Comparison of ventral organ development across Pycnogonida (Arthropoda, Chelicerata) provides evidence for a plesiomorphic mode of late neurogenesis in sea spiders and myriapods
Comparative studies of neuroanatomy and neurodevelopment provide valuable information for phylogenetic inference. Beyond that, they reveal transformations of neuroanatomical structures during animal evolution and modifications in the developmental processes that have shaped these structures. In the extremely diverse Arthropoda, such comparative studies contribute with ever-increasing structural resolution and taxon coverage to our understanding of nervous system evolution. However, at the neurodevelopmental level, in-depth data remain still largely confined to comparably few laboratory model organisms. Therefore, we studied postembryonic neurogenesis in six species of the bizarre Pycnogonida (sea spiders), which – as the likely sister group of all remaining chelicerates – promise to illuminate neurodevelopmental changes in the chelicerate lineage.
We performed in vivo cell proliferation experiments with the thymidine analogs 5-bromo-2′-deoxyuridine and 5-ethynl-2′-deoxyuridine coupled to fluorescent histochemical staining and immunolabeling, in order to compare ventral nerve cord anatomy and to localize and characterize centers of postembryonic neurogenesis. We report interspecific differences in the architecture of the subesophageal ganglion (SEG) and show the presence of segmental “ventral organs” (VOs) that act as centers of neural cell production during gangliogenesis. These VOs are either incorporated into the ganglionic soma cortex or found on the external ganglion surface. Despite this difference, several shared features support homology of the two VO types, including (1) a specific arrangement of the cells around a small central cavity, (2) the presence of asymmetrically dividing neural stem cell-like precursors, (3) the migration of newborn cells along corresponding pathways into the cortex, and (4) the same VO origin and formation earlier in development.
Evaluation of our findings relative to current hypotheses on pycnogonid phylogeny resolves a bipartite SEG and internal VOs as plesiomorphic conditions in pycnogonids. Although chelicerate taxa other than Pycnogonida lack comparable VOs, they are a characteristic feature of myriapod gangliogenesis. Accordingly, we propose internal VOs with neurogenic function to be part of the ground pattern of Arthropoda. Further, our findings illustrate the importance of dense sampling in old arthropod lineages – even if as gross-anatomically uniform as Pycnogonida – in order to reliably differentiate plesiomorphic from apomorphic neurodevelopmental characteristics prior to outgroup comparison.
KeywordsNervous system Ventral nerve cord Cell proliferation Evolution 5-bromo-2′-deoxyuridine 5-ethynyl-2′-deoxyuridine Callipallenidae Ammotheidae Pycnogonidae Phoxichilidiidae
central nervous system
phosphorylated histone H3
ventral longitudinal tract
ventral nerve cord
walking leg ganglion
walking leg neuromere
The astounding species diversity of arthropods and the many different forms of their segmented bodies have fascinated zoologists for many centuries. Resolving the phylogenetic relationships within this taxon and understanding the evolutionary transformations leading to its extreme morphological diversification has proved to be an enormous challenge, with some issues persisting to this day [1, 2]. With the advent of molecular phylogenetics and its rapid advances in the last two decades, some traditional morphology-based arthropod relationships have been seriously questioned. As a result of discrepancies between morphological and molecular analyses, (re)investigation of morphological character complexes has been intensified in order to critically evaluate the evidence for debated nodes in the arthropod tree. For instance, scrutiny of different features pertaining to nervous system development and adult neuroanatomy (e.g., [3, 4, 5, 6]) helped to clarify the relationships of the major mandibulate groups Myriapoda, Hexapoda and crustaceans: in agreement with molecular evidence (e.g., [7, 8, 9]), the neural characters consolidated support for the taxon Tetraconata (paraphyletic crustaceans + Hexapoda), thereby overhauling the traditionally advocated clade of myriapods and hexapods (see  for review). Over the last decades, the assessment and interpretation of neuroanatomical and neurodevelopmental character complexes in the context of arthropod phylogeny and evolution – more recently even in fossil representatives – has morphed into a prospering field, occasionally referred to as “neurophylogeny” or “neural cladistics” [10, 11, 12, 13, 14, 15, 16, 17].
In this context, numerous studies on initial embryonic neurogenesis in chelicerate and mandibulate arthropods and their close relatives Onychophora have provided us with many details on the cell types and dynamics involved, as well as the gene network governing the neurogenic processes (see [18, 19, 20] for reviews). Yet, recent investigations of neurogenic processes in more advanced developmental stages – including free-living postembryonic instars in groups with anamorphic development – are still restricted to few taxa outside of Hexapoda. Especially the paucity of data in representatives of the chelicerate and myriapod lineages hampers well-founded comparison of late neurogenesis across arthropods.
Hence, we set out to expand our understanding of the pycnogonid VOs and their function during postembryonic development in five pycnogonid species in addition to Meridionale sp., all of which belonging to different genera (Fig. 1). Two of these species are thought to be closely related to Meridionale sp. (Fig. 1a, b) with which they are traditionally placed in the Callipallenidae that may, however, represent a paraphyletic assemblage (Fig. 1g) (e.g., [40, 41]). The remaining three species are representatives of different taxa that are well-separated in the pycnogonid tree (Fig. 1c, d, f, g). In an attempt to maximize direct comparability of the semaphoronts examined, we focused on late postlarval and juvenile instars that are part of the postembryonic developmental pathways of all six species, in spite of distinct differences in earlier developmental periods (see  for review). Animals were exposed to in vivo cell proliferation markers followed by immunolabeling procedures to test whether or not (1) the presence and location of the VOs can be confirmed in the additional five species, (2) the characterization of postembryonic VOs as niches of neurogenic cell proliferation can be validated beyond the genus Meridionale, (3) any indications for specific neural precursor cell types can be found, and (4) cell migration patterns of newborn cells can be reconstructed. Results from these experiments are evaluated in the context of current hypotheses on sea spider phylogeny and plesiomorphic character states for crown group Pycnogonida are deduced and discussed within an arthropod-wide perspective.
All species names of Pycnogonida have been updated according to the latest suggestions by Bamber and colleagues . The designation of postembryonic stages as specific postlarval or juvenile instars follows the general definitions recently suggested by Brenneis and colleagues .
