Neurogenesis suggests independent evolution of opercula in serpulid polychaetes
© Brinkmann and Wanninger. 2009
Received: 8 May 2009
Accepted: 23 November 2009
Published: 23 November 2009
Skip to main content
© Brinkmann and Wanninger. 2009
Received: 8 May 2009
Accepted: 23 November 2009
Published: 23 November 2009
The internal phylogenetic relationships of Annelida, one of the key lophotrochozoan lineages, are still heavily debated. Recent molecular analyses suggest that morphologically distinct groups, such as the polychaetes, are paraphyletic assemblages, thus questioning the homology of a number of polychaete morphological characters. Serpulid polychaetes are typically recognized by having fused anterior ends bearing a tentacular crown and an operculum. The latter is commonly viewed as a modified tentacle (= radiole) and is often used as an important diagnostic character in serpulid systematics.
By reconstructing the developmental neuroanatomy of the serpulid polychaete Spirorbis cf. spirorbis (Spirorbinae), we found striking differences in the overall neural architecture, the innervation pattern, and the ontogenetic establishment of the nervous supply of the operculum and the radioles in this species. Accordingly, the spirorbin operculum might not be homologous to the radioles or to the opercula of other serpulid taxa such as Serpula and Pomatoceros and is thus probably not a part of the tentacular crown.
We demonstrate that common morphological traits such as the prostomial appendages may have evolved independently in respective serpulid sublineages and therefore require reassessment before being used in phylogenetic analyses. Our findings corroborate recent molecular studies that argue for a revision of serpulid systematics. In addition, our data on Spirorbis neurogenesis provide a novel set of characters that highlight the developmental plasticity of the segmented annelid nervous system.
A number of classical and recent studies have shown that structural similarities among organisms are not necessarily based on homologous characters, but may instead be the result of convergent (analogous) evolution. Thereby, these matching phenotypes have emerged independently in distantly related organisms [1–6]. Accordingly, if misinterpreted as sharing common ancestry, convergently evolved characters might lead to false conclusions in phylogenetic reconstructions.
In the age of genomics, convergent evolution is mostly inferred based on existing phylogenies (i.e., after the actual analysis) and is often used to explain incongruencies between the phylogenetic signal obtained from molecular versus morphological datasets. In this respect, neuronal innervation patterns of organ systems may be used as an independent test of homology prior to phylogenetic analysis, as illustrated by classical studies on the structure of the polychaete nervous system [7–9]. Along these lines, it seems reasonable to assume that investigations of neuronal innervation patterns and neurogenesis pathways might also aid in resolving long standing questions regarding the internal phylogenetic relationships of Annelida, one major lophotrochozoan lineage. In particular, the basal branches of the annelid tree remain unresolved, despite the additional use of large-scale molecular analyses [10–19]. Current interest has therefore partly shifted to lower taxonomic levels, taking a "top down" approach to reconstruct annelid phylogeny.
In order to test the proposed homology - and thus the potential usefulness for phylogenetic analyses - of one of the most crucial morphological characters in the serpulid bodyplan, the operculum, and to assess the hypothesis that all spirorbin opercula evolved from a cephalic radiole, we investigated the ontogeny of the innervation pattern of the prostomial appendages of the spirorbin Spirorbis cf. spirorbis. In addition, we provide a new set of characters for a broad comparison of neurogenesis in polychaete species.
The neurotransmitter serotonin is present in Spirorbis cf. spirorbis from the early trochophore stage onwards (Figure 3; Figure 4). The first serotonergic signals appear more or less simultaneously with the FMRFamidergic immunoreactivity. Initially, serotonin is found in the early anlagen of the adult central nervous system, such as the cerebral ganglion, the circumesophageal connectives, the ventral nerve cords, and the subesophageal commissure (Figure 3A). In addition, serotonin is distributed in a nerve ring underlying the larval prototroch. This serotonergic innervation of the ciliated prototroch changes considerably over time (Figure 3; Figure 4). The primary prototroch nerve ring expands on its ventral side shortly before hatching and forms a network of cross-linked serotonergic fibers (Figure 3B, C, D). Moreover, two clusters of accessory prototroch neurons develop dorsolaterally in the free-swimming larva (Figure 4A, B), and a broad gap appears in the serotonergic prototroch innervation (Figure 3D; Figure 4B). After settlement, the neurons of the prototroch are reduced to a semi-circle on the dorsal side and degenerate completely during metamorphosis (Figure 3E, F).
