Nervous system development in the Pacific oyster, Crassostrea gigas (Mollusca: Bivalvia)
Bivalves comprise a large, highly diverse taxon of invertebrate species. Developmental studies of neurogenesis among species of Bivalvia are limited. Due to a lack of neurogenesis information, it is difficult to infer a ground pattern for Bivalvia. To provide more comprehensive morphogenetic data on bivalve molluscs and relationships among molluscan clades, we investigated neurogenesis in the Pacific oyster, Crassostrea gigas, from the appearance of the first sensory cells to the formation of the larval ganglionic nervous system by co-immunocytochemistry of the neuronal markers FMRFamide or 5-HT and vesicular acetylcholine transporter (VAChT).
Neurogenesis begins with the emergence of the apical serotonin-immunoreactive (5-HT-ir) sensory cells and paired sensory posttrochal dorsal and ventral FMRFamide-immunoreactive (FMRFamide-ir) cells at the early trochophore stage. Later, at the early veliger stage, the apical organ (AO) includes 5-HT-ir, FMRFamide-ir, and VAChT-ir cells. At the same stage, VAChT-ir cells appear in the posterior region of larvae and send axons towards the AO. Thus, FMRFamide-ir neurites and VAChT-ir processes form scaffolds for longitudinal neurite bundles develop into the paired ventral nerve cords (VNC). Later-appearing axons from the AO/CG neurons join the neurite bundles comprising the VNC. All larval ganglia appear along the VNC as paired or fused (epiathroid) clusters in late veliger and pediveliger larvae. We observed the transformation of the AO into the cerebral ganglia, which abundantly innervated the velum, and the transformation of ventral neurons into the pedal ganglia, innervating the foot, gills, and anterior adductor muscle. The visceral ganglia appear last in the pediveliger oyster and innervate the visceral mass and posterior adductor of premetamorphic larvae. In addition, a local FMRFamide-ir network was detected in the digestive system of pediveliger larvae. We identified VAChT-ir nervous elements in oyster larvae, which have not been observed previously in molluscs. Finally, we performed a morphology-based comparative analysis of neuronal structures among bivalve, conchiferan, and aculiferan species.
We described the development of the nervous system during the larval development in Crassostrea gigas. These data greatly advance the currently limited understanding of neurodevelopment in bivalves and mollusks, which has hampered the generation of a ground pattern reconstruction of the last common ancestor of Mollusca. Our morphological data support phylogenomic data indicating a closer Bivalvia-Gastropoda sister group relationship than the Bivalvia-Scaphopoda (Diasoma) group relationship.
KeywordsMollusca Larvae Neurogenesis Evolution Serotonin FMRFamide Acetylcholine
Bivalves comprise a large taxon of invertebrates with a high degree of variation in development (planktotrophic and lecithotrophic larvae), life style (sedentary or sessile), and morphology (adaptive reduction of inner organs), as well as ecological diversity (marine, freshwater, tropics and Arctic waters). Recent transcriptomic characterizations of bivalves using RNA-seq data has arranged them into five clades (Protobranchia, Pteriomorpha, Palaeoheterodonta, Archiheterodonta, and Euheterodonta) . Most bivalves exhibit a biphasic life cycle that includes actively swimming plankton larval stages and sedentary benthic adult animals . Neurogenesis in this animal group has been explored to a limited extent by zoologists and morphologists, especially in the larval stage, with only a few detailed descriptions in the literature [3, 4, 5, 6].
The subject of the present study is the Pacific oyster, Crassostrea gigas (Pteriomorphia: Ostreida, Thunberg, 1793), which is one of the commonly found molluscs in the world . The nervous system of the adult oyster Crassostrea virginica consists of central and peripheral branches. The central nervous system comprises paired cerebral ganglia lying symmetrically on both sides of the molluscan body and a huge visceral ganglion in which the right and left components are fused into a single organ . Cerebral ganglia are located in the esophageal region and are connected by a U-shaped commissure. The visceral ganglia, which are the major constituent of the oyster nervous system, are located in the most caudal part of the body and are connected to the cerebral ganglia via long cerebro-visceral. Cerebro-pedal and pedal-visceral connectives form paired ventral nerve cords (VNC) described in molluscs and billaterians. In the adult oyster, the pedal ganglia and the cerebro-pedal connections are reduced due to the loss of the foot after metamorphosis . The peripheral nervous system includes numerous nerves that extend from the ganglia that innervate the mantle edge, gills, and other parts of the body.
Data on neurodevelopment in larval bivalves are sparse, particularly for oysters [3, 9, 10], and in many cases, such investigations have been restricted to studies of a single neuroactive substance , a strongly modified mode of development , or late developmental stages [4, 12, 13]. Further, the different methodological approaches used make it difficult to perform a comparative analysis of the neurodevelopment of various groups or species. For example, histological data have been used to provide a detailed description of neurogenesis in the oyster Crassostrea virginica , whereas only immunochemical data are available for the Pteriomorpha (Mytillus trossulus and Mytillus edulis) [4, 5] and Imparidentia (Dreissena polymorpha, Spisula solidissima) [15, 16].
Knowledge of the emergence, development, and organization of the larval oyster nervous system is essential for understanding the role of the nervous system in larval survival and adaptation. Moreover, developmental ontogenetic data are necessary to demonstrate the degree to which neurogenesis reflects ancestral traits. Therefore, the search for the evolutionary origin and phylogenetic sister-groups of bivalves, as well as Mollusca in general, is ongoing. Based on segmentation there are known to be numerous synapomorphies that share neuro-developmental characteristics (e.g., structure of the larval apical organ and nerve cords) in larval entoprocts, aplacophorans, and polyplacophorans as well as and similarities in bivalve and scaphopod neurogenesis. Considering these observations along with anatomical, embryological and paleontological data, several phylogenetic hypotheses and concepts have been suggested, including the diasoma concept , the annelid-mollusk hypothesis [18, 19], and the tetraneuralia concept . These competing ideas are difficult to confirm and reconcile due to the lack of information provided by fossil records and insufficient morphological data related to neurogenesis in mollusks, particularly bivalve species.
