Urbilaterian origin and evolution of sNPF-type neuropeptide signalling

Physiology and behaviour are controlled by neuropeptide signalling systems comprising peptide ligands and cognate receptors. Molecular phylogenetics combined with experimental identification of neuropeptide-receptor pairs has revealed that many neuropeptide signalling systems originated in the urbilaterian common ancestor of protostomes and deuterostomes. Neuropeptide-Y/neuropeptide-F (NPY/NPF)-type signalling is one such example, whereas NPY/NPF-related short-NPF (sNPF)-type signalling has hitherto only been identified in protostomes. Here we report the discovery of a neuropeptide (pQDRSKAMQAERTGQLRRLNPRF-NH2) that is the ligand for an sNPF-type receptor in a deuterostome, the starfish Asterias rubens (Phylum Echinodermata). Informed by phylogenetic analysis of sequence data, we conclude that the paralogous NPY/NPF-type and sNPF-type signalling systems originated in Urbilateria but NPY/NPF-type signalling was lost in echinoderms. Furthermore, we present evidence that sNPF-type peptides are orthologs of vertebrate prolactin-releasing peptides. Our findings demonstrate the importance of experimental studies on echinoderms for reconstructing the evolutionary history of neuropeptide signalling systems.

Physiology and behaviour are controlled by neuropeptide signalling systems comprising 32 peptide ligands and cognate receptors. Molecular phylogenetics combined with experimental 33 identification of neuropeptide-receptor pairs has revealed that many neuropeptide signalling 34 systems originated in the urbilaterian common ancestor of protostomes and deuterostomes. 35 Neuropeptide-Y/neuropeptide-F (NPY/NPF)-type signalling is one such example, whereas 36 NPY/NPF-related short-NPF (sNPF)-type signalling has hitherto only been identified in 37 protostomes.
Here we report the discovery of a neuropeptide 38 (pQDRSKAMQAERTGQLRRLNPRF-NH 2 ) that is the ligand for an sNPF-type receptor in 39 a deuterostome, the starfish Asterias rubens (Phylum Echinodermata). Informed by 40 phylogenetic analysis of sequence data, we conclude that the paralogous NPY/NPF-type and 41 sNPF-type signalling systems originated in Urbilateria but NPY/NPF-type signalling was lost 42 in echinoderms. Furthermore, we present evidence that sNPF-type peptides are orthologs of 43 vertebrate prolactin-releasing peptides. Our findings demonstrate the importance of 44 experimental studies on echinoderms for reconstructing the evolutionary history of 45 neuropeptide signalling systems. 46 Introduction 47 Neuropeptides are neuronally secreted signalling molecules that regulate many 48 physiological processes and behaviours in animals, including feeding, digestion, reproduction 49 and social behaviour. They typically exert effects by binding to cognate G-protein coupled 50 receptors (GPCRs) on target cells, which leads to changes in the activity of downstream 51 effectors (e.g. ion channels, enzymes) (Jékely et al. 2018). Investigation of the evolution of 52 neuropeptide signalling has revealed that many of the neuropeptide systems found in 53 vertebrates have orthologs in invertebrate deuterostomes (urochordates, cephalochordates, 54 hemichordates, echinoderms) and protostomes (e.g. arthropods, nematodes, molluscs, 55 annelids, platyhelminthes). Thus, the evolutionary origin of over thirty neuropeptide 56 signalling systems has been traced back to the common ancestor of the Bilateria (Urbilateria) 57 (Mirabeau and  One of the neuropeptide signalling systems that originated in Urbilateria is 59 neuropeptide Y (NPY)-type signalling. NPY is a 36-residue peptide that was first isolated 60 from the porcine hypothalamus Tatemoto 1982) but that is also 61 expressed by neurons in many other regions of the nervous system (Adrian et al. 1983; 62 Morris 1989) and in peripheral organs such as the gut and cardiovascular system (Holzer et 63 al. 2012; Farzi et al. 2015). Accordingly, NPY is pleiotropic (Pedrazzini et al. 2003), but it is 64 perhaps most widely known as a potent stimulant of food intake in mammals (Minor et al. 65 2009; Zhang et al. 2011). 