Evolution of neurohormone function revealed by actions of kisspeptin-type peptides in an echinoderm

TL;DR

This paper explores how kisspeptin-type peptides in starfish regulate various physiological processes, revealing their evolutionary origins and diverse functions.

Contribution

The study identifies two kisspeptin precursor proteins in starfish and shows their neuropeptides have opposing effects on muscle activity and behavior.

Findings

Kisspeptin-type neuropeptides in starfish have myoinhibitory and myoexcitatory effects on gonads and stomach.

ArKP2.2 acts as a ligand for a receptor related to vertebrate kisspeptin receptors.

Phylogenetic analysis suggests inhibitory kisspeptin signaling evolved uniquely in Ambulacraria or was lost in chordates.

Abstract

The neurohormone kisspeptin regulates reproductive maturation and function in mammals by stimulating hypothalamic production and release of gonadotropin-releasing hormone. However, little is known about kisspeptin-type neuropeptide function in invertebrates and the evolution of kisspeptin signalling as a regulator of physiological processes. Here, we address these issues in a deuterostome invertebrate — the starfish Asterias rubens (phylum Echinodermata).

Unlike mammals that have one kisspeptin precursor protein, in A. rubens, two precursor proteins (ArKPP1, ArKPP2) give rise to four kisspeptin-type neuropeptides (ArKP1.1, ArKP1.2, ArKP2.1, ArKP2.2). Use of mRNA in situ hybridisation and immunohistochemistry revealed widespread but different patterns of expression of KP1-type and KP2-type neuropeptides in the central nervous system, locomotory organs, sensory organs, reproductive system, and digestive system of A. rubens. In vitro experiments revealed that KP1-type and KP2-type neuropeptides have opposing myoinhibitory and myoexcitatory effects, respectively, on starfish gonad and stomach preparations. When injected in vivo, both KP1-type and KP2-type neuropeptides trigger stomach eversion and ArKP1.2 affects righting behaviour.

This study has revealed that kisspeptin-type neuropeptides are evolutionarily ancient and pleiotropic regulators of processes that extend beyond reproductive physiology. Furthermore, the excitatory actions of ArKP2.2 in A. rubens are consistent with stimulatory effects of kisspeptins in vertebrates; accordingly, ArKP2.2 acts as a ligand for a receptor (ArKPR3) that is closely related to vertebrate kisspeptin receptors. On the other hand, phylogenetic analysis of receptors for ArKP1.1 and ArKP1.2 indicates that inhibitory kisspeptin signalling either evolved uniquely in Ambulacraria (echinoderms, hemichordates) or originated in Urbilateria but was lost in chordates.

The online version contains supplementary material available at 10.1186/s12915-026-02555-1.

Figures

Actions of kisspeptin-type neuropeptides in starfish provide insights into the evolution of kisspeptin signalling. The dendrogram on the left shows relationships between kisspeptin-type receptors in the starfish *A. rubens* and kisspeptin-type receptors in chordates (e.g. human) and protostomes (e.g. molluscs) based on a phylogenetic analysis of sequence data reported previously [[12](#CR12)]. Neuropeptides that have been identified as ligands for kisspeptin-type receptors in chordates and in *A. rubens* [[12](#CR12)] are listed, and the actions (excitatory or inhibitory) of these neuropeptide

Sequences of ArKPP1 and ArKPP2 and alignment of neuropeptides derived from these precursors with other kisspeptin-type peptides. A Sequences of ArKPP1 and ArKPP2. The predicted N-terminal signal peptide is shown in blue, neuropeptides derived from ArKPP1 or ArKPP2 are shown in red (with C-terminal glycine that is a substrate for amidation shown in orange), and monobasic or dibasic cleavage sites are shown in green. B Alignment of ArKP1.1, ArKP1.2, and KP1-type neuropeptides from other echinoderms (light blue) with human kisspeptin (KP-14, light peach). C Alignment of ArKP2.1, ArKP2.2, and KP2-

Localisation of ArKPP1 expression in *Asterias rubens* using mRNA in situ hybridisation. A Transverse section of radial nerve cord showing ArKPP1-expressing cells in both the ectoneural (arrowheads) and hyponeural (arrow) regions. The inset shows the absence of stained cells in radial nerve cord incubated with sense probes. B Higher magnification of the left side boxed area in A, showing stained cells (arrowheads) in the ectoneural region of radial nerve cord adjacent to a tube foot. C Higher magnification of the right-side boxed area in A, showing stained cells (arrows) in hyponeural region o

Immunohistochemical localisation of ArKP1.1 and ArKP2.1 in central nervous system, tube feet, and arm tip. A, B Horizontal sections of juvenile starfish showing immunostaining in radial nerve cords, circumoral nerve ring, and tube feet. C, D Immunostaining in transverse sections of radial nerve cord, which is abolished by pre-absorption with antigen peptides (insets). E, F Boxed regions in C and D, respectively, showing immunoreactivity in ectoneural cells (black arrowheads) and neuropile (asterisks). ArKP1.1-, but not ArKP2.1, immunoreactive cells (black arrows) are present in hyponeural regi

Immunohistochemical localisation of ArKP1.1 and ArKP2.1 in central nervous system, tube feet, and arm tip. A, B Horizontal sections of juvenile starfish showing immunostaining in radial nerve cords, circumoral nerve ring, and tube feet. C, D Immunostaining in transverse sections of radial nerve cord, which is abolished by pre-absorption with antigen peptides (insets). E, F Boxed regions in C and D, respectively, showing immunoreactivity in ectoneural cells (black arrowheads) and neuropile (asterisks). ArKP1.1-, but not ArKP2.1, immunoreactive cells (black arrows) are present in hyponeural regi

Opposing effects of ArKPP1- and ArKPP2-derived neuropeptides on in vitro organ preparations from *A. rubens*. A, B ArKP1.1 and ArKP1.2 cause relaxation of gonad preparations precontracted with acetylcholine, whereas ArKP2.2 causes contraction of gonad preparations (see also Additional file 5: Fig. S4). C Graph comparing relaxing effects of neuropeptides on gonad preparations (normalised as percentage reversal of contracting effect of 10^−6^ M acetylcholine). D Graph comparing contracting effects of ArKP2.2 and acetylcholine (both at 10^−6^ M) on gonad preparations (normalised to contracting ef

Opposing effects of ArKPP1- and ArKPP2-derived neuropeptides on in vitro organ preparations from *A. rubens*. A, B ArKP1.1 and ArKP1.2 cause relaxation of gonad preparations precontracted with acetylcholine, whereas ArKP2.2 causes contraction of gonad preparations (see also Additional file 5: Fig. S4). C Graph comparing relaxing effects of neuropeptides on gonad preparations (normalised as percentage reversal of contracting effect of 10^−6^ M acetylcholine). D Graph comparing contracting effects of ArKP2.2 and acetylcholine (both at 10^−6^ M) on gonad preparations (normalised to contracting ef

In vivo injection of ArKPP1- and ArKPP2-derived neuropeptides triggers partial eversion of the cardiac stomach in *A. rubens*. A Photographs of starfish injected with 10 µl water, 10 µl 10^−3^ M asterotocin, 100 µl 10^−3^ M S2, 100 µl aqueous solution containing 0.5 × 10^−3^ M ArKP1.1 and 0.5 × 10^−3^ M ArKP1.2, 10 µl 10% DMSO, and 10 µl 10^−3^ M ArKP2.2 in 10% DMSO at 0, 5, and 10 min after injection. Cardiac stomach eversion is observed after injection of asterotocin, S2, ArKP1.1 + ArKP1.2, and ArKP2.2, and the area of everted stomach is labelled with black dashed lines. Note that asterotoci

Analysis of the effects of ArKPP1- and ArKPP2-derived neuropeptides on righting behaviour in *A. rubens*. A Graph showing time taken (mean ± SEM) for starfish to right without injection (*n* = 46) or after injection of water (*n* = 10, 10 μl), asterotocin (*n* = 10, 10 μl 10^−3^ M), S2 (*n* = 10, 100 μl 10^−3^ M), ArKP1.1 (*n* = 8, 100 μl 10^−3^ M), ArKP1.2 (*n* = 8, 100 μl 10^−3^ M), or ArKP1.1 + ArKP1.2 (*n* = 10, 100-μl aqueous solution containing 0.5 × 10^−3^ M ArKP1.1 and 0.5 × 10^−3^ M ArKP1.2). Statistical analysis reveals significant differences in time taken to right between non-injec

Full text

Background

Maturation and function of the reproductive system in mammals are controlled by pulsatile release of gonadotropin-releasing hormone (GnRH) from hypothalamic neurons. Furthermore, genetic analysis of patients with delayed onset of puberty led to the discovery of kisspeptins, neuropeptides that stimulate GnRH synthesis and release. In humans, the KiSS1 gene encodes a 145-amino acid residue precursor that gives rise to kisspeptins with an identical C-terminal region (KP10, KP13, KP14, KP54), and these neuropeptides exert their effects by binding to the G-protein coupled receptor GPR54/KiSS1R [1–8].

Greater diversity in kisspeptin-type signalling systems has been revealed among nonmammalian vertebrates, including three paralogous kisspeptin precursor genes and four paralogous kisspeptin receptor genes that are thought to have evolved as a consequence of two rounds of whole-genome duplication [9–11]. However, in some extant vertebrates (e.g. birds), there has been complete loss of genes encoding kisspeptin-type precursors and receptors [10–13]. Furthermore, lineage-specific gene duplication has also given rise to multiple kisspeptin-type precursors and receptors in the invertebrate chordate Branchiostoma floridae (phylum Cephalochordata) [9, 12, 14], and evidence of a conserved role of kisspeptin signalling in regulation of reproductive physiology has been obtained from experimental studies on Branchiostoma japonicum [15].

Broader analysis of the phylogenetic distribution of kisspeptin-type receptors in non-chordate animals has revealed their occurrence in other deuterostomes (hemichordates, echinoderms) and in some protostome invertebrates (e.g. molluscs, annelids), indicating that the origin of the kisspeptin-type signalling is evolutionarily ancient and can be traced back to the common ancestor of bilaterian animals [12, 14, 16]. However, little is known about the physiological roles of kisspeptin signalling in non-chordate animals. A key breakthrough was the discovery of genes encoding kisspeptin-type neuropeptides in the starfish Asterias rubens and other echinoderms. Two kisspeptin-type precursors were identified in A. rubens: ArKPP1, which is the precursor of the neuropeptides ArKP1.1 and ArKP1.2, and ArKPP2, which is the precursor of the neuropeptides ArKP2.1 and ArKP2.2 [12, 17] (Fig. 1, Additional file 1: Dataset S1). Furthermore, experimental studies revealed that these neuropeptides act as potent ligands for members of an expanded family of kisspeptin-type receptors in A. rubens. Thus, ArKP1.2 acts as a ligand for ArKPR1 and ArKPR9, ArKP1.1 acts as a ligand for ArKPR8, and ArKP2.2 acts as a ligand for ArKPR3 [12]. However, nothing is known about the physiological roles of these kisspeptin-type neuropeptides in starfish. Therefore, the aim here was to address this issue by using mRNA in situ hybridisation and immunohistochemistry to map and compare the expression patterns in A. rubens of ArKPP1 and ArKPP2 and neuropeptides derived from these precursors (ArKP1.1, ArKP2.1). Then in vitro and in vivo pharmacological experiments were performed to investigate and compare the actions of ArKPP1- and ArKPP2-derived neuropeptides in A. rubens. Our functional characterisation of kisspeptin-type neuropeptides in A. rubens provides new insights into the evolution of kisspeptin signalling as a regulator of reproductive and nonreproductive processes in animals.Fig. 1. Sequences of ArKPP1 and ArKPP2 and alignment of neuropeptides derived from these precursors with other kisspeptin-type peptides. A Sequences of ArKPP1 and ArKPP2. The predicted N-terminal signal peptide is shown in blue, neuropeptides derived from ArKPP1 or ArKPP2 are shown in red (with C-terminal glycine that is a substrate for amidation shown in orange), and monobasic or dibasic cleavage sites are shown in green. B Alignment of ArKP1.1, ArKP1.2, and KP1-type neuropeptides from other echinoderms (light blue) with human kisspeptin (KP-14, light peach). C Alignment of ArKP2.1, ArKP2.2, and KP2-type neuropeptides from other echinoderms (light blue) with human kisspeptin (KP-14, light peach). Abbreviations: Ar, Asterias rubens; Ov, Ophionotus victoriae; Sp, Strongylocentrotus purpuratus; Aj, Apostichopus japonicus; Am, Antedon mediterranea; Hs, Homo sapiens. The accession numbers for the sequences included in this figure are listed in Additional file 1: Dataset S1

Results

Localisation of ArKPP1 and ArKPP2 expression in A. rubens using mRNA in situ hybridisation