The aim of the present study was to elucidate details about the main events in larval bivalve neurogenesis and to discuss the possible defining characteristics of the last common ancestor of Bivalvia and Conchifera. Therefore, we examined neuronal development in Crassostrea gigas larvae from the appearance of the first sensory cells and their neurite pathways to the formation of the larval nervous system, including innervation of inner organs in oyster pediveligers. Characterizations were made with the neuronal markers serotonin (5-HT) and Phe-Met-Arg-Phe amide (FMRFamide). In addition, we examined immunoreactivity against vesicular acetylcholine transporter (VAChT) and choline acetyltransferase (ChAT) as presumptive markers of acetylcholine-containing neurons.
Mature oysters (Crassostrea gigas) were collected from the estuary at the “Vostok” marine biological station of the National Scientific Center of Marine Biology of the Far East Branch of the Russian Academy of Sciences (NSCMB FEB RAS) (Vostok Bay and Peter the Great Bay of the Sea of Japan 42°48′34,9″ W, 142°54′.18,9″ L, 10 m depth), during July 2014 and July 2015. Standard techniques , with some modifications, were used to obtain larval cultures. Briefly, spermatozoa and mature oocytes were obtained by gonad stripping of adult oysters. Larval cultures were kept at 20 °C in 5-L beakers containing filtered sea water with constant agitation of the water column by an air jet directed at the water surface; the water was changed every 3 days. Starting from 36 h post-fertilization (hpf), larvae were fed microalgae (Isochrysis galbana and Chaetoceros muelleri; 100,000 cells/mL), and an additional mixture of Dunaliella salina and Phaeodactylum tricornutum (50,000 cells/mL) was added to the larval rations after 10 days post-fertilization (dpf). The subsequent stages of larval development were examined under a Zeiss Axio Imager Z2 light microscope (Carl Zeiss, Jena, Germany) equipped with a digital camera (Axio Cam Hrc, Carl Zeiss) using bright field or differential interference contrast techniques (Far East Center of Electron Microscopy, NSCMB, FEB RAS).
For morphological studies, oyster larvae were fixed at the following stages: blastula (12 hpf); early, middle, and late trochophore stages (20, 24, and 28 hpf, respectively); and the early (D-) (36–52 hpf), middle (92–96 hpf), and late veliger stages (5–9, and 15 dpf), as well as the pediveliger stage (28–35 dpf).
Primary and secondary antibodies used in the study
Source, Cat. #
ChAT (choline acetyltransferase)
VAChT (vesicular acetylcholine transporter)
AlexaFluor 555 anti-goat
Life Technologies, A21432
AlexaFluor 488 anti-rabbit
Life Technologies, A21206
AlexaFluor 555 anti-rabbit
Life Technologies, A31572
AlexaFluor 647 anti-mouse
Life Technologies, A31571
AlexaFluor 555 anti-mouse
Life Technologies, A31570
Morphological and Anatomical Features of Crassostrea gigas Larval Development Stages
Time of developmentb
Spheroidal form, slightly conical at the basal region with two obvious invaginations (shell gland and presumptive mouth opening) and prototroch (Fig. 1a). Long cilia form an apical tuft that is present at the apical pole.
Spheroidal form with one invagination (mouth). Shell gland everted (Fig. 1b).
Spheroidal form with well-developed prototroch and telotroch. Shell gland begins to secrete a shell (Fig. 1c). Locomotory organ—prototroch.
Early veliger (D-hinge-stage veliger)
36 hpf—72 hpf
Two D-shaped shells surround the larval body. Digestive tract is complete and consists of mouth, esophagus, stomach, intestine, and anus. Elements of the muscular system (retractors) are well developed. Locomotory organ—velum (Fig. 1d).
92 hpf-96 hpf
Shells increase in size and are slightly elongated. Retractors and two adductors are well developed (Fig. 1e).
after 5 dpf
Right and left shells first demonstrate asymmetry (Fig. 1f). Velum is markedly larger than in the previous stage.
Pediveliger (umbo stage)
after 28 dpf
Umbo is distinguished as highest and most prominent part of each valve of the larval shell. Velum is markedly larger than at the previous stage.
Prior to fixation, larvae were relaxed in a solution of 7% MgCI2 in 0.1 M phosphate-buffered saline (PBS, pH 7.4). After relaxation, larvae were fixed in 4% paraformaldehyde in 1 × PBS for 5 h at 4 °C and rinsed for 15 min in 0.1 M PBS with 0.03% NaN3 three times. Immediately thereafter, larvae at veliger stages were treated with 0.1 M EDTA for 1 h at room temperature and washed in PBS three times to decalcify their shells. Larvae were permeabilized in 1× PBS with 0.03% NaN3 and 1% Triton X-100 (PBST) for 15 min at room temperature and then incubated overnight in blocking solution (10% normal goat serum, 1% bovine serum albumin [BSA] in PBST) at 4 °C to eliminate non-specific binding sites. The larvae were subsequently incubated with primary antibodies (Table 1) at a final dilution of 1:2000 in blocking solution for 3–5 days at 4 °C using a monoclonal antibody against acetylated α-tubulin (Sigma, St Louis, MO, USA) or polyclonal antibodies against FMRFamide or 5-HT (Immunostar, Hudson, WI, USA), and VAChT or ChAT (Millipore, Darmstadt, Germany). After triple rinsing in PBST, the larvae were incubated overnight at 4 °C with secondary antibodies (Table 2), including AlexaFluor 488 goat anti-rabbit IgG (GAR), AlexaFluor 546 goat anti-mouse IgG (GAM), and AlexaFluor 633 GAM (Molecular Probes, Waltham, MA, USA) that had been diluted to 1:1000 in 1× PBS. To detect the larval muscle system, AlexaFluor 555-labelled phalloidin (Molecular Probes, diluted to 1:1000) was used.
For co-localization analysis of FMRFamide and 5-HT, larvae were first incubated with the FMRFamide antibody followed by the AlexaFluor 546 GAR secondary antibody. The larvae were then incubated with a mixture of the anti-5-HT and anti-acetylated α-tubulin antibodies followed by incubation with the AlexaFluor 488 GAR and AlexaFluor 633 GAM antibodies. Each incubation was followed by three 15-min washes in PBST.
