Molecular Identification and Cellular Localization of a Corticotropin-Releasing Hormone-Type Neuropeptide in an Echinoderm

Background: Corticotropin-releasing hormone (CRH) mediates physiological responses to stressors in mammals by triggering pituitary secretion of adrenocorticotropic hormone, which stimulates adrenal release of cortisol. CRH belongs to a family of related neuropeptides that include sauvagine, urotensin-I, and urocortins in vertebrates and the diuretic hormone DH44 in insects, indicating that the evolutionary origin of this neuropeptide family can be traced to the common ancestor of the Bilateria. However, little is known about CRH-type neuropeptides in deuterostome invertebrates. Methods: Here, we used mass spectrometry, mRNA in situ hybridization, and immunohistochemistry to investigate the structure and expression of a CRH-type neuropeptide (ArCRH) in the starfish Asterias rubens (phylum Echinodermata). Results: ArCRH is a 40-residue peptide with N-terminal pyroglutamylation and C-terminal amidation, and it has a widespread pattern of expression in A. rubens. In the central nervous system comprising the circumoral nerve ring and 5 radial nerve cords, ArCRH-expressing cells and fibres were revealed in both the ectoneural region and the hyponeural region, which contains the cell bodies of motoneurons. Accordingly, ArCRH immunoreactivity was detected in innervation of the ampulla and podium of locomotory organs (tube feet), and ArCRH is the first neuropeptide to be identified as a marker for nerve fibres located in the muscle layer of these organs. ArCRH immunoreactivity was also revealed in protractile organs that mediate gas exchange (papulae), the apical muscle, and the digestive system. Conclusions: Our findings provide the first insights into CRH-type neuropeptide expression and function in the unique context of the pentaradially symmetrical body plan of an echinoderm.


Introduction
The existence of a hypothalamic neurohormone that stimulates pituitary secretion of adrenocorticotropic hormone was first postulated in the 1950s [1][2][3]. However, it was not until 1981 that this corticotropin-releasing hormone (CRH) was identified as an amidated 41-residue This article is licensed under the Creative Commons Attribution 4.0 International License (CC BY) (http://www.karger.com/Services/ OpenAccessLicense). Usage, derivative works and distribution are permitted provided that proper credit is given to the author and the original publisher. DOI: 10.1159/000517087 peptide [4], which is derived from the C-terminal region of a 196-residue precursor protein [5,6]. Sequencing of CRH revealed that it is a homologue of sauvagine and urotensin-I, vasoactive peptides isolated from the skin of the frog Phyllomedusa sauvagii [7] and the urophysis of teleost fish, respectively [8,9]. Thus, a family of related bioactive peptides was discovered in vertebrates, and more recently genes encoding CRH/sauvagine/urotensin-1-like peptides named urocortin 1 (UCN1), urocortin 2 (UCN2), and urocortin 3 (UCN3) were identified in humans and other mammals [10][11][12]. CRH/UCN-type peptides exert effects via 2 G-protein-coupled receptors in humans known as CRHR1 and CRHR2, with CRHR1 selectively activated by CRH and UCN1 and CRHR2 non-selectively activated by all 4 peptides [13][14][15]. Furthermore, investigation of the phylogenetic distribution of CRH/UCN-type neuropeptide signalling in vertebrates by comparative analysis of genome sequence data indicates that a common ancestor of vertebrates would have had 1 gene encoding a CRH/UCN1-type neuropeptide, 1 gene encoding a UCN2/UCN3-type neuropeptide, and 1 gene encoding a receptor for these 2 neuropeptides. Then, genome duplications followed by gene loss gave rise to variety in the complement of genes encoding CRH/UCNtype precursors and receptors in different vertebrate lineages [15,16]. Genes encoding CRH-type precursors and CRH-type receptors have also been identified in the invertebrate chordates Ciona intestinalis (sub-phylum Urochordata) and Branchiostoma floridae (sub-phylum Cephalochordata), providing insights into the pre-vertebrate origins of CRH-type neuropeptide signalling in the phylum Chordata [17,18].
A key advance in our knowledge and understanding of the evolution of CRH/UCN-type signalling was made with the discovery of a 44-residue diuretic hormone (DH44) in the insect Manduca sexta that shares sequence similarity with CRH and UCNs [19]. Subsequently, DH44-type peptides were identified in other insects, and further evidence that DH44 is a homologue of CRH/urocortin-type peptides in vertebrates was obtained with the identification of DH44 receptors as homologues of vertebrate CRH/UCN-type receptors [20][21][22]. Thus, the discovery of the DH44 signalling system in insects revealed that the evolutionary origin of CRH/UCN-type signalling can be traced back to the urbilaterian common ancestor of protostome invertebrates (e.g., insects) and vertebrates. Accordingly, precursors of DH44-like neuropeptides have also been identified in other protostomes, which include egg-laying hormone (ELH) in gastropod molluscs [18,23]. Important insights into the evolution and comparative physiology of neuropeptide signalling systems have been obtained recently from studies on non-chordate deuterostomes (echinoderms and hemichordates), which occupy an "intermediate" phylogenetic position with respect to chordates and protostomes [24]. In particular, the starfish Asterias rubens (class Asteroidea) has been used extensively as an experimental system for molecular and functional characterization of neuropeptide signalling systems [25]. For example, molecular characterization of neuropeptide signalling systems in A. rubens has provided key insights for reconstruction of the evolutionary history of neuropeptide-S/crustacean cardioactive peptidetype [26,27], gonadotropin-releasing hormone (GnRH)/ corazonin-type [28,29], and prolactin-releasing peptide/ short neuropeptide-F-type neuropeptide signalling [30]. Furthermore, functional characterization of neuropeptides in A. rubens has provided new insights into the comparative physiology of, for example, GnRH/corazonintype [31], calcitonin-type [32], vasopressin/oxytocintype [33], and somatostatin-type [34] neuropeptides.