Phoxichilidium femoratum (Rathke, 1799), Tanystylum orbiculare Wilson, 1878, and Callipallene brevirostris (Johnston, 1837)
Specimens of Phoxichilidium femoratum (Phoxichilidiidae) were collected in June and July 2015 close to Prescott Park in Portsmouth, NH, USA. During the summer months, adults, juveniles and the last postlarval instar of this species cling to colonies of the hydrozoan Tubularia larynx . Specimens of Tanystylum orbiculare (Ammotheidae) and Callipallene brevirostris (Callipallenidae) were collected in July 2015 and 2016 near Vineyard Haven and Menemsha on Martha’s Vineyard, MA, USA. At this time of year, these species are found in high abundances on colonies of the hydrozoan genus Pennaria as previously reported by Morgan . Patches of the respective hydrozoan colonies were manually removed from floating docks and maintained for the duration of the experiments in the Animal Care Facility at Wellesley College. Small pieces of the colonies were screened under a stereomicroscope and pycnogonids together with some hydrozoans were manually transferred into 50 mL Falcon tubes with cut-out “windows” covered by a fine mesh to be left freely floating in a recirculating artificial sea water system.
Stylopallene cheilorhynchus Clark, 1963, and Meridionale sp.
Specimens of both callipallenid species were collected in November 2015 in coastal waters near Eaglehawk Neck, Tasmania, Australia. During SCUBA diving, animals were collected by hand from their bryozoan prey Orthoscuticella sp. For the duration of the experiments, pycnogonids were kept with patches of bryozoan colonies in small mesh cages submerged in aerated tanks filled with local natural sea water. All Meridionale specimens investigated are part of the “variabilis”-complex, the taxonomy of which being still unsatisfactorily resolved at the species level . Only specimens of the morphotype with blackened chela and proboscis tips were used.
Pycnogonum litorale Strøm, 1762
Postlarval instars and juveniles of P. litorale (Pycnogonidae) were lab-reared in a recirculating artificial sea water system in Berlin, the setup of which being adapted from previous successful studies [46, 47, 48, 49].
In vivo cell proliferation experiments
Pycnogonids were transferred into 6-well plates filled with filtered natural sea water containing either the thymidine analog 5-bromo-2′-deoxyuridine (BrdU, 5 mg/mL; Sigma-Aldrich, #B5002) or 5-ethynyl-2′-deoxyuridine (EdU, 100 μM, provided in Click-iT® EdU Alexa Fluor® 488 or 594® Imaging Kits; Invitrogen – Molecular Probes® #C10337 or #C10339, respectively). The plates were slowly rotated on a horizontal shaker for the duration of the nucleoside exposure. Exposure times differed between experiments (see results for specific time periods). In double-nucleoside labeling experiments, BrdU was used as first marker, followed by EdU after a variable intervening time period in sea water tanks.
Specimen fixation and pre-staining processing
After nucleoside exposure, animals were fixed for 1 to 2 nights at 4 °C in 4% PFA/SW (16% formaldehyde in double-distilled H2O (methanol-free, Electron Microscopy Sciences, #15710) diluted 1:4 in filtered natural sea water) and subsequently either directly processed or alternatively stored at − 20 °C in cryoprotectant buffer (0.05 M sodium phosphate buffer, 300 g/L sucrose, 30% (v/v) ethylene glycol, 10 g/L polyvinylpyrrolidone) until further use. In the majority of cases, the complete CNS was dissected in phosphate-buffered saline (PBS; 75 mM Na2HPO4, 20 mM KH2PO4, 100 mM NaCl, pH 7.4). In some specimens of Meridionale sp. and P. litorale, the walking legs, ovigers and the strongly sclerotized portions of the cheliphores (if present) were removed, the remaining animal briefly coated in 0.01% Poly-L-Lysine and embedded in 5% agar (dissolved in double-distilled H2O) for sectioning. Sagittal sections of 50 μm thickness were produced with a Leica VT 1000S vibratome.
Dissected CNSs and sections were thereafter exposed to 10 μg/mL Proteinase K (Ambion, #AM2546) in PBS for 10–15 min at 37 °C, followed by several quick rinses in PBS under gentle shaking at room temperature (RT). In the case of subsequent BrdU labeling, an additional incubation step in 2 N HCl for 20 min at RT was performed.
Immunohistochemistry and fluorescent histochemistry
Prior to antibody exposure, samples were permeabilized in PBS + 0.3% Triton-X (PBTx) for at least 2 hours with several changes of the buffer. The primary and secondary antibodies were diluted in PBTx, with incubation times lasting 12–48 h at 4 °C. In developing and adult pycnogonid ganglia, immunolabeling of the structural protein acetylated α-tubulin can be used to visualize the neurites and part of the cortical cytoskeleton of the neuronal somata, thus giving an excellent overview of the ganglionic architecture (e.g., [22, 50]). A primary monoclonal mouse antibody (anti-ac-α-tub IgG 2b Isotype, clone 6-11 B-1, Sigma T6793, dilution 1:200) was used in conjunction with a Cy3- or Alexa Fluor®647-coupled secondary goat antibody (anti-mouse IgG (H + L), Jackson/Dianova, #115-165-146 (Cy3) & #115-605-166 (A647), dilution 1:200). For detection of BrdU-positive cells, a monoclonal rat anti-BrdU primary antibody was applied (Accurate Chemical & Scientific Corporation, clone BU1/75 (ICR1), # OBT0030G, dilution 1:50) followed by a Alexa Fluor®488-coupled goat anti-rat secondary antibody (IgG(H + L), Jackson/Dianova, #112-545-167, dilution 1:200). Detection of cells in M phase was performed with a polyclonal rabbit anti-phosphorylated histone H3 antibody (Upstate Biotechnology, Lake Placid, NY, Cat# 07-492, dilution 1:200) in combination with a Cy3-coupled goat anti-rabbit secondary antibody (IgG(H + L), Jackson/Dianova, #111-165-144, dilution 1:200). All antibody incubations were followed by rinsing in PBTx for at least 4 h at RT with gentle agitation on a horizontal shaker. As control for non-specific binding of the secondary antibodies, primary antibodies were omitted; this resulted in complete loss of signal.