As a framework for interpretation of the expression patterns of ArKPP1 and ArKPP2 in A. rubens, a diagram illustrating starfish anatomy is shown in Additional file 2: Fig. S1. The starfish central nervous system comprises radial nerve cords linked by a circumoral nerve ring in the central disc region. Two rows of locomotory tube feet are located on each side of the radial nerve cord, and perioral tube feet are located adjacent to the circumoral nerve ring. Thus, coordinated locomotor activity mediated by tube feet on the underside of each arm requires communication between neurons located in the radial nerve cords in each of the five arms. The mouth is located on the underside of the central disc region and opens into an oesophagus and a large and highly folded cardiac stomach that is everted out of the mouth when starfish feed on prey [18]. Use of mRNA in situ hybridisation revealed the anatomical distribution of cells expressing ArKPP1 and ArKPP2 in A. rubens, as illustrated in Figs. 2 and 3, respectively, and described below.Fig. 2. Localisation of ArKPP1 expression in Asterias rubens using mRNA in situ hybridisation. A Transverse section of radial nerve cord showing ArKPP1-expressing cells in both the ectoneural (arrowheads) and hyponeural (arrow) regions. The inset shows the absence of stained cells in radial nerve cord incubated with sense probes. B Higher magnification of the left side boxed area in A, showing stained cells (arrowheads) in the ectoneural region of radial nerve cord adjacent to a tube foot. C Higher magnification of the right-side boxed area in A, showing stained cells (arrows) in hyponeural region of radial nerve cord. D Transverse section of circumoral nerve ring showing ArKPP1-expressing cells (arrowheads) in ectoneural region. E Higher magnification of boxed area in D, showing a cluster of stained cells at the margin of circumoral nerve ring. F ArKPP1-expressing cells (arrowheads) in epidermis. G Longitudinal section of a tube foot showing ArKPP1-expressing cells (boxed region, arrowheads) in the stem and disc regions. H Higher magnification of boxed area in G showing ArKPP1-expressing cells (arrowheads). I Transverse section of central disk region showing stained cells in mucosal layer of cardiac stomach (arrowheads). The inset shows higher magnification of boxed area. Scale bars: 200 µm (A, D, G, H) and 60 µm (B, C, E, F, H). Insets: 200 µm (A) and 60 µm (I)Fig. 3. Localisation of ArKPP2 expression in Asterias rubens using mRNA in situ hybridisation. A Transverse section of radial nerve cord showing ArKPP2-expressing cells in ectoneural region (arrowheads, boxed area) proximal to junction with adjacent tube feet. Inset shows the absence of stained cells in radial nerve cord incubated with sense probes. B Higher magnification of boxed area in A. C Transverse section of circumoral nerve ring showing ArKPP2-expressing cells (arrowheads) proximal to adjacent peri-oral tube feet and peristomial membrane (boxed area). D Higher magnification of boxed area in C. E ArKPP2-expressing cells in a tube foot (boxed area) and at junction between adjacent tube feet (arrowhead). F Higher magnification of boxed area in E. G Longitudinal section of a tube foot showing ArKPP2-expressing cells (boxed area) in disk region. H Higher magnification of boxed area in G. I ArKPP2-expressing cells (arrowhead) in marginal nerve. J ArKPP2-expressing cells (arrowhead, boxed area) in peristomial membrane. K Higher magnification of boxed area in J. L Longitudinal section of oesophagus showing an ArKPP2-expressing cell (boxed area). Inset shows a higher magnification of boxed area. M ArKPP2-expressing cells (arrowheads, boxed area) in cardiac stomach. Inset shows a higher magnification of boxed area. N Transverse section of apical muscle showing ArKPP2-expressing cells (arrowheads, boxed area) in coelomic epithelium. O Higher magnification of boxed area in N. P ArKPP2-expressing cells (arrowheads, boxed area) in a pedicellaria, with stained cells located in epidermis and proximal to adductor muscles. Q Higher magnification of boxed area in P. Scale bars: 200 µm (A, C, J, G, L, M), 100 µm (E, P), 60 µm (B, D, K), 30 µm (F, H, I), and 6 µm (O, Q). Insets: 200 µm (A) and 60 µm (L, M)

Analysis of the distribution of ArKPP1 transcripts in the central nervous system using antisense probes revealed stained cells in both the ectoneural and hyponeural regions of the radial nerve cords (Fig. 2A, B, C), and the specificity of this staining was confirmed by the absence of stained cells in radial nerve cord sections incubated with sense probes (Fig. 2A inset). Accordingly, stained cells were also revealed in the ectoneural region of the circumoral nerve ring (Fig. 2D, E). Analysis of peripheral expression of ArKPP1 revealed stained cells in the body wall (Fig. 2F), tube feet (Fig. 2G, H), and cardiac stomach (Fig. 2I).

Analysis of the distribution of ArKPP2 transcripts in the central nervous system using antisense probes revealed stained cells in the ectoneural region of the radial nerve cords, but not in the hyponeural region (Fig. 3A, B), and the specificity of staining was confirmed by the absence of stained cells in radial nerve cord sections incubated with sense probes (Fig. 3A inset). ArKPP2-expressing cells in the ectoneural region of the radial nerve cords were found to be mainly concentrated laterally, proximal to the adjacent tube feet (Fig. 3A, B). Accordingly, ArKPP2-expressing cells were also revealed in the ectoneural region of the circumoral nerve ring, concentrated proximal to the perioral tube feet and the peristomial membrane (Fig. 3C, D). ArKPP2-expressing cells were also revealed in the tube feet, both at the base of the stem (Fig. 3E, F) and in the disk region (Fig. 3G, H), and in the marginal nerve (Fig. 3I), which is located lateral to the outer rows of tube feet. ArKPP2-expressing cells were revealed in several regions of the digestive system, including the peristomial membrane, oesophagus, and cardiac stomach (Fig. 3J, K, L, M). Associated with the body wall, ArKPP2-expressing cells (neurons) were revealed in the coelomic epithelial layer of the apical muscle, a strand of muscle that runs longitudinally along the inner surface of the body wall in each arm (Fig. 3N, O) and in pedicellariae, pincer-like organs on the external surface of the body wall that protect against encrusting organisms.

Comparative immunohistochemical analysis of ArKP1.1 and ArKP2.1 expression in A. rubens

Having localised cells expressing ArKPP1 and ArKPP2 transcripts in A. rubens using mRNA in situ hybridisation (Figs. 2 and 3), immunohistochemistry was then employed to compare the distribution of neuropeptides derived from ArKPP1 and ArKPP2. To enable this, antibodies to ArKP1.1 and ArKP2.1 were generated using antigen peptides corresponding to the C-terminal regions of these peptides. Antisera were analysed for the presence of antibodies to ArKP1.1 and ArKP2.1 antigen peptides using an enzyme-linked immunosorbent assay (ELISA). This revealed the presence of antibodies to ArKP1.1 and ArKP2.1 in the ArKP1.1 antiserum and the ArKP2.1 antiserum, respectively. Furthermore, the specificity of these antibodies for their respective antigens was investigated, revealing that the ArKP1.1 antiserum does not cross-react with the ArKP2.1 antigen peptide and native peptide, and the ArKP2.1 antiserum does not cross-react with the ArKP1.1 antigen peptide and native peptide (Additional file 3: Fig. S2).

Preliminary immunohistochemical tests revealed that the ArKP1.1 antiserum could be used for specific visualisation of ArKP1.1 expression in A. rubens at a dilution of 1:1000. Pre-absorption tests revealed that immunostaining was abolished by incubation of the ArKP1.1 antiserum with ArKP1.1 antigen peptide (Fig. 4) but not with ArKP1.2 or ArKP2.2 (Additional file 4: Fig. S3), demonstrating the specificity of the antiserum for ArKP1.1. Immunostaining was also observed with the ArKP2.1 antiserum when tested at a dilution of 1:1000, and this staining was not abolished by preincubation of the antiserum with ArKP1.1 or ArKP2.2 (Additional file 4: Fig. S3). However, affinity purification of antibodies from the ArKP2.1 antiserum was required to enable specific visualisation of ArKP2.1 expression in A. rubens, and, accordingly, pre-absorption of the affinity-purified antibodies with the ArKP2.1 antigen peptide abolished immunostaining (Fig. 4D). The results of our comparative immunohistochemical analysis of ArKP1.1 and ArKP2.1 expression in A. rubens are illustrated in Figs. 4, 5, and 6, where the distribution of ArKP1.1 and ArKP2.1 immunoreactivity is compared in adjacent sections of the arms or central disc region of starfish, as described below.Fig. 4. Immunohistochemical localisation of ArKP1.1 and ArKP2.1 in central nervous system, tube feet, and arm tip. A, B Horizontal sections of juvenile starfish showing immunostaining in radial nerve cords, circumoral nerve ring, and tube feet. C, D Immunostaining in transverse sections of radial nerve cord, which is abolished by pre-absorption with antigen peptides (insets). E, F Boxed regions in C and D, respectively, showing immunoreactivity in ectoneural cells (black arrowheads) and neuropile (asterisks). ArKP1.1-, but not ArKP2.1, immunoreactive cells (black arrows) are present in hyponeural region. Both ArKP1.1- and ArKP2.1-immunoreactive fibres are present in hyponeural region (grey arrows). G, H Transverse sections of circumoral nerve ring showing ArKP1.1 and ArKP2.1 immunoreactivity in ectoneural region, with variation in density of neuropile (black and white asterisks). Insets are higher magnifications of lower boxed regions, showing ArKP1.1- and ArKP2.1-immunoreactive cells (black arrowheads). I, J Higher magnifications of upper boxed regions in G and H, respectively, showing immunostaining in neuropile of ectoneural region (asterisks). ArKP1.1 immunoreactivity (black arrows), but not ArKP2.1 immunoreactivity, is present in cells located in hyponeural region. However, ArKP2.1-immunoreactive fibres (grey arrows) are present in hyponeural region. K, L ArKP1.1 and ArKP2.1 immunoreactivity in marginal nerve, showing immunostained cells (black arrowheads) and fibres (asterisks). M, N Longitudinal sections of a tube foot showing ArKP1.1 and ArKP2.1 immunoreactivity in subepithelial nerve plexus in the stem and extending into basal nerve ring. Upper insets are higher magnifications of boxed regions, showing immunostaining in subepithelial nerve plexus (black arrowheads); lower insets are transverse sections of a tube foot, with ArKP1.1 and ArKP2.1 immunoreactivity present in basal nerve ring and subepithelial nerve plexus. O, P Transverse sections of arm tip region showing ArKP1.1 and ArKP2.1 immunoreactivity in terminal tentacle, optic cushion, tube feet, and body wall epidermis. Q, R Higher magnifications of regions in O and P boxed with a dashed line, showing ArKP1.1-immunoreactive cells (arrowheads) and fibres (asterisk) in optic cushion, which is largely void of ArKP2.1 immunoreactivity. S, T Higher magnification of larger boxed regions in O and P showing immunostained cells (arrowheads) and fibres (asterisks) in terminal tentacle. U, V Higher magnification of smaller boxed regions in O and P showing immunostained cells (arrowheads) and fibres (asterisks) in terminal tentacle. Scale bars: 1250 μm (A, B), 200 μm (C, D, G, H, M, N, O, P), 100 μm (Q, R), 60 μm (E, F, K, L, S, T), 6 μm (I, J, U, V) Insets: 200 μm (C, D bottom inset in M and N), 100 μm (top inset in M, N), and 30 μm (G, H)Fig. 5. Immunohistochemical localisation of ArKP1.1 and ArKP2.1 in digestive system. A, B ArKP1.1 and ArKP2.1 immunoreactivity in basiepithelial nerve plexus of peristomial membrane (arrowheads); insets are higher magnifications of boxed regions in A and B showing ArKP1.1- and ArKP2.1-immunoreactive cells (black arrowheads) adjacent to basiepithelial nerve plexus. C, D ArKP1.1 and ArKP2.1 immunoreactivity in basiepithelial nerve plexus of oesophagus. E, F ArKP1.1 and ArKP2.1 immunoreactivity in cardiac stomach, pyloric stomach, and apical muscle (see Fig. 6). G, H Higher magnifications of upper boxed regions in E and F showing immunoreactivity in the basiepithelial nerve plexus (both ArKP1.1 and ArKP2.1) and in the visceral muscle layer nerve plexus (only ArKP2.1) of pyloric stomach. I, J. Higher magnifications of lower boxed regions in E and F showing immunoreactivity in basiepithelial nerve plexus (both ArKP1.1 and ArKP2.1) and in visceral muscle layer nerve plexus (only ArKP2.1) of cardiac stomach. Immunostained cells (arrowheads) are located in mucosal layer. K, L ArKP1.1 and ArKP2.1 immunoreactivity in basiepithelial nerve plexus (ArKP1.1 and ArKP2.1) and visceral muscle layer nerve plexus (ArKP2.1) of cardiac stomach. Immunostained cells (black arrowheads) are located in mucosal layer. M, N Higher magnifications of upper boxed regions in K and L, respectively, showing immunoreactivity in the basiepithelial nerve plexus (both ArKP1.1 and ArKP2.1) and in the visceral muscle layer nerve plexus (only ArKP2.1). Immunostained cells (black arrowheads) are located in mucosal layer. O, P Higher magnifications of lower boxed regions in K and L, respectively, showing ArKP1.1- and ArKP2.1-immunoreactive cells (black arrowheads) in mucosal layer and ArKP1.1- and ArKP2.1-immunoreactive fibres in basiepithelial nerve plexus. Q, R ArKP2.1 immunoreactivity in intrinsic retractor strand (boxed region), nodule, and extrinsic retractor strand of cardiac stomach; no ArKP1.1 immunoreactivity is observed in these structures. Both ArKP1.1 and ArKP2.1 immunoreactivity are in basiepithelial nerve plexus of cardiac stomach. S, T Higher magnifications of boxed regions in Q and R, respectively, showing the absence of ArKP1.1-immunoreactive fibres and the presence of ArKP2.1-immunoreactive fibres (arrows) in an intrinsic retractor strand and nodule. U, V Adjacent sections of a pyloric caecum showing immunoreactivity in basiepithelial nerve plexus (ArKP1.1 and ArKP2.1) and in visceral muscle layer nerve plexus (ArKP2.1). W, X Higher magnifications of boxed regions in U and V, respectively, showing ArKP1.1- and ArKP2.1 immunoreactive cells and fibres in the mucosa and basiepithelial nerve plexus, respectively, and ArKP2.1-immunoreactive fibres in visceral nerve plexus. Scale bars: 500 μm (E, F, Q, R), 200 μm (C, D, U, V), 100 μm (G, H, I, J, K, L), and 60 μm (M, N, O, P, S, T, W, X). Insets: 6 μm (A, B)Fig. 6. Immunohistochemical localisation of ArKP1.1 and ArKP2.1 in reproductive system and body wall. A, B ArKP2.1 immunoreactivity, but not ArKP1.1 immunoreactivity, is present in nerve plexuses adjacent to both the lumenal and coelomic epithelia of a gonoduct in male starfish. C, D ArKP1.1 and ArKP2.1 immunoreactivity (black arrowhead) in coelomic epithelium of testis. E, F ArKP1.1 immunoreactivity (black arrowhead), but not ArKP2.1 immunoreactivity, is present in a nerve plexus adjacent to lumenal epithelium of gonoduct in female starfish; inset shows higher magnification of ArKP1.1 immunoreactivity (black arrowhead) in a nerve plexus adjacent to lumenal epithelium of gonoduct in female starfish. G, H ArKP1.1 and ArKP2.1 immunoreactivity (black arrowheads) in coelomic epithelial layer of ovary. I, J ArKP1.1 and ArKP2.1 immunoreactivity in the subepithelial nerve plexus (arrows) and in cells (black arrowhead) located in epidermis. K, L ArKP1.1- and ArKP2.1-immunoreactive fibres (black arrowheads) in apical muscle. M, N ArKP1.1- and ArKP2.1-immunoreactive fibres in the subepithelial nerve plexus of body wall (black arrowheads); insets show higher magnifications of boxed regions in M and N. O, P ArKP1.1 and ArKP2.1 immunoreactivity in subepithelial nerve plexus (black arrowheads) of a pedicellaria. Insets show higher magnifications of boxed regions in O and P, respectively, showing ArKP2.1 immunoreactivity (arrowheads), but not ArKP1.1 immunoreactivity, in nerve fibres associated with adductor muscles. Q, R ArKP1.1- and ArKP2.1-immunoreactive fibres in subepithelial nerve plexus of body wall proximal to a papula; insets are higher magnifications of boxed regions in Q and R, respectively, showing ArKP1.1 and ArKP2.1 immunoreactivity (white asterisks) in subepithelial nerve plexus of body wall proximal to a papula. Scale bars: 500 μm (A, M, N), 200 μm (B, E, F, Q, R), 100 μm (K, L, O, P), and 60 μm (C, D, G, H, I, J). Insets: 30 μm (E, M, N, O, P, Q, R)