Analysis of A. rubens neural transcriptome sequence data recently enabled identification of a transcript encoding the first precursor of a CRH-type peptide to be discovered in an echinoderm, which is referred to as ArCRH [35]. The discovery of this precursor provides an opportunity for characterization of a CRH-type peptide in an echinoderm. Accordingly, here, we report use of mass spectrometry to determine the structure of the CRH-like neuropeptide derived from the ArCRH precursor protein. Furthermore, we report use of mRNA in situ hybridization and immunohistochemistry, employing novel antibodies to ArCRH, to enable a detailed analysis of Ar-CRH expression in A. rubens. scriptome sequence data [35] and a cDNA encoding ArCRHP has been cloned and sequenced [36]. The predicted CRH-type peptide derived from ArCRHP is a 40-residue peptide with an amidated C-terminus. To determine if this predicted structure of ArCRH is correct, extracts of radial nerve cords from A. rubens were prepared and analysed using mass spectrometry (Nano-LC-ESI-MS/ MS), employing methods described in detail previously [32,37]. Briefly, extracts were fractionated by reversed-phase chromatography on an Ultimate 3000 RSLCnano system (Dionex) and then injected directly via a TriVersa NanoMate nanospray source (Advion Bio Sciences, Ithaca, NY, USA) into a Thermo Orbitrap Fusion (Q-OT-qIT; Thermo Scientific) mass spectrometer. Data analysis was performed using Proteome Discoverer 2.2 (Thermo Fisher Scientific) with an ion mass tolerance of 0.050 Da and a parent ion tolerance of 10.0 ppm. C-terminal amidation and N-terminal pyroglutamylation were specified as potential post-translational modifications. Furthermore, differentiating the presence of isobaric residues (leucine and isoleucine) and near isobaric residues (lysine and glutamine) in peptide fragments was enabled by the availability of the ArCRH sequence predicted from cDNA sequencing. In addition, synthetic ArCRH with the predicted structure pQGLSVSPIFPIQRIRLNAIERDRQDQVDQAEANQGLFQ-IA-NH 2 was synthesized and purified (>95%) by Peptide Protein Research Ltd. (Fareham, Hampshire, UK) (see online suppl. Fig. 1; see www.karger.com/doi/10.1159/000517087 for all online suppl. material) and then also analysed using mass spectrometry to enable comparison of spectra with those obtained from analysis of radial nerve cord extracts. Having determined the structure of the mature ArCRH peptide, its sequence was aligned with CRH and CRH-related peptides from other species using the Clustal Omega multiple sequence alignment program (https://www.ebi.ac.uk/ Tools/msa/clustalo/).

Localization of ArCRHP Expression in A. rubens Using mRNA in situ Hybridization
Digoxigenin-labelled RNA antisense probes complementary to ArCRHP transcripts and corresponding sense probes were synthesized, as reported previously [36]. The methods employed for visualization of ArCRHP expression in sections of the arms, central disc, or whole body of A. rubens were the same as those reported previously for analysis of the expression of the calcitonin-type precursor ArCTP [32], the relaxin-type precursor ArRGPP [37], the GnRH type precursor ArGnRHP, the corazonin-type precursor ArCRZP [31], the pedal peptide-type precursor ArPPLNP1 [38], and somatostatin-type precursor ArSSP2 [34].

Production and Characterization of a Rabbit Antiserum to the C-Terminal Region of ArCRH
To generate a rabbit antiserum against ArCRH (pQGLSVSPIF-PIQRIRLNAIERDRQDQVDQAEAN QGLFQIA-NH 2 ), a peptide containing the C-terminal 10 amino acids of ArCRH (KEAN-QGLFQIA-NH 2 ) was synthesized and purified (>95%) by Peptide Protein Research Ltd. (Fareham, Hampshire, UK) and then used as a peptide antigen (online suppl. Fig. 2). The N-terminal lysine residue was incorporated into the antigen peptide to provide a reactive site for conjugation with a carrier protein (porcine thyroglobulin; Sigma Aldrich). The conjugate of the antigen peptide and carrier protein was prepared by using 5% glutaraldehyde in phosphate buffer (0.1 M, pH 7.2) as a coupling reagent [39]. Then, the solution was dialysed in distilled water to remove glutaraldehyde and any uncoupled peptide and then was divided into aliquots containing approximately 50 nmol conjugated antigen in each tube. Antiserum production in a male rabbit was carried out by Charles River Ltd. (Margate, UK; project code number 17582) using the same immunization and bleeding protocol reported previously for generation of antisera to ArPPLN1b and ArCT [32,38]. To assess the presence and titre of antibodies to the antigen peptide, pre-immune serum and antiserum from the final bleed were analysed using an ELISA, employing protocols similar to those reported previously for antisera to ArPPLN1b and ArCT [32,38].

Immunohistochemical Localization of ArCRH in A. rubens
The methods employed for immunohistochemical localization of ArCRH in A. rubens were the same as those reported previously for ArPPLN1b and ArCT [32,38], but with the ArCRH antiserum being used at a dilution of 1:32,000.