EdU-positive cells were detected with Click-iT® EdU Alexa Fluor® 488 or 594® Imaging Kits (Invitrogen – Molecular Probes®, #C10337 or #C10339, respectively) according to the guidelines of the manufacturer’s protocol but with an extended Click-iT® reaction time of 2–3 h.
F-actin labeling for visualization of the cortical cytoskeleton in Ph. femoratum was performed with Alexa Fluor®488 phalloidin (Invitrogen – Molecular Probes® #A12379, 1:50 in PBTx) overnight at 4 °C.
Dissected CNSs of 70% ethanol-preserved adult specimens of the species Nymphon gracile Leach, 1814, Parapallene australiensis (Hoek, 1881), Endeis spinosa Montagu, 1808, and Cilunculus japonicus (Turpaeva, 1990) were labeled overnight at 4 °C with the lipophilic marker FM 1-43FX (Invitrogen Molecular Probes®, #F35355, 5 μg/mL in double-distilled H2O).
Nuclear counterstaining was performed with Hoechst (H33342, Invitrogen Molecular Probes®, #H1399, 1 μg/mL in PBS) after all other labeling procedures. Incubation lasted at least 1 h (sections) and was occasionally extended overnight at 4 °C (complete CNSs). After final rinsing in PBS, samples were cleared in Vectashield® Mounting Medium (Vector Laboratories, Inc.) and mounted on microscopic slides with tiny plasticine pieces attached to the corners of the cover slips to prevent compression of the samples.
Data documentation, analysis and presentation
Images of adult specimens preserved in 70% ethanol are based on z-stacks that were manually taken with a Leica M165C FC stereomicroscope equipped with a Leica DFC310 FX camera. Images were thereafter aligned in Adobe Photoshop CS3. Fluorescent overview images of CNSs were taken with a Zeiss Lumar V12, aligned z-stacks being automatically created with Zeiss AxioVision software (Version 4.7.10). Each z-stack was merged to a single image with extended depth of field using Helicon Focus software (Heliconsoft, Version 6.7.1).
Confocal laser scanning microscopy of fluorescent labeling was performed with a Leica DMI 6000 CS microscope coupled to a Leica TCS SP5 II scan unit. Based on the excitation spectra of the applied fluorochromes, a combination of UV laser (405 nm → Hoechst), argon laser (488 nm → Alexa Fluor® 488) and helium-neon laser (543 nm → Cy™3; 594 nm → Alexa Fluor® 594; 633 nm → Alexa Fluor® 647) was selected for imaging.
The 3D reconstruction software Imaris (Bitplane AG, Version 7.0.0) was used for subsequent analyses. Software tools were applied as previously described (e.g., ). Global contrast and brightness values of some of the images were adjusted using Adobe Photoshop CS3. All figures were compiled with Adobe Illustrator CS3. Supplementary movies were generated in Imaris (Animation module) and subsequently transformed into MP4-format using the freeware FormatFactory (version 2.96, www.pcfreetime.com).
Gross architecture of the VNC and its development in postlarval and juvenile instars
In the last postlarval and the first juvenile instars, the developing ganglia of the walking leg segments are well-defined morphological units (e.g., Fig. 3b). However, as postembryonic development of all species – with the exception of C. brevirostris – is anamorphic, the size of the growing ganglia decreases along an anterior-posterior (a-p) developmental gradient, which is especially evident in the last postlarval instar (e.g., Fig. 3b). In this instar, the a-p gradient is also externally recognizable by the incompletely differentiated limb buds of walking leg pair 4 (Fig. 3a).
In addition to the walking leg ganglia, one or two small posterior ganglion anlagen are transiently formed during the postlarval phase but subsequently fuse with the last walking leg ganglion in juveniles (e.g., Figs. 2a and 3b; Additional file 1; see  for more details).
Confirmation of cell proliferation in the external VOs of Meridionale sp.
In Meridionale sp., only advanced juvenile instars were investigated (originally described as postembryonic stage 6, although this stage may encompass a number of subsequent instars ). Our previous study on this species described the development of the VOs and resolved their function via documentation of mitoses and cell counts in the ganglionic soma cortex . Here, cell proliferation experiments were performed to independently validate previous results, study the direction of cell movement and enable direct comparison to the other pycnogonid species.
In corroboration of the earlier results, double-nucleoside labeling with BrdU and EdU (4 h BrdU exposure, 92 h sea water, 4 h EdU exposure, 20 h sea water, sacrifice = 5 d survival time) reveals the expected cell proliferation in the external VO-clusters and along the streams that penetrate through the neural sheath into the ganglionic soma cortex (Fig. 4; Additional file 2). Beyond that, the labeling profiles of nuclei within the VOs provide strong evidence that cell migration proceeds from the clusters into the soma cortex: Labeling for BrdU – the marker that was presented first – is found in the external VO-cluster, but in addition to this, almost a third of all BrdU-positive (BrdU+) cells are located in the streams and even deeper within the soma cortex (258 out of 857 BrdU+ cells [= 30.1%] in streams and cortices of six specimens) (Fig. 4b–d; Additional file 2). In contrast, the great majority of cells labeling for EdU – the marker presented second – remain confined to the external VO-clusters (352 out of 383 EdU+ cells [= 91.9%] in the clusters of six specimens) and only a low number of them is encountered in the proximal parts of the streams (Fig. 4e; Additional file 2). This pattern with its more restricted distribution of the EdU+ cells confirms the external VO-cluster as the predominant proliferation center, acting as a neurogenic niche for new neural precursor cells that subsequently migrate into the interior of the ganglion.
Centers of cell proliferation in the postlarval and juvenile VNC of other pycnogonid species
In addition to juveniles of Meridionale sp., we conducted in vivo cell proliferation experiments with postlarval and/or juvenile instars of all other species except P. litorale.