66 NPY belongs to a family of related signalling molecules in vertebrates, including 67 peptide YY (PYY) and pancreatic polypeptide (PP), that evolved from a common ancestral 68 peptide by gene/genome duplication (Larhammar et al. 1993;Larhammar 1996;Elphick et al. 69 2018). Furthermore, the sequences of NPY-type peptides are highly conserved across the 70 vertebrates, sharing up to 92% identity between mammals and cartilaginous fish (Larhammar 71 et al. 1993;Larhammar 1996;Cerdá-Reverter et al. 2000). A neuropeptide in vertebrates that 72 is evolutionarily related to NPY/PYY/PP-type peptides is prolactin-releasing peptide (PrRP), 73 which was first discovered as a ligand for the orphan receptor hGR3 (Hinuma et al. 1998). 74 Phylogenetic analysis has revealed that PrRP-type receptors are paralogs of NPY/PYY/PP-75 type receptors and it has been proposed that PrRP-type signalling originated in the vertebrate 76 lineage (Lagerström et al. 2005). However, more recently, orthologs of vertebrate PrRP-type 77 receptors have been identified in invertebrate deuterostomes -the cephalochordate 78 Branchiostoma floridae and the hemichordate Saccoglossus kowalevskii -indicating that 79 PrRP-type signalling may have originated in a common ancestor of the deuterostomes 80 (Mirabeau and Joly 2013). 81 An important insight into the evolutionary history of NPY-type peptides was obtained 82 by purification from extracts of a protostome invertebtate, the platyhelminth Moniezia 83 expansa, of a peptide immunoreactive with antibodies to the C-terminal hexapeptide of PP 84 (Maule et al. 1991). Sequencing revealed a 39-residue peptide with a similar structure to 85 NPY, but with the C-terminal tyrosine (Y) substituted with a phenylalanine (F). Hence, this 86 invertebrate NPY homolog was named neuropeptide F (NPF) (Maule et al. 1991). 87 Subsequently, NPF-type neuropeptides have been identified in other protostomian 88 invertebrates, including other platyhelminths (Curry et al. 1992), molluscs (Leung et al. 1992; 89 Rajpara et al. 1992), annelids (Díaz-Miranda et al. 1991;Veenstra 2011;Conzelmann et al. 90 2013; Bauknecht and Jékely 2015) and arthropods (Brown et al. 1999), and these peptides 91 typically have a conserved C-terminal RPRFamide motif and range in length from 36 to 43 92 residues. 93 Following the discovery of M. expansa NPF, antibodies to this peptide were 94 generated and used to assay for related peptides in other invertebrates. Interestingly, this 95 resulted in the discovery of two novel peptides, ARGPQLRLRFamide and 96 APSLRLRFamide, in brain extracts from the Colorado potato beetle Leptinotarsa 97 decemlineata (Spittaels et al. 1996). As these peptides were isolated using antibodies to M. 98 expansa NPF, they were originally referred to as NPF-related peptides. However, because 99 they are much shorter in length than NPF, they were later renamed as short neuropeptide F 100 (sNPF) (Vanden Broeck 2001) and homologs were identified in other insects (Schoofs et al. 101 2001). Furthermore, alignment of NPY-type peptides and precursors from vertebrates with 102 NPF-type and sNPF-type peptides and precursors from protostomes revealed that whilst NPF 103 peptides are clearly closely related (orthologous) to vertebrate NPY peptides, sNPF peptides 104 and precursors exhibit too many differences to be considered orthologs of NPY/NPF-type 105 peptides and precursors (Nässel and Wegener 2011). Further evidence that chordate NPY-106 type and invertebrate NPF-type neuropeptides are orthologous has been provided by 107 similarity-based clustering methods, showing that the NPY-type and NPF-type precursors 108 form a pan-bilaterian cluster, whereas sNPF-type precursors form a separate cluster (Jékely 109 2013). Thus, sNPF-type peptides are considered to be a family of neuropeptides that is 110 distinct from the NPY/NPF-type family of neuropeptides. 111 A receptor for sNPF-type peptides was first identified in the fruit fly Drosophila 112 melanogaster with the deorphanisation of the G-protein coupled receptor CG7395 (Mertens 113 et al. 2002), which was previously annotated as a homolog of mammalian NPY-type 114 receptors. 2013). In the cockroach Periplaneta americana, starvation followed by feeding increases and 130 then decreases, respectively, the number of sNPF-immunoreactive cells in the midgut 131 epithelium. 132 Since sNPF signalling was discovered in insects, it was initially thought that this 133 neuropeptide system may be unique to arthropods (Nässel and Wegener 2011 frame encodes a 108-residue protein comprising a predicted 19-residue signal peptide, a 23-187 residue NPY-like peptide sequence with an N-terminal glutamine residue and a C-terminal 188 glycine residue, followed by a putative monobasic cleavage site (Supplementary Figure 1A). 