Nervous system, tube feet, and arm tip

ArKP1.1 and ArKP2.1 immunoreactivity was revealed in the nervous system of A. rubens, including the radial nerve cords, circumoral nerve ring, and marginal nerves (Fig. 4). In the radial nerve cords, extensive ArKP1.1 and ArKP2.1 immunoreactivity was revealed in the neuropile of the ectoneural region (Fig. 4C and D), and, accordingly, bipolar-shaped ArKP1.1- and ArKP2.1-immunoreactive cells are present in the subcuticular epithelium of the ectoneural region (Fig. 4C, D and E, F). In the hyponeural region of the radial nerve cords, ArKP1.1-immunoreactive cells were observed, but no ArKP2.2-immunoreactive cells were observed in the hyponeural region. However, ArKP2.2-immunoreactive fibres appear to be present in the hyponeural region of the radial nerve cords (Fig. 4E and F). The patterns of ArKP1.1 and ArKP2.1 expression in the circumoral nerve ring were consistent with those observed in the radial nerve cords. Thus, extensive ArKP1.1 and ArKP2.1 immunoreactivity is observed in the ectoneural neuropile of the circumoral nerve ring but with regional variation in the density of immunostaining. Accordingly, bipolar-shaped ArKP1.1- and ArKP2.1-immunoreactive cells were revealed in the subcuticular epithelium of the ectoneural region (Fig. 4G and H). In the hyponeural region of the circumoral nerve ring, ArKP1.1-immunoreactive cells were observed, but no ArKP2.1-immunoreactive cells were observed. However, as in the radial nerve cords, ArKP2.2-immunoreactive fibres appear to be present in the hyponeural region of the circumoral nerve ring (Fig. 4I and J). ArKP1.1 and ArKP2.1 immunoreactivity was also revealed in the marginal nerves, which are thickenings of the subepithelial nerve plexus that run along the length of the arms adjacent to the outer row of the tube feet (Fig. 4K and L).

In the tube feet, both ArKP1.1 and ArKP2.1 immunoreactivity was observed in the subepithelial nerve plexus along the length of the stem and extending into the basal nerve ring of the disk region (Fig. 4M and N). In the arm tips, both ArKP1.1 and ArKP2.1 immunoreactivity were observed in tube feet, the terminal tentacle, and the subepithelial nerve plexus of adjacent body wall. However, a striking difference in expression was observed in the optic cushion, with extensive ArKP1.1 immunoreactivity observed but little or no ArKP1.2 immunoreactivity detected (Fig. 4O, P, Q, R, S, T, U, V).

Digestive system

Both ArKP1.1 and ArKP2.1 immunoreactivity were detected in several regions of the digestive system of A. rubens, including the peristomial membrane, oesophagus, cardiac stomach, pyloric stomach, and pyloric caeca (Fig. 5A, B, C, D, E, F, G, H, I, J, K, L, M, N, O, P, Q, R, S, T, U, V, W, X). In the peristomial membrane and oesophagus, ArKP1.1 and ArKP2.1 immunoreactivity are present in the basiepithelial nerve plexus (Fig. 5A, B, C, D). In the pyloric stomach and cardiac stomach, there are both similarities and differences in the distribution of ArKP1.1 and ArKP2.1 (Fig. 5E, F, G, H, I, J, K, L, M, N, O, P). Thus, both ArKP1.1 and ArKP2.1 immunoreactivity is present in the basiepithelial nerve plexus of the pyloric stomach and cardiac stomach, with associated immunoreactive cells revealed in the mucosal layer of the cardiac stomach. However, only ArKP2.1 immunoreactivity was detected in the nerve plexus associated with the visceral muscle layer of the cardiac stomach and pyloric stomach (Fig. 5G, H, I, J, K, L, M, N). Accordingly, ArKP2.1 immunoreactivity, but not ArKP1.1 immunoreactivity, was revealed in the intrinsic retractor strands, nodules, and the extrinsic retractor strands, which mediate attachment of the cardiac stomach to the ambulacrum in each arm of the starfish body (Fig. 5Q, R, S, T). In pyloric caeca, both ArKP1.1 and ArKP2.1 immunoreactivity was observed in the basiepithelial nerve plexus (Fig. 5U, V, W, X).

Reproductive system and body wall-associated structures

In the reproductive system, sex-dependent patterns of expression were observed in the gonoducts. Thus, ArKP1.1 immunoreactivity was revealed in females (associated with the lumenal epithelium; Fig. 6A) but not in males (Fig. 6E). Conversely, ArKP2.1 immunoreactivity was revealed in males (associated with both the lumenal and coelomic epithelia; Fig. 6B) but not in females (Fig. 6F). In the gonads, both ArKP1.1 immunoreactivity and ArKP2.1 immunoreactivity were revealed in the coelomic epithelial layer of males (testis; Fig. 6C, D) and females (ovary; Fig. 6G, H).

In the body wall, both ArKP1.1 immunoreactivity and ArKP2.1 immunoreactivity were detected in the subepithelial nerve plexus (Fig. 6I, J, M, N, Q, R, O, P) and in associated cells in the epidermis (Fig. 6I). Both ArKP1.1 and ArKP2.1 immunoreactivities were also revealed in the apical muscle, a longitudinally oriented strand of muscle located along the midline of the coelomic lining in each arm (Fig. 6K, L). However, a striking difference in ArKP1.1 and ArKP2.1 expression was observed in the body wall-associated pedicellariae, with ArKP2.1 immunoreactivity, but not ArKP1.1 immunoreactivity, located in nerve fibres proximal to the adductor muscles (Fig. 6O, P**)**.

Opposing myoregulatory effects of ArKPP1- and ArKPP2-derived neuropeptides in A. rubens

Informed by the expression pattern of ArKPP1 and ArKPP2 transcripts and ArKPP1- and ArKPP2-derived neuropeptides, as described above, ArKP1.1, ArKP1.2, and ArKP2.2 were tested on in vitro preparations of gonad, cardiac stomach, tube foot, and apical muscle from A. rubens (Fig. 7, Additional file 5: Dataset S2). Due to insolubility of ArKP2.1, it was not possible to test this peptide on these preparations.Fig. 7. Opposing effects of ArKPP1- and ArKPP2-derived neuropeptides on in vitro organ preparations from A. rubens. A, B ArKP1.1 and ArKP1.2 cause relaxation of gonad preparations precontracted with acetylcholine, whereas ArKP2.2 causes contraction of gonad preparations (see also Additional file 5: Fig. S4). C Graph comparing relaxing effects of neuropeptides on gonad preparations (normalised as percentage reversal of contracting effect of 10^−6^ M acetylcholine). D Graph comparing contracting effects of ArKP2.2 and acetylcholine (both at 10^−6^ M) on gonad preparations (normalised to contracting effect of 10^−6^ M acetylcholine). E ArKP1.1 and ArKP1.2 cause relaxation of cardiac stomach preparations precontracted with KCl-supplemented seawater; the two vertical lines represent interruptions in the recordings when other peptides were tested. F ArKP2.2 causes contraction of a cardiac stomach preparation. G Graph comparing dose-dependent relaxing effects of ArKP1.1 and ArKP1.2 on cardiac stomach preparations, normalised as percentage reversal of KCl-induced contraction (mean ± SEM, n = 8). ArKP1.2 is more potent than ArKP1.1, and, accordingly, significant differences in effects of the peptides are observed at 10^−8^ M and 10^−9^ M. H Graph comparing relaxing effects of ArKP1.1, ArKP1.2, and S2 (all at 10^−6^ M) on cardiac stomach preparations, normalised as percentage reversal of KCl-induced contraction (mean ± SEM, n = 8). Effect of ArKP1.2, but not ArKP1.1, is significantly larger than effect of S2. I Graph showing dose-dependent contracting effect of ArKP2.2 on cardiac stomach preparations, normalised to effect of 10^−6^ M acetylcholine (mean ± SEM, n = 8). J Contracting effects of acetylcholine and ArKP2.2 (both at 10^−6^ M) on a tube foot preparation. K Graph showing dose-dependent contracting effect of ArKP2.2 on tube foot preparations, normalised to effect of 10^−6^ M acetylcholine (mean ± SEM, n = 6). Statistical analysis is as follows: ****p ≤ 0.0001, ***p ≤ 0.001, **p ≤ 0.01, *p ≤ 0.05, ns — not significant; see Additional file 5: Dataset S2 for details of statistical tests

When tested on gonad preparations that had been precontracted with 10^−6^ M acetylcholine, both ArKP1.1 (10^−6^ M) and ArKP1.2 (10^−6^ M) caused relaxation (Fig. 7A). Relaxing effects of ArKP1.1 and ArKP1.2 on gonad preparations were observed consistently but were quite small. Therefore, in Additional file 6: Fig. S4, expanded versions of the traces in Fig. 7A are shown alongside a control recording of the same preparation where acetylcholine was applied without subsequent addition of ArKP1.1 or ArKP1.2. This shows that after application of acetylcholine, the preparation maintains a contracted state until the preparation is washed, contrasting with experiments where the preparation slowly relaxes after application of ArKP1.1 or ArKP1.2. ArKP2.2 (10^−6^ M) caused contraction of gonad preparations (Fig. 7B).

A previous study revealed that the pedal peptide-type neuropeptide ArPPLN2h causes relaxation of gonad preparations from A. rubens [19]. Comparison of the effects ArKP1.1, ArKP1.2, and ArPPLN2h (all at 10^−6^ M) on gonad preparations revealed that the relaxing effects of ArKP1.1 and ArKP1.2 were significantly smaller than the effect of ArPPLN2h, whereas the effects of ArKP1.1 and ArKP1.2 were not significantly different (Fig. 7C). Conversely, acetylcholine causes contraction of gonad preparations from A. rubens [19], but the contracting effect of ArKP2.2 on gonad preparations was significantly smaller than the effect of acetylcholine (10^−6^ M) (Fig. 7D).

Both ArKP1.1 and ArKP1.2 caused relaxation of cardiac stomach preparations precontracted with 30-mM KCl (Fig. 7E), whereas ArKP2.2 caused contraction of cardiac stomach preparations (Fig. 7F). Comparison of the potency of ArKP1.1 and ArKP1.2 when tested over a range of concentrations from 10^−12^ to 10^−6^ M revealed that ArKP1.2 was more potent than ArKP1.1, and the EC_50_ value for ArKP1.2 was determined to be 5.18 × 10^−10^ M. However, it was not possible to determine an EC_50_ value for ArKP1.1 because its lower potency prevented plotting of a sigmoidal dose–response curve (Fig. 7G). Consistent with the lower potency of ArKP1.1, comparison of the effects of ArKP1.1 and ArKP1.2 at each concentration tested revealed significant differences at 10^−8^ M and 10^−9^ M (mean ± SEM, n = 8, Students’ t-test, p = 0.0004 and p = 0.0221, respectively; Fig. 7G). Previous studies have revealed that the SALMFamide-type neuropeptide S2 (SGPYSFNSGLTF-NH_2_) acts as a cardiac stomach relaxant in A. rubens [20, 21]. Therefore, here the effect of S2 was compared with the effects of ArKP1.1 and ArKP1.2, when tested at 10^−6^ M; this revealed that there was no significant difference in the effects of S2 and ArKP1.1, but the effects of ArKP1.2 were significantly larger than the effects of S2 (n = 8; Dunnett’s multiple comparison test, p = 0.0583 and p = 0.0366, respectively) (Fig. 7H). The ArKPP2-derived neuropeptide ArKP2.2 caused dose-dependent contraction of cardiac stomach preparations when tested at concentrations ranging from 10^−10^ to 10^−6^ M (EC_50_ = 7.0 × 10^−9^ M), normalising with the effects of 10^−6^ M acetylcholine (mean ± SEM, n = 8, Fig. 7I). Statistical analysis revealed that the magnitude of the contracting effects of 10^−6^ M ArKP2.2 and 10^−6^ M acetylcholine on cardiac stomach preparations was not significantly different (n = 8, Mann–Whitney test, p = 0.4042).

ArKP1.1, ArKP1.2, and ArKP2.2 were also tested on preparations of body wall-associated muscles/organs, the apical muscle, and tube feet. None of the three peptides exerted any observable effects on apical muscle preparations (Additional file 7: Fig. S5). Likewise, ArKP1.1 and ArKP1.2 had no observable effects on tube foot preparations (Additional file 7: Fig. S5). However, consistent with its contracting effect on cardiac stomach preparations, ArKP2.2 caused dose-dependent contraction of tube foot preparations with an EC_50_ of 4.44 × 10^−9^ M when normalised with the effects of 10^−6^ M acetylcholine (mean ± SEM, n = 6; Fig. 7J, K). Furthermore, statistical analysis revealed that the contracting effect of 10^−6^ M ArKP2.2 was significantly larger than the contracting effect of 10^−6^ M acetylcholine (n = 6, Mann–Whitney test, **p = 0.0022).