Overview of species, postembryonic stages and number of specimens analyzed in the course of this study
# scanned & analyzed specimens
penultimate postlarval instar
last postlarval instar
1st juvenile instar
later juvenile instar
(“variabilis” complex sensu Arango & Brenneis, 2013 )
In order to further characterize the paired ganglionic regions with the concentrated cell proliferation, EdU detection was coupled with tubulin immunolabeling. The EdU+ cells are located ventrally in the soma cortex, show more intense tubulin labeling than the surrounding somata and are arranged in conspicuous formations of spherical to ellipsoid shape. Within these formations, centrally directed processes of the intrinsic cells converge around a tiny cavity (Figs. 5b, b’, d, d’, 6b, b’, d, d’ and 7b, c; Additional file 1). This specific cell arrangement identifies these structures as the internal VOs that have been previously reported in some pycnogonids [33, 36, 38] and is strikingly similar to the arrangement in the external VOs of Meridionale sp. ([19, 22], see above). In further correspondence to the external VOs, the internal VOs undergo characteristic changes during development. The older a developing ganglion is, the more compact the internal VO that houses the EdU+ cells tends to be (Figs. 5b, d, 6b, d and 7b). Early stages of the internal VOs encompass many cells that feature a large nucleus with euchromatic DNA (judging from staining intensity), being distinctly larger than the nuclei of adjacent VO cells and the nuclei of cells in the surrounding cortex (e.g., Additional file 1). In contrast, the nuclei in more advanced (i.e., older) internal VOs are smaller and more densely packed than the ones in the surrounding cortical regions (Figs. 5b, d, 6b, d and 7b, c; Additional file 4). In accordance with the age-related changes in nuclear morphology, the number of EdU+ cells changes over time. In the ganglia of postlarval instars, higher numbers of EdU+ nuclei are found than in the ganglia of the older juvenile instars (Figs. 5a, c, 6a, c and 7a). This suggests a decrease in cell production with ongoing postembryonic development. Similarly, the intra-individual developmental gradient is reflected in higher numbers of EdU+ cells in the more posterior walking leg ganglia with their more voluminous VOs. Interestingly, while this gradient along the VNC appears rather linear in the postlarval instars of Ph. femoratum and T. orbiculare (Figs. 5a and 6a), a sharp increase in EdU+ cells is detectable between walking leg ganglia 3 and 4 of C. brevirostris (Fig. 7a, b). Further, the internal VOs of the palpal and ovigeral neuromeres in Ph. femoratum and T. orbiculare break the pattern, since they continue to exhibit comparably high cell proliferation in juveniles, even as the numbers of EdU+ cells in the posteriorly adjacent walking leg neuromeres decrease (Figs. 5c and 6c).
In contrast to the external VO-clusters of Meridionale sp., the internal VOs are not anatomically distinct from the surrounding soma cortex and any directed movement of cells between VOs and the cortex is more challenging to reconstruct in fixed samples (see below for T. orbiculare). However, it is interesting to note that anterior and posterior cell bands extend dorsally from each VO through the soma cortex towards the ganglionic neuropil, similar to the cell streams found in Meridionale sp. [19, 22]. While these strongly tubulin-positive cell bands are still rather difficult to delimit from the surrounding cells in earlier stages, they condense to more compact strands with accompanying small nuclei in later stages (Figs. 5b’, 6b’ and 7b; Additional files 1, 4). These strands are positioned roughly at the same level as the prominent ventral longitudinal tract that encircles the neuropil.
In S. cheilorhynchus, only first juvenile instars and a single specimen of the second juvenile instar were available for in vivo cell proliferation studies (12 h BrdU, 12 h sea water, sacrifice) (Table 1). The general findings for this callipallenid species are identical to the other three species (Fig. 7d–f’) including well-defined fibrous strands emanating from the compact internal VOs into the surrounding cortical areas (Fig. 7e’–f’). In line with the juveniles of the other species, the developmental gradient along the a-p axis is not as pronounced in the first and second juvenile instars of S. cheilorhynchus.
Asymmetric divisions in internal VOs and first insights into the duration of the S/G2/M phase
Recent investigations on the neurogenic processes in Meridionale sp. [19, 22] have supported classical histological works on pycnogonid development that claimed a neural stem cell (NSC)-like precursor type to reside in the segmental VOs of pycnogonids [33, 36, 39]. After confirming the presence of the internal VOs, we further scrutinized them in three of the species (Ph. femoratum, C. brevirostris, T. orbiculare) to elucidate whether specific cell types can be morphologically distinguished and to gain more insights into their cell cycle and cell division characteristics.
In T. orbiculare, the VOs of the last postlarval instar were studied after EdU exposure (4 h exposure, sacrifice). Not all of the EdU+ cells in the VOs show notable size differences compared to the somata in the surrounding cortex (Fig. 10c, d). Nonetheless, several of the observed mitotic stages in larger cells show the same indications of morphological asymmetry with shifted metaphase plates (Figs. 6b and 10c) and persisting slight differences in later mitotic stages (Fig. 10d). Notably, very strong EdU labeling was encountered in cells in all mitotic phases, including the telophase (Fig. 10c, d). At the same time, however, some mitotic cells were devoid of EdU labeling (Fig. 6b). Accordingly, the cycling times of the proliferating VO cells in T. orbiculare seem to be more heterogeneous, comprising faster (S phase to M phase < 4 h) and slower (S phase to M phase > 4 h) cycling cells. With the available data it cannot be determined whether these cell cycle differences relate to different neural precursor types in the VOs.
Double-nucleoside labeling using BrdU and EdU – estimation of BrdU clearing time and further evidence for NSC-like precursor types in pycnogonid VOs
During the postembryonic development of Meridionale sp., the direction of cell movement from external VO-clusters via the streams into the soma cortex has been previously reconstructed  and independently validated here with double-nucleoside cell proliferation experiments (Fig. 4; Additional file 2). Aiming to investigate whether cell migration occurs in species with internal VOs, we conducted similar experiments with postembryonic instars of Ph. femoratum and T. orbiculare. Experiments with postlarval instars and juveniles of T. orbiculare were successful and lasted 3 days in total (see next section). In Ph. femoratum, by contrast, survival of postembryonic instars in good condition could only be accomplished with short time windows of less than 24 h. This duration was too short to extract cell migration patterns with any confidence (Additional file 4). However, although the experiments failed in this respect, the relatively short time intervals between the two proliferation marker pulses provided insights into the marker clearing time in pycnogonid instars falling in the studied size range: The shortest applied sea water interval between BrdU and EdU exposures was 12 h. After this experiment, the VOs of Ph. femoratum showed a mixed pattern of BrdU+/EdU- and BrdU+/EdU+ cells, but importantly also some BrdU-/EdU+ cells (Fig. 9e; Additional file 4). The presence of EdU-only labeled cells indicates that BrdU was no longer available at detectable levels after the intervening 12 h in sea water, resulting in the exclusive incorporation of EdU in cells passing through S phase during the EdU exposure period. Accordingly, 12 h can be deduced as a conservative estimate for the clearing time of BrdU from the animals. This may still be an overestimation since shorter inter-pulse intervals have not been tested.