The apical organ of free-swimming Spirorbis larvae initially comprises four serotonergic flask-shaped perikarya (Figure 3B). Three of these cells lie in one plane, with the fourth cell body being positioned medially above them. All apical perikarya possess projections that seem to be connected to the rudiment of the adult cerebral ganglion. Interestingly, the number and position of these apical serotonergic somata varies over time as well as between individuals with a maximum of seven being present in specimens competent to settle.
The cerebral ganglion is one of the first adult features that are established during neurogenesis. It is divided into two distinct serotonergic hemispheres (left and right) in the encapsulated and free-swimming larval stages. Posterior to the two domains, the cephalic commissures appear as one strand which is directly connected to the circumesophageal connectives (Figure 4B). After settlement, the latter split anteriorly into the so-called dorsal and ventral roots. Each root differentiates into a dorsal and a ventral commissure. Accordingly, four main commissures, all derived from the circumesophageal connectives, traverse the cerebral ganglion (Figure 4C). The cerebral commissures come to lie in close proximity to the subesophageal commissure, because the circumesophageal connectives are relatively reduced in length in juvenile specimens (Figure 3F; Figure 4D; Figure 5C).
Two thin axons represent the first rudiments of the developing serotonergic ventral nerve cords while the larva is still within the egg capsule (Figure 3A). In later larval stages, the signal of the ventral nerve cords is discontinuous in the second segment. Anteriorly, two axons are present on each side, followed posteriorly by a discrete perikaryon with one long axon pointing medially towards the posterior end (Figure 3B). The ventral nerve cords consist of multiple fibers in the collar region during the free-swimming larval stage and are interconnected by an additional commissure just posterior to the already established subesophageal commissure (Figure 3C, D, E; Figure 4A, B, C). In the abdomen, the ventral nerve cords show a different organization. Thereby, it appears that the axons of the inner ventral nerve cords (i-v) derive from the posterior ectoderm, whereas the ones of the outer part (o-v) are extensions from the anterior thorax region (Figure 3C, D). After settlement, a second commissure is formed in the region where the most anterior neurites of the ventral cords terminate (Figure 4C). The architecture of the ventral nervous system changes significantly with the onset of metamorphosis, eventually resulting in the ventral nerve cords being located on the left side in early juvenile specimens due to the rotated body axis (Figure 3F; Figure 5B, C). Another alteration relates to the shift of the ventral commissures in an anterior direction, resulting in the first ventral commissure merging into the newly formed subesophageal ganglion (Figure 4D; Figure 5C). Furthermore, the two strands of the ventral nerve cords fuse in the segmented setigerous abdomen, except for the posterior-most end (Figure 3F; Figure 5C). Several serotonergic perikarya are associated with the ventral nerve cords from the free-swimming larval stage onwards (Figure 4A, B, C). The number and position of these cells, which are arranged in three clusters, is not consistent among individuals of the same stage. On the opposite side of the abdomen, dorsal longitudinal neurons are present. Interestingly, these two widely separated strands of the peripheral nervous system have a commissural neuron (Figure 3F).
The development of the peripheral nervous system starts with a network of collar neurons in the encapsulated larva (Figure 3B-F). In a second step, dorsal peripheral neurons that innervate the abdomen appear during the planktonic larval phase (Figure 3C, D, E). After settlement, these are directly connected to a transversal thorax nerve which crosses the dorsal body region (Figure 3E; Figure 4C). One lateral longitudinal nerve runs along each side of the thorax from the transversal thorax nerve to the main collar nerve (Figure 3E; Figure 4C). Here, anterior segmental neurons branch off. They are associated with the first ventral commissure of the paired nerve cords (Figure 4C; Figure 5B). These anterior segmental neurons are connected to the subesophageal ganglion after metamorphosis, due to a forward migration of the first ventral commissure (Figure 4D; Figure 5C). Moreover, the lateral longitudinal nerves are shortened compared to the pre-metamorphic condition, and the transversal thorax nerve is connected to the ventral nerve cords at the position of the second ventral commissure (Figure 4D).
