2010).
One earlier study showed that the larva of the palaeonemertean Tubulanus punctatus may also possess some sort of a transitory epidermis (Iwata 1960). Although this report has never been independently confirmed, it led to the hypothesis that the metamorphic type of life history, as found in pilidiophorans, may be ancestral for the phylum (Jägersten 1972; Maslakova and Malakhov 1999). With the idea of testing this hypothesis, the author set out to examine the larval development of the palaeonemertean Carinoma tremaphoros, looking for transitory larval epidermis, and a possible homology to pilidial metamorphosis. Instead, we discovered, by tracing the epidermal cell outlines and the cell lineage, that what appeared at first as a transitory larval epidermis, was in fact, a vestigial prototroch, composed of 40 cells derived from the classical spiralian trochoblast lineage (Maslakova et al. 2004a, 2004b). This prototroch is inconspicuous because Carinoma’s planuliform larva is uniformly ciliated, but the cells of the prototroch are distinguished by their large size, arrangement in a discrete pre-oral ring, cleavage arrest, and ultimate degenerative fate (Fig. 5C). The presence of a “hidden trochophore” in a basal nemertean supports the notion that pilidial development is derived in nemerteans, and provides a useful reference point when comparing nemertean larvae to the larvae of other spiralians.
It remains to be tested (e.g., by cell lineage analysis) whether the transitory epidermis of the hoplonemertean planuliform larvae is homologous to the prototroch in Carinoma and other spiralians, or whether it may be homologous to the pilidial larval body. Morphologically, this transitory or “deciduous” epidermis is different from the vestigial prototroch in Carinoma in several ways. First of all, more cells (80–90) are participating in the formation of the hoplonemertean transitory epidermis, than in the hidden prototroch of Carinoma. Second, the cells of the hoplonemertean transitory epidermis become separated from each other by the intercalating cells of the definitive epidermis (Fig. 5B), as opposed to histolysing as a contiguous domain (Fig. 5C) in Carinoma (Maslakova et al. 2004a, 2004b; Maslakova and von Döhren 2009; Hiebert et al. 2010). To emphasize the difference between the palaeonemertean and the hoplonemertean larvae, the author proposes a new name for the hoplonemertean planuliform larva—the decidula (Fig. 5B), from the Latin deciduus, meaning “to shed”.
The most characteristic feature of pilidial development, the imaginal discs, might also have a counterpart in planuliform development. Invaginated rudiments of one kind or another have been described in development of almost every non-pilidiophoran nemertean species. At the very least, two lateral transitory invaginations are observed at the anterior end on each side of the apical plate. These invaginations, found in both palaeonemerteans and hoplonemerteans, have been variously interpreted as the rudiments of the nervous system, the cerebral organs, or other structures (Maslakova and von Döhren 2009 and references therein). In the hoplonemertean Pantinonemertes californiensis, in addition to the anterior invaginations, there is a pair of transitory postero-lateral invaginations (Fig. 6; Hiebert et al. 2010), which resemble the imaginal discs of encapsulated Desor’s and Schmidt’s larvae of the pilidiophorans Lineus viridis and L. ruber (Schmidt 1964).
Hoplonemertean development is of particular interest to the origin of the pilidium larva, because the hoplonemerteans represent a sister group to the Pilidiophora (Thollesson and Norenburg 2003), and the hoplonemertean decidula larva is clearly different from the hidden trochophore larva of basal nemerteans. Nothing is known about the cell lineage of hoplonemerteans beyond the larval contribution of the four embryonic quadrants (Henry and Martindale 1994). At the moment, therefore, it is not clear what relationship the transitory epidermis and the various invaginated rudiments in hoplonemerteans might have to the pilidial epidermis and the imaginal discs, or to what extent the trochoblast lineage participates in forming the transitory epidermis of the decidula. A cell lineage analysis of hoplonemertean development, as well as application of long-term lineage markers to furthering an understanding of pilidial development will help to shed light on the developmental homologies in this phylum, and lay the foundation for future comparative studies of genetic regulatory mechanisms in nemertean development.
Friday Harbor Laboratories postdoctoral fellowship to S.A.M; the Society for Integrative and Comparative Biology and the Society for Developmental Biology (Symposium on Spiralian Development).
I thank the researchers and staff of the Friday Harbor Laboratories and the Center for Cell Dynamics for resources, encouragement and support throughout the years. I also thank George von Dassow for feedback on the article, and Dave Lambert and Elaine Seaver for organizing a very stimulating symposium on Spiralian Development.