Regardless of the uncertainty pertaining to shorter inter-pulse intervals, this finding has important implications for experiments using longer inter-pulse intervals with the similar-sized postembryonic instars of Meridionale sp. The intervening sea water interval in our double-labeling experiments on Meridionale sp. juveniles lasted almost 4 days (92 h), that is, almost 8-times the 12 h clearing time established in Ph. femoratum. Therefore, it seems highly unlikely that any of the double-labeled cells observed in Meridionale sp. (Fig. 4b–e; Additional file 2) incorporated BrdU during the EdU exposure period. As a consequence, double labeling with both proliferation markers suggests that cells have repeatedly passed through S phase during the experiment. Such double-labeled cells are almost exclusively found in the external VO-clusters of Meridionale sp. and only occasionally in the proximal portions of the migratory streams (Fig. 4b–e; Additional file 2). Further, although BrdU-only labeled cells were readily identifiable in the external VO-clusters and deep within the soma cortex, EdU labeling is detected in the overwhelming majority of cases in combination with BrdU labeling (381 out of 383 EdU+ cells [= 99.5%] with BrdU co-labeling in six specimens). This pattern strongly suggests that repeatedly dividing neural precursor cells reside within the external VO-clusters, their progeny migrating through the streams into the cortex. Morphological delimitation of different cell types is not possible in the external VOs of juvenile instars of Meridionale sp. . However, since the external clusters are derivatives of the earlier VO-stages that house large asymmetrically dividing NSC-like precursors , it is feasible that the double-labeled cells are of the same type, having decreased in size with ongoing cycling during postembryonic development.
Double-nucleoside labeling in T. orbiculare: cell migration from the VOs into the soma cortex
In T. orbiculare, 3-day-long double-nucleoside labeling experiments were conducted (4 h BrdU exposure, 64 h sea water, 4 h EdU exposure, sacrifice) with the penultimate and last postlarval instars as well as with the first juvenile instar. Investigation of this developmental series enabled us to evaluate labeling patterns intra-individually along the a-p developmental gradient but also between specific segmental ganglia across different instars.
In contrast to the EdU labeling, BrdU+ cells are spread far beyond the VOs into the growing soma cortex of the developing ganglia. They do not only surround the VOs but also reach further dorsally towards the forming ganglionic neuropil. Especially in advanced stages of ganglion development, the BrdU+ cells extend anteriorly and posteriorly from the neuropil, leading in lateral view to a sickle-like arrangement embedded in the cortex (Fig. 11a, a’, b, b’, c, c’; Additional file 3). This is strongly suggestive of active cell migration. Also in ventral view, the overall pattern of BrdU+ cells changes over time. During the early phases of rapid ganglion growth, they are spread out in broad, almost concentric rings around the VOs (Fig. 11a, a’, b, b’) as would be expected from the migration of newborn cells in all directions from the main centers of cell proliferation. In later stages, however, the BrdU+ cells are arranged in more elongated band-like patterns with an oblique a-p axis (Fig. 11c, c’). The VOs are still in the center of these cell bands but migration of newborn cells has become restricted laterally and is instead concentrated along two pathways that encircle the central neuropil anteriorly and posteriorly, in close proximity to the ventral longitudinal tract (Additional file 3). Incidentally, already in earlier stages, cells that have recently undergone DNA synthesis (i.e., are EdU+) and have started to migrate dorsally are encountered in close vicinity to the same tract (Fig. 12d).
Importantly, the considerable number of BrdU+ cells labeled during the rapid ganglion growth may not be exclusively attributable to cell proliferation in the VOs but may in part be explained by additional cell divisions during migration. Although this aspect cannot be conclusively resolved with the methods applied here, the almost complete lack of EdU labeling in deeper layers (e.g., Fig. 12c, d) suggests that such divisions of newborn cells leaving the VOs would have to occur in early stages of migration. The majority of BrdU+ cells located deeper in the cortex are therefore more likely to be differentiating/differentiated postmitotic neural cells, which notion receives further support from their similar nuclear morphology compared to the surrounding BrdU-negative neuronal somata (Fig. 12d, e).
Gross architecture of the VNC: independent inclusion of walking leg neuromere 1 into the SEG of different pycnogonid lineages
Our study highlights one of the few taxon-specific architectural differences of the pycnogonid VNC: the three callipallenid representatives possess an anatomically separate walking leg ganglion 1 posterior to a bipartite SEG that comprises the palpal and ovigeral neuromeres. By contrast, the remaining three species have a tripartite adult SEG that includes the neuromere of walking leg segment 1. Importantly, however, outside of Phoxichilidiidae, Pycnogonidae and the genus Tanystylum, all pycnogonid taxa hitherto studied have a bipartite SEG as found in the three callipallenids (Fig. 13a–f, also [30, 52]). This taxon list includes other ammotheid genera (e.g., Fig. 13d) as well as groups which have been indicated to branch off close to the base of the pycnogonid tree in the most exhaustive phylogenetic analyses [40, 41, 53] (Fig. 13g), such as Austrodecidae (Fig. 13f) and Colossendeidae (e.g., [31, 54]). Further, it also encompasses Ascorhynchidae (e.g., [31, 55]) which has been proposed as the earliest (paraphyletic) offshoot in one 18S rRNA-based molecular analysis  (but see, e.g.,  for Ascorhynchidae well-nested in the pycnogonid tree). Hence, when mapping the SEG architecture on currently discussed topologies of pycnogonid interrelationships, a bipartite composition is resolved as the most parsimonious plesiomorphic state (Fig. 13g). The tripartite SEG of some extant taxa, on the other hand, is indicated to go back to three independent events in pycnogonid crown group evolution, being the result of the fusion of the palpal, ovigeral and walking leg 1 neuromeres during postembryonic gangliogenesis. Notably, this interpretation holds true regardless of the morphologically questionable splitting of the Phoxichilidiidae, which has been recovered due to the rather surprising placement of the genus Phoxichilidium in one of the recent analyses  (Fig. 13g).