The earliest serotonergic signals from the two neuronal growth cones of the branchial crown are present in the encapsulated larva, and additional branchial neurons appear in the free-swimming larval stage (Figure 3B, C, D; Figure 4A, B; Figure 6B). Several axons outline the contour of the operculum shortly after settlement (Figure 3E; Figure 6B). At the same time, the two main nerves of the tentacular crown are already established and split dichotomously at their anterior end (Figure 6B). During metamorphosis, two branchial crown ganglia develop at the base of the tentacles (Figure 6C). From there, neurons branch off and project into the radioli. Exactly the same number of branchial filaments arises on either side. At this point of development, it is not possible to distinguish pinnules from radioles. Eventually, three distinct nerves innervate each individual branchial filament. Accordingly, all radioli and pinnules have one median nerve and two lateral branchial nerves (Figure 6C). The latter are located in a different focal plane than the internal nerve. By contrast, the opercular peduncle and lid are only lined by a single, loop-like opercular nerve. The tissue in the middle of the peduncle is innervated by axons that are irregularly linked to the opercular nerve (Figure 6C). The neurons at the base of the peduncle emerge from the dorsal cerebral commissures, whereas the branchial crown nerves emerge from a more posterior region (Figure 6C). Moreover, the branchial crown nerves reach almost to the subesophageal ganglia. Accordingly, the neuronal innervation of the branchial crown and the operculum, respectively, arises from two different roots which emerge from two distinct regions of the cerebral ganglion in post-metamorphic Spirorbis specimens.
Despite a significant increase of data on lophotrochozoan neurogenesis and larval neuroanatomy [31–50] and the crucial role of polychaetes for evolutionary inferences, detailed studies on the ontogeny of their nervous system employing immunocytochemical methods are as of yet only available for four species [31, 40, 43, 45]. While final conclusions concerning the ancestral neural bodyplan of Annelida can therefore not yet be made, a comparison of the data currently available hints towards certain evolutionary trends which may have considerable bearing for our understanding of the origin of the polychaete (and annelid) nervous system.
Within Annelida sensu lato, it appears that a segmented body organization has been lost independently multiple times, but that it has been at least partly retained in the larval nervous system of some taxa that show no external signs of metamerism in the adult stage, such as echiurans [34, 35, 39] and sipunculans [47, 51]. Similarly, multiple variations of the conventional rope-ladder-like nervous system have been described for various polychaetes. Accordingly, the number of connectives may range from one to five in adult polychaete annelids . The presence of only one nerve in the setigerous abdomen of the adult and a reduced number of ventral commissures illustrates that the metameric arrangement of the nervous system is only weakly expressed in Spirorbis as well. These findings are consistent with an earlier description of another spirorbid polychaete, Spirorbis moerchi . In this species, three pairs of ganglia and three ventral commissures, which are associated with the three larval segments, are formed, and these constitute the only neural components that hint towards a segmented nervous system . Similar to the condition found in S. spirorbis, the ventral nerve cords are widely separated in the anterior body region before they fuse into one strand in the setigerous abdomen. Although the latter is externally segmented in the adult, this condition is not reflected in the nervous system [53–56]. The plasticity of annelid neural segmentation is also demonstrated by the variability concerning the neurogenesis pathways that may lead to the metameric arrangement of individual components of the segmented nervous system. As such, in the sabellid polychaete Sabellaria alveolata, two distinct modes of neural development are found, namely the simultaneous formation of the first three pairs of peripheral segmental neurons on the one hand and a strictly progressive formation in anterior-posterior direction of the ventral commissures on the other . Taken together, these data indicate that the ontogenetic establishment of the segmented annelid nervous system is a highly dynamic process that must have undergone significant modifications from its ancestral developmental pathway in the various annelid lineages, and additional future studies on the subject are needed before we are able to fully understand the mechanisms that underlie annelid segmentation and nervous system evolution.