Finally, larvae were immersed in glycerol with 2.5% DAPCO antifade (Sigma). At least 50 embryos were examined at each stage using each of the staining protocols, and negative controls were obtained by omitting the primary antibodies to demonstrate absence of specific staining with secondary antibodies. All samples were viewed under Leica SPE (Wetzlar, Germany) or Zeiss LSM 810 confocal laser scanning microscopes. Images were processed with Illustrator CS6 (Karolinska Institute, Solna, Sweden), and 3D reconstructions were created using Imaris 7.0 (Karolinska Institute). Adult oyster tissues were prepared for immunohistochemistry using a previously described protocol .
To describe larval neurogenesis in the present study, we used antibodies (Table 1) against rabbit polyclonal FMRFamide (Cardio-excitatory Peptide, Immunostar) and rabbit polyclonal serotonin (5-HT, Immunostar). FMRFamide and serotonin antibodies from Immunostar were quality control tested using standard immunohistochemical methods. The antiserum demonstrated strongly positive labeling of rat hypothalamus and spinal cord in indirect immunofluorescence assays, and the specificity of FMRFamide and serotonin antibodies has repeatedly been confirmed in bivalve species using immunohistochemistry [5, 15, 23, 24]. FMRFamide and serotonin antibodies are widely used markers for detecting neuronal elements in different invertebrate taxa. Details of the primary and secondary antibodies used are presented in Table 1.
Cholinergic neuronal elements were detected using goat polyclonal choline acetyltransferase antibody (ChAT, Millipore) and goat polyclonal vesicular acetylcholine transporter antibody (VAChT, Millipore). As this antibody has not been previously used in bivalve tissues, we first tested whether the antibody specifically recognized Crassostrea ChAT and VAChT in western blotting and immunohistochemistry of oyster adult tissues and larval materials. As a positive control for ChAT and VAChT labeling, we used total protein extracts from mouse spinal cord. VAChT, but not the ChAT antibody, recognized a band of the expected molecular weight in adult tissues and in whole-mount immunostaining oyster preparations.
All neuronal antibodies used here were combined with antibodies against monoclonal acetylated tubulin (Abcam, Cambridge, MA, USA). This antibody is widely used to detect microtubules in the ciliary systems and nervous system elements of invertebrate taxa .
Morphology of the main developmental stages of Crassostrea gigas
At the middle trochophore stage (24 hpf), faint dotted immunoreactivity appears at the apical region (Fig. 2b), and the number and position of the dorsal cells do not change (Fig. 2b1 and b2). An anteriorly running axon projects further into the ventral side from each dorsal sensory FMRFa-ir cell, following the curvature of the anterior hemisphere, and a total of four cells (2 flask-shape with cilia and two round cells) are now detected in the ventral group (Fig. 2b3-b4 and j).
In the late trochophore stage (28 hpf), FMRFamide-ir becomes more prominent within the bodies and thick apical dendrites of the anterior apical cells, forming the apical organ (AO) (Fig. 2c, c5, c6 and j). At this stage, the positions of the paired sensory dorsal (Fig. 2c1 and c2) and ventral cells (Fig. 2c3, 2c4) do not change. Long axons of dorsal cells run underneath the flask-shaped anterior cells (Fig. 2c), and a brightly visible growth cone of the right dorsal cell axon reaches the ipsilateral ventral cell group (Fig. 2c7 and j).
By the early D-hinge veliger stage (36 hpf), the cell bodies of the apical cells and their basal neurites comprise a compact AO that has no connections with dorsal or ventral cells and their processes (Fig. 2d, d1, and e); the dendrites of the apical cells bear long cilia (Fig. 2e1). At this stage, two cells are visible in each of the ventral groups (Fig. 2d2, d3, and E insets), and long neurites of both the right and left dorsal cells reach the ipsilateral groups of the ventral cells (Fig. 2d2 and d3). Thus, two parallel (left and right) processes of dorsal and ventral FMRFa-ir cells form the anlagen of the paired ventral nerve cords (Fig. 2d, d2, d3 and j).
In the early veliger stage (52 hpf), FMRFa-ir labeling within the cell bodies of both the early dorsal and ventral cells becomes punctate (Fig. 2f), while the long axons of the dorsal neurons are still visible. Processes emanating from the AO run along this connection (Fig. 2f1-f3), and solitary small immunopositive neurons appear posteriorly (Fig. 2f-f3 and j). Later, (60 hpf) four flask-shaped apical neurons (asterisks) of AO were detected (Fig. 2f, insert).
At the middle veliger stage (96 hpf), the cell bodies of the dorsal groups of early-appearing cells are no longer detected (Fig. 2g), and there is also no sign of the dendritic knob of the dorsal cells on the dorsal side of the larval body. The basal processes of the AO neurons are organized into a compact central neuropil, and six flask-shaped cells form the complex spatial structure, i.e., the apical/cerebral ganglion (AO/CG) (Fig. 2g1, 2g2 and j). A single neurite extends from the neuropil of the AO/CG toward the velum, and two commissures connect the right and left ventral cords (Fig. 2g). The anlagen of the pedal ganglia (PG) appear in the region of the developing foot and along each ventral cord (Fig. 2g3 and g4) enveloping the esophagus. Solitary posterior cells are located at the caudal end of each ventral cord (Fig. 2g5).
At the late veliger stage (9 days post fertilization, dpf), FMRFa-ir cells within the apical/cerebral ganglion become more compact (Fig. 2h and h1), and neurites extending from the AO/CG run to the dorsal side of the velum. The anlagen of the PG (Fig. 2h2 and h3) are located adjacent to the paired ventral cord, which is connected by two commissures (Fig. 2h) and terminates with posterior neurons (Fig. 2h4 and h5). Thus, the 9-dpf late veliger stage possesses the AO/CG complex, paired ventral cords with interconnecting commissures, paired pedal ganglia, and posterior neurons (Fig. 2i). We further characterized the neuroanatomy of the FMRFamide system and innervation patterns in the pediveliger, as is detailed later in the paper (Fig. 6).