Determination of the Structure of the A. rubens CRH-Type Neuropeptide ArCRH and Comparison with the Sequences of CRH and CRH-Related Peptides in Other Taxa
A cDNA encoding the ArCRH precursor (ArCRHP) has been cloned and sequenced [36], confirming a previously reported transcript sequence obtained from analysis of A. rubens neural transcriptome sequence data [35] (GenBank accession number KT601710.1). ArCRHP is a 130-residue protein comprising a predicted 28-residue N-terminal signal peptide (residues 1-28) followed by a pro-peptide, which includes a putative 41-residue CRHlike peptide sequence (QGLSVSPIFPIQRIRLNAIER-DRQDQVDQAEANQGLFQIAG, residues 87-127) that is bounded by putative dibasic/tribasic cleavage sites Nterminally and C-terminally (Fig. 1a). The N-terminal glutamine (Q) and C-terminal glycine (G) residues of the 41-residue peptide are noteworthy as potential substrates for post-translational pyroglutamylation (pQ) and amidation (NH 2 ), respectively. Therefore, the predicted mature structure of ArCRH is pQGLSVSPIFPIQRIRLNAI-ERDRQDQVDQAEANQGLFQIA-NH 2 .
To determine the mature structure of ArCRH, mass spectrometric methods were employed. To facilitate this, a synthetic peptide with the predicted structure of Ar-CRH was analysed in parallel with analysis of A. rubens radial nerve cord extracts. Synthetic ArCRH was detected with a monoisotopic mass of m/z 1,129.85 (4+) (online suppl. Determination of the structure of the A. rubens CRH-type neuropeptide ArCRH and comparison of its sequence with CRHrelated peptides in other taxa. a Amino acid sequence of the Ar-CRH precursor (GenBank accession number: KT601710.1); the predicted signal peptide is shown in blue, predicted dibasic/tribasic cleavage sites are shown in green, and the predicted CRH-type peptide is shown in red, with its C-terminal glycine (G) residue that is a putative substrate for amidation shown in orange. b The mature structure of ArCRH as determined by mass spectrometric analysis of A. rubens radial nerve cord extracts, with post-translational conversion of an N-terminal glutamine to pyroglutamate shown as pQ and conversion of a C-terminal glycine residue to an amide (a) shown in lowercase. Fragments of the ArCRH peptide for which structures were determined by mass spectrometry are indicated by the lines underneath the peptide sequence, with (brown) or without (purple) trypsin treatment of nerve cord extracts. The molecular mass divided by the charge state (m/z) of the peptide ions is shown for each fragment, with the charge state in parentheses. The mass spectrometric data for these peptide fragments are shown in online suppl. Fig. 3. c Alignment of ArCRH with CRH-related peptides in other taxa. Conserved residues high-lighted in black or grey, with black signifying a higher level of sequence conservation than grey. Several of the conserved residues and the C-terminal amide group in ArCRH are highlighted with asterisks above the alignment. Underlining of the N-terminal glutamine (Q) in A. rubens CRH and P. sauvagii sauvagine signifies that this residue has been shown to be post-translationally converted to a pyroglutamate in the mature peptides. An N-terminal Q in two of the CRH-type peptides in the brittle star Ophionotus victoriae is likewise a potential substrate for pyroglutamylation, but this has not been proven experimentally so the Q is not underlined. The species/name for each peptide is highlighted with one of the following phylum-specific colours: yellow (Echinodermata), purple (Hemichordata), light blue (Chordata), and olive (Arthropoda  Fig. 3b, c). Analysis of radial nerve cord extracts that had been treated with trypsin, but without dithiothreitol, revealed 2 peptide fragments of ArCRH, pQGLSVSPIFPIQR and DRQDQVDQAEAN-QGLFQIA-NH 2 , with monoisotopic masses of 712.89 m/z (2+) and 715.68 m/z (3+), respectively ( Fig. 1b; online suppl. Fig. 3d, e). In summary, the mass spectrometric analysis of A. rubens radial nerve cord extracts under different experimental conditions confirmed the predicted sequence of ArCRH, with an N-terminal pyroglutamate and a C-terminal amide identified as post-translational modifications (Fig. 1b).
Having determined the structure of ArCRH, we compared it with the sequences of CRH and CRH-related peptides in other taxa, as shown in the alignment in Figa b ure 1c. As a 40-residue peptide, ArCRH has a typical length for a CRH-type peptide as illustrated by the peptides included in the alignment in Figure 1c, which range in length from 38 to 48 residues. C-terminal amidation is another characteristic of ArCRH, which is a conserved characteristic of CRH-related peptides. Furthermore, other evolutionarily conserved features highlighted with asterisks in Figure 1c include the following residues in ArCRH, with positions in parentheses: LSV (3-5), I (11), R (13), L (16), QAE (29)(30)(31), N (33), and L (36).

Cellular Localization of ArCRHP Transcripts in A. rubens Using mRNA in situ Hybridization
Here, mRNA in situ hybridization was employed to analyse expression of ArCRHP in A. rubens, revealing a widespread pattern of expression including in the central nervous system (Fig. 2), the digestive system ( Fig. 3), and body wall-associated structures (Fig. 4).