The serpulid operculum and its peduncle are commonly regarded as a modified tentacle (= radiolus) of the branchial crown [57–61]. Ever since Müller's  observation that the operculum forms at the tip of a pinnulate radiolus in individuals of the genus Serpula, the tentacles have been considered homologous to the operculum. This notion is still widely accepted, despite considerable phenotypic and developmental variation of serpulid opercula. As such, the operculum may either develop directly at the distal tip of a smooth peduncle without pinnules (e.g., in Pomatoceros and Spirorbis) or indirectly (e.g., in Serpula). In the latter case, the peduncle retains its pinnules only in some filogranin species. In addition, the Filograninae encompass a few non-operculate forms such as Protula (see Figure 1B).
Serpulid phylogeny has traditionally been founded to a large extent on these opercular characters . For example, Pillai  proposed that the spirorbin opercula are not homologous to the ones of other serpulid species, and accordingly suggested a new classification of the spirorbin taxa, while ten Hove  proposed an evolutionary hypothesis for serpulid phylogeny based on a transformation series of the branchial crown alone. However, the latter author stressed at the same time the lack of other conclusive characters to substantiate serpulid interrelationships. To further assess this issue, neuronal innervation patterns of organ systems may be used as an independent test of homology prior to phylogenetic analysis, as illustrated by recent neuroanatomical studies on euchelicerates and pycnogonids (seaspiders) [63–65]. In addition, the comprehensive studies on the cephalic nervous system of polychaetes by Orrhage [7–9] highlight the relevance of such investigations for the homologization of anterior appendages in annelids.
The internal anatomy of the radioles has been analysed in several different serpulid species [60, 66–68] including Spirorbis [53, 68, 69]. However, the present study on neurogenesis in S. spirorbis is the first one to raise serious doubts on the proposed homology of opercula and radioles in Spirorbinae based on differences concerning (i) the overall neural architecture, (ii) the innervation pattern, and (iii) the ontogenetic establishment of the nervous supply of the operculum and the radioles. In Spirorbis, the number of tentacles increases during development, resulting in an asymmetric branchial crown comprising five radioles on the right side and four on the left side in adult specimens [53, 70]. In addition, classical studies have shown that development of the operculum is retarded compared to the radioles [71–73]. Indeed, the neuronal rudiments of the branchial crown are formed prior to the ones of the operculum. However, the differentiation of the opercular neurons advances more rapidly during subsequent development, with the opercular nerve lining the entire peduncle even before the radiole nerves differentiate (Figure 6B). In addition, exactly the same number of radioles and pinnules arises on both sides (Figure 6C). Accordingly, we regard the asymmetric arrangement in the adult a secondary condition, probably constrained by limited space in the cephalic region rather than being the result of evolutionary transformation events of a radiole into an operculum.
The general neural architecture of Spirorbis shows that each radiole comprises three nerves [, present study]. A similar condition has been described for Pomatoceros, where two nerves are situated in the abfrontal corners ("external branchial nerves") and a third one frontally in close proximity to the food groove ("internal branchial nerve"; [60, 68]). Thereby, the abfrontal nerves lie outside the basement membrane of the epidermis, whereas the frontal nerve lies just inside the basement membrane. All three branchial nerves send branches into the pinnules. This tripartite arrangement has been documented for Pomatoceros not only for the radioles and pinnules, but also for the peduncle . At the top of the peduncle the internal and external peduncular nerves enter the operculum through a small, single gap. There, all three nerves furcate, thus giving rise to numerous small branches which can be traced to the rim surrounding the distal pole of the operculum . This tripartite arrangement in both the radioles and the operculum is in accordance with the hypothesis of Müller . Hanson  described three branchial nerves also for other serpulid species such as Vermiliopsis, Protula, Salmacina and Hydroides, but he did not mention explicitly the neuronal innervation of the peduncle. Only for Hydroides he clearly stated that the internal and external peduncular nerves are similar to those of the filaments of Pomatoceros. However, in juvenile S. spirorbis, there are no three parallel nerves present in the peduncle. Instead, a single nerve loop lines the margin of the opercular stalk. Furthermore, the two strands of this nerve loop do not lie close together at the transition between the peduncle and the operculum. Accordingly, the number, spatial distribution, and site of emergence of the opercular nerves in Spirorbis are unique and do not correspond to the situation found in the radioles and in the opercular peduncle of Pomatoceros and Hydroides.