In the early D-hinge veliger stage (36 hpf), three 5-HT-ir cells are present within the AO (two flask-shaped and one round, Fig. 3b, b1-b2 and g), and tiny neurites extend from the 5-HT-ir apical cells and run towards the velum. By 52 hpf, long immunopositive neurites extend anterior−dorsally and posterior−ventrally from the AO and run along the ventral sides of the larval body (Fig. 3c and g). Later, two posterior−ventral neurites run along the ventral part of the larval body. The AO consists of three flask-shaped and two round cells, and basal processes of AO neurons are organized into a compact central neuropil (Fig. 3c and c1-c3). Two posterior−ventral neurites from cells of the AO run in parallel along the ventral part of the larval body organizing the ventral nerve cords (VNC) (Fig. 3d and g).
By the late veliger stage (9 dpf), the main 5-HT-ir structures, i.e., the cerebral ganglion and the pedal ganglia, are located along the ventral nerve cords. In addition, small immunoreactive posterior neurons are located near the caudal end of each ventral cord. At this and later stages, CG neurites strongly innervate the velum region (Fig. 3e).
In the later veliger stage (15 dpf), commissures connecting the left and right nerve cords are visible, and as in the previous stage, clearly visible cerebral and pleural parts of fused central ganglia are located on top of the ventral cords (Fig. 3f, f1-2 and g). Peripheral innervation of visceral organs is extensive, including cerebro−pleural ganglia innervation of the velum and pedal ganglia innervation of the foot and the rest of the visceral mass (Fig. 3f and f2).
Characterization of VAChT antibodies and VAChT-immunoreactivity in Crassostrea gigas tissues
To examine the spatiotemporal distribution of VAChT-ir nerves, we monitored its expression pattern from the trochophore (24 hpf) to veliger (7 dpf) stages. No VAChT-ir elements are observed in trochophores (data not shown); instead, they first appear in cell bodies and processes of the apical cells and in solitary posterior cells at the veliger stage (44 hpf; Fig. 4c, c1-2 and f). Cell processes run toward each other along the ventral edge of the larvae. At the 92 hpf veliger stage, VAChT-ir become more intense within the AO/CG complex, pedal ganglion, and along the caudal portion of the ventral cords, and single fibers appear to innervate the velum (Fig. 4d and f). In the 7-dpf veliger stage, VAChT-ir cells are found in the main larval neuronal structures: the AO/CG complex, ventral cords, PG, and posterior neurons. Immunoreactive fibers appear in the velum, developing foot, and around the mouth (Fig. 4e and f). In contrast, no positive ChAT bands and no immunopositive signal were found in any tested adult or larval oyster tissues (Additional file 2: Figure S2).
Mutual arrangement of FMRFa-ir, 5-HT-ir, and VAChT-ir
Double-labeling demonstrated the absence of FMRFa-ir and VAChT-ir co-localization within cell bodies in the veliger stages (Fig. 5e and f). In the 44-hpf veliger stage, the VAChT-ir cells are surrounded by FMRFa-ir cells in the AO (Fig. 5e, e1 and k), and small, solitary caudal VAChT-neurons send the process anteriorly along the ventral FMRFa-ir fiber. The locations where FMRFa-ir and VAChT-ir processes intertwine appear partly co-localized (Fig. 5e2 and f1). In the 7-dpf veliger stage, both VAChT-ir and FMRFa-ir are detected in the neurons of the AO/CG complex and PG as well as in the bodies of solitary posterior neurons (Fig. 5f1-3, g, front view and Fig. 5k). Partial co-localization of FMRFamide immunoreactivity and VAChT immunoreactivity within the ventral cords, commissures, and apical neuropil is due to tightly contiguous processes exhibiting different immunoreactivities (Fig. 5f1, f2 and Fig. 5g). Double-labeling indicates the absence of 5-HT-ir and VAChT-ir co-localization within early neurons in the veliger stages. At the 44-hpf veliger stage, the AO contains closely compacted VAChT-ir and 5-HT-ir neurons, while solitary posterior cells only exhibit VAChT-ir (Fig. 5h and k). In the 92-hpf and 7-dpf veliger stages, VAChT and 5-HT are expressed within the cell bodies of the AO/CG, and their basal fibers come into close contact in the neuropil (Fig. 5i, j inset, k). VAChT- and 5-HT are detected within different cell bodies of the PG (Fig. 5i1, j1 and k). In the 7-dpf veliger, VAChT-ir and 5-HT-ir fibers emanating from the AO/CG innervate different portions of the velum (Fig. 5k). Only VAChT-ir, but not 5-HT-ir, is detected in the caudal posterior cells at the 7-dpf stage (Fig. 5j1).
Neuromorphology of the pediveliger and peripheral innervation of larval tissues
5-HT immunostaining can be observed in the main ganglia (cerebro-pleural, pedal, and visceral). FMRFa-ir and 5-HT-ir cells are equally present within the cerebro-pleural ganglia, while 5-HT-ir cells are prevalent within the pedal ganglia (Fig. 6b). In the 35-dpf pediveliger, the cerebral and pleural regions of the cerebro-pleural ganglion are clearly visible (Fig. 6c inset), and the velum receives rich innervations from the cerebro-pleural and pedal ganglia (Fig. 6b, c and c1). Nerves emerging from the pedal and visceral ganglia show intense varicose ramifications within the anterior and posterior adductors, respectively (Fig. 6c and c1).
Comparative neuromorphology of Bivalvia larvae
Data on bivalve neurogenesis are limited and often restricted to accounts of single species and stages of development performed by light microscopy [8, 27, 28, 29]. Comparative analysis of immunocytochemical data on bivalves are still scare and restricted to the analysis of the development of serotonin-, FMRFamide- and catecholamine-ir, small cardioactive peptide-like nerve elements in larvae of the mussels Mytilus trossulus  and Mytilus edulis , scallop Placopecten magellanicus , oyster Crassostrea virginica , clam Spisula solidissima  and mussel Dreissena polymorpha . Unfortunately, no published neurogenesis data have been reported for the Protobranchia, Palaeoheterodonta, Archiheterodonta, and Anomalodesmata clades. This lack of data makes it difficult to identify the ground pattern of the larval nervous system and to clarify phylogenetic relationships between bivalve species. Nevertheless, using published, non-published data and novel neuroanatomical studies presented here, we have inferred a ground pattern of the larval nerve system inherent to some Bivalvia clades: Pteriomorpha and Euheterodonta (Fig. 7).