189 Analysis of radial nerve cord extracts using mass spectrometry (LC-MS-MS) revealed the 190 presence of a peptide with the structure pQDRSKAMQAERTGQLRRLNPRF-NH 2 , 191 showing that the N-terminal glutamine and C-terminal glycine in the precursor peptide are 192 post-translationally converted to a pyroglutamate residue and amide, respectively 193 (Supplementary Figure 1B). Having determined the structure of this peptide, we provisionally 194 named it A. rubens NPY-like peptide or ArNPYLP. 195 The ArNPYLP sequence was aligned with related peptides from other echinoderms 196 and with NPY/NPF-type peptides from other phyla ( Figure 1). This revealed that ArNPYLP 197 and a closely related peptide in the starfish Acanthaster planci both share a C-terminal 198 PRFamide sequence with several protostome NPF-type peptides. In contrast, related peptides 199 in two other echinoderms, a brittle star and a sea urchin, have a C-terminal RYamide motif, 200 which is a characteristic of vertebrate NPY-type peptides. However, the alignment also 201 revealed that the echinoderm NPY-like peptides are shorter (22-25 residues) than NPY/NPF-202 type peptides in other taxa (30-41 residues). Furthermore, the echinoderm peptides lack two 203 proline (P) residues that are a conserved feature of the N-terminal region of many NPY/NPF-204 type peptides, with the exception some peptides that have only one of these proline residues 205 and a peptide in the cephalochordate Branchiostoma floridae that has neither ( Figure 1). 206 Furthermore, there are four other residues that are highly conserved in bilaterian NPY/NPF 207 peptides -tyrosine (Y), leucine (L), tyrosine (Y) and isoleucine (I) residues, which are 208 marked with asterisks in Figure 1. Importantly, none of these residues are present in the 209 echinoderm NPY-like peptides. It is noteworthy, however, that all but one of the 210 aforementioned six conserved residues in NPY/NPF-type peptides are present in a peptide 211 from a species belonging to a sister phylum of the echinoderms -the hemichordate 212 Saccoglossus kowalevskii (Figure 1) (Mirabeau and Joly 2013; Elphick and Mirabeau 2014). 213 Collectively these findings indicated that ArNPYLP and related peptides in other 214 echinoderms may not be orthologs of NPY/NPF-type peptides. 215

231
The exon-intron structure of echinoderm NPYLP genes is different to NPY/NPF genes 232 To investigate further our proposition that echinoderm NPY-like neuropeptides may 233 not be orthologs of NPY/NPF-type neuropeptides, we compared the exon-intron structure of 234 genes encoding these peptides. Previous studies have reported that a conserved feature of 235 NPY/NPF genes is an intron that interrupts the coding sequence for NPY/NPF-type peptides, 236 with the intron located between the second and third nucleotide of the codon for the arginine 237 residue of the C-terminal RF or RY dipeptide (Mair et al. 2000). Here we show this 238 conserved feature in NPY/NPF genes in species from several animal phyla, including a 239 hemichordate (sister phylum to the echinoderms), chordates, molluscs, an annelid, a 240

Deuterostomes
Protostomes priapulid, an arthropod and a nematode ( Figure 2). Because genome sequence data are 241 currently not available for the starfish A. rubens, we examined the structure of the NPYLP 242 gene in echinoderm species where genome sequences have been obtained -the starfish A. 243 planci and the sea urchin S. purpuratus. This revealed that in the echinoderm NPYLP genes 244 the coding sequence for NPYLP is interrupted by an intron, but it is located in a different 245 position to the intron that interrupts the coding sequence for NPY/NPF-type peptides. Thus, it 246 does not interrupt the codon for the arginine of the C-terminal RF or RY motif, but instead it 247 is located between the first and second nucleotide of the codon for an alanine (A. planci) or 248 glycine (S. purpuratus) residue located in the N-terminal or central regions, respectively, of 249 the NPYLPs (Figure 2). Another difference is that typically in NPY/NPF genes there is 250 another intron that interrupts the coding sequence in the C-terminal region of the precursor 251 protein, whereas in the echinoderm NPYLP genes the coding sequence for the C-terminal 252 region of the precursor protein is not interrupted by an intron. Collectively, these findings 253 provide further evidence that the echinoderm NPY-like peptides are not orthologs of 254 NPY/NPF-type neuropeptides.