Probably the most striking argument against a shared evolutionary ancestry of the radioles and the operculum in Spirorbis is revealed when the cerebral innervation sites of both structures are compared. While the opercular nerves appear to emerge from the dorsal commissures of the cerebral ganglion, the two main nerve bundles of the radioles are associated with the ventral part of the cerebral ganglion. The nerves of the radioles show no connection to the opercular nerves. The innervation of anterior appendages in the Serpulidae has been already investigated in detail for selected species, albeit not in Spirorbis, and used for homology assessments . Based on his findings, Orrhage described ten different nerve roots of the branchial crown that partly originate from the ventral and partly from the dorsal commissures on either side of the cerebrum, before finally forming one anterior branchial crown nerve. It might be that one or more of these nerve roots have gained new functions in spirorbin taxa that brood in their operculum . The resolution of the data presented herein does not allow for unambiguous distinction of individual nerve roots. The question whether this is due to limitations of the used immunolabeling techniques or a lack of differentiation in early developmental stages has to be left open. Orrhage  does not explicitly describe the innervation of the operculum in Spirorbis. Nethertheless, we expect that the neuronal root of the operculum would be merged with the other nerve roots to form a single branchial crown nerve in adults, in case that the operculum represents indeed a modified radiole.
The most recent phylogenetic studies propose that Hydroides belongs to the Serpula- clade, the sister group of the Pomatoceros -clade (Figure 1B). This implies that the operculum with three peduncular nerves, that evolved from a radiole of the branchial crown, is most likely a shared character of these two sister clades. So far, nothing is known about the neuronal innervation of the operculum in species of the Protula -clade, where some of its members do not even have an operculum. Therefore, any final conclusions about the ancestral condition of the operculum at the base of the serpulid tree are premature. In principal, several scenarios appear possible in the light of the current phylogenetic hypothesis (Figure 1B): 1) opercula evolved at least twice independently, once in the Serpula/Pomatoceros -clade and once in the Protula/Spirorbinae -clade, 2) the operculum was lost at the base of the Spirorbinae, and secondarily evolved again within this group as an independent opercular structure, or (3) the operculum is a shared ancestral feature of the Serpulidae that was - together with its neural innervation pattern - secondarily modified in the descending lineages. Based on our data on the neural innervation pattern of the operculum in Spirorbis we favour the former alternative, although additional comparative data on the developmental programme (e.g., gene expression analysis) that underlies radiole and operculum formation in the respective serpulid clades are needed to finally settle this issue.
The operculum of Spirorbis is a prostomial appendage and does not belong to the branchial crown. It is thus not homologous to the radioles.
The operculum of Spirorbis is not homologous to the opercula of Serpula, Hydroides and Pomatoceros. In the latter three, the operculum is indeed likely to have a shared evolutionary origin with the radioles. Therefore, the opercula of Serpula, Hydroides and Pomatoceros are considered homologous to each other.
Based on our data we predict that future neuroanatomical and developmental studies of the branchial crown of selected species, especially of the Protula -clade (including the former Filograninae), will further contribute to our understanding of both the evolution of opercular structures in serpulid polychaetes and will also aid in our quest for the eventual placement of this group within the annelid phylogenetic tree.