Central Nervous System The central nervous system of A. rubens comprises 5 radial nerve cords linked by a circumoral nerve ring situated in the central disc region. Both the radial nerve cord and circumoral nerve ring comprise 2 distinct regions: the ectoneural region and the hyponeural region, which contains the cell bodies of motoneurons [40]. Anti-sense probes revealed ArCRHP-expressing cells in both the ectoneural and hyponeural regions of the radial nerve cords (Fig. 2a-e), and the specificity of staining observed with anti-sense probes was confirmed by control experiments using sense probes, where no staining was observed (Fig. 2a, inset). Examination of the distribution of Ar-CRHP-expressing cells in transverse sections of the radial nerve cord revealed that stained cells are distributed throughout the epithelial layer of the ectoneural region, whereas in the hyponeural region, only a single stained cell was observed ( Fig. 2a-c). Accordingly, examination Localization of ArCRHP mRNA in the radial nerve cords and circumoral nerve ring of A. rubens using in situ hybridization. a Transverse section of a radial nerve cord that was incubated with antisense probes showing stained cells in both the ectoneural (arrowheads) and the hyponeural (arrow) regions. In the ectoneural region, stained cells are distributed throughout the sub-cuticular epithelium, whereas only 1 stained cell can be seen in the hyponeural region. Higher magnification images of the boxed regions are shown in panels (b, c). The inset in (a) shows an absence of stained cells in a transverse section of a radial nerve cord incubated with sense probes, demonstrating the specificity of staining observed with antisense probes. b High-magnification image of the apex of the V-shaped radial nerve cord showing that this region is largely devoid of stained cells; distal to the apex a single stained cell (arrowhead) can be seen in the ectoneural epithelium. c High-magnification image of the lateral region of the radial nerve cord showing stained cells in both the ectoneural region (arrowheads) and the hyponeural region (arrow). d Longitudinal parasagittal section of a radial nerve cord showing that stained cells are distributed along its length in the ectoneural region (arrowheads). In the hyponeural region, a single stained cell can be seen (arrow). A higher magnification image of the boxed region is shown in (e). e Stained cells in the ectoneural epithelium (arrowheads) and hyponeural region (arrow) of the radial nerve cord. f Transverse section of the central disc region showing stained cells in both the ectoneural (arrowheads) and the hyponeural (arrows) regions of the circumoral nerve ring. High-magnification images of the boxed regions are displayed in panels (g, h). g Stained cells located in the ectoneural region of the circumoral nerve ring (arrowheads) close to its junction with the peristomial membrane, which also contains stained cells. h Stained cells in the ectoneural epithelium (arrowheads) and hyponeural region (arrows) of the circumoral nerve ring. Scale bars: 100 μm in (a, d, f), 50 μm in inset (a), and 25 μm in (b, c, e, g, h). CONR, circumoral nerve ring; Ec, ectoneural region of radial nerve cord; Hy, hyponeural region of radial nerve cord; PM, peristomial membrane; RHS, radial haemal strand; RNC, radial nerve cord; TF, tube foot; THS, transverse haemal strand; Ar-CRHP, A. rubens CRH-type precursor; CRH, corticotropin-releasing hormone. of longitudinal sections of the radial nerve cord revealed stained cells along the length of the ectoneural epithelium but with only a single stained cell observed in the hyponeural region (Fig. 2d, e). Consistent with the pattern of ArCRHP expression in the radial nerve cords, stained cells were revealed in both the ectoneural and hyponeural regions of the circumoral nerve ring. However, the relative abundance of stained cells in the circumoral nerve ring was higher than in the radial nerve cords, with densely packed clusters of stained cells in the ectoneural region and several stained cells in the hyponeural region being revealed in transverse sections ( Fig. 2f-h).
Digestive System Cells expressing ArCRHP were revealed in the cardiac stomach (Fig. 3a, b) and pyloric stomach (Fig. 3c, d), and in both regions of the stomach, the stained cells are located within the mucosal layer but in close proximity to the basiepithelial nerve plexus (BNP) layer (Fig. 3b, d). Stained cells were also revealed in the pyloric ducts that link the pyloric stomach with the paired pyloric caeca (digestive organs) located in each arm, with the ArCRHPexpressing cells located on the aboral side of the pyloric duct (Fig. 3e, f). Stained cells were also revealed in the pyloric caeca (Fig. 3g, h), located in the duct region of each pyloric caecum diverticulum.

Apical Muscle
The apical muscle is located in a sagittal and aboral position along the coelomic lining of each arm in A. rubens. ArCRHP-expressing cells were revealed in the coelomic epithelium of the apical muscle, as shown here in both transverse (Fig. 4a, b) and longitudinal (Fig. 4c, d) sections of arms.

Characterization of a Rabbit Antiserum to ArCRH
Using ELISA To test for the presence of antibodies to ArCRH in antiserum from a rabbit that had been immunized with a conjugate of thyroglobulin and an ArCRH antigen peptide (ArCRH-ag), serum from the second bleed was prepared at dilutions ranging from 1:500 to 1:128,000 and incubated with a fixed amount of ArCRH-ag in a microtitre plate (1 × 10 −10 moles/well). In parallel, pre-immune serum was tested in the same way and experiments without any added serum were also performed. No immunoreactivity was detected in wells containing pre-immune serum, but immunoreactivity was detected with the anti- serum at dilutions ranging from 1:500 to 1: 32,000 (online suppl. Fig. 4a). To assess the sensitivity of the antiserum for detection of ArCRH-ag, the antiserum (diluted at 1:16,000) was incubated with different amounts of Ar-CRH-ag ranging from 1 × 10 −9 to 1 × 10 −16 moles. At this antiserum dilution, 1 × 10 −9 to 1 × 10 −11 moles of the antigen peptide could be detected (online suppl. Fig. 4b).
Collectively, these ELISA results demonstrated that antibodies to ArCRH were generated successfully. Furthermore, the finding that even at antiserum dilutions as low as 1:32,000 ArCRH-ag could be detected indicates that the antiserum contains a high titre of antibodies to the ArCRH. Accordingly, the ArCRH antiserum was used for immunohistochemical analysis of ArCRH expression in A. rubens (see below) at a dilution of 1:32,000.