ArKPP1- and ArKPP2-derived neuropeptides cause cardiac stomach eversion in vivo

Cardiac stomach eversion during feeding in starfish requires the relaxation of the musculature of the stomach. Therefore, if neuropeptides cause relaxation of the cardiac stomach in vitro, it is of interest to investigate if they trigger stomach eversion in vivo. Accordingly, previous studies have reported that the SALMFamide neuropeptide S2 and the vasopressin-/oxytocin-type neuropeptide asterotocin trigger cardiac stomach relaxation in vitro and eversion in vivo when tested in A. rubens, with asterotocin being more effective than S2 [18, 20]. Because ArKP1.1 and ArKP1.2 were found to cause relaxation of cardiac stomach preparations in vitro, here they were also tested to determine if they trigger cardiac stomach eversion in vivo (Fig. 8, Additional file 5: Dataset S2). Asterotocin and S2 were also tested as positive controls to enable comparison of the effectiveness of different neuropeptides in inducing cardiac stomach eversion. Both asterotocin (10 µl of 10^−3^ M) and S2 (100 µl of 10^−3^ M) induced cardiac stomach eversion (Fig. 8A, B), and, consistent with previously reported findings [18], the magnitude of the effect asterotocin was larger than that of S2. Furthermore, whilst cardiac stomach eversion was observed in all animals injected with asterotocin, cardiac stomach eversion was only observed in 6 out of 10 animals injected with S2. When ArKP1.1 and ArKP1.2 were tested, neither peptide induced cardiac stomach eversion when injecting 10 µl of 10^−3^ M (n = 10) or 100 µl of 10^−3^ M (n = 12) solutions of each peptide individually. However, injection of 100 µl of an aqueous solution containing 0.5 × 10^−3^ M ArKP1.1 and 0.5 × 10^−3^ M ArKP1.2 triggered eversion of the cardiac stomach (n = 6) (Fig. 8A, B). Comparison of the magnitude of the effect of the mixture of ArKP1.1 and ArKP1.2 with the effect of asterotocin and S2 revealed that it was significantly smaller than the effect of asterotocin but not significantly different to the effect of S2 (Fig. 8C, D).Fig. 8. In vivo injection of ArKPP1- and ArKPP2-derived neuropeptides triggers partial eversion of the cardiac stomach in A. rubens. A Photographs of starfish injected with 10 µl water, 10 µl 10^−3^ M asterotocin, 100 µl 10^−3^ M S2, 100 µl aqueous solution containing 0.5 × 10^−3^ M ArKP1.1 and 0.5 × 10^−3^ M ArKP1.2, 10 µl 10% DMSO, and 10 µl 10^−3^ M ArKP2.2 in 10% DMSO at 0, 5, and 10 min after injection. Cardiac stomach eversion is observed after injection of asterotocin, S2, ArKP1.1 + ArKP1.2, and ArKP2.2, and the area of everted stomach is labelled with black dashed lines. Note that asterotocin also induces flexion of arms, as reported previously [18]. B Graph comparing the area of cardiac stomach everted, expressed as a percentage of area of central disk region (mean ± SEM), at 1-min intervals over a 10-min period following injection of asterotocin (n = 6), S2 (n = 6), ArKP1.1 (100 µl 10^−3^ M, n = 12), ArKP1.2 (100 µl 10^−3^ M, n = 12), ArKP1.1 + ArKP1.2 (n = 6), and water (n = 6). Note that injection of water, ArKP1.1 alone, or ArKP1.2 alone had no effect, and the symbols for these treatments are overlapped. C, D Graphs comparing the area of the everted stomach at 5 (C) and 10 (D) min post-injection of asterotocin, S2, or ArKP1.1 + ArKP1.2. E Graph comparing the area of cardiac stomach everted, expressed as a percentage of the area of the central disk region (mean ± SEM), at 1-min intervals over a 10-min period following injection of asterotocin (n = 6), S2 (n = 6), water (n = 6), ArKP2.2 (n = 11), and 10% DMSO (n = 10). Note that injection of water or 10% DMSO had no effect, and the symbols for these treatments are overlapped. F, G Graphs comparing the area of everted stomach at 5 (F) and 10 (G) min post-injection of asterotocin, S2, or ArKP2.2. Statistical analysis is as follows: ****p ≤ 0.0001, ***p ≤ 0.001, **p ≤ 0.01, *p ≤ 0.05, ns — not significant; see Additional file 5: Dataset S2 for details of statistical tests

After extraoral feeding in starfish is completed, the cardiac stomach is withdrawn back into the central disk region of the animal, and this requires contraction of the stomach musculature. Accordingly, previous studies have revealed that neuropeptides that cause cardiac stomach contraction in vitro can also trigger cardiac stomach retraction in vivo [22, 23]. Because ArKP2.2 caused concentration-dependent contraction of the cardiac stomach preparations in vitro (Fig. 7I), this neuropeptide was tested to investigate if it triggers cardiac stomach retraction in vivo. As reported previously, cardiac stomach eversion was induced by placing starfish in seawater containing 2% added MgCl_2_ [23]. Unexpectedly, injection of 10 µl of 10^−3^ M ArKP2.2 slightly enlarged the area of the everted stomach in these experiments (Additional file 8: Fig. S6). Therefore, this neuropeptide was then tested to determine if it induces cardiac stomach eversion, employing the same method described above for asterotocin, S2, ArKP1.1, and ArKP1.2. ArKP2.2 triggered partial eversion of the cardiac stomach (Fig. 8A xvi, xvii, xviii, E), but cardiac stomach eversion was not observed after injection of 10% DMSO, the vehicle used for preparing a solution of ArKP2.2 (Fig. 8E). The magnitude of the stomach-everting effect of ArKP2.2 (10 µl of 10^-3^ M) was significantly smaller than the effect of asterotocin (10 µl of 10^−3^ M), but it was not significantly different to the effect of S2 (100 µl of 10^−3^ M) (Fig. 8F, G).

ArKP1.2 affects righting behaviour in starfish

A previous study revealed that the vasopressin-/oxytocin-type neuropeptide asterotocin not only induces cardiac stomach eversion in A. rubens but also causes flexion of the arms. In so doing, asterotocin-injected starfish adopt a posture similar to the humped posture that starfish have when feeding on prey. Thus, asterotocin appears to trigger ‘fictive feeding’ in starfish [18]. Furthermore, asterotocin also affects righting behaviour in A. rubens, increasing the time taken to right when animals are upturned, and it is inferred that this effect of asterotocin on righting behaviour is a consequence of asterotocin-induced changes in body posture (arm flexion) [18]. Here, we investigated if ArKPP1- and ArKPP2-derived neuropeptides affect righting behaviour in A. rubens (Fig. 9, Additional file 5: Dataset S2). Animals were starved for 1 week to normalise their physiological condition and then were upturned so that the aboral body surface was in contact with the floor of a testing tank containing artificial seawater. First righting time was recorded without injection of a test substance, and then righting time was recorded after injection of water (negative control), asterotocin (10 μl of 10^−3^ M, positive control), S2 (100 μl of 10^−3^ M), or ArKPP1-/ArKPP2-derived neuropeptides. Consistent with previous findings, injection of water had no effect on righting time, and asterotocin caused a significant increase in the time taken to right. Furthermore, S2 also caused a significant increase in the time taken to right (Fig. 9A, B, C, D).Fig. 9. Analysis of the effects of ArKPP1- and ArKPP2-derived neuropeptides on righting behaviour in A. rubens. A Graph showing time taken (mean ± SEM) for starfish to right without injection (n = 46) or after injection of water (n = 10, 10 μl), asterotocin (n = 10, 10 μl 10^−3^ M), S2 (n = 10, 100 μl 10^−3^ M), ArKP1.1 (n = 8, 100 μl 10^−3^ M), ArKP1.2 (n = 8, 100 μl 10^−3^ M), or ArKP1.1 + ArKP1.2 (n = 10, 100-μl aqueous solution containing 0.5 × 10^−3^ M ArKP1.1 and 0.5 × 10^−3^ M ArKP1.2). Statistical analysis reveals significant differences in time taken to right between non-injection and injection with asterotocin or S2 or ArKP1.2 or ArKP1.1 + ArKP1.2. B Graph showing percentage difference in righting time (mean ± SEM) without injection and with injection of water (n = 10, 10 μl), asterotocin (n = 10, 10 μl 10^−3^ M) or S2 (n = 10, 100 μl 10^−3^ M), ArKP1.1 (n = 8, 100 μl 10^−3^ M) or ArKP1.2 (n = 8, 100 μl 10^−3^ M), or ArKP1.1 + ArKP1.2 (n = 10, 100 μl aqueous solution containing 0.5 × 10^−3^ M ArKP1.1 and 0.5 × 10^−3^ M ArKP1.2). In comparison with non-injection versus water injection, the mean percentage difference in righting time is statistically significant for non-injection versus injection with asterotocin or S2 or ArKP1.2 or ArKP1.1 + ArKP1.2. C Graph showing time taken (mean ± SEM) for starfish to right without injection (n = 50) or after injection of water (n = 10, 10 μl), asterotocin (n = 10; 10 μl 10^−3^ M), S2 (n = 10, 100 μl 10^−3^ M), ArKP2.2 (n = 10, 10 μl 10^−3^ M), or 10% DMSO (n = 10). Statistical analysis revealed significant differences in time taken to right between non-injection and injection with asterotocin or S2 or ArKP2.2. D Graph showing the percentage difference in righting time (mean ± SEM) without injection and with injection of water (n = 10, 10 μl), asterotocin (n = 10, 10 μl 10^−3^ M), S2 (n = 10, 100 μl 10^−3^ M), or ArKP2.2 (n = 10, 10 μl 10^−3^ M). In comparison with non-injection versus water injection, the mean percentage difference in righting time is statistically significant for non-injection versus injection with asterotocin or S2 or 10% DMSO against non-injected. Statistical analysis is as follows: ****p ≤ 0.0001, ***p ≤ 0.001, **p ≤ 0.01, *p ≤ 0.05, ns — not significant; see Additional file 5: Dataset S2 for details of statistical tests

The ArKPP1-derived neuropeptide ArKP1.2 caused a statistically significant increase in righting time when tested alone or when tested in combination with ArKP1.1, but ArKP1.1 did not have a statistically significant effect on righting time when tested alone (Fig. 9A, B). The ArKPP2-derived neuropeptide ArKP2.2 also appeared to cause a significant increase in righting time (Fig. 9C). However, when pairwise differences in righting time (non-injected versus injected) were analysed, there was not a statistically significant difference between treatment with 10% DMSO (vehicle) and treatment with ArKP2.2 dissolved in 10% DMSO (Fig. 9D). In conclusion, these findings indicate that injection of ArKP1.2 affects righting time in A. rubens when tested alone or in combination with ArKP1.1, but injection of ArKP1.1 or ArKP2.2 does not affect righting time in A. rubens.

Discussion

The roles of kisspeptin-type neuropeptides as regulators of reproductive and other physiological processes in mammals and other vertebrates have been investigated extensively [24, 25]. However, little is known about the functions of kisspeptin-type neuropeptides in invertebrates. Investigation of the physiological roles of kisspeptin-type neuropeptides in invertebrates is necessary if we are to obtain insights into the evolution of kisspeptin signalling. Discovery of kisspeptin-type neuropeptides and their cognate receptors in the starfish A. rubens (phylum Echinodermata) [12, 17] has enabled us to address this issue, as reported here.

Distribution and actions of kisspeptin-type neuropeptides derived from two precursor proteins in the starfish A. rubens

In mammals, a single precursor protein (KISS1) is subject to variable proteolysis to give rise to kisspeptins with an identical amidated C-terminal decapeptide sequence but with differences in length N-terminally [26, 27]. In contrast, in the starfish A. rubens, two kisspeptin-type precursors, ArKPP1 and ArKPP2, give rise to the neuropeptides ArKP1.1 and ArKP1.2 and ArKP2.1 and ArKP2.2, respectively (Fig. 1) [12, 17]. It is of interest, therefore, to analyse and compare the patterns of expression of these precursors and the neuropeptides derived from them. Here, this was accomplished using mRNA in situ hybridisation and immunohistochemistry, which employed specific antibodies to ArKP1.1 and ArKP2.1. Informed by the patterns of expression revealed, we then investigated the in vitro and in vivo actions of ArKPP1- and ArKPP2-derived neuropeptides in A. rubens.

Extensive expression of ArKPP1 and ArKPP2 transcripts was revealed in the central nervous system of A. rubens, which comprises the circumoral nerve ring and five radial nerve cords. Accordingly, immunohistochemistry revealed ArKP1.1- and ArKP2.1-immunoreactive cells and fibres in the circumoral nerve ring and radial nerve cords. These findings indicate that kisspeptin-type neuropeptides act as mediators of neuronal signalling in the central nervous system of starfish. More specifically, analysis of adjacent sections revealed that ArKPP1-derived and ArKPP2-derived neuropeptides are expressed in different populations of neurons without evidence of co-expression. The clearest evidence of this was seen in the hyponeural region of the radial nerve cords and circumoral nerve ring, which contains the cell bodies of somatic motoneurons, where ArKPP1-/ArKP1.1-expressing cells were detected, but ArKPP2/ArKP2.1 expression was not detected in hyponeural cells. Relevant to this finding, here we investigated the effects of ArKPP1- and ArKPP2-derived neuropeptides on righting in starfish, a behaviour that requires coordinated control of the body wall musculature by the central nervous system [28]. Interestingly, we found that in animals injected with ArKP1.2 or with both ArKP1.1 and ArKP1.2, the time taken to right was significantly longer than in control animals injected with water. However, in animals injected with ArKP2.2, the time taken to right was not significantly different to that in control animals that were injected with the vehicle, 10% DMSO. This difference in the in vivo behavioural effects of kisspeptin-type neuropeptides may be reflective of their expression in the central nervous system of A. rubens, with ArKPP1-derived neuropeptides expressed by hyponeural motoneurons and ArKPP2-derived neuropeptides not expressed by hyponeural motoneurons.

Closely associated with the central nervous system of starfish are locomotory organs known as tube feet, which are located in rows on both sides of the radial nerve cords. Each tube foot is under motor control of neurons located in segmental ganglia in the hyponeural region of the radial nerve cords. Furthermore, the stepping activity of tube feet during locomotion involves extension and retraction of the tube foot podium, with associated respective relaxation and contraction of longitudinally orientated muscle [29, 30]. Immunohistochemistry revealed expression of ArKP1.1 and ArKP2.1 in the subepithelial nerve plexus of tube foot podia and in the basal nerve ring of the tube foot adhesive disk. Therefore, we investigated the effects of ArKP1.1, ArKP1.2, and ArKP2.2 on the contractility of isolated tube foot preparations. As reported previously, it was not possible to test ArKP2.1 on tube feet or other starfish neuromuscular preparations because of its insolubility. ArKP2.2 caused concentration-dependent contraction of tube foot preparations and with greater efficacy than acetylcholine, which acts as a general myoexcitatory neurotransmitter in starfish. In contrast, ArKP1.1 and ArKP1.2 had no observable effects on tube foot contractility. The myoexcitatory effect of ArKP2.2 on tube feet in vitro indicates that this neuropeptide may mediate neural control of tube feet during locomotion and other behaviours (e.g. feeding, see below) where contraction of the tube foot podia occurs. In addition to acetylcholine, ArKP2.2 joins a list of several other neuropeptides that have been found to cause contraction of tube feet in A. rubens, including NGFFYamide, the gonadotropin-releasing hormone-type neuropeptide ArGnRH, the corazonin-type neuropeptide ArCRZ, the sulfakinin-type neuropeptides ArSK/CCK1&2, the somatostatin/allatostatin-C-type neuropeptide ArSS1, and the bombesin-type neuropeptide ArBN [22, 31–34]. Thus, neural control of tube foot contractility appears to be mediated by a complex combination of neuropeptides that includes ArKP2.2.

Expression of ArKPP1, ArKPP2, and/or neuropeptides derived from these precursors was also revealed in visceral organs of A. rubens, including the gonads. Detection of ArKP1.1 and ArKP2.1 immunoreactivity in gonad tissue was of particular interest because of the role of kisspeptins as regulators of reproductive maturation and function in vertebrates [35–37]. Consistent with the myoexcitatory effect of ArKP2.2 on tube feet, in vitro experiments revealed that ArKP2.2 causes contraction of gonad preparations from A. rubens. Furthermore, ArKP1.1 and ArKP1.2 caused relaxation of gonad preparations, partially reversing the contracting effect of acetylcholine. Thus, ArKPP1-derived and ArKPP2-derived neuropeptides have opposing myoregulatory effects on gonad preparations from A. rubens. Other neuropeptides that affect the contractility of the gonads in A. rubens have also been identified, including ArSS1, which causes gonadal contraction, and the pedal peptide/orcokinin-type neuropeptide ArPPLN2h, which causes gonadal relaxation [19]. Thus, it appears that a variety of neuropeptides, including the kisspeptin-type neuropeptides that are the focus of this study, may mediate neural regulation of gonad contractility in A. rubens. Furthermore, we speculate that these actions may be physiologically relevant during spawning in starfish when gametes are expelled from the gonads via the gonoducts to the external environment.

Expression of ArKPP1, ArKPP2, and neuropeptides derived from these precursors was revealed in several regions of the digestive system in A. rubens. To investigate the functional relevance of these findings, we tested ArKP1.1, ArKP1.2, and ArKP2.2 on in vitro preparations of the cardiac stomach. Consistent with the effects of these neuropeptides on gonad preparations, ArKP1.1 and ArKP1.2 caused concentration-dependent relaxation of cardiac stomach preparations, and ArKP2.2 caused concentration-dependent contraction of cardiac stomach preparations. Comparison with a known cardiac stomach relaxant, the SALMFamide-type neuropeptide S2 [20, 21], revealed that ArKP1.1 and ArKP1.2 were as effective or more effective than S2, respectively. Likewise, the contracting effect of ArKP2.2 on cardiac stomach preparations was similar in magnitude to the effect of acetylcholine when tested at the same concentration (10^−6^ M). Thus, ArKPP1-derived and ArKPP2-derived neuropeptides have opposing myoexcitatory and myoinhibitory effects on cardiac stomach preparations from A. rubens. As such, these kisspeptin-type neuropeptides join a variety of neuropeptides that have been found to cause relaxation or contraction of the cardiac stomach in A. rubens. Neuropeptides that cause cardiac stomach relaxation include the SALMFamide-type neuropeptides S1 and S2, the pedal peptide-/orcokinin-type neuropeptides ArPPLN1b and ArPPLN2h, the vasopressin-/oxytocin-type neuropeptide asterotocin, and the somatostatin-/allatostatin-C-type neuropeptide ArSS2 [18, 20, 38, 39], whereas neuropeptides that cause cardiac stomach contraction include NGFFYamide, ArGnRH, ArCRZ, ArSK/CCK1&2, ArSS1, and ArBN [22, 31–34, 40].

Discovery of neuropeptides that cause relaxation or contraction of in vitro preparations of the cardiac stomach is of interest with respect to the neural mechanisms that control feeding behaviour in starfish. Feeding in A. rubens involves eversion of the cardiac stomach out of the mouth and over the digestible soft tissue of prey such as the mussel Mytilus edulis. Partially digested tissue is carried by ciliary action back into the pyloric stomach and pyloric caeca, where further digestion and absorption occurs. When all of the soft tissue of the mussel has been removed, the cardiac stomach is withdrawn back into the starfish body [41]. Cardiac stomach eversion and retraction requires relaxation and contraction, respectively, of muscular tissue in the stomach wall. Accordingly, we have discovered that some neuropeptides that cause cardiac stomach relaxation in vitro also trigger cardiac stomach eversion in vivo, including the SALMFamide-type neuropeptide S2, asterotocin and ArSS2 [18, 20, 39], whilst conversely some neuropeptides that cause cardiac stomach contraction in vitro also trigger cardiac stomach retraction, including NGFFYamide, ArSK/CCK1&2, and ArBN [22, 31, 33]. Therefore, informed by the in vitro effects of ArKP1.1 and ArKP1.2 as cardiac stomach relaxants and the in vitro effects of ArKP2.2 as a cardiac stomach contractant, we investigated the in vivo effects of these neuropeptides in A. rubens. Consistent with their in vitro effects as relaxants, an equimolar mixture of ArKP1.1 and ArKP1.2 triggered cardiac stomach eversion. Furthermore, the magnitude of ArKP1.1- and ArKP1.2-induced cardiac stomach eversion was comparable to that observed with the SALMFamide-type neuropeptide S2. Surprisingly, ArKP2.2 also induced cardiac stomach eversion in A. rubens, which was unexpected because ArKP2.2 causes cardiac stomach contraction in vitro. Indeed, ArKP2.2 is the first neuropeptide that we have found to have what appear to be incongruous in vitro and in vivo effects. So, how can the in vivo effect of ArKP2.2 be explained? One possible explanation is that it triggers cardiac stomach eversion by causing mouth opening mediated by ArKP2.2-induced contraction of radial orientated muscle fibres in the peristomial membrane that surrounds the mouth, resulting in partial eversion of the cardiac stomach. The detection of ArKPP2 transcripts and ArKP2.1 immunoreactivity in the peristomial membrane (Figs. 3J, K; 5B) provides evidence in support of this hypothesis. Collectively, our data indicate that both ArKPP1-derived and ArKPP2-derived neuropeptides are involved in regulation of cardiac stomach contractility and may contribute to neural mechanisms controlling feeding behaviour in starfish.

Comparative physiology of kisspeptin-type neuropeptide signalling

Functional characterisation of ArKPP1- and ArKPP2-derived neuropeptides in the starfish A. rubens, as discussed above, has provided new insights into the comparative physiology of kisspeptins. Furthermore, it has provided a basis for comparative functional studies on other echinoderms. Currently, nothing is known about the action of kisspeptin-type neuropeptides in other starfish species. However, analysis of transcriptomic and/or mass spectrometric data from the crown-of-thorn starfish Acanthaster cf solaris has revealed expression of the KP1-type precursor/neuropeptides in the radial nerve cords, tube feet, gonads, digestive system, and body wall and associated appendages in this species [42, 43], consistent with the expression pattern of ArKPP1/ArKP1.1 in A. rubens reported here. Furthermore, expression of the KP2-type precursor/neuropeptides (referred to as tachykinin type based on the original annotation of ArKPP2 in A. rubens [17]) was detected in the radial nerve cords, tube feet, digestive system, and body wall and associated appendages in A. cf solaris [42, 43] which is also consistent with the expression pattern of ArKPP2/ArKP2.1 in A. rubens reported here.

Turning to other echinoderm classes, we have identified KP1-type and KP2-type precursors in ophiuroids (brittle stars), echinoids (sea urchins and sand dollars), holothurians (sea cucumbers), and crinoids (feather stars and sea lilies) [12, 44–46]. Therefore, opportunities exist for analysis of the expression and action of kisspeptin-type neuropeptides in these other echinoderm taxa. Expression of a KPP1-type precursor has been detected in the digestive system of larvae of the sea urchin Strongylocentrotus purpuratus [47], consistent with the extensive expression of ArKPP1 in the digestive system of adult starfish, as reported here. However, nothing is known about the physiological actions of neuropeptides derived from the KPP1-type precursor in sea urchins. Furthermore, there are currently no reports on the functional analysis of KP2-type neuropeptides in S. purpuratus, which include SpKP2.1 and SpKP2.2 derived from the precursor SpKPP2, as well as SpKP3 derived from a third kisspeptin-type precursor (SpKPP3) in this species [12].

Progress has been made in functional analysis of kisspeptin-type signalling in the sea cucumber Apostichopus japonicus. Both KPP1-type and KPP2-type precursors have been identified in this species [12, 45], but only the KPP1-type precursor (AjKPP1) and neuropeptides derived from this precursor have been investigated experimentally. Analysis of the expression of AjKPP1 (referred to as AjpreKiss) using quantitative PCR revealed seasonal changes in expression in A. japonicus, with the highest levels of expression detected in the anterior region of animals (probably including the circumoral nerve ring) between December and March and in the testis during April, which corresponds to the period when the gonads are approaching maturity prior to spawning [48]. Thus, this finding is consistent with the reproductive functions of kisspeptins in chordates and our detection of the gonadal expression and action of KP1-type neuropeptides in A. rubens. In a separate study, the in vivo pharmacological effects of KP1-type neuropeptides (AjK1 or AjKP1.1, AjK2 or AjKP1.2) in A. japonicus were investigated, and AjK1 caused a significant reduction in locomotor activity and changes in the metabolomic profile of longitudinal body wall muscles [49]. These findings may be relevant to the myoinhibitory effects of ArKP1.1 and ArKP1.2 in A. rubens, as reported in this study, and therefore, it would be interesting to determine if KP1-type neuropeptides also cause muscle relaxation in sea cucumbers. Nothing is known, however, about the actions of KP2-type neuropeptides in A. japonicus or other sea cucumbers, and therefore, addressing this issue will be an important objective for future studies.

Hemichordates are a sister phylum to the echinoderms in the superphylum Ambulacraria, and therefore, investigation of kisspeptin signalling in these animals will also be of particular interest from a comparative perspective. However, whilst genes encoding kisspeptin-type receptors have been identified in the hemichordate Saccoglossus kowalevskii [12, 14], the neuropeptides that act as ligands for these receptors have yet to be discovered. Aside from echinoderms, the only other invertebrates in which kisspeptin-type neuropeptides and receptors have been characterised are the cephalochordates Branchiostoma floridae and B. japonicum, with evidence of roles in regulation of reproduction reported [14, 15]. These findings are consistent with the actions of kisspeptins as regulators of reproduction in vertebrates and the gonadal expression and actions of KP1-type and KP2-type neuropeptides in starfish, as reported here. Thus, collectively, these findings indicate that kisspeptin-type neuropeptides are evolutionarily ancient regulators of reproductive processes in deuterostomes. Whether or not this role extends to the protostomes remains to be determined because, as discussed below, kisspeptin-type neuropeptides have yet to identified in protostomes.

Turning to non-reproductive functions of kisspeptins, the expression and action of kisspeptin-type neuropeptides in the digestive system of starfish are of interest from a comparative perspective because it has been reported that kisspeptins inhibit food intake in mammals [50, 51]. Thus, our discovery that ArKPP1-derived and ArKPP2-derived neuropeptides regulate cardiac stomach contractility and eversion in starfish indicates that kisspeptins are evolutionarily ancient regulators of feeding processes. However, whilst kisspeptins inhibit food intake in mammals, the effects of ArKPP1-/ArKPP2-derived neuropeptides on the cardiac stomach of starfish are consistent with a role in promotion of feeding. Interestingly, we reported similar lineage-specific differences in neuropeptide function previously for vasopressin-/oxytocin-type peptides, where oxytocin inhibits feeding in mammals, and the vasopressin-/oxytocin-type neuropeptide asterotocin triggers fictive feeding in starfish [18]. Thus, whilst there is evolutionary conservation in the types of physiological processes regulated by neuropeptides (e.g. feeding, reproduction), there may be clade-specific differences in the directionality of their effects (inhibitory/stimulatory).

Evolution of kisspeptin signalling: evidence of clade-specific expansion or loss of inhibitory kisspeptin signalling

The functional characterisation of neuropeptides derived from ArKPP1 and ArKPP2 that have opposing myoinhibitory or myoexcitatory effects, respectively, in the starfish A. rubens raises questions regarding their relatedness to kisspeptin-type neuropeptides in chordates. Alignment of echinoderm KP1-type and KP2-type neuropeptides with chordate kisspeptins does not reveal significant differences in the level of sequence similarity (Fig. 1). An alternative approach to address this issue is to examine the characteristics of the receptors that mediate the effects of KP1-type and KP2-type neuropeptides. In A. rubens, and in other echinoderms, there is an expanded family of kisspeptin-type receptors, which contrasts with the occurrence of a single kisspeptin receptor (GPR54) in mammals. Thus, in A. rubens, there are 11 kisspeptin-type receptors (ArKPR1-11) and neuropeptides that act as ligands for ArKPR1, 3, 6, 7, 8, and 9 have been identified [12]. Phylogenetic analysis has revealed relationships of these receptors with kisspeptin-type receptors in other taxa, including chordates and protostomes, as illustrated here in a simplified summary diagram (Fig. 10). The receptors are positioned in one of three clades: Clade 1 includes ArKPR2-4 and chordate kisspeptin receptors, clade 2 includes ArKPR1 and protostome kisspeptin-type receptors, and clade 3 includes ArKPR6-9.Fig. 10. Actions of kisspeptin-type neuropeptides in starfish provide insights into the evolution of kisspeptin signalling. The dendrogram on the left shows relationships between kisspeptin-type receptors in the starfish A. rubens and kisspeptin-type receptors in chordates (e.g. human) and protostomes (e.g. molluscs) based on a phylogenetic analysis of sequence data reported previously [12]. Neuropeptides that have been identified as ligands for kisspeptin-type receptors in chordates and in A. rubens [12] are listed, and the actions (excitatory or inhibitory) of these neuropeptides are stated. Clade 1 comprises chordate kisspeptin-type receptors and A. rubens receptors ArKPR2-4. These receptors are activated by kisspeptin-type peptides (ArKP2.2 and chordate kisspeptins) that have excitatory effects and have only been identified in deuterostomes, indicating this type of signalling evolved uniquely in deuterostomes or evolved in Urbilateria but with subsequent loss in protostomes. Clade 2 comprises protostome kisspeptin-type receptors and the A. rubens receptor ArKPR1, which is activated by a neuropeptide (ArKP1.2) that has inhibitory effects in A. rubens. These findings suggest that clade 2-type kisspeptin receptors originated in Urbilateria but were lost in chordates. However, neuropeptides that act as ligands for protostome kisspeptin-type receptors have yet to be identified, and so it remains to be determined if they have inhibitory effects in protostomes. Clade 3 comprises a family of kisspeptin-type receptors only found in ambulacrarians, which are represented here by ArKPR6-9. It is noteworthy that the clade 3 kisspeptin-type receptors are activated by different types of neuropeptides that have inhibitory effects in A. rubens: ArKP1.1 and ArKP1.2, as reported in this paper, and both L-type and F-type SALMFamides, as reported previously [52]

It is noteworthy that the clade 1 receptor ArKPR3 is selectively activated by ArKP2.2, and we hypothesise that ArKP2.1 may act as a ligand for ArKPR2, which is a paralog of ArKPR3, but we have been unable to test this hypothesis because of the insolubility of ArKP2.1 [12]. Informed by these findings, we conclude that ArKP2.1-ArKPR2 and ArKP2.2-ArKPR3 are closely related to the kisspeptin signalling system in chordates. From a functional perspective, this is interesting because of the myoexcitatory effects of ArKP2.2 that we observe in A. rubens. Accordingly, kisspeptins typically exert excitatory effects in mammals, as exemplified by the canonical action of kisspeptins in stimulating hypothalamic pulsatile release of GnRH [24, 53, 54]. We conclude, therefore, that chordate kisspeptins, echinoderm KP2-type neuropeptides, and their cognate receptors may have originated in a common ancestor of the deuterostomes and then retained an evolutionarily ancient role as excitatory-type kisspeptin signalling systems.

Turning now to clade 2, which includes ArKPR1 and kisspeptin-type receptors in lophotrochozoan protostomes (annelids, molluscs), it is noteworthy that the myoinhibitory neuropeptide ArKP1.2 acts as ligand for ArKPR1 in A. rubens [12]. However, ArKP1.2 also acts as a ligand for the clade 3 receptor ArKPR9, and it is not known whether the myoinhibitory action of this peptide is mediated by ArKPR1 and/or ArKPR9 in vivo. Furthermore, neuropeptides that act as ligands for kisspeptin-type receptors in lophotrochozoans have yet to be discovered. Therefore, it remains to be determined if there has been evolutionary conservation in the physiological effects of kisspeptin-type neuropeptides mediated by the clade 2-type kisspeptin receptors. What can be inferred, however, is that the occurrence of clade 2-type kisspeptin receptors in both deuterostome (echinoderms and hemichordates) and protostome (annelids and molluscs) phyla indicates that the evolutionary origin of this receptor type can be traced back to the common ancestor of the Bilateria. Furthermore, phylogenetic analysis of genome sequence data indicates that this receptor type has been lost in chordates (Fig. 10) and ecdysozoan protostomes [12]. Therefore, discovery of neuropeptides that act as ligands for the clade 2-type kisspeptin receptors in lophotrochozoans is critical if we are to gain a deeper understanding of the evolutionary history and physiological roles this type of kisspeptin signalling.

Lastly, turning to clade 3, this clade only comprises kisspeptin-type receptors that occur in ambulacrarians (echinoderms and hemichordates). These include ArKPR8, which is activated by ArKP1.1, and ArKPR9, which is activated by ArKP1.2. We infer that the myoinhibitory effects of ArKP1.1 are likely to be mediated by ArKPR8 because this is the only kisspeptin-type receptor identified in A. rubens for which ArKP1.1 is a potent ligand. Furthermore, because ArKPR9 is closely related to ArKPR8, the myoinhibitory effects of ArKP1.2 may be mediated by this receptor. However, as discussed above, it remains to be determined if the myoinhibitory effects of ArKP1.2 in A. rubens are mediated by ArKPR1 and/or ArKPR9. Also positioned in clade 3 are kisspeptin-type receptors that are activated by peptides belonging to a family of neuropeptides in echinoderms known as SALMFamides [52]. Thus, ArKPR7 is activated by the SALMFamide neuropeptide S1 and a cocktail of S1 and six other L-type SALMFamides derived from the S1 precursor, whereas ArKPR6 is activated by the SALMFamide neuropeptide S2 and a cocktail of S2 and seven F-type SALMFamides derived from the S2 precursor [12]. The discovery that SALMFamides act as ligands for kisspeptin-type receptors was an important finding because it revealed a hitherto unknown relationship between kisspeptins and SALMFamides, which we were identified in starfish 10 years before the discovery of kisspeptins [52, 55]. In retrospect, a relationship between SALMFamides and kisspeptins is apparent in the sequence similarity that these neuropeptides share [12]. Furthermore, it is of interest from an evolutionary perspective to assess what is known about the physiological roles of SALMFamides in starfish and other echinoderms. Importantly, both S1 and S2 act as myorelaxants in A. rubens, causing relaxation of in vitro preparations of the cardiac stomach, tube feet, and apical muscle, with S2 being more potent/effective than S1 [20, 56]. Moreover, this myoinhibitory action of SALMFamides appears to be an evolutionarily conserved role in echinoderms because in vitro pharmacological experiments have revealed that SALMFamide-type neuropeptides also act myorelaxants in sea cucumbers and sea urchins [57, 58]. Relevant to this, it is noteworthy that all of the A. rubens receptors in clade 3 are activated by neuropeptides that act as myorelaxants in this species — ArKP1.1, ArKP1.2, S1, and S2 (Fig. 10). This indicates that the clade 3-type kisspeptin receptors are a family of receptors that have uniquely evolved in the Ambulacraria, and based on our findings from A. rubens, these receptors may have specifically evolved as mediators of the inhibitory effects of KP1/SALMFamide-type neuropeptides. Further insights into this issue could be obtained by investigating if KP1-type neuropeptides act as myorelaxants in other echinoderms and by discovering and functionally characterising kisspeptin-/SALMFamide-type neuropeptides in hemichordates.

Conclusions

The key findings of this study are that kisspeptin-type neuropeptides derived from different precursor proteins (ArKPP1, ArKPP2) have opposing myoinhibitory (KP1-type) and myoexcitatory (KP2-type) effects in the starfish A. rubens, and these effects are associated with neural regulation of the reproductive and non-reproductive (e.g. digestive) systems. The excitatory effects of a KP2-type neuropeptide in A. rubens are accordant with the stimulatory effects of kisspeptins in vertebrates, and, based on the phylogenetic distribution of receptors that mediate these effects, the evolutionary origin of excitatory kisspeptin signalling can be traced back to the deuterostome common ancestor of echinoderms and chordates. Inhibitory kisspeptin-type signalling systems may have uniquely evolved in the ambulacrarian clade (echinoderms and hemichordates) of the animal kingdom via gene duplication giving rise to an expanded family of precursors of kisspeptin-type peptides (KP1-type and SALMFamides) and their cognate receptors (e.g. ArKP6-9 in A. rubens). Alternatively, inhibitory kisspeptin signalling may have originated in a common ancestor of the Bilateria and then was lost in chordates. Characterisation of kisspeptin signalling in lophotrochozoan invertebrates is now needed to provide further insights in this fascinating area of investigation into the evolution of neurohormone function.

Methods

Animals

Starfish (common European starfish, Asterias rubens) were obtained from a fisherman based at Whitstable (Kent, UK) or were collected at low tide near Margate (Kent, UK). The starfish were maintained in an aquarium containing circulating seawater at a temperature of ~ 12 °C and salinity of 32‰ under a 12-h light–dark cycle, located in the School of Biological and Behavioural Sciences at Queen Mary University of London. Animals were fed on mussels (Mytilus edulis) that were collected at low tide near Margate (Kent, UK). Juvenile specimens of A. rubens used for immunohistochemistry were collected and fixed at the University of Gothenburg Sven Lovén Centre for Marine Infrastructure (Kristineberg, Sweden).

Localisation of ArKPP1 and ArKPP2 transcripts in A. rubens using mRNA in situ hybridisation

The sequences of the genes/transcripts encoding the A. rubens kisspeptin-type precursors ArKPP1 and ArKPP2 have been reported previously based on analysis of transcriptome and genome sequence data [12, 17]. To facilitate analysis of the expression of these precursors in A. rubens using mRNA in situ hybridisation, cDNAs encoding ArKPP1 and ArKPP2 were synthesised or cloned. A cDNA encoding ArKPP1 was custom synthesised and incorporated (via blunt-end cloning with EcoRV) into pBluescript II (SK +) (GenScript). For ArKPP2, total RNA was isolated from A. rubens radial nerve cords using the SV Total RNA Isolation System (Promega, Southampton, UK), and then the QuantiTect® Reverse Transcription Kit (QIAGEN, Manchester, UK) was used to synthesise cDNA. A cDNA comprising the ArKPP2 protein-coding region was amplified by PCR using Phusion high-fidelity PCR master mix (NEB, Hitchin, Hertfordshire, UK) with specific oligonucleotide primers (5’-ATGCTTCTTGCTATGGCG-3′ and 5′-ATTCTATTCGTATCTCTTGG −3’), which were designed using Primer3 software (http://bioinfo.ut.ee/primer3-0.4.0/) with reference to the ArKPP2 precursor transcript sequence (GenBank accession number KT601707 [17]). The amplified cDNAs were gel extracted and later ligated into the pCR-Blunt II TOPO plasmid and transformed into top 10 chemically competent E. coli cells using the Zero Blunt TOPO PCR kit (Thermo Fisher Scientific, Waltham, MA, USA). Successful cloning of cDNAs was confirmed by sequencing (Eurofins Genomics GmbH, Ebersberg, Germany).

To enable localisation of ArKPP1 and ArKPP2 transcripts in A. rubens using mRNA in situ hybridisation, digoxigenin-labelled RNA probes were synthesied. A pBluescript II (SK +) plasmid containing the ArKPP1 cDNA sequence was linearised using the restriction endonucleases BamHI (for sense probe) or KpnI (for antisense probe) (NEB). Linearised plasmids were then purified using phenol–chloroform/chloroform-isomyl alcohol (Sigma-Aldrich) extraction. Plasmids containing the cloned and sequenced ArKPP2 precursor cDNA were used as templates to synthesise RNA probes. First, the insert was amplified by performing a routine PCR with M13 universal primers and Phusion High-Fidelity PCR Master mix (NEB, Cat. No. M0531S). The resultant amplicon was gel extracted and purified using the QIAquick gel extraction kit (QIAGEN, Cat. No. 28704). Sense and antisense RNA probes were synthesised using linearised plasmid (ArKPP1) or 1000 ng of the purified PCR products (ArKPP2), 0.5 mM of digoxigenin RNA labelling mix (Roche, Cat. No. 11277073910), transcription buffer (NEB, Cat. No. M0251S), 5-mM dithiothreitol (Promega, Cat. No. P1171), 1 U/μl placental ribonuclease inhibitor (NEB, Cat. No. M0307S), and 5 U/μl of SP6 RNA polymerase (NEB, Cat. No. M0251S) for the antisense probe or 5 U/μl of T7 RNA polymerase for the sense probe (NEB, Cat. No. M0207S). After subsequent ethanol precipitation and centrifugation, the final concentration of the probes was determined using a NanoDrop One UV–visible spectrophotometer (Thermo Scientific, Cat. No. 13–400–518). Finally, RNase inhibitor (NEB, Cat. No. M0307S) was added to the mixture at a concentration of 1 unit/µl to prevent degradation, and the probes were stored at − 80 °C until used.

Small specimens of A. rubens (diameter ⁓4 cm) were dissected to separate arms from the central disk region and then immersed in 4% paraformaldehyde diluted in phosphate buffer saline (PBS) for 16 h at 4 °C. The body parts were decalcified in Morse’s solution (10% sodium citrate, 22.5% formic acid) for 2–8 h and then were washed in PBS, dehydrated through a graded ethanol series into water-free xylene (Honeywell, Fisher Scientific UK Ltd, Loughborough, UK) before proceeding to embedding in paraffin wax. The embedded arms and central disks were sectioned at a thickness of 14 μm using a microtome (RM 2145, Leica Microsystems [UK], Milton Keynes, UK). The sections were placed on poly-L-lysine-coated slides (VWR Chemicals, Lutterworth, Leicestershire, UK, Cat. No. 100500–998) covered with distilled water on a hot plate at 55 °C to stretch the sections and then were left to dry.

Slides were deparaffinised in xylene and rehydrated through a graded ethanol series and then washed with PBS-Tween. After rehydration, the tissue sections were permeabilised at 37 °C with proteinase K (QIAGEN, Cat. No. 19131) diluted in PBS buffer to a final concentration of 10 μg/mL. The permeabilised tissue sections were post-fixed using 4% PFA/PBS following a PBS wash before submerging in acetylation solution (1.325% triethanolamine, 0.25% acetic anhydride, and 0.175% acetic acid) for 10 min. After washing with PBS/Tween 0.1%, prehybridisation was done using 500-µl hybridisation buffer (50% formamide, 5 × SSC, 500 μg/ml yeast RNA, 50 μg/ml heparin, 0.1% Tween 20 in dH_2_O) in a humidified chamber for 2 h. Then, DIG-labelled ArKPP1 or ArKPP2 sense or antisense probes (750 ng/ml) made up in hybridisation buffer were added to slides for overnight incubation in a humidified chamber at 65 °C. Sections were washed twice with 0.25 × SSC buffer (0.03 M NaCl, 0.003 M sodium citrate, pH 7.0) before equilibration in buffer B1 (0.1 M Tris pH 7.5, 0.15 M NaCl) and subsequent incubation in a humidified chamber with B1/5% goat serum for 2 h at room temperature. Slides were then incubated with alkaline phosphatase-conjugated anti-DIG antibody (1:3000, Roche, Cat. No. 11093274910) in 2.5% goat serum/B1 in a humidified chamber overnight at 4 °C. Slides were washed twice with B1 buffer following equilibration in buffer B3 (0.08-M Tris pH 9.5, 0.08 M NaCl, 0.04 M MgCl_2_) and a staining solution comprising nitro-blue tetrazolium chloride (Sigma-Aldrich, Gillingham, UK) (75 mg/ml in 70% dimethylformamide), and 5-bromo-4-chloro-3′-indolyl phosphate-p-toluidine salt (BCIP; Sigma-Aldrich, Gillingham, UK) (50 mg/ml BCIP in buffer B3) was added to the slides until strong staining was observed. Slides were then washed with distilled water and allowed to dry on a hot plate at 65 °C, cleared in xylene twice, and mounted with coverslips using histomount mounting media (Sigma-Aldrich Cat. No. 06522).

Generation and characterisation of antibodies to ArKP1.1 and ArKP2.1

To facilitate immunohistochemical analysis of the expression of ArKPP1- and ArKPP2-derived neuropeptides in A. rubens, specific antibodies to ArKP1.1 and ArKP2.1 were generated. Peptides comprising the C-terminal region of ArKP1.1 and ArKP2.1 (KNTASRVLPF-NH_2_ and KYANQQSGL-NH_2_, respectively) were custom synthesised (Peptide Protein Research Ltd., Hampshire, UK) as antigens, with the addition of an N-terminal lysine to enable glutaraldehyde-mediated coupling to a carrier protein. A tyrosine residue was also introduced at position 2 in the ArKP2.1 antigen peptide so that the peptide could be radiolabelled with iodine-125, if required. These antigen peptides were coupled with thyroglobulin (Sigma-Aldrich, Gillingham, UK) as a carrier protein by the addition of 5% glutaraldehyde (Sigma-Aldrich, Gillingham, UK). The coupled antigens were dialyzed for 48 h in 4 l of water at 4 °C using SnakeSkin dialysis tubing (3.5 MWCO; Thermo Scientific, Cat. No. 88242). Both the ArKP1.1- and the ArKP2.1-conjugated antigens were then used to immunise rabbits following a 70-day protocol (Charles River Laboratories, France). Freund’s complete adjuvant and Freund’s incomplete adjuvant were used to emulsify the antigens for the primary and the subsequent booster immunisations, respectively.

The presence of antibodies to ArKP1.1 (GPNPNTASRVLPF-NH_2_), ArKP1.2 (GPPKNSRARGGRTLLPF-NH_2_), ArKP2.1 (QLWANQQSGLF-NH_2_), and ArKP2.2 (GGGVPHVFQSGGIF-NH_2_) in antisera (bleeds from days 37, 51, and 70) was assessed using enzyme-linked immunosorbent assays (ELISA) in three different experimental setups. In the first experiment, a fixed amount (100 μl of 10^−6^ M, 10^−10^ mol) of the antigen peptides ArKP1.1-ag (KNTASRVLPF-NH_2_) or ArKP2.1-ag (KYANQQSGLF-NH_2_) was added to the wells and then incubated with varying concentrations of pre-immune and antiserum samples (10^−3^–10^−8^, diluted in 5% goat serum/PBS). In the second experiment, a fixed amount (100 μl of 10^−6^ M, 10^−10^ mol) of the peptides ArKP1.1, ArKP1.2, or ArKP2.2 (due to insolubility of ArKP2.1, ArKP2.2 was used instead since both are derived from the same precursor) was added to each well of a microtitre plate (Microlon®, Greiner Bio-One International) and then incubated with varying concentrations of pre-immune and antiserum samples (10^−3^–10^−8^ diluted in 5% goat serum/PBS). In a third experiment, wells were pre-coated with varying quantities (100 µl 10^−6^–10^−13^ M, 10^−10^–10^−17^ mol) of the peptides ArKP1.1, ArKP1.2, or ArKP2.2 and then incubated with pre-immune and antiserum at a dilution of 10^−3^. In the first and third experiments, antisera raised to ArKP1.1-ag and ArKP2.1-ag were tested for cross-reactivity with antigen peptides (ArKP2.1-Ag and ArKP1.1-Ag) and peptides derived from ArKPP2 and ArKPP1, respectively. In all experiments described above, the stock solutions of both native and antigen peptides were diluted in freshly prepared carbonate/bicarbonate buffer (50 mM anhydrous sodium carbonate, 50 mM sodium bicarbonate, pH 9.8) to pre-coat the wells of a polystyrene microtiter plate (Microlon®, Greiner Bio-One International). The plates were then covered with parafilm (Bemis Company, Cat. No. P7793-1EA) and incubated overnight at 4 °C. On the following day, the contents of the plates were disposed of, and the wells were rinsed with approximately 200 μl of PBS (1 × PBS, PH 7.4) three times (5-min incubation for each wash). The plates were dried by tapping several times on tissue paper. After the plates were fully dried, 200 μl of blocking buffer (5% goat serum (Sigma-Aldrich)/PBS) was added to each well and incubated for 2 h at 37 °C. After incubation, the blocking buffer was drained, and plates were washed with 0.1% PBS-Tween 20 three times (5-min incubation for each wash). The plates were then dried by tapping several times on tissue paper. After the plates were fully dried, antiserum diluted in 5% goat serum (Sigma-Aldrich)/PBS was added to each well and incubated overnight at 4 °C. On the third day, the contents of the plates were disposed of, and the wells were rinsed with approximately 200 μl of 0.1% PBS-Tween 20 three times (5-min incubation for each wash). The plates were dried by tapping several times on tissue paper. A total of 100 μl of alkaline phosphatase-conjugated goat anti-rabbit IgG secondary antibodies (Vector Laboratories, diluted 1:3000 in 5% normal goat serum/PBST) was added to each well and incubated for 3 h at 37 °C, and then plates were washed with 0.1% PBS-Tween 20 four times (5-min incubation in each wash). The contents were drained, and plates were dried by tapping several times on tissue paper. The alkaline phosphatase substrate (p-nitrophenyl phosphate, pNPP, Vector Laboratories) prepared in carbonate/bicarbonate buffer was added to each well at a fixed volume of 100 μl, and the plate was then incubated at room temperature for 30 min. Absorbance was measured at 415 nm using a VANTAstar (BMG LABTECH — The Microplate Reader Company) for experiment 1 and a FLUOstar Omega (BMG LABTECH — The Microplate Reader Company) for experiments 2 and 3. Mean absorbance values were calculated, and graphs were plotted using GraphPad Prism 10.5.

Affinity purification of antibodies to ArKP2.1

Affinity purification of antibodies to ArKP2.1-ag peptide (KYANQQSGLF-NH_2_) was performed using the AminoLink Plus Immobilization Kit (Thermo Scientific, Cat. No. 44894). First, the affinity column was equilibrated using 25–30 ml of PBS (pH 7.4), and the contents were drained by gravity. A 1-mM stock solution of ArKP2.1-ag peptide diluted in 4 ml of PBS (pH 7.4) was added to the affinity column, and the contents of the column were mixed for 1–2 min to form a slurry of the resin and peptide solution. A total of 40 µl of cyanoborohydride solution (5 M NaCNBH_3_ dissolved in 1 M NaOH) was added to the slurry, and then the column was incubated for 3 h at room temperature with end-over-end rocking. The column was then stored upright at 4 °C, and on the following day, the contents of the column were drained and washed with 4 ml of quenching buffer (1 M Tris–HCl, pH 7.4). Then, 2 ml of quenching buffer and 40 µl of cyanoborohydride solution were added to the column followed by 30 min of end-over-end rocking at room temperature. After this, the column was washed five times with 4 ml of wash solution (1 M NaCl) to remove any uncoupled/unbound antigen peptide. Then, 1 ml of ArKP2.1 antiserum from the terminal bleed (day 70) was added to the column, which was then subject to end-over-end rocking at room temperature for 90 min. Then, the column was drained (contents were collected and stored at − 20 °C) and washed five times with 5 ml of PBS. The purified antibodies were eluted using two elution buffers. The first flow-through was collected by adding 4 ml of 100 mM glycine to the column and later mixed with 1 ml of 1 M Tris buffer (pH 7.5). The column was then rinsed with several washes of PBS. The second flow-through was collected in another tube by adding 4 ml of 100 mM triethylamine to the column and later mixed with 1 ml of 1 M Tris buffer (pH 7.4). Both samples of eluted affinity-purified antibodies were dialysed overnight in 4 L of 1 × PBS (pH 7.4) at 4 °C using SnakeSkin dialysis tubing 3.5 MWCO (Thermo Scientific, Cat. No. 88242). Lastly, the affinity-purified antibodies were stored at 4 °C after adding 0.1% sodium azide as a preservative.

Immunohistochemical localisation of ArKP1.1 and ArKP2.1 in A. rubens

Small specimens (diameter ⁓4–6 cm) of A. rubens (arms and the central disc) were fixed in Bouin’s solution (75 ml of saturated picric acid in distilled water, 25 ml of 37% formaldehyde, 5 ml of acetic acid) at 4 °C for around 24–48 h and then were decalcified over a period of 2 weeks with a 1:1 solution of 4% ascorbic acid and 0.3-M sodium chloride. Specimens were dehydrated through a graded ethanol series into xylene and then were embedded in paraffin wax. The embedded samples were sectioned at 12-µm thickness using a microtome (RM 2145, Leica Microsystems [UK], Milton Keynes, UK). The sections were then mounted on chrome alum-gelatin coated glass slides. The slides were placed in xylene to remove paraffin wax from sections followed by 100% ethanol and then were incubated in 0.3% hydrogen peroxide (VWT Chemicals, Leicestershire, UK)/methanol for 30 min in room temperature to quench endogenous peroxidases. Then, the sections were rehydrated through a graded ethanol series (90%, 70%, and 50%) into distilled water followed by PBS and then were blocked in 5% goat serum (Sigma-Aldrich, Gillingham, UK) made up in PBS containing 0.1% Tween. Sections were incubated overnight with rabbit antisera to ArKP1.1 or ArKP2.1 (1:1000 in 5% goat serum/PBST) or affinity-purified antibodies to ArKP2.1 (diluted 1:5 in PBST). Slides were washed in PBS-Tween before incubation with the Peroxidase-AffiniPure goat anti-rabbit horseradish peroxidase-conjugated immunoglobulins (Jackson ImmunoResearch via Stratech Scientific, Newmarket, Suffolk, UK) diluted 1:1000 in 2% goat serum/PBST). Immunolabelling was revealed using a solution containing 0.015% hydrogen peroxide, 0.05% diaminobenzidine (VWR Chemicals, Leicestershire, UK), and 0.05% nickel chloride (Sigma-Aldrich, Gillingham, UK) in PBS. When staining was observed, slides were washed in distilled water, dehydrated through a graded ethanol series (50%, 70%, 90%, and 100%), and cleared in xylene before being mounted with coverslips on DPX mounting medium (Thermo Fisher Scientific, Waltham, MA, USA) [40].

Larger mature specimens of A. rubens (15.2 ± 1.32 cm) were also analysed using immunohistochemistry to investigate ArKP1.1 and ArKP2.1 expression in the reproductive system. Tissue specimens including the gonad and gonoduct were collected from both male and female animals during the mid-seasonal reproductive cycle (December 2021, mid-gametogenic stage), as described in [19].

The specificity of immunostaining was assessed by pre-absorption of antisera/affinity-purified antibodies with antigen peptides (ArKP1.1-ag, ArKP2.1-ag) and native peptides (ArKP1.1, ArKP2.2). Antisera (at 1:100 or 1:1000 dilution in PBS) or affinity-purified antibodies (undiluted) were incubated with peptides (200 µM) or an equivalent volume of added solvent (water or 10% DMSO, for positive controls) at room temperature for 1–2 h on a rocking shaker. Then, antisera or affinity-purified antibodies were diluted tenfold or fivefold, respectively, and tested on sections of starfish arms using the immunohistochemical method described above.

Imaging

Images of sections processed for mRNA in situ hybridisation and immunohistochemistry were captured using a QIClick CCD colour camera (Qimaging, British Columbia, Canada) linked to a DMRA2 light microscope (Leica), utilising Velocity version 6.3.1 image analysis software (PerkinElmer, Seer Green, UK) running on iMac computer (27-inch, Late 2013 model with OS X Yosemite, version 10.10). Images were adjusted for brightness and contrast and annotated using Adobe Photoshop CC (RRID:SCR_014199; version 26.8, × 64) and then assembled into montages using Adobe Illustrator CC (RRID:SCR_010279; version 29.6, × 64) running on an Asus computer (15.6 inch, version X509JP.307). Representative images were captured based on analysis of sections from at least two animals for each technique.

Investigation of the in vitro pharmacological effects of ArKPP1- and ArKPP2-derived peptides in A. rubens

ArKPP1- and ArKPP2-derived peptides were tested on organ preparations from A. rubens, including gonads, cardiac stomach, tube feet, and apical muscle. However, as reported previously [12], it was not possible to test ArKP2.1 due to its insolubility. Therefore, ArKP1.1, ArKP1.2, and ArKP2.2 were tested for myoactivity. Preparations of the cardiac stomach, tube foot, and apical muscle were dissected from specimens of A. rubens and set up in a 20-ml aerated organ bath, as described previously [20, 21, 32, 56]. Gonad preparations were dissected and set up in a 20 ml aerated organ bath, as described previously [19]. The contractile state of preparations was recorded using an isotonic transducer (MLT0015, ADInstruments Pty Ltd.) connected to a bridge amplifier (FE221 Bridge Amp, ADInstruments Pty Ltd.) linked to PowerLab data acquisition hardware (2/36, ADInstruments Pty Ltd). Data were recorded using LabChart (v8.0.7) software installed on a laptop computer (Lenovo E540, Windows 7 Professional) and collected using LabChart reader. In recordings, changes in the length of preparations caused by contraction or relaxation are converted to electrical signals quantified as a change in potential difference (mV). This enabled quantitative comparison of the magnitude of contraction/relaxation in response to application of substances tested at different concentrations.

Stock solutions of the synthetic peptides ArKP1.1 (GPNPNTASRVLPF-NH_2_), ArKP1.2 (GPPKNSRARGGRTLLPF-NH_2_), and ArKP2.2 (GGGVPHVFQSGGIF-NH_2_) (synthesised by PPR Ltd., Fareham, UK) were prepared in distilled water (ArKP1.1 and ArKP1.2) and 10% DMSO (ArKP2.2) because ArKP2.2 exhibits low aqueous solubility. Preliminary tests on cardiac stomach preparations revealed that ArKP1.1 and ArKP1.2 acted as muscle relaxants, so to optimise detection of their effects, these preparations were maintained under depolarising conditions using 30-mM KCl-supplemented seawater [20] when testing these peptides (10^−9^–10^−6^ M). SALMFamide-2 (S2, SGPYSFNSGLTF-NH_2_), which is a known starfish cardiac stomach relaxant [20, 21], was used to compare the effects of ArKP1.1 and ArKP1.2 by normalising the responses as a percentage reversal of the 100% contracted state induced by 30 mM KCl [59]. When ArKP1.1 and ArKP1.2 were tested on tube foot, apical muscle, and gonad preparations, 10^−5^ M acetylcholine was used to precontract these preparations and to normalise the peptide-induced responses, as described previously [18, 19, 22].

Preliminary tests revealed that ArKP2.2 caused contraction of cardiac stomach preparations, so this peptide was tested on preparations without using KCl-supplemented seawater and by applying it to preparations in seawater to achieve final concentrations ranging from 10^−10^ to 10^−6^ M. Acetylcholine (10^−6^ M and/or 10^−5^ M), which causes contraction of cardiac stomach, tube foot, apical muscle, and gonad preparations, was tested as a positive control to assess the viability of these preparations and to enable normalisation of responses to ArKP2.2.

Data were analysed using a variety of statistical tests, as detailed in Additional file 5: Dataset S2.

Investigation of the in vivo pharmacological effects of ArKPP1- and ArKPP2-derived peptides on cardiac stomach eversion/retraction in A. rubens

ArKP1.1 and ArKP1.2 were tested to determine if they induce stomach eversion when injected in vivo. Twenty-four specimens of A. rubens were starved for 1 week and were injected with 10 µl or 100 µl of 1 mM ArKP1.1 (n = 10 or 12, respectively) or 10 µl or 100 µl of 1 mM ArKP1.2 (n = 10 or 12, respectively). Hamilton 75 N 10-μl syringes (Sigma-Aldrich, Gillingham, UK) were used to inject the peptides or distilled water (negative control) through the aboral body wall of arms into the perivisceral coelom, proximal to the junctions of arms with the central disk region. Experiments were also performed where 100 µl of an equimolar mixture of 10^−3^ M ArKP1.1 and 10^−3^ M ArKP1.2 dissolved in distilled water was injected into six animals following a week of starvation. Control experiments were done by injecting 10 µl of distilled water (n = 6) and 10 µl of 10^−3^ M asterotocin (n = 6) into animals that had been starved for 1 week following the same method described above. To correspond with in vitro experiments, 100 µl 10^−3^ M S2 (n = 10) was also tested in vivo for stomach-everting effects.

As ArKP2.2 caused contraction of cardiac stomach preparations in vitro (Fig. 7B, E), this neuropeptide was tested to determine if it induces stomach retraction in animals in which stomach eversion had been induced with 2% MgCl_2_, as described previously [23]. Surprisingly, injection of 10 µl of 10^−3^ M ArKP2.2 caused slight enlargement of the area of the everted stomach (Additional file 8: Fig. S6). Therefore, experiments were performed to investigate if ArKP2.2 triggers stomach eversion by injecting 10 µl of 10^−3^ M ArKP2.2 (n = 8) and testing 10% DMSO in control experiments (n = 6), using animals starved for 1-week following the method described above and as reported in [18].

Animals injected with distilled water or 10% DMSO or peptides were video recorded for 10 min using a video camera (Canon EOS 700D) positioned underneath a petri dish containing the tested animal. Static images from video recordings were captured at 1-min intervals from the time of injection. Then, the two-dimensional area of the everted cardiac stomach was measured from the images using the ImageJ software (RRID:SCR_003070; version 1.53q, × 64) and normalised as a percentage of the area of the central disc region.

Data were analysed using a variety of statistical tests, as detailed in Additional file 5: Dataset S2.

Investigation of the in vivo effects of ArKPP1- and ArKPP2-derived peptides on righting behaviour in A. rubens

A previous study revealed that a vasopressin-/oxytocin-type neuropeptide (asterotocin) that induces cardiac stomach eversion in A. rubens also affects righting behaviour in this species [18]. Therefore, it was of interest to investigate if ArKPP1- and/or ArKPP2-derived peptides also affect righting behaviour [60]. To do this, 60 adult animals (10.4–15.7 cm in diameter) were initially starved for 1 week prior to testing to normalise their physiological condition. First, all starfish were tested for righting behaviour without injection, and the animals that did not right themselves within 5 min of inversion were considered to be unhealthy and were excluded from the experiment. Animals that passed this initial righting test were allowed 30 min to recover, and then they were either injected with 10 μl of 10^−3^ M of ArKP1.1 (n = 8) or ArKP1.2 (n = 8), equimolar ArKP1.1 and ArKP1.2 (n = 10), ArKP2.2 (n = 10) or asterotocin (n = 10), S2 (n = 10) or 10 μl of 10% DMSO (n = 10), or 10 μl of distilled water (n = 10). Ten minutes after injection, starfish were placed with their aboral side lowermost in a circular glass tank containing ⁓6 L of artificial seawater, and then the time taken for righting to occur was recorded, with completion of righting determined by the oral surface of the five arms being in contact with the floor or the wall of the testing tank. The time taken to right (in seconds) and percentage differences in righting time between animals under each testing condition were calculated, and graphs were generated as scatter plots and box and violin graphs, respectively, using Prism 10.5 (GraphPad). Data were analysed statistically using the Kruskal–Wallis followed by Dunn’s multiple comparisons tests and ordinary one-way ANOVA followed by Dunnett’s multiple comparisons tests (see also Additional file 5: Dataset S2).

Supplementary Information

Additional file 1: Dataset S1. Sequences and accession numbers of kisspeptin-type precursors used for alignments shown in Fig. 1.Additional file 2: Fig.S1 Diagram showing anatomy of starfish. A. Photograph of a specimen of Asterias rubens. The lines labelled B, C, and D show the position and orientation of the diagrams in panels B, C, and D, respectively. B. Schematic diagram of the central disc and proximal region of an arm. C. Schematic diagram of a transverse section of an arm. D. Schematic diagram of a transverse section through the tip of an arm. The coloured key identifies regions of the body shown in panels B and C. Abbreviations: a, anus; am, apical muscle; amp, ampulla; ce, coelomic epithelium; conr, circumoral nerve ring; cs, cardiac stomach; g, gonad; gcc, general coelomic cavity; ll, lateral lappet; m, mouth. md, madreporite; o, ossicle; oc, optic cusion; p, papula; pc, pyloric caeca; pd., pyloric duct; ped, pedicellaria; pm, peristomial membrane; ps, pyloric stomach; rc, rectal caeca; rca, ring canal; rnc, radial nerve cord; rwv, radial water vascular canal; sc, stone canal; sp, spine; tb, Tiedemann's body; tf, tube foot; to, terminal ossicle; tt, terminal tentacle. The photograph in panel A was taken using an iPhone14 pro max with a 48‐megapixel sensor camera and then background removal and labelling were accomplished using Adobe Photoshop CC 2025. Images B and C were adapted from [61] and were created with BioRender.com. Image D was adapted from [62] and was drawn in Adobe Illustrator CC 2025 using the Apple pencil connected to a iPad Air 2022.Additional file 3: Fig. S2. Characterisation of rabbit antisera to ArKP1.1 and ArKP2.1 antigen peptides using an enzyme-linked immunosorbent assay, demonstrating that specific antibodies to ArKP1.1 and ArKP2.1 were successfully generated. A. Graph showing that the antigen peptide ArKP1.1-ag is detected by the ArKP1.1 antiserum at dilutions between 1:1000 and 1:100,000, whereas no immunoreaction is observed with pre-immune serum or the ArKP2.1 antiserum. All data points are mean values from three replicates. B. Graph showing that the antigen peptide is detected by the ArKP2.1 antiserum at dilutions between 1:1000 and 1:100,000, whereas no immunoreaction is observed with pre-immune serum or the ArKP1.1 antiserum. All data points are mean values from three replicates. C. Graph showing that the native peptide ArKP1.1 is detected by the ArKP1.1 antiserum at dilutions between 1:1000 and 1:100,000, whereas no immunoreaction is observed with pre-immune serum. All data points are mean values from three replicates. D. Graph showing that the native peptide is detected by the ArKP2.1 antiserum at dilutions between 1:1000 and 1:100,000, whereas no immunoreaction is observed with pre-immune serum. All data points are mean values from three replicates. E. Graph showing incubation of the ArKP1.1 antiserum, the ArKP2.1 antiserum and pre-immune serum at a 1:1000 dilution with addition of between 10^–17^ mol and 10^–10^ mol of ArKP1.1 per well. With the ArKP1.1 antiserum ArKP1.1 is detected at well above the background OD with 10^–12^ to 10^–10^ mol of ArKP1.1 per well. No immunoreaction with ArKP1.1 is observed with the pre-immune serum and the ArKP2.1 antiserum. All data points are mean values from two replicates. F. Graph showing incubation of ArKP2.1 antiserum, ArKP1.1 antiserum and pre-immune serum at 1:1000 dilution with between 10^–17^ mol and 10^–10^ mol of ArKP2.2 per well. With the ArKP2.1 antiserum ArKP2.2 is detected at well above the background OD with 10^–10^ and 10^–11^ mol per well. No immunoreaction with ArKP2.2 is observed with the pre-immune serum and the ArKP1.1 antiserum. All data points are mean values from two replicates. * Due to insolubility of ArKP2.1, ArKP2.2 was used instead of ArKP2.1 in the experiment shown in panel E. ArKP2.1 and ArKP2.2 share sequence similarity in their C-terminal regions and the experiments in panels D and F show that this sequence similarity is sufficient to enable antibodies to the ArKP2.1-ag peptide to recognise ArKP2.2.Additional file 4: Fig. S3. Pre-absorption control experiments on adjacent sections of *A. rubens *radial nerve cord showing specificity of antibodies to ArKP1.1 and ArKP2.1 for their respective antigen peptides. A. Immunostaining with ArKP1.1 antiserum. B. Immunostaining is not abolished by pre-absorption of the ArKP1.1 antiserum with ArKP2.1-antigen peptide. C. Immunostaining is not abolished by pre-absorption of the ArKP1.1 antiserum with ArKP2.2 native peptide. D. Immunostaining with ArKP2.1 antiserum. E. Immunostaining is not abolished by pre-absorption of the ArKP2.1 antiserum with ArKP1.1 native peptide. F. Immunostaining is not abolished by pre-absorption of the ArKP2.1 antiserum with ArKP2.2 native peptide. Thus, although ArKP2.2 is derived from the same precursor as ArKP2.1 and shares some sequence similarity with ArKP2.1, it does block immunostaining, which demonstrates the specificity of the antibodies to ArKP2.1. Scale bar: A-F, 100 μm.Additional file 5: Dataset S2. Expanded figure legends for Figs. 7, 8 and 9 that include details of statistical analysis of data.Additional file 6: Fig. S4. Representative recordings of the effects of acetylcholine, ArKP1.1 and ArKP1.2 on a gonad preparation from A. rubens. A. Acetylcholine causes sustained contraction of the gonad preparation for a period of ~ 7 min, which is reversed by washing with artificial seawater. B. Following pre-contraction with ACh, application of ArKP1.1 has a relaxing effect on the gonad preparation, causing partial but sustained reversal of ACh-induced contraction. C. Following pre-contraction with ACh, application of ArKP1.2 has a relaxing effect on the gonad preparation, causing partial but sustained reversal of ACh-induced contraction.Additional file 7: Fig. S5. The ArKPP1-derived peptides ArKP1.1 and ArKP1.2 have no effect on the contractility of tube foot and apical muscle preparations and the ArKPP2-derived peptide ArKP2.2 has no effect on apical muscle preparations from A. rubens. A. Representative recording showing that acetylcholine causes contraction of a tube foot preparation but ArKP1.1 has no effect. The black downward pointing arrow shows when the preparation was washed with artificial sea water. Scale bar: vertical 0.02 mV; horizontal 1 min. B. Representative recording showing that acetylcholine causes contraction of a tube foot preparation but ArKP1.2 has no effect. The black downward pointing arrow shows when the preparation was washed with artificial sea water. Scale bar: vertical 0.05 mV; horizontal 1 min. C. Representative recording showing that acetylcholine causes contraction of an apical muscle preparation but ArKP1.1 has no effect. The black downward pointing arrow shows when the preparation was washed with artificial sea water. Scale bar: vertical 0.2 mV; horizontal 2 min. D. Representative recording showing that acetylcholine causes contraction of an apical muscle preparation but ArKP1.2 has no effect. The black downward pointing arrow shows when the preparation was washed with artificial sea water. Scale bar: vertical 0.2 mV; horizontal 2 min. E. Representative recording showing that acetylcholine causes contraction of an apical muscle preparation. The two black vertical lines represent an interruption in recording when other substances were tested prior to washing with artificial sea water. Scale bar: vertical 0.2 mV; horizontal 1 min. F. Representative recording showing that ArKP2.2 has no effect on an apical muscle preparation. The black downward pointing arrow shows when the preparation was washed with artificial sea water. Scale bar: horizontal 1 min.Additional file 8: Fig. S6. ArKP2.2 does not cause cardiac stomach retraction in A. rubens. Representative photographs of starfish injected with water, 10% DMSO, NGFFYamide and ArKP2.2, n = 3, xiii-xvi) are shown. The starfish were immersed in 2% MgCl_2_ to induce stomach eversion and then test substances were injected to investigate if they trigger stomach retraction. Static images from videos were captured at 0-, 1-, 3- and 6-min after injection. Water and 10% DMSO had no effect. Consistent with previously reported findings [23], NGFFYamide caused partial reversal of MgCl_2_-induced stomach eversion. In contrast, ArKP2.2 appears to cause a slight increase in stomach eversion and therefore it was also tested to determine if it induces stomach eversion when injected without MgCl_2_-induced stomach eversion.