Adult specimens of Spirorbis cf. spirorbis were collected in the intertidal around Roscoff, France, in June 2006 and 2007, respectively. The calcareous tubes were crushed with tweezers under a stereo microscope in order to obtain the egg strings which contained embryos of various developmental stages (Figure 1A). Developmental stages (encapsulated embryos, free-swimming larvae, settled larvae, and juveniles) were cultured in embryo dishes at 17-19°C in Millipore-filtered seawater (MFSW). Prior to fixation, the specimens were anesthetized in a 1:1 dilution of MFSW and MgCl2 (7%). They were then fixed at room temperature in 4% paraformaldehyde in 0.1 M phosphate buffer (PB) for 1.5-3 h or overnight at 4°C, washed three times in PB, and stored at 4°C in PB containing 0.1% NaN3. Tubed juveniles were decalcified in 50 mM EGTA and washed thrice in PB prior to storage at 4°C.
Stored animals were washed in distilled water, postfixed in 1% osmium tetroxide, washed twice in distilled water, and dehydrated in a graded alcohol series. After critical point drying in acetone, the samples were mounted on SEM stubs, sputter-coated with gold-palladium, and analyzed with a JEOL JSM-6335-F SEM (JEOL, Tokyo, Japan).
The following steps were all performed at 4°C. Antibody staining was preceded by tissue permeabilization for 1 h in 0.1 M PB with 0.1% NaN3 and 0.1% Triton X-100 (PTA), followed by incubation in block-PTA [6% normal goat serum (Sigma-Aldrich, St. Louis, MO, USA) in PTA] overnight. The primary antibodies, polyclonal rabbit anti-serotonin (Zymed, San Francisco, CA, USA, dilution 1:800), polyclonal rabbit anti-FMRFamide (Chemicon, Temecula, CA, USA, dilution 1:400), and monoclonal mouse anti-acetylated α-tubulin (Sigma-Aldrich, dilution 1:1000), all in block-PTA, were either applied separately or in a mixed cocktail for 24 h. Subsequently, specimens were rinsed in block-PTA with three changes over 6 h and incubated thereafter with 4'6-diamidino-2-phenyl-indole [DAPI (Invitrogen, Eugene, OR, USA)] and secondary fluorochrome-conjugated antibodies [goat anti-rabbit FITC (Sigma-Aldrich), dilution 1:400; goat anti-rabbit Alexa Fluor 594 (Invitrogen), dilution 1:1000; goat anti-mouse FITC (Sigma-Aldrich), dilution 1:400] in block-PTA for 12-20 h. Finally, the specimens were washed three times in PB without NaN3, once in distilled water, and thereafter dehydrated through a graded ethanol series (30%, 50%, 70%, 80%, 90%, 3 × 100%), cleared with benzyl benzoate:benzyl alcohol (2:1), and mounted on glass slides.
Immunolabeled specimens were analyzed for each antibody separately for a minimum of 10 individuals per developmental stage. Altogether, about 60 image stacks of optical sections were recorded as Z-wide-projections with 0.1-0.5 μm step size using a Leica DM IRE2 fluorescence microscope equipped with a Leica TCS SP 2 confocal laserscanning unit (Leica, Wetzlar, Germany). The Z-stacks were projected into maximum intensity pixel as well as depth-coded images. In addition, corresponding light micrographs were recorded for most specimens. The three-dimensional computer reconstructions were generated with the imaging software Imaris v. 5.5.3 (Bitplane, Zürich, Switzerland) using surface rendering algorithms. Image adjustment was done with Adobe Photoshop CS2 and arrangements of plates with Adobe Illustrator CS3 (Adobe Systems, San Jose, CA, USA).
We are grateful to the staff of the Station Biologique de Roscoff (France) for their hospitality. We thank Lisbeth Haukrogh (Copenhagen) and Ricardo Neves (Aveiro and Copenhagen) for help with SEM preparation. NB is the recipient of an EU fellowship within the MOLMORPH Network under the 6th Framework Program "Marie Curie Host Fellowships for Early Stage Research Training" (contract number MEST-CT-2005-020542), which is coordinated by AW.
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.