Variations of a common theme – external versus internal pycnogonid VOs
The deviating position of the external and internal VOs raises the questions whether they serve the same function and if they are homologous structures across pycnogonids. With regard to the functional role, our cell proliferation experiments and mitosis marker labeling unambiguously characterize internal VOs as centers of significant cell proliferation during VNC development, in correspondence to the external VOs of Meridionale sp. [19, 22]. Further, regardless of VO position, our double-nucleoside labeling approach illustrates migration of significant numbers of newborn cells from the VOs into the growing soma cortex of the respective ganglion. Accordingly, this indicates that external and internal VOs act as postembryonic neurogenic niches in the ventral ganglia – and refutes previous speculations on the predominantly neurosecretory nature of the internal VOs [37, 38].
In an evolutionary framework, it remains then to be clarified which of the two VO types represents the plesiomorphic condition of the taxon Pycnogonida. Across the groups studied so far, external VOs have been exclusively discovered in Nymphonidae (e.g., [35, 36, 58]) and the callipallenid genera Meridionale (, present study) and Pseudopallene . In all other taxa investigated, including the callipallenid genera Stylopallene and Callipallene ([33, 38, 39], present study), the VOs are incorporated into the soma cortex (Fig. 14b). Irrespective of which phylogenetic hypothesis is favored [40, 41, 53, 56], a parsimony-based assessment of the distribution of VO types suggests the external VOs to be an apomorphic condition that has emerged within the callipallenid-nymphonid lineage (Fig. 14b). To date, the exact relationships of the nymphonid and callipallenid taxa are still unsatisfactorily resolved. However, with further progress in this field, the presence of external VOs may prove phylogenetically informative, as a potential apomorphic character state of a monophylum comprised of a subset of callipallenid species and nymphonids.
Interspecific differences and deviations from the a-p gradient of VO cell proliferation correlate to specific events of postembryonic development
Compared to the rather linear a-p gradient of ventral ganglion development in Ph. femoratum and T. orbiculare, the last postlarval instar of C. brevirostris shows an abrupt increase in cell proliferation between the VOs of walking leg ganglia 3 and 4 (Fig. 7a). This observation coincides with a different mode of postembryonic development. The genus Tanystylum shows a pronounced anamorphic development with a minute protonymphon larva as hatching stage and strictly sequential addition of the walking leg segments [33, 59]. This developmental type is widely considered to be plesiomorphic for Pycnogonida (e.g., ). Likewise, all members of the Phoxichilidiidae follow an anamorphic – albeit accelerated – mode with a slight but perceivable a-p gradient in walking leg segment formation [44, 60]. By contrast, the two genera Callipallene and Neopallene feature the most pronounced embryonization of development known in Pycnogonida [33, 38, 39, 51, 54]. One hallmark of their development is the synchronized differentiation of walking leg segments 1–3 early during embryogenesis, whereas walking leg segment 4 is formed distinctly later . As a consequence, the first three walking leg segments and their ganglia are well-differentiated in the postlarval instar, but walking leg segment 4 is still in the period of major ganglion growth with VOs that house numerous large NSC-like precursors (Fig. 7b, c). Accordingly, the observed differences in overall cell proliferation patterns between postlarval instars of the different species examined reflect deviations in the timing of walking leg segment formation during their development.
Another case of proliferation patterns deviating from a monotonous a-p developmental gradient is encountered in the anterior palpal and ovigeral VOs that are affiliated with the SEG. While cell proliferation of the posteriorly following VOs in walking leg segments 1 and 2 decreases between the last postlarval and the first juvenile instar, the number of EdU+ cells in the palpal and ovigeral VOs was found to remain distinctly higher in Ph. femoratum and T. orbiculare (Figs. 5A, c and 6a, c). Again, these observations correlate with developmental events in this particular region: all pycnogonid taxa with a protonymphon larva as hatching stage possess small three-segmented palpal and ovigeral larval limb pairs at the beginning of the postlarval phase [57, 58, 61, 62, 63]. However, one of the most peculiar aspects of pycnogonid development is the metamorphosis of these limbs into the adult palps and ovigers (if present) which starts at the end of the postlarval phase. In the case of the oviger, this metamorphosis even includes the almost complete atrophy of the larval limb prior to a de novo outgrowth of the appendage during the first instars of the juvenile phase (see  for Tanystylum,  for review). Accordingly, major neuroarchitectural changes are to be expected in the SEG during this period, owing to the loss of neural elements of the palpal and ovigeral appendage nerves of the larva (Fig. 15a, a’) as well as the integration of new afferent and efferent neurons into the existing circuits as the adult limb pairs differentiate. Embedded in this developmental framework, the higher cell proliferation rates in the corresponding VOs appear a logical consequence, and may serve as additional evidence for the VOs' indicated neurogenic role.
Beyond sea spiders: Internal VOs as a plesiomorphic feature of arthropod gangliogenesis
Our new data on the internal VOs in some pycnogonid taxa and the reconstruction of this VO type as the plesiomorphic condition of Pycnogonida has implications in a broader arthropod-wide framework.
Similarities and differences between neurogenic processes in the major arthropod groups and their closest relatives have been discussed in depth in numerous contributions of the last decades (e.g., [18, 19, 20]), often with focus on the initial embryonic phase and to the inclusion of the involved gene networks. In this context, the discovery of detailed similarities between the neural precursor types and neural cell lineages underlying early embryonic neurogenesis in hexapods and (some) crustaceans (see [3, 6, 18, 64]) has contributed compelling neurodevelopmental arguments in favor of the taxon Tetraconata that is strongly supported in molecular analyses (e.g., [7, 8, 9]). In similar fashion, correspondences between the early neurogenic processes in the non-pycnogonid chelicerates (represented by horseshoe crabs and spiders) and myriapods have been briefly discussed as supporting a sister group relationship of both lineages [65, 66, 67] but are now more generally considered as the plesiomorphic condition of Arthropoda (e.g., [19, 20]).