Discovery of orthologs of sNPF-type receptors in A. rubens and other echinoderms 279
Having obtained evidence that the echinoderm NPY-like peptides are not orthologs 280 of NPY/NPF-type neuropeptides in other bilaterians, we then investigated the occurrence 281 in A. rubens and other echinoderms of proteins related to G-protein coupled receptors that 282 mediate effects of NPY/NPF-type peptides and sNPF-type peptides in other bilaterians. 283 Using receptor sequences for the H. sapiens NPY-type, D. melanogaster NPF-type and D. 284 melanogaster sNPF-type receptors as queries for similarity-based analysis of A. rubens 285 neural transcriptome sequence data, a transcript (contig 1120879) encoding a 386-residue 286 protein was identified as the best hit (Supplementary Figure 2). Homologs of this protein 287 were also identified in other echinoderms, including the starfish A. planci, the sea urchin 288 S. purpuratus and the sea cucumber A. japonicus. To determine the relationship of these 289 echinoderm receptors with other bilaterian neuropeptide receptors, we performed a 290 phylogenetic analysis using the maximum likelihood method. For this analysis, in addition 291 to bilaterian NPY/NPF-type receptors and protostome sNPF-type receptors, we also 292 included receptors that are closely related to NPY/NPF-type and sNPF-type receptors -293 prolactin-releasing peptide-type, GPR83-type, tachykinin (TK)-type and luqin (LQ)-type 294 receptors. This revealed that the echinoderm receptors are positioned within a branch of the 295 phylogenetic tree that comprises NPY/NPF-type and sNPF-type receptors, with the other 296 receptor types included in the analysis occupying an outgroup position ( Figure 3). 297 Furthermore, the echinoderm receptors are not positioned in a clade comprising NPY/NPF-298 type receptors but instead they are positioned in a clade comprising sNPF-type receptors, 299 with bootstrap support of >90 %. Thus, we conclude that the echinoderm receptors are 300 orthologs of protostome sNPF-type receptors and accordingly we named the A. rubens 301 receptor Ar-sNPFR. Furthermore, we hypothesised that this protein may be the receptor for 302 the A. rubens peptide that we have referred to hitherto as ArNPYLP. 303 sNPF-type NPY/F-type co-expressed with Gα16 in CHO-K1 cells expressing apoaequorin to produce the cell system 333 CHO-Ar-sNPFR. Synthetic ArNPYLP (pQDRSKAMQAERTGQLRRLNPRF-NH 2 ) was 334 then tested as a candidate ligand for Ar-sNPFR at concentrations ranging from 10 −14 M to 335 10 −5 M, comparing with cells incubated in assay media without the addition of the peptide. 336 This revealed that ArNPYLP at a concentration of 10 −5 M triggers luminescence responses 337 (defined as 100%) in CHO-Ar-sNPFR cells that were approximately five times the 338 background luminescence detected with the assay media used to dissolve the peptide ( Figure  339 4A), demonstrating that ArNPYLP acts as a ligand for the receptor. Furthermore, ArNPYLP 340 induced dose-dependent luminescence in CHO-Ar-sNPFR cells with a half-maximal response 341 concentration (EC 50 ) of 1.5 × 10 −10 M ( Figure 4B). Importantly, no response to ArNPYLP 342 was observed in CHO-K1 cells transfected with the vector alone, demonstrating that the 343 signal observed in CHO-Ar-sNPFR cells exposed to ArNPYLP can be attributed to activation 344 of the transfected receptor (Supplementary Figure 4). Because ArNPYLP contains a potential 345 dibasic cleavage site (see underlined arginine residues in its sequence: 346 pQDRSKAMQAERTGQLRRLNPRF-NH 2 ), we hypothesised that the C-terminal 347 pentapeptide of ArNPYLP (LNPRFamide) may also be generated from ArNPYLP in vivo.

348
Therefore, we also tested synthetic LNPRFamide as a candidate ligand for Ar-sNPFR. 349 However, this peptide did not induce luminescence responses in CHO-Ar-sNPFR cells 350 ( Figure 4B). So we conclude that the 22-residue amidated peptide ArNPYLP is the natural 351 ligand for Ar-sNPFR in A. rubens. Furthermore, on this basis we changed the name of this 352 peptide from ArNPYLP to Ar-sNPF. 353 354

366
Comparison of the sequences of Ar-sNPF and orthologs in other echinoderms with 367 sNPF-type peptides from other taxa. Luminescence (raw value) -15 -14 -13 -12 -11 -10 -9 -8 -7 peptides and sNPF-type peptides in other phyla. For example, the echinoderm peptides 383 typically have a serine-glycine (SG) motif (or TG in Ar-sNPF, which represents a 384 conservative substitution) in their central region and this aligns with an N-terminal SG motif 385 in a sNPF-type peptide from the insect Tribolium castaneum and with a serine or glycine 386 residue in the N-terminal region of other protostome sNPF-type peptides ( Figure 5). 