Cell composition of apical organ in bivalves
The apical organ (AO) is the most conserved larval sensory structure and has plesiomorphic features of high importance for studies of animal evolution and relationships. In the bivalve species (Pteriomorpha and Imparidentia), the AO develops from early larval stages (early trochophore) and consists of ciliated flask-shaped and rounded 5-HT-ir as well as flask-shaped FMRFamide-ir cells [5, 16].
In addition to serotonin-ir cells, the AO of Crassostrea gigas larvae contain two, four, and later six FMRFamide-ir flask-shaped cells in veliger larvae (Fig. 8). Five FMRFamide-ir cells were detected in Mytilus trossulus AO at D-veliger stage . It is possible that this large number of FMRF-ir cells in the AO represent a common morphological criterion for bivalves or Pteriomorpha species. However, without additional information of the FMRFamide containing cells within AO in Palaeoheterodonta, Archiheterodonta, and Anomalodesmata clades, it is difficult to reconstruct the ground pattern of the nervous system for Bivalvia.
Ventral nerve cord development and main ganglia
A distinctive feature of early neurogenesis of the oyster C. gigas is that the first-appearing peripheral sensory FMRF-ir cells are located posttrochally at the dorsal and ventral sites of the larvae (Fig. 8). All cells have short sensory dendrites and the long axons form a scaffold of two paralleled ventral nerve cords (VNC), which will be followed by growing axons of caudally positioned VAChT-expressing cells. Such morphology is very prominent and is characteristic for most of the pioneer neurons described in the development of other trochophore animals (Nezlin and Voronezhskaya ). As in other described trochophore larvae, both early dorsal and ventral groups are probably transient and resorbed after fulfilling their routing function. In a previous investigation of Mytilus trossulus larvae, all early peripheral FMRF-ir cells were found in the episphere at the trochophore stage, with the FMRFa-ir pioneering axons projecting to the ventral part of the larvae to form the VNC in the veliger . Thus, in both oyster and mussel larvae, FMRFa-ir processes of early cells track in a dorso-ventral direction to form the anlagen of the VNC, which will be followed by neurites of cells differentiating later within the AO and main ganglia rudiments. Later commissures connect left and right longitudinal cords producing a rope-ladder-like (segmented) ventral nerve structure. Therefore, while there appear to be differences in the number and position of early peripheral FMRFa-ir cells and 5-HT-ir cells between bivalves and in spite of species-specific early patterns of neurodevelopment, the same final result of an AO and paired VNC is realized (Fig. 9).
Interestingly, serotonin Abs also mark the VNC in oyster with two commissures appearing at the late veliger stage, whereas serotonin immunostaining is restricted to the cerebral ganglion in mussel M. trossulus veliger/pediveliger . In contrast to M. trossulus, Dreissena polymorpha has a 5-HT-ir VNC and cerebro-visceral connectives starting from the veliger stage . In addition, in the oyster Crassostrea virginica, small cardioactive peptide (SCP-like neuropeptide) also label the VNC with two commissures at the late veliger stage . These also confirm the VNC as a common neuronal morphological structure for all investigated Bivalvia larvae (Fig. 9) and a feature for comparative analysis within lophotrochozoan clades.
All larval ganglia appear along the VNC as paired or fused (epiathroid) clusters in late veliger and pediveliger larvae. The maturation of the anlagen of the cerebral ganglion (CG) is characterized by loss of flask morphology of FMRFa-ir and 5-HT-ir cells in the AO and the appearance of abundant serotonin innervation of the velum in the late oyster veliger (Fig. 8). Whether the apical organ cells are precursors of neurons of the CG remains unclear. Based on our data, we cautiously assume that cells of the AO are incorporated within the CG and became a part of the juvenile oysters CG. This suggestion is supported by data from M. trossulus neurogenesis representing continuity of cilia flaked-shaped cells of the AO into the cerebral ganglion , and by recent reconstructions of the serotoninergic nervous system in Dreissena polymorpha .
The anlagen of the pedal ganglia (PG) appears along each ventral cord in the region of the developing foot in the oyster Crassostrea gigas and mussel M. trossulus veliger [present data and 5]. It seems that the precursors of PG neurons are rounded FMRFa-ir and 5-HT-ir ventral neurons appearing in early and late oyster veligers (Fig. 8). PG in the late veliger of Crassostrea virginica contains SCP-like neuropeptide-positive cells . In M. trossulus, neurons of PG include catecholamine-ir and FMRFa-ir cells, whereas in D. polymorpha and Spisula solidissima, PG neurons were not detected at any developmental stages using 5-HT and tubulin antibodies. If true, the absence of pedal neurons could be an apomorphic character of the Heterodonta clade. However, the absence of PG neurons (or positive immunostainings) in Spusula solidissima and D. polymorpha veligers is unconvincing, because both species have a foot that exhibits activity in pediveliger larvae [32, 33]. Future studies of neurogenesis using FMRFamide and VAChT Abs, as introduced herein, as well as other neuronal markers are needed to determine conclusively if there is a PG neurons in these species.
Visceral ganglia (VG) are the last ganglia appearing in oyster development (Fig. 6). These paired ganglia contain FMRFamide-ir and 5-HT-ir neurons. In M. trossulus, visceral neurons express FMRFamide-ir and catecholamine-ir, but not 5-HT-ir. VG in D. polymorpha was not found in the veligers (144 hpf), because of absent or belated ganglia at this stage . The data on the presence or absence of the VG in S. solidissima at the early veliger stage (48 hpf) is needed to verify the neuronal ground pattern for Bivalvia. Thus, VG is a common morphological structure appearing at late developmental stages and is one of the main criteria defining the Pteriomorpha species.