When putting the focus on later phases of neurogenesis, VO-like structures as centers of neural cell production have not been reported in any of the recently investigated chelicerate taxa other than Pycnogonida [19, 68, 69], nor are they known from any tetraconate taxon. Intriguingly, however, VO-like cell formations are invariably formed during advanced neurogenesis of the myriapod taxa [70, 71, 72, 73, 74], in some studies having even been designated as “ventral organs” [71, 72, 74]. As is the case in pycnogonids, these myriapod VOs develop via an invagination of the central area of the neuroectoderm of each hemisegment. Moreover, in accordance with the here deduced pycnogonid ground pattern, myriapod VOs are subsequently incorporated into the ventral soma cortex of the ganglia, where they have been reported to produce additional ganglion cells [70, 71, 72, 73]. In light of increasingly better support for myriapods as sister group of Tetraconata within the taxon Mandibulata (e.g., [7, 27, 75]), we therefore propose internal VOs as centers of late neurogenesis to be a symplesiomorphic feature of gangliogenesis in pycnogonids and myriapods. In other words, we suggest such internalized VOs to be part of the ground pattern of Arthropoda.
Also in the likely arthropod sister group Onychophora, conspicuous segmental thickenings of the ventral ectoderm are formed during advanced development. As for Myriapoda, histological studies have termed these structures “ventral organs” (e.g., [76, 77, 78]) and in part suggested their involvement in late embryonic neurogenesis (e.g., ). However, several recent investigations have refuted this role (e.g., [79, 18]), as the onychophoran VOs neither invaginate, nor retain a connection to the underlying CNS. Instead, they gradually atrophy to tiny sclerotized epidermal structures that act as muscle attachment sites for limb muscles in the adult animals [79, 80]. Owing to these differences, homologization of these onychophoran structures with the pycnogonid and myriapod VOs seems not tenable. A notable exception to this may be found in the protocerebral neuroectoderm of onychophorans. Here, the central area invaginates during development to form the hypocerebral organ – an external cell cluster that is attached to the ventral side of brain. To this day, the function of the hypocerebral organ remains largely unresolved, and apart from a (neuro)secretory nature (e.g., [37, 81, 82]), a role in late and perhaps even adult neurogenic processes has been considered . Clearly, further studies on this organ are needed for a better understanding of its constituting cell types and their function.
Similarly, it will be the task of future studies to further dissect the cellular processes and their underlying genetic networks in the internal VOs of pycnogonids and myriapods in more detail. For instance, the presence of specialized NSC-like precursors comparable to those in pycnogonids needs to be critically reinvestigated in myriapod taxa, as the currently available data are in this respect still contradictory and inconclusive (e.g., [65, 74, 83]). Beyond that, the potential persistence of internal VOs in the ganglia of adult pycnogonids and myriapods – as shown for the external VOs of Meridionale sp.  – and the related issue of life-long neurogenic processes in these two arthropod groups remain to be clarified.
Our comparative study on six different pycnogonid species provides new insights into the evolution of neurodevelopmental processes and adult neuroanatomy in one of the oldest extant arthropod lineages. By evaluating our findings in light of current hypotheses on pycnogonid phylogeny, we reveal a bipartite SEG as the plesiomorphic condition of sea spiders and find evidence for the convergent fusion of walking leg neuromere 1 to this SEG in three different pycnogonid lineages. These fusion events are clearly independent of the apomorphic postoral CNS condensation in the prosoma of euchelicerates (e.g., [12, 84]), as well as of the different SEG formation/expansion events discussed for mandibulate taxa (e.g., ). Further, we reconstruct the presence of internal VOs instead of external VOs as plesiomorphic character state of sea spider gangliogenesis. Although chelicerate taxa other than Pycnogonida lack comparable VOs, they are a characteristic feature of myriapod gangliogenesis and, accordingly, we propose internal VOs with neurogenic function to be part of the ground pattern of Arthropoda. Finally, our study demonstrates the importance of an expanded taxon sampling in neurodevelopmental and neuroanatomical investigations, even when dealing with arthropod lineages that show such a conserved basic body organization as Pycnogonida. This enables the documentation of ingroup variation and more reliable differentiation of plesiomorphic from apomorphic character states prior to outgroup comparison.
We would like to thank Eric Lovely for sharing the location for the collection of Phoxichilidium femoratum. Karen Gowlett-Holmes, Mick Baron and Pauline and Ben Nuttens of the Eaglehawk Neck Dive Centre significantly contributed to the successful collection of Meridionale sp. and Stylopallene cheilorhynchus. David Staples, Prashant P. Sharma as well as Katsumi Miyazaki and Günter Pass are thanked for providing specimens of Parapallene australiensis, Pantopipetta armoricana and Cilunculus japonicus, respectively. Jeanne L. Benton gave invaluable advice regarding the BrdU and EdU labeling procedures. The assistance of Julia S. Lord and Sharon Gobes with the vibratome sectioning is gratefully acknowledged. Pat Carey and Valerie LePage of the Animal Care Facility at Wellesley College contributed to the well-being of the live animals during the experiments. We thank two anonymous reviewers for their helpful comments. Collection of Australian sea spiders was made possible by a permit of the Tasmanian Department of Primary Industries, Parks, Water and Environment (permit no. 15111).
GB is funded by the Deutsche Forschungsgemeinschaft (BR 5039/1-1, BR 5039/2-1). Data generation on P. litorale was funded by a grant of the Deutsche Forchungsgemeinschaft awarded to GS (Scho 442/9-1); during part of this time GB was supported by the Studienstiftung des deutschen Volkes. BSB received funding from the National Science Foundation (NSF-IOS-1656103).
Availability of data and materials
Raw data generated in this study are in the care of the first author (GB).
GB conceived and designed the experiments, performed the experimental work, data acquisition, data analysis and interpretation, and wrote the first draft of the manuscript. GS and BSB provided lab space and contributed to data interpretation and manuscript writing. All authors read and approved the final manuscript.
The animals studied are non-regulated invertebrates.
Consent for publication
The authors declare that they have no competing interests.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
- 3.Dohle W. Are the insects terrestrial crustaceans? A discussion of some new facts and arguments and the proposal of the proper name ‘Tetraconata’ for the monophyletic unit Crustacea + Hexapoda. Ann Soc Entomol Fr. 2001;37:85–103.Google Scholar
- 8.von Reumont BM, Jenner RA, Wills MA, Dell’Ampio E, Pass G, Ebersberger I, Meyer B, Koenemann S, Iliffe TM, Stamatakis A, et al. Pancrustacean phylogeny in the light of new phylogenomic data: support for Remipedia as the possible sister group of Hexapoda. Mol Biol Evol. 2012;29:1031–45.CrossRefPubMedGoogle Scholar
- 12.Strausfeld NJ. Arthropod brains. Evolution, functional elegance, and historical significance. Cambridge (Massachusetts). London: Belknap Press of Harvard University Press; 2012.Google Scholar
- 13.Loesel R, Wolf H, Kenning M, Harzsch S, Sombke A. Architectural principles and evolution of the arthropod central nervous system. In: Minelli A, Boxshall G, Fusco G, editors. Arthropod biology and evolution molecules, development, morphology. Berlin, Heidelberg: Springer; 2013. p. 299–342.CrossRefGoogle Scholar
- 30.Helfer H, Schlottke E. Pantopoda. Bronns Klassen und Ordnungen des Tierreichs, Bd. 5, Abt. IV, Buch 2. Leipzig: Akademische Verlagsgesellschaft m.b.H; 1935.Google Scholar
- 31.Henry LM. The nervous system of the Pycnogonida. Microentomology. 1953;18:16–36.Google Scholar
- 32.Brenneis G. Pycnogonida (Pantopoda). In: Schmidt-Rhaesa A, Harzsch S, Purschke G, editors. Structure and evolution of invertebrate nervous systems. Oxford: Oxford University Press; 2016. p. 419–27.Google Scholar
- 33.Morgan TH. A contribution to the embryology and phylogeny of the pycnogonids, Studies from the biological Laboratory of the Johns Hopkins University Baltimore, vol. 5; 1891. p. 1–76.Google Scholar
- 34.Meisenheimer J. Beiträge zur Entwicklungsgeschichte der Pantopoden. I. Die Entwicklung von Ammothea echinata Hodge bis zur Ausbildung der Larvenform. Z Wiss Zool. 1902;72:191–248.Google Scholar
- 35.Dogiel V. Studien über die Entwicklungsgeschichte der Pantopoden. Nervensystem und Drüsen der Pantopodenlarven. Z Wiss Zool. 1911;99:109–46.Google Scholar
- 36.Dogiel V. Embryologische Studien an Pantopoden. Z Wiss Zool. 1913;107:575–741.Google Scholar
- 37.Sanchez S. Cellules neurosécrétrices et organes infracérébraux de Peripatopsis moseleyi Wood (Onychophores) et neurosécrétion chez Nymphon gracile Leach (Pycnogonides). Arch Zool Exp Gén Notes Revue. 1958;96:57–62.Google Scholar
- 38.Sanchez S. Le développement des Pycnogonides et leurs affinités avec les Arachnides. Arch Zool Exp Gén. 1959;98:1–102.Google Scholar
- 43.Bamber RN, El Nagar A, Arango CP. Pycnobase: World Pycnogonida Database. Available online at http://www.marinespecies.org/pycnobase. Accessed 13 March 2018.
- 52.Loman JCC. Beiträge zur Anatomie und Biologie der Pantopoden. Tijdschr Ned Dierkundige Vereeniging Ser 2. 1917;16:53–102.Google Scholar
- 53.Sabroux R, Corbari L, Krapp F, Bonillo C, Le Prieur S, Hassanin A. Biodiversity and phylogeny of Ammotheidae (Arthropoda: Pycnogonida). Eur J Taxon. 2017;286:1–33.Google Scholar
- 54.Hoek PPC. Report on the Pycnogonida, dredged by H.M.S. challenger during the years 1873-76. Challenger report. Zoology. 1881;3:1–167.Google Scholar
- 55.Dohrn A. Die Pantopoden des Golfes von Neapel und der angrenzenden Meeres-Abschnitte. Leipzig: Verlag von Wilhelm Engelmann; 1881.Google Scholar
- 58.Bogomolova EV, Malakhov VV. Larvae of sea spiders (Arthropoda, Pycnogonida) from the White Sea. Entomol Rev. 2003;83:222–36.Google Scholar
- 60.Maxmen A. Pycnogonid development and the evolution of the arthropod body plan. Cambridge: PhD thesis, Harvard University; 2006.Google Scholar
- 63.Hübner J, Wagner P, Lehmann T, Melzer RR. Testing species delimitation with larval morphology: scanning electron microscopy analysis of protonymphon larvae of two closely related sea spiders, Pallenopsis patagonica (Hoek) and Pallenopsis yepayekae Weis. Invertebr Syst. 2017;31:363–74.CrossRefGoogle Scholar
- 70.Heymons R. Die Entwicklungsgeschichte der Scolopender. Zoologica. 1901;33:1–244.Google Scholar
- 71.Tiegs OW. The embryology and affinities of the Symphyla, based on a study of Hanseniella agilis. Q J Microsc Sci. 1940;82:1–225.Google Scholar
- 73.Dohle W. Die Embryonalentwicklung von Glomeris marginata (Villers) im Vergleich zur Entwicklung anderer Diplopoden. Zool Jahrb Anat. 1964;81:241–310.Google Scholar
- 74.Knoll HJ. Untersuchungen zur Entwicklungsgeschichte von Scutigera coleoptrata L. (Chilopoda). Zool Jahrb Anat. 1974;92:47–132.Google Scholar
- 76.Kennel J. Entwicklungsgeschichte von Peripatus edwardsii Blanch. und Peripatus torquatus n.sp, Arbeiten aus dem zoologisch-zootomischen Institut Würzburg. 1886;8:1–93.Google Scholar
- 77.Sedgwick A. A monograph of the development of Peripatus capensis. In: Studies from the morphological Laboratory in the University of Cambridge. Cambridge: Cambridge University Press; 1888. p. 1–146.Google Scholar
- 78.Pflugfelder O. Entwicklung von Paraperipatus amboinensis n. spec. Zool Jahrb Anat. 1948;69:443–92.Google Scholar
- 84.Scholtz G. Perspective---heads and brains in arthropods: 40 years after the ‘Endless dispute’. In: Schmidt-Rhaesa A, Harzsch S, Purschke G, editors. Structure and evolution of invertebrate nervous systems. Oxford: Oxford University Press; 2016.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.