387 Furthermore, the C-terminal region of the echinoderm peptides also shares sequence 388 similarity with the C-terminal region of protostome sNPF-type peptides. Thus, Ar-sNPF has 389 the C-terminal sequence LNPRFamide and likewise sNPF-type peptides with a C-terminal 390 LxxRFamide motif occur in some insect species and sNPF-type peptides with a C-terminal 391 LxRFamide or LxRYamide (with Y being a conservative substitution) occur in some 392 molluscan and annelid species. There are, however, also notable differences between the 393 echinoderm peptides and the protostome sNPF-type peptides. Thus, in addition to obvious 394 differences in peptide length, many protostome sNPF-type peptides have a conserved proline 395 residue but this is not a feature of the echinoderm peptides ( Figure 5). Finally, a noteworthy 396 highly variable feature of protostome sNPF-type precursors is the number of neuropeptides 397 they give rise to. Thus, the echinoderm precursors contain a single neuropeptide, whereas the 398 number of sNPF-type peptides derived from protostome precursors range from one (C. 399 elegans FLP-21, T. castaneum), to three or four (e.g. S. mediterranea, D. melanogaster, C. 400 gigas) to as many as seven (e.g. C. elegans FLP3, P. dumerilii). 401 Comparison of the structure of genes encoding precursors of sNPF-type peptides 415 Having identified  the  22-residue  amidated  peptide  416 pQDRSKAMQAERTGQLRRLNPRF-NH 2 (Ar-sNPF) as the ligand for Ar-sNPFR, it was 417 also of interest to compare the structure of genes encoding orthologs of this peptide in 418 echinoderms for which genome sequence data are available with the structure of genes 419 encoding sNPF-type peptides in protostomes ( Figure 6). Consistent with the variability in the 420 number of neuropeptides derived from sNPF-type precursors, we found that the structure of 421 the genes encoding these proteins was also highly variable. Thus, the number of introns 422 interrupting the coding sequence ranges from one in the starfish A. planci and in the mollusc 423 C. gigas to as many as five in the C. elegans FLP-15 precursor gene. However, a consistent 424 feature is the presence of an intron located after the protein-coding exon(s) that encode the N-425 terminal signal peptide. It is noteworthy that in the echinoderm precursor genes this intron 426 interrupts the coding sequence for the sNPF-type peptide, whereas in protostome sNPF-type 427 genes the coding sequences for sNPF-type peptides are located 3' to this intron. This intron 428 may be an evolutionarily conserved feature of sNPF-type precursor genes in the Bilateria, but 429 with Based on a cluster analysis of neuropeptide receptor relationships, it has been 453 proposed previously that protostome sNPF-type signalling may be orthologous to vertebrate 454 prolactin-releasing peptide (PrRP)-type signalling (Jékely 2013 Having found that echinoderm sNPF-type peptides share sequence similarity with 476 vertebrate PrRPs, we also compared the structure of genes encoding the precursors of these 477 peptides ( Figure 7B). This revealed that a common characteristic is the presence of an intron 478 that interrupts the coding sequence at a position corresponding to the N-terminal or central 479 region of the echinoderm sNPFs and vertebrate PrRPs. Furthermore, in both echinoderm 480 sNPF-type genes and vertebrate PrRP genes the intron interrupts the coding sequence in the 481 same frame, at a position between the first and second nucleotide of the interrupted codon, 482 which is denoted by +1 in Figure 7B. Precursors of peptides that share sequence similarity with members of the bilaterian 520 NPY/NPF-type neuropeptide family were discovered recently in the phylum Echinodermata 521 (Zandawala et al. 2017). These include proteins identified in several brittle star species (class 522 Ophiuroidea) and the starfish species Asterias rubens and Patiria miniata (class Asteroidea). 523 Here we report the cloning and sequencing of a cDNA encoding the precursor of the NPY-524 like peptide in A. rubens. Furthermore, the primary structure of this peptide (ArNPYLP) was 525 determined using mass spectrometry, demonstrating that it is a twenty two-residue amidated 526 peptide with an N-terminal pyroglutamate. However, comparison of the sequences of 527 ArNPYLP and related peptides from other echinoderms with NPY/NPF-type neuropeptides 528 from other bilaterians revealed some striking differences. Most notable is that the echinoderm 529 peptides lack two conserved proline residues that are present in the majority of NPY/NPF-530 type neuropeptides that have been identified in other bilaterians. These prolines form part of 531 what is known as the polyproline-helix or polyproline-fold, which interacts with other 532 conserved residues (a leucine residue and two tyrosine residues) that have been shown to be 533 important in determining the three-dimensional structure of NPY-type peptides in vertebrates 534 ( bioactivity. Furthermore, these conserved residues are also present in an NPY/NPF-type 545 peptide that has been identified in S. kowalevskii, a species belonging to the phylum 546 Hemichordata, which is a sister phylum to the Echinodermata in the ambulacrarian clade of 547 the Bilateria ( Specifically, comparison of the structure of the human NPY precursor gene with the structure 556 of the gene encoding