Immunohistochemical Localization of ArCRH in A. rubens
Immunohistochemical analysis using the ArCRH antiserum revealed that ArCRH-immunoreactive cells and processes are widely distributed in A. rubens, including in the nervous system (Fig. 5), tube feet (Fig. 6), the digestive system (Fig. 7, 8), and body wall-associated structures (Fig. 9).

Nervous System
Intense ArCRH immunoreactivity was detected in the radial nerve cords and circumoral nerve ring of A. rubens (Fig. 5a, b). Importantly, the specificity of the immunostaining observed with the ArCRH antiserum was confirmed by pre-absorption control experiments (Fig. 5b,   inset). Figure 5a provides an overview of the distribution of ArCRH in the nervous system in a horizontal section of a juvenile starfish, with immunostaining revealed in the circumoral nerve, the radial nerve cords, and the marginal nerves, which run parallel to the outer row of tube feet on each side of the arms.
Analysis of immunostaining in transverse sections of the radial nerve cord revealed that the distribution of stained cells (Fig. 5b) is highly consistent with the distribution of ArCRHP transcripts revealed using mRNA in situ hybridization (see Fig. 2). Thus, stained cells can be seen throughout much of the epithelial layer of the ectoneural region but are absent at the apex of the V-shaped nerve cord. Furthermore, use of immunohistochemistry enabled visualization of ArCRH in the dense fibre network of the underlying neuropile of the ectoneural region (Fig. 5b, c). Also, the bipolar shape of the ectoneural Ar-CRH-immunoreactive cells can be seen in Figure 5c. In the hyponeural region, only a single stained cell can be seen in the transverse section shown in Figure 5b and at higher magnification in Figure 5d, consistent with findings from use of mRNA in situ hybridization for localization of ArCRHP transcripts.
At the margins of the radial nerve cord, immunostaining in the ectoneural neuropile is continuous with immunostaining in the sub-epithelial nerve plexus (SNP) of adjacent tube feet, while immunostained processes derived from the hyponeural region project around the margin of the perihaemal canal (Fig. 5e). High-magnification images of transverse sections of the marginal nerve cord reveal that immunostaining in the neuropile of the mar- In the ectoneural epithelial layer, the stained cells (arrowheads) are concentrated laterally and no stained cells are present in apical region. A single stained cell can be seen in the hyponeural region (arrow). The inset shows an absence of staining in a section of a radial nerve cord incubated with antiserum that was pre-absorbed with the ArCRH peptide antigen, demonstrating the specificity of immunostaining observed in sections incubated with the antiserum. The boxed regions are displayed at high magnification in (c, d). c Immunostained bipolar-shaped cells in the ectoneural epithelium with stained processes (arrowheads) projecting into the underlying stained neuropile (*). d Immunostained cell in the hyponeural region and immunostained axon profiles in the neuropile of the adjacent ectoneural region. e Immunostaining at the junction between a radial nerve cord and an adjacent tube foot. Immunostained processes (arrowhead) appear to project from the ectoneural neuropile of the radial nerve cord into the BNP of the tube foot. Immunostained processes (arrow) derived from the hyponeural region project around the margin of the perihaemal canal. f Immunostaining in the marginal nerve and in the BNP (arrowheads) of an adjacent tube foot. Immunostained processes (arrow) can also be seen here in the muscle layer the tube foot. g Immunostaining in a longitudinal section of the circumoral nerve ring showing immunostained cells in ectoneural epithelium (arrowheads) and immunostained processes in the underlying ectoneural neuropile (asterisk) and in the hyponeural region (arrows). Scale bars: 250 μm in (a, f), 100 μm in (b, e), 25 μm in (g), 50 μm in inset (b), and 12.5 μm in (c, d). CONR, circumoral nerve ring; Ec, ectoneural region of radial nerve cord; Es, oesophagus; Hy, hyponeural region of radial nerve cord; MN, marginal nerve; RHS, radial haemal strand; RNC, radial nerve cord; TF, tube foot; ArCRHP, A. rubens CRH-type precursor; CRH, corticotropin-releasing hormone; BNP, basiepithelial nerve plexus.  , c). b Immunostaining (arrow) in the SNP in the stem of a tube foot. Immunostained fibres (arrowhead) can also be seen here in the muscle layer. c Immunostaining in the basal nerve ring of a tube foot disc. d Immunostaining in the ampulla of a tube foot. The boxed region is shown at higher magnification in (e), which shows that the immunostaining is present in fibres located beneath the coelomic lining of the ampulla (arrowhead) and in fibres located in the muscle layer (arrow). Tube Feet Immunostaining is present throughout the SNP of tube foot podia and extending into the basal nerve ring (Fig. 5f, 6a-c). No immunostained cell bodies were observed in tube feet, indicating that the immunostaining in the SNP is associated with fibres derived from cells located in the adjacent ectoneural epithelium of the radial nerve cord. Immunostained fibres were also revealed in the muscle layer of tube foot podia (Fig. 5f,  6b). Immunostaining is also present in tube foot ampullae, and here it is located in fibres underlying the coelomic epithelial layer and in fibres in the muscle layer (Fig. 6d, e).

Digestive System
Immunostaining was revealed in the peristomial membrane (Fig. 7a), where it is localized in the BNP located beneath the external epithelial layer and in a nerve plexus located beneath the coelomic epithelial lining (Fig. 7b). Accordingly, immunostained cells were also observed in close proximity to the immunostained plexi (Fig. 7b). A similar pattern of immunostaining to that observed in the peristomial membrane is seen in the oesophagus (Fig. 7c). Immunostaining was also observed in the cardiac stomach, most notably in the highly folded lateral pouches proximal to intrinsic retractor strands (Fig. 7d-h) [41]. Furthermore, immunostaining was not evenly distributed throughout the folds of the cardiac stomach, with immunostaining most intense in the mucosal basiepithelial plexus and associated cells adjacent to intrinsic retractor strands and with immunostaining least intense in inter-vening sections of the stomach wall (Fig. 7d-f). Accordingly, in small juvenile starfish where it is possible to view an entire horizontal section of the cardiac stomach, immunostaining can be seen to be concentrated in regions of the BNP adjacent to sites of attachment of the extrinsic retractor strands (Fig. 7g, h).
In the pyloric stomach, immunostaining was widely distributed throughout the BNP, as seen in both transverse (Fig. 7d) and horizontal (Fig. 8a) sections of the central disc region. At high magnification, it can be observed that immunostained fibres in the BNP are derived from immunostained cells located at the boundary between the nerve plexus and the mucosa (Fig. 8b). Immunostaining associated with the BNP also extends into the pyloric ducts, which link the pyloric stomach to the paired pyloric caeca located in each arm (Fig. 8a, c). In the pyloric caeca, the pattern of immunostaining is inhomogeneous, and this can be observed in both horizontal (Fig. 8a, d) and transverse (Fig. 8f, i) sections of pyloric caeca. Immunostaining is most prominent at the apex of the oral side of the pyloric duct regions of the pyloric caeca, whereas in diverticula, immunostaining is less prominent (Fig. 8a, arrow and arrowhead). Furthermore, high-magnification images of the ducts reveal that immunostaining is localized in both the BNP of the mucosal layer and the nerve plexus located beneath the coelomic epithelial lining (Fig. 8h, arrow and arrowhead). Last, no staining was observed in the rectal caeca (Fig. 8j), but immunostaining was observed in the BNP of the rectum (Fig. 8j, k) and in both the BNP and sub-coelomic nerve plexus of the intestine (Fig. 8j, l). Immunohistochemical localization of ArCRH in the peristomial membrane, oesophagus, and cardiac stomach of A. rubens. a Transverse section of the central disc region showing immunostaining in circumoral nerve ring and peristomial membrane. The boxed region is displayed at higher magnification in (b), which shows an immunostained cell (arrowhead) in the external epithelium of the peristomial membrane and immunostained fibres in the underlying BNP and an immunostained cell (arrow) beneath the coelomic epithelium and immunostained fibres in the underlying nerve plexus. c Horizontal section of the central disc region in a small juvenile specimen showing immunostained fibres in nerve plexi of the peristomial membrane and the oesophagus. d Transverse section of the central disc region showing immunostaining in both the cardiac stomach and the pyloric stomach. Note that there is variation in the density of immunostaining in the BNP in the folds of the cardiac stomach wall, as exemplified by the 2 boxed areas that are shown at high magnification in (e, f). e Region of the cardiac stomach wall adjacent to an intrinsic retractor strand; here, a stained bipolar cell can be seen in the mucosal wall and the BNP is intensely stained (********). Immunostained fibres can also be seen in an intrinsic retractor strand. f Region of the cardiac stomach, where staining of the BNP is less intense (……….) than in the adjacent region shown in (e). g Horizontal section of the central disc region of a juvenile starfish showing immunostaining in the cardiac stomach. The boxed region is displayed at higher magnification in (h), which shows variation in the intensity of immunostaining, with a region adjacent to sites of attachment of extrinsic retractor strands containing stained cells and a thickened and intensely stained BNP (********), while a neighbouring region is largely void of immunostaining (…….). Scale bars: 250 μm in (a), 200 μm in (d, g), 25 μm in (e, f, h), and 12. 5 μm in (b, c). BNP, basiepithelial nerve plexus; CS, cardiac stomach; CONR, circumoral nerve ring; CE, coelomic epithelium; CT, collagenous tissue; Es, oesophagus; ERS, extrinsic retractor strand; IRS, intrinsic retractor strand; Lu, lumen; Mu, mucosa; PM, peristomial membrane; PS, pyloric stomach; RNC, radial nerve cord; ArCRHP, A. rubens CRH-type precursor; CRH, corticotropin-releasing hormone.

Body Wall-Associated Structures
In the body wall, immunostained processes are present proximal to the circular muscle layer and in the coelomic BNP (Fig. 9a, b). Immunostained cells are present in the coelomic epithelial layer of the apical muscle (Fig. 9b) and immunostained processes derived from these cells ramify among the longitudinally orientated muscle fibres of the apical muscle (Fig. 9b). The immunostained nerve plexi associated with the longitudinally and circularly orientated muscle layers of the body wall also extend into the walls of papulae, finger-shaped appendages that enable gas exchange between the coelomic fluid and the external seawater (Fig. 9a, c). Immunostaining is also present in the SNP of the external body wall epithelium (Fig. 9d).

Discussion
In this paper, we report the first analysis of the molecular structure and expression of a CRH-type neuropeptide in an echinoderm -the starfish A. rubens. Previous studies have reported the sequencing of a cDNA encoding the precursor of an A. rubens CRH-type neuropeptide (ArCRH) [35,36]. Here, mass spectrometric analysis of extracts of radial nerve cords from A. rubens enabled determination of the mature structure of ArCRH, revealing that it is a 40-residue peptide with Nterminal pyroglutamylation and C-terminal amidation. The frog skin peptide sauvagine, which was the first CRHtype peptide to be sequenced, also has an N-terminal pyroglutamate [7]. However, this is not a conserved feature of CRH-type peptides; for example, human CRH and urocortins do not have an N-terminal pyroglutamate (Fig. 1c). In contrast, C-terminal amidation is a conserved feature of CRH-type peptides (Fig. 1c). Furthermore, comparison of the sequence of ArCRH with CRH and related neuropeptides in other taxa revealed other conserved structural characteristics, as highlighted with asterisks in Figure 1c. It is noteworthy that the conserved residues are predominantly located within the N-terminal and C-terminal regions of the peptide, with the central region of the peptide (residues 17-28) exhibiting less sequence conservation with other CRH-type peptides. This is interesting because the N-and C-terminal regions of mammalian CRH-type peptides are involved in receptor binding/activation [42]. Therefore, it is likely that these regions are also important for the bioactivity of Ar-CRH.
Cloning and sequencing of a cDNA encoding the Ar-CRH precursor (ArCRH) has previously enabled generation of labelled RNA probes for ArCRHP transcripts, which have been employed for cellular localization of Ar-CRHP expression in A. rubens larvae using mRNA in situ hybridization techniques [36]. No expression of ArCRHP was observed during early larval development in the 2-armed bipinnariae, but in brachiolaria larvae, Ar-CRHP-expressing cells were observed in the brachia and in tissue adjacent to the adhesive disc [36]. The adhesive disc mediates larval attachment to the substratum prior to metamorphosis of the bilaterally symmetrical larval stage into a pentaradially symmetrical juvenile [43,44]. Therefore, we speculated that ArCRH may be involved in regulation of physiological processes associated with larval attachment [36].
Here, mRNA in situ hybridization was employed to investigate ArCRHP expression for the first time in post- metamorphic juvenile and adult starfish. Furthermore, determination of the mature structure of ArCRH provided a basis for generation of specific antibodies to the Cterminal region of this peptide. Characterization of an ArCRH rabbit antiserum using ELISA revealed a high titre of antibodies to the C-terminal region of ArCRH, enabling use of a highly diluted antiserum (1:32,000) for immunohistochemical localization of ArCRH. Analysis of the distribution of ArCRHP transcripts and ArCRH peptide in A. rubens using mRNA in situ hybridization and immunohistochemistry, respectively, revealed mutually consistent patterns of stained cells. Furthermore, immu-nohistochemistry enabled visualization of the stained axonal processes of ArCRH-expressing neurons in A. rubens. Informed by the patterns of ArCRHP/ArCRH expression in the A. rubens, we present below a functional interpretation and discussion of our findings.
In the central nervous system of A. rubens, which comprises the circumoral nerve ring and 5 radial nerve cords, ArCRH-expressing cells were revealed in both the ectoneural and hyponeural regions. The ectoneural region receives and integrates input from sensory cells located throughout the external epithelium of the body surface, and accordingly, it has an extensive neuropile layer [40].  ArCRH-expressing cells were revealed in the epithelium of the ectoneural region, and in the underlying neuropile, a dense population of ArCRH-immunoreactive fibres was revealed. These findings are generally consistent with previous analyses of the expression of other neuropeptides, although specific differences in the patterns of ectoneural neuropeptide expression in A. rubens are observed. The hyponeural region of the CNS in starfish and other echinoderms contains segmental clusters (ganglia) of motoneuronal cell bodies, and the activity of these neurons is driven by input from ectoneural neurons [40,45]. In this context, it is noteworthy that typically only a single ArCRH-immunoreactive hyponeural cell body was observed in transverse sections of the radial nerve cord in A.
rubens. This contrasts with other neuropeptides such as pedal peptide/orcokinin-type, GnRH-type, calcitonintype, and somatostatin-type peptides, where several cell bodies expressing these neuropeptides were observed in transverse sections of the radial nerve cord in A. rubens [31,32,34,38,46]. However, in sections of the circumoral nerve ring of A. rubens, several ArCRH-immunoreactive cell bodies were typically revealed in the hyponeural region, indicating regional differences in hyponeural Ar-CRH expression in the A. rubens CNS. The functional significance of differences in patterns of hyponeural neuropeptide expression in A. rubens is not known. However, analysis of peripheral expression of neuropeptides in the axons of hyponeural motoneurons may be informative. For example, one interesting feature of ArCRH expression in A. rubens is the presence of ArCRH-immunoreactive fibres in the muscle layer of the ampulla and podium of tube feet. This is noteworthy because hitherto we have not observed expression of other neuropeptides in nerve fibres located within the muscle layer of tube feet. Conversely, motor fibres innervating interossicular muscles of the body wall in A. rubens are immunoreactive with antibodies to other neuropeptides (e.g., pedal peptide/orcokinin-type and calcitonin-type) [32,38,46], but these nerve fibres were not observed to be immunoreactive with ArCRH antibodies in this study. Thus, comparative analysis of neuropeptide expression provides a basis for identification of different populations of hyponeural neurons, and our findings reported here indicate that Ar-CRH is expressed in a sub-population of hyponeural motoneurons that directly innervate the muscle layer in the podium and ampulla of tube feet in A. rubens. Furthermore, ArCRH immunoreactivity is also present in other compartments of the nervous system associated with tube feet, including nerve fibres located within the coelomic lining of the ampulla and nerve fibres located in the SNP and basal nerve ring of the podium. Therefore, ArCRH may participate in a variety of neural mechanisms associated with regulation of tube foot activity, with potential relevance to behaviours that involve tube feet (e.g., locomotion and feeding).
Gas exchange in starfish is mediated by protractible organs called papulae, which penetrate through the body wall skeleton and provide a thin-walled surface for gas exchange between external seawater and coelomic fluid that bathes visceral organs [47]. ArCRH immunoreactivity was revealed in the innervation of papulae, which is derived from the axons of hyponeural motoneurons associated with the circular muscle layer of the body wall and the apical nervous system associated with the longitudinal muscle layer of the body wall. Consistent with the latter, ArCRHP/ArCRH-expressing cells were revealed in aboral coelomic lining of the body wall and ArCRH-immunoreactive nerve fibres were revealed within the apical muscle, a thickened strand of longitudinally orientated muscle that facilitates arm flexion. Therefore, ArCRH may participate in neural mechanisms associated with regulation of papula protraction/retraction and control of the circularly and longitudinally orientated muscle layers of the body wall in A. rubens.
A widespread pattern of ArCRH expression was observed in the digestive system of A. rubens, including the peristomial membrane that surrounds the mouth, oesophagus, cardiac stomach, pyloric stomach, pyloric ducts, pyloric caeca, intestine, and rectum. ArCRH-immunoreactive cells were revealed in the gut mucosal layer, with immunostained processes contributing to staining in the BNP. Based on these characteristics, the ArCRH-immunoreactive cells in the gut mucosa can be classified as enteric neurons with axonal processes and/or endocrine cells with neuropods [48], but detailed ultrastructural analysis will be required for a more precise identification of cell types expressing ArCRH in the starfish digestive system. ArCRH immunoreactivity was also revealed in the visceral nerve plexus, which is located beneath the coelomic epithelium of the gut. The extensive expression of Ar-CRH throughout much of the digestive system in A. rubens is indicative of a general role in regulation of gut function. Noteworthy in this respect is the variation in the intensity or density of immunostaining in the BNP within regions of the gut. For example, in the cardiac stomach, immunostaining is stronger in regions adjacent to the intrinsic retractor strands, and in the pyloric caeca immunostaining is stronger in the pyloric duct region than in the diverticula. Variation in the intensity/density of immunostaining within regions of the gut has been also ob-DOI: 10.1159/000517087 served for other neuropeptides in A. rubens; for example, the calcitonin-type neuropeptide ArCT [32]. However, its functional significance remains to be elucidated. The higher concentration of ArCRH immunoreactivity in the pyloric duct region of the pyloric caeca could be informative in this respect because this region of the gut is involved in ciliary-mediated transport of food material [49]. Therefore, ArCRH may be involved in neural regulation of these processes in the digestive system of A. rubens.
The extensive expression of ArCRH in A. rubens is clearly indicative of roles in regulation of a variety of physiological processes, as discussed above. Therefore, it is of interest to consider more broadly what is known about the physiological roles of CRH-related peptides in other taxa. In mammals, CRH is, of course, known for its role in mediating physiological adaptations to stressors, stimulating pituitary secretion of adrenocorticotropic hormone, which leads to adrenal release of cortisol. However, CRH and its paralogue UCN also have behavioural effects associated with adaptation to stressors and novel environments, including increased arousal, decreased food intake, reduced sexual and reproductive activity, and increased grooming. Thus, it is proposed that CRH/ UCN acting via the CRF1 receptor in the CNS integrates physiological and behavioural responses to stressors [50]. Interestingly, it has been proposed that these actions of CRH/UCN may be reflective of a highly specialized role in mammals, with the actions of UCN2 and UNC3, perhaps being more representative of the neuropeptide family as a whole. In this context, it is noteworthy that UCN2/ UCN3 acting via CRF2 have anxiolytic effects and roles in energy homeostasis, with UCN2 increasing tissue sensitivity to insulin and UCN3 increasing pancreatic secretion of insulin [50]. Furthermore, investigation of the expression and actions of a CRH-type peptide in the invertebrate chordate C. intestinalis has revealed evidence of roles in inhibitory regulation of feeding behaviour [51].
Turning now to protostome invertebrates, here our knowledge of CRH-type neuropeptide function draws largely from investigation of the physiological roles of the CRH-like diuretic hormone DH44 in Drosophila melanogaster. This has revealed that, in addition to its diuretic action [52][53][54], DH44 has roles in regulation of a variety of processes, including circadian locomotor activity [55][56][57], feeding [53,54,58,59], and sperm retention/storage in post-copulatory females [60]. With respect to the latter function, it is noteworthy that ELH in Aplysia californica and in other gastropod molluscs has been identified as a DH44-like peptide [18,[61][62][63][64]. Therefore, DH44/ELHtype neuropeptides may have evolutionarily conserved roles as regulators of reproductive processes in protostomes. Further studies on a wider range of taxa are now needed to gain broader insights into the physiological roles of DH44-type neuropeptides in protostomes.
In conclusion, our detailed analysis of the expression pattern of the CRH-type neuropeptide ArCRH in an echinoderm, the starfish A. rubens, provides an anatomical basis for experimental investigation of its physiological roles and comparison with the findings from chordates and protostome invertebrates discussed above. It will also be of interest to investigate the expression and physiological roles of CRH-type neuropeptides in other echinoderms. Thus far, precursors of ArCRH-like neuropeptides have not been identified in echinoids (e.g., sea urchins), holothurians (sea cucumbers), or crinoids (e.g., feather stars). However, 4 precursors of ArCRH-like neuropeptides have been identified in an ophiuroid species, the brittle star Ophionotus victoriae [65]. Investigation of the expression of CRH-type precursors in O. victoriae and/or other brittle stars would enable comparison with findings reported here for ArCRH in the starfish A. rubens and exploration of the physiological significance of gene duplication giving rise to multiple CRH-type neuropeptide precursors in these animals.