VAChT-ir nerve elements
Using a VAChT antibody, this study provides the first characterization of the acetylcholine (ACh)-ir nervous system in bivalves. We found VAChT-ir in the AO, posterior neurons and, later, in the cerebral, pedal ganglia, and VNC (Fig. 8). Acetylcholine has been detected in the larvae of M. edulis using a histochemical technique, as well as in the larvae of the annelid Platynereis dumerilii by whole-mount in situ hybridization (WMISH) [3, 10, 34, 35, 36]. Acetylcholinesterase-positive ciliary sensory-like cells have been identified as the earliest rudiments of the pedal ganglia and primary byssus glands of mussels by Rainery and Ospovat. In addition, Raineri identified three pairs of posttrochal ganglia rudiments, which were considered to be the visceral, parietal, and pleural ganglia located along the longitudinal nerve cords in the trochophore larva. The author claimed that the cerebral ganglia differentiate adjacent to AO, and the nerve network appears at the velum after the differentiation of the first pedal nerve rudiments. While histochemical method used in this study did not allow unequivocally identify nerve cells and their processes, the main pattern of the mussel nervous system was similar to that observed later using immunochemical markers. It is known that the early larval neurons in the annelid P. dumerilii express the cholinergic marker VAChT, as determined by WMISH at the trochophore stage, and VAChT expression later expands to the VNC and brain, as well as the neurons of the peripheral nervous system (which is associated with the appendages) [35, 36]. Further research on Bivalvia neurogenesis, with particular focus on VAChT-ir structures, is required in order to understand the cell composition of the AO and central and peripheral nervous systems in veligers, and to expand understanding to their role in the regulation of physiological functions.
Innervation of the velum by 5-HT-ir and FMRFamide-ir neurites (which form the velar nerve ring or plexus pattern) originating from the AO/CG has been demonstrated in various gastropods [37, 38, 39, 40] and bivalves [2, 5, 41]. Early veligers of Crassostrea gigas present dorsal velum innervation by both FMRFamide-ir and 5-HT-ir neurites projecting from the AO/CG. VAChT-ir neurites form the neural ring from the ventral side of the AO. Recent data show that Dreissena polymorpha has 5-HT-ir dorsal velum innervation  very similar to what is observed in C.gigas. Later, C. gigas larvae demonstrate an abundant 5-HT-ir plexus-like innervation pattern similar to that detected in the pediveliger of Mytilus edulis . Catecholaminergic cells and their neurites are also found in the velum of the veligers and pediveligers of Placopecten magellanicus, M. trossulus, and M. edulis [4, 5]. In addition, SCP-like peptides label processes in velum that originate from cerebral ganglion (CG) . Thus, the velum of Bivalvia larvae is under neuronal control by different neuropeptides and neurotransmitters in veligers and receives abundant innervation in pediveligers of Pteriomorpha species. Unfortunately, there are not enough data for comparative morphological analysis of velum innervation patterns among other Bivalve clades and future investigations are needed, including investigations of Hererodonta species (D. polymorpha and S. solidissima,) in late developmental stages.
We found that smooth adductor muscles are innervated by serotonergic and FMRFamidergic neurons in the pediveliger larvae of the oyster C. gigas. The same innervation patterns were shown in Mytilus trossulus pediveligers . In C. gigas and M. trossulus, FMRFamide-ir fibers originate from the PG and innervate the anterior adductor in pediveligers, while processes from the visceral ganglia innervate the posterior adductor in C. gigas. 5-HT, as a key regulator catch-contraction of adult adductors, was detected in oyster pediveligers innervating anterior and posterior adductors. Interestingly, 5-HT innervation was not detected in the anterior adductor of the mussel M. trossulus pediveliger [5, 23]. Further research on neuromuscular interactions in pediveligers is required for an understanding of the neuronal control of catch contractions and innervation patterns after metamorphosis.
Here, we made the novel observation of FMRFamide immunoreactivity throughout the enteric nervous system, including the digestive system, and innervations of paired gill rudiments in oyster pediveligers. We suggest the FMRFamide-ir cell bodies and fibers either originate from the VG or arise independently of ganglia (autonomous or local genesis) from ganglia manner. It should be noted, that 5-HT does not innervate these organs in late larvae. However, VAChT-ir neurites also participate in innervation of the mouth region in veligers.
To summarize the common characteristics of the bivalves studied herein: (1) bivalve larvae show an apical organ composed of three 5-HT-ir flask-shaped cells, which is an ancestral feature of three to five subclasses of Bivalvia, and two 5-HT-ir round cells that are detected in Pteriomorpha and Imparidentia; (2) the AO in Pteriomorpha larvae show six FMRFamide-ir flask-shaped cells (apomorphic feature); (3) in the post AO-period, the larval nervous system of bivalves have ganglionic-like structure characteristic of the adult nervous system; (4) a paired ventral nerve cord is a common feature (plesiomorphic) of the bivalve larvae investigated herein (Figs. 7 and 8). We suggested these features were inherent to the larva of the last common bivalve ancestor (LCBA). To understand additional morphological criteria, such as the VAChT-ir nervous system, innervation patterns, enteric nerve system, more data are necessary for all clades of Bivalvia.
Comparison of larval neuromorphological structures in Mollusca
Phylogenomic analysis has divided all Mollusca into two main Subphyla: Aculifera, which comprises Neomeniomorpha, Chaetodermomorpha, Polyplacophora, and Conchifera, which comprises Monoplacophora, Cephalopoda, Scaphopoda, Gastropoda, and Bivalvia . Despite extensive molecular and morphological investigations of conciferan taxons, our understanding of their positions on the phylogenetic tree and their interrelationships remain unclear. Here, we detail the main morphological criteria.
Apical organ (AO) in Mollusca
Among larval structures, the AO is a temporary sensory organ found in the early stages of development among phylogenetically diverse animal groups that presumably acts as a chemosensory structure that resorbs before or after metamorphosis. An AO consisting of flask-shaped cells has also been reported in most other lophotrochozoans, such as annelids, nemerteans, ectoprocts and the brachiopods, and phoronids [43, 44, 45, 46, 47, 48, 49]. 5-HT- and/or FMRFamide-ir cells can be found in the AO of both Aculifera and Conchifera groups of Mollusca, but their morphology (round, flask-shaped cells, cilia) and their number are variable among mollusc groups. We found/hypothesize that the presiomorphic feature in bivalves is three 5-HT-ir flask-shaped cells in the AO, which has also been described for basal (Protobranchia) and more advanced groups (Euheterodonta). The same number of 5-HT-ir flask-shaped cells in the AO was found in basal gastropod molluscs, patellogastropods Tectura scutum , and Lottia cf. kogamogai . Moreover, in addition to flask-shaped cells in the AO, Lottia cf. kogamogai have two 5-HT-ir round cells, as is observed in the Crassostrea gigas AO. The same cellular composition of the AO is present in the larval abalone Haliotis kamtschatkana, which is a Vetigastropoda . In opistobranch gastropod larvae, such as Aplysia californica (Anaspidea) [37, 53], Aeolidiella stephania and Phestilla sibogae, Berghia verrucicornis, Melibe leonina, and Tritonia diomedea (Nudibranchia) [39, 54, 55], three 5-HT-ir flask-shaped and two non-sensory round cells were detected. The same cellular composition of the AO (three flask-shaped + two round 5-HT-ir cells) was found in the caenogastropod Euspira lewisii (Naticoidea)  while other species of this clade have differences in the AO cell composition (3 + 3; 2 + 2; 1 + 2), . Although data on FMRFamide-ir cells of the AO are scant for gastropods, bivalves and other clades of molluscs, the number of FMRFamide-ir cells in the AO corresponds to those in some caenogastropods and nudibranchs [38, 55]. Therefore, most gastropod groups have three 5-HT-ir cells in the AO with or without round cells and show a similar arrangement and number as that in Bivalvia AO found in larvae of the protobranchs, pteriomorphs, and imparidents (Fig. 9).
Larval neuromorphology of bivalves and scaphopods (Antalis entails) have fewer similarities in the 5-HT-ir system: the number of 5-HT-ir flask-shaped cells in Antalis entails larva start from two and reach four cells in the AO (plus two lateral cells) . This number of serotonin cells was found in Bivalvia and gastropods larvae only in early trochophores. There are no published data on the AO cell composition in Monoplacophora. Thus, the ground pattern of AO inherent to Conchifera species’ sensory organ includes two to three flask-shaped 5-HT-ir cells.
Among Aculifera taxons, the most studies Polyplacophora species are Mopalia muscosa and Ischnochiton hakodadensis [18, 57], which have 8–10 flask-shaped 5-HT-ir cells in the AO. These cells are surrounded by dorsal and ventral cells, and six FMRFamide-ir AO cells and peripheral cells . However, early stages of Mopalia muscosa (55 hpf) clearly show three flask-shaped 5-HT-ir cells in the AO and additional peripheral sensory cells [18, 58]. In the Neomeniomorpha (Solenogastres) Wirenia argentea and Gymnomenia pellucida, the AO contains only two 5-HT-ir cells and two FMRFa-ir cells . These data support there being a simple AO structure in the aculiferan ancestor, and are contrary to postulates of a complex structure of the AO as an ancestral condition for Aculifera (8–10 cells in polyplacophores). For the most complete comparative analysis of neurogenesis in bivalves plus Aculifera and Conchifera and to reconstruct the putative ground pattern of the last common ancestor of Mollusca (LCAM), further detailed and high-quality studies of Neomeniomorpha and Chaetodermomorpha are required.
Morphology of larval nervous system and reconstruction of possible ground pattern
Comparative morphology and recently emerging genomic data continue to be important tools in our efforts to determine the putative pattern of the Last Common Ancestor of Bivalves (LCAB), Conchifera (LCAC), Aculifera (LCAA), and Last Common Ancestor of Mollusca (LCAM), and to better define phylogenetic relationships between molluscan clades. Comparative analysis of larval neuronal structures showed that Bivalvia larvae have striking similarities in organization and arrangement of the AO (three flask-shaped cells), the position of the main ganglia (cerebral/pleural, pedal, and visceral) along the VNC, and the presence of commissures forming a rope-ladder-like (segmented) nervous system. These features are shared by gastropod larvae with minor modifications (additional ganglia) and are a unifying features of the node-base group Pleistomollusca. Thus, morphology-based analysis supports phylogenomic data claiming that Bivalvia and Gastropods are sister-taxons  having a common ancestor (Pleistomollusca ancestor).
The position of Scaphopoda on the phylogenic tree of Conchifera is not clear. Scaphopoda are considered a potential sister group of Pleistomollusca  and Bivalvia (Diasomal concept [17, 61]. There may also be a cephalopod-scaphopod relationship . The larval nervous system of scaphopods is ganglion type with a ventral (pedal) nerve cord . From the above, the nervous system of the LCAC presumably had a paired ventral nerve cord with ganglia located along it. The morphology of the LCAA remains elusive and developmental evidence of neurogenesis from recent aplacophoran representatives is still lacking. Popyplacophora and Neomeniomorpha species show differences in cellular composition of the AO and ganglia, as well as in order of appearance of the nerve cords [57, 58, 59]. However, despite these differences, we can conclude that the ground pattern of the nervous system of the LCAA is an AO containing two to three 5-HT-ir flask-shaped cells (2 flask-shaped cells in Neomeniomorpha, as we do not take into account alpha-tubulin ciliary staining ), three flask-shaped cells at early stages of development , and a paired VNC with commissures (segmented) and perikarya positioned along it. If we consider that the polyclocophoran AO of Aculifera and Mollusca generally consists of 8–10 5-HT-ir cells as a basal pattern, then secondary simplification of the cellular composition of the AO in Neomeniomorpha and all Conchifera clades is a general evolutionary trend for Mollusca. To determine the LCAA with greater accuracy and to finally shed light on molluscan ancestor neuromorphology, future investigations on mollusc neurogenesis are needed.
The question of the origin of the VNC as a ground trait for the LCAM is one the main and most interesting question in the evolution of bilaterians. In the oyster C. gigas, we found that FMRFamide-ir sensory peripheral cells appear posttrochaly and send their neurites from the dorsal to ventral side of early larvae, thereby playing the role of a scaffold for the VNC (Fig. 9). Pioneer peripheral cells and their neurites in the primordium of the VNC were detected in several lophotrochozoa, with the cell bodies located in the episphere in bivalves (scenario 1 on Fig. 9 ), in the hyposphere (dorsally) in annelids and bivalves (scenario 2 on Fig. 9 ([31, 62, 63, 64], present data)), and in the hyposphere (caudally or caudo-laterally) in gastropods and polyplacophores (scenario 3 on Fig. 9 [57, 65, 66]). Important, the primary location of the sensory pioneer neurons in larvae is largely irrelevant, while the way of their neurites will ultimately mark a position of VNC, albeit through different pathways (dorso-ventral direction or posterior-anterior direction). Moreover, the presence of a similar dorsoventral arrangement of transcription factors along the nerve cords of Spiralia supports the hypothesis of ancestrality of the VNC in Bilateria .
Here, we described the dynamics of neurogenesis during the development of the Pacific oyster Crassostrea gigas using 5-HT and FMRFamide antibodies, as well as VAChT antibodies for detection of cholinergic neurons (a first in bivalves). We conducted detailed neuroanatomical and axon tracing studies through the early trochophore and veliger stages until the pediveliger stage. Comparative morphological analysis of Crassostrea gigas neurogenesis with other Bivalvia, revealed common conserved characteristics as well distinctions in early neuronal specialization of the FMRFamide-ir and 5-HT-ir cells in trochophore and veliger larvae. We conclude that the sensory AO in bivalves, as well as in Conchifera, consists of three flask-shaped cells, and that the larval nervous systems are of the ganglion type and have paired VNC with commissures. We propose that axons of early peripheral neurons laying down the pathway for VNC, which is the most conservative larval structure. Our morphological data support phylogenomic data indicating a closer Bivalvia-Gastropoda sister group relationship than the Bivalvia-Scaphopoda (Diasoma) group relationship and raise questions about the validity of the Tetraneuralia concept for bivalves.
We are grateful to the staff of the Vostok Biological Station (NSCMB FEB RAS) and the staff of the Far East Centre of Electron Microscopy and the Optical Research Group of IDB RAS for their generous assistance. We also thank Olga Kharchenko for the drawings.
This work was supported by the Russian Science Foundation, grant № 14-50-00034 (Immunocytochemistry), President scholarship (СП-2890.2016.4) (Western blot). The work of EEV was supported by Russian Foundation for Basic Research, № 18-04-01213 (Early sensory neurons).
Availability of data and materials
All authors had full access to all the data in the study and take responsibility for its integrity and the accuracy of the analysis. Study concept and design: OY, EV, and VD. Data acquisition: OY and VD. Western blot data: OS. Data analysis and interpretation: OY, EV, and VD. Drafting of the manuscript: OY, EV, and VD. Drawings: EV. Administrative, technical, and material support: VD and OY. Study supervision: VD. All authors read and approved the final manuscript.
The field studies did not involve endangered or rare invertebrate species. No specific permissions were required to access the marine area, as it falls within Russian state-owned land.
Consent for publication
All authors agree on the submission and publication of this paper and its included figures and tables.
The authors declare that they have no competing interests.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
- 2.Dame RF. Ecology of marine bivalves: an ecosystem approach. 2nd ed. New York: CRC press; 2011.Google Scholar
- 8.Galtsoff PS. The American oyster. Fish Bull (Wash DC). 1964;64:1–480.Google Scholar
- 9.Koolakovsky EE, Phlyachinskaya LP. Formation of elements of the regulatory systems during larval development of Mytilus edulis. Zoologichesky Zhurnal. 1993;72:20–8. (in Russian)Google Scholar
- 12.Flyachinskaya LP, Raikova OI. Investigation of the pediveliger's nervous system in bivalve mollusc Mya arenaria. Zoologichesky Zhurnal. 2014;93:489–96. (in Russian)Google Scholar
- 13.Hickman RW, Gruffydd LLD. The histology of the larva of Ostrea edulis during metamorphosis. In: Crisp DJ, editor. The Fourth European Marine Biology Symposium. London: Cambridge University Press; 1971. p. 281–94.Google Scholar
- 16.Pavlicek A, Schwaha T, Wanninger A. Towards a ground pattern reconstruction of bivalve nervous systems: neurogenesis in the zebra mussel Dreissena polymorpha. Org Divers Evol. 2018;18:101–14.Google Scholar
- 17.Salvini-Plawen LV, Steiner G. Synapomorphies and plesiomorphies in higher classification of Mollusca. In: Taylor JD, editor. Origin and evolutionary radiation of the Mollusca. New York: Oxford University Press; 1996. p. 29–51.Google Scholar
- 25.Schmidt-Rhaesa A, Harzsch S, Purschke G. Structure and evolution of invertebrate nervous systems. 1st ed. Oxford: Oxford University Press; 2016.Google Scholar
- 27.Erdmann W . Untersuchungen iiber die Lebensgeschichte der Auster. Nr. 5. Ober die Entwicklung und die Anatomie der ansatzreifen Larve von Ostrea edulis mit Bemerkungen iiber die Lebensgeschichte der Auster." Wiss. Meeresunters. N.F., Abt. Helgoland, Bd. XIX, Abh. Nr. 6, S. 1—25, 1934.Google Scholar
- 28.Elston R. Functional anatomy, histology, and ultrastructure of the soft tissues of the larval American oyster, Crassostrea virginica. Proc Natl Shellfish Ass. 1980;70:65–93.Google Scholar
- 29.Bayne BL. The biology of mussel larvae. In: Bayne BL, editor. Marine mussels: their ecology and physiology. Cambridge: Cambridge University Press; 1976. p. 81–120.Google Scholar
- 30.Ramsmayer DP. Neurogenesis in Nucula tumidula and Kurtiella bidentata (Mollusca: Bivalvia) as revealed by immunocytochemistry and confocal laser scanning microscopy. Masterarbeit: Universität Wien; 2014.Google Scholar
- 32.Walker RL, O'Beirn FX. Embryonic and larval development of Spisula solidissima similis (Say, 1822) (Bivalvia: Mactridae). Veliger. 1996;39:60–4.Google Scholar
- 41.Plummer JT. Bivalve neural development. MSc. Thesis, Department of Physiology and Biophysics, Dalhousie University, Halifax, Nova Scotia; 2002.Google Scholar
- 56.Wanninger A, Haszprunar G. The development of the serotonergic and FMRF-amidergic nervous system in Antalis entalis (Mollusca, Scaphopoda). Zoomorphology. 2003;122:77–85.Google Scholar
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.