the NPF precursor in the platyhelminth Moniezia expansa revealed that 557 in both genes the first protein-coding exon encodes the N-terminal signal peptide and most of 558 the NPY/NPF-type peptide through to the first two nucleotides of the codon for the 559 penultimate residue, an arginine residue. The next exon contains the third nucleotide of the 560 arginine codon and codons for i) a C-terminal tyrosine (in the case of the human NPY gene) 561 or a C-terminal phenylalanine (in the case of M. expansa NPF gene), ii). a glycine residue 562 that is a substrate for C-terminal amidation, iii). a dibasic cleavage site (KR) and iv). part of 563 the C-terminal region of the precursor protein (Mair et al. 2000). Here we expanded 564 comparative analysis of NPY/NPF gene structure to include other bilaterians. We found that 565 a gene structure in which most of NPY/NPF-type neuropeptide sequence is encoded in one 566 exon and the C-terminal F or Y and the amidation and cleavage sites are in the next exon is a 567 highly conserved feature of NPY/NPF genes, which is seen in vertebrates, cephalochordates, 568 hemichordates, lophotrocozoans, priapulids, arthropods and nematodes. Therefore, our 569 finding that this is not a feature of genes encoding the NPY/NPF-like peptides in 570 echinoderms (the starfish A. planci and the sea urchin S. purpuratus) provides important 571 further evidence that these peptides are not orthologs of NPY/NPF-type neuropeptides. 572 573 Discovery of sNPF-type neuropeptide signalling in echinoderms 574 If the echinoderm NPY/NPF-like peptides are not orthologs of the NPY/NPF-type 575 neuropeptide family, then a logical prediction would be that orthologs of receptors for 576 NPY/NPF-type neuropeptides are also absent in echinoderms, because analysis of 577 neuropeptide-receptor co-evolution in the Bilateria has revealed that loss of a neuropeptide in 578 an animal lineage is invariably accompanied by loss of its cognate receptor (Mirabeau and 579 Joly 2013). Therefore, we performed a detailed phylogenetic analysis of sequence data to 580 address this issue. Consistent with our prediction, orthologs of bilaterian NPY/NPF-type 581 receptors were not found in any of the echinoderm species analysed. However, we discovered 582 that A. rubens and other echinoderms do have orthologs of sNPF-type receptors, paralogs of 583 the NPY-type receptors that hitherto have only been characterised in protostomes. Therefore, 584 we hypothesised that the echinoderm NPY-like peptides may act as ligands for sNPF-type 585 receptors and performed experimental studies to test this hypothesis. Having identified a 586 transcript encoding a sNPF-type receptor in the starfish A. rubens (Ar-sNPFR), we cloned a 587 cDNA encoding this receptor and expressed it in CHO-K1 cells. Then the A. rubens NPY-588 like peptide (ArNPYLP) was tested as a candidate ligand for Ar-sNPFR. This revealed that 589 ArNPYLP causes dose-dependent activation of the Ar-sNPFR with an EC 50 value of 0.15 590 nM, demonstrating that it is a potent ligand for this receptor. Evidence of the specificity of 591 peptide-receptor pairing was established by our finding that other peptides, including a C-592 terminal fragment of ArNPYLP (LNPRFamide) and the A. rubens luqin-type neuropeptide 593 ArLQ (Yañez-Guerra et al. 2018), do not act as ligands for Ar-sNPFR. Therefore, we 594 conclude that the twenty two-residue neuropeptide formerly referred to as ArNPYLP is the 595 natural ligand for the A. rubens sNPF-type receptor Ar-sNPFR and therefore this peptide 596 should be renamed Ar-sNPF. Our discovery of the Ar-sNPF -Ar-sNPFR signalling system in 597 A. rubens is important because this is the first sNPF-type signalling system to be identified in 598 a deuterostome. Thus, sNPF-type signalling is not unique to protostomes, as has been 599 suggested previously, and the evolutionary origin of this signalling system can be traced back 600 to the common ancestor of the Bilateria. 601 Our discovery of sNPF-type signalling in a deuterostome, the starfish A. rubens 602 (Phylum Echinodermata) and our and previous (Mirabeau and Joly 2013) phylogenetic 603 analyses of neuropeptide receptor relationships indicates that sNPF-type and NPY/NPF-type 604 signalling are paralogous. Thus, we can infer that gene duplication in a common ancestor of 605 the Bilateria gave rise to paralogous NPY/NPF-type and sNPF-type precursor genes and 606 paralogous NPY/NPF-type and sNPF-type receptor genes. In this context, by analysing the 607 phylogenetic distribution and sequences of NPY/NPF-type and sNPF-type precursors and 608 receptors, the evolutionary history of these signalling systems in the Bilateria can be 609 examined and reconstructed. 610 611 Reconstructing the evolution of NPY/NPF-type neuropeptide signalling 612 Our analysis and previous analysis (Mirabeau and Joly 2013) of the phylogenetic 613 distribution of NPY/NPF-type signalling indicates that this neuropeptide system has been 614 widely preserved and is highly conserved in the Bilateria, with relatively few instances of 615 loss based on the data currently available (Figure 8) based on similarity-based sequence alignments, it has been suggested that the mature peptide 631 derived from the C. elegans protein FLP-27 may be an ortholog of NPY/NPF-type peptides 632 (Fadda et al. 2019). Here, our analysis of the structure of the gene encoding the FLP-27 633 precursor has revealed that it has the characteristic structure of NPY/NPF-type genes, with an 634 intron interrupting the codon for the C-terminal arginine of the NPF-type peptide sequence. 635 Thus, based on our analysis of C. elegans sequence data, we conclude that the NPY/NPF-636 type peptide derived from the FLP-27 precursor protein is likely to act as a ligand for the 637 NPR-11 and/or NPR-12 receptors. 638 The first NPF-type peptide was discovered in the platyhelminth Monieza expansa 639 (Maule et al. 1991), but the receptor for this ligand has not been identified. We were unable 640 to identify a candidate receptor for NPF in M. expansa, which probably reflects the limited 641 availability of sequence data for this species. However, genome/transcriptome sequence data 642 are available for the flatworm species S. mediterranea and an expanded family of sixteen 643 putative NPY/NPF-type receptors (Smed-NPYR1 -Smed-NPYR16) in this species has been 644 reported (Saberi et al. 2016). Conversely, our phylogenetic analysis (Figure 2) has revealed 645 that only Smed-NPYR1, Smed-NPYR3, Smed-NPYR5 and Smed-NPYR6, are orthologs of 646 the NPY/NPF-type family of receptors. Accordingly, it has been shown that Smed-NPYR1 is 647 activated by a peptide that has a characteristic NPY/NPF-type structure (Saberi et al. 2016). 648 In the annelid P. dumerilii, a receptor named NPY-4 receptor 1 that is activated by 649 three NPY/NPF-type peptides (NPY1, NPY3 and NPY4) has been reported previously. 650 However, cluster analysis indicated that this receptor may not be an NPY/NPF-type receptor 651 (Bauknecht and Jékely 2015). Interestingly, we have identified an NPF-type precursor and 652 peptide in P. dumerilii that has not been reported previously and which contains residues that 653 are conserved in NPF-type peptides from other protostomes (Figure 1). Furthermore, the 654 exon/intron structure of a gene encoding an ortholog of the P. dumerilii NPF-type precursor 655 in the annelid Helobdella robusta is consistent with NPY/NPF-type precursor genes ( Figure  656 2). It is likely, therefore, that the NPF-type peptide in P. dumerilii is the ligand for the orphan 657 receptor GPR62, which is clearly an ortholog of NPY/NPF-type receptors (Figure 3). 658 Although NPY/NPF-type signalling has been retained in the majority of phyla, as 659 discussed above, it has been reported previously that NPY/NPF-type signalling has been lost 660 in urochordates (Mirabeau and Joly 2013). Furthermore, here we present evidence for the 661 first time indicating that NPY/NPF-type signalling has also been lost in echinoderms. The 662 functional significance of the loss of NPY/NPF-type signalling in urochordates and 663 echinoderms is unknown. However, insights into this issue may emerge as we learn more 664 about the physiological roles of NPY/NPF-type signalling in a variety of invertebrate taxa. 665 666 Reconstructing the evolution of sNPF-type neuropeptide signalling 667 Discovery of sNPF-type neuropeptide signalling in echinoderms is interesting because 668 orthologs of protostome sNPF-type receptors have not been identified in other deuterostome 669 phyla -Chordata and Hemichordata. Conversely, both peptides and receptors of the sNPF-670 type signalling system have been identified in several protostome phyla (Figure 8). In this 671 context, it is of interest to first review here what is currently known about the molecular 672 components of the sNPF-type signalling system in protostomes. 673 Starting with the ecdysozoan protostomes, it was originally thought that sNPF-type 674 signalling may be arthropod-specific, reflecting the original discovery of this signalling sNPF-type receptors. Hitherto the existence of sNPF-type signalling in priapulids has not 689 been reported. Here our analysis of sequence data from Priapulus caudatus has identified an 690 sNPF-type receptor but we did not identify a precursor protein that gives rise to a candidate 691 ligand for this receptor and therefore this is an objective for future work. 692 Turning to the spiralian protostomes, sNPF-type receptors were identified in molluscs, 693 annelids and platyhelminths (Figure 2, 8) clade comprising spiralian molluscan sNPF-type receptors also contains a receptor from the 699 annelid Platynereis dumerilli that has been experimentally characterised as a receptor that is 700 activated by the amidated tetrapeptide FMRFamide and a peptide known as NKY (Bauknecht 701 and Jékely 2015). Furthermore, the NKY peptide and the NKY receptors have been described 702 as paralogs of NPY-type peptides and NPY-type receptors, respectively (Bauknecht and 703 Jékely 2015). Although our phylogenetic analysis indicates that the C. gigas sNPF receptor 704 and the P. dumerilli NKY receptor are orthologs, there is a discrepancy in the ligands that 705 activate these two receptors. The C. gigas sNPF-type receptor is activated by a sNPF-type 706 peptide comprising five to six residues and with a C-terminal LFRFamide sequence (Bigot et  707 al. 2014), whereas the P. dumerilii NKY receptor was shown to be activated by NKY-type 708 peptides that are typically up to forty-three residues in length and with a C-terminal 709 LLRYamide sequence (Bauknecht and Jékely 2015). Therefore, although it was not the 710 primary purpose of this study, we investigated this anomaly by comparing the ability of three 711 peptides to act as ligands for the C. gigas sNPF receptor: i). the peptide GSLFRFamide, 712 which has been shown previously to act as a ligand for this receptor (Bigot et al. 2014), ii). 713 the amidated tetrapeptide FMRFamide and iii). a C. gigas NKY-type peptide. This 714 experiment revealed that GSLFRFamide is the most potent ligand of this receptor, with an 715 EC 50 value of 31 nM (Supplementary Figure 5). Interestingly, however, we found that the C. 716 gigas NKY-type peptide and FMRFamide also cause activation of the receptor, but only at 717 relatively high concentrations. Thus, the EC 50 for FMRFamide was 3.4 µM and the EC 50 for 718 the C. gigas NKY-type peptide was 3.02 µM. We conclude from this that GSLFRFamide is a 719 natural ligand for the C. gigas sNPF receptor, consistent with the findings of (Bigot et al. 720 2014), whereas the ability of FMRFamide and the C. gigas NKY-type peptide to activate the 721 C. gigas sNPF-type receptor may reflect non-physiological neuropeptide-receptor cross-talk. 722 Accordingly, the P. dumerilii receptor identified as a receptor for NKY (Bauknecht and  723 Jékely 2015) may be activated physiologically by shorter sNPF-type GSLFRFamide-like 724 peptides, the sequences of which we show in the alignment in Figure 5 (e.g. GTLLRYamide, 725 GSLMRYamide etc.). It is noteworthy that the C-terminal tetrapeptide of GTLLRYamide is 726 similar to the C-terminal tetrapeptide of P. dumerilli NKY-1 (IMRYamide), which likely 727 explains why NKY was found to act as a ligand, albeit with an EC 50 of 420 nM, for a P. 728 dumerilli NKY/sNPF receptor (Bauknecht and Jékely 2015). Further studies are now needed 729 to investigate the ligand-binding properties of the P. dumerilli NKY/sNPF receptor in more 730 detail. 731 As highlighted above, an expanded family of sixteen putative NPY/NPF-type receptors 732 (Smed-NPYR1 -Smed-NPYR16) has been identified in the platyhelminth S. mediterranea 733 (Saberi et al. 2016). However, our phylogenetic analysis indicates that four of these receptors 734 (Smed-NPYR7, Smed-NPYR8, Smed-NPYR9, and Smed-NPYR10) are orthologs of sNPF-735 type receptors. Therefore, it would be expected that the peptide ligands for these receptors are 736 similar to the peptides that have been identified as ligands for sNPF-type receptors in another 737 spiralian -the mollusc C. gigas (Bigot et al. 2014). On this basis, we have identified 738 candidate ligands for S. mediterranea sNPF-type receptors, which are included in the 739 alignment shown in Figure 5 (SSVFRFamide, RGVAFRFamide and GSVFRYamide). 740 Having reviewed the characteristics of sNPF-type signalling in protostomes, it is of 741 interest to make comparisons with the sNPF-type signalling system that has been identified 742 here for the first time in a deuterostome phylum -the Echinodermata. Alignment of the 743 sequences of protostome sNPF-type peptides with the echinoderm sNPF-type peptides 744 reveals modest C-terminal sequence similarity, as shown in Figure 5 and as described in the 745 results section of this paper. Furthermore, the echinoderm sNPF-type peptides are much 746 longer than protostome sNPF-type peptides. Another difference is that protostome sNPF-type 747 neuropeptide precursors typically give rise to multiple sNPF-type peptides, whereas in 748 echinoderms the sNPF-type precursor contains a single sNPF-type peptide that is located 749 adjacent to the signal peptide. Likewise, comparison of the structure of the genes encoding 750 sNPF-type precursors in protostomes and echinoderms reveals limited similarity ( Figure 6). 751 Thus, there is little evidence of orthology from comparison of the neuropeptide, precursor 752 and gene sequences in protostomes and echinoderms. Consequently, our conclusion that the 753 echinoderm NPY-like peptides are orthologs of protostome sNPF-type peptides is principally 754 based on the orthology of their receptors, as shown in Figure 3. It is important to note, 755 however, that this is not unprecedented in investigations of the evolution neuropeptide 756 signalling. Thus, whilst the sequences of some neuropeptides and neuropeptide precursors are 757 highly conserved throughout the Bilateria, others are so divergent that they can be 758 unrecognisable as orthologs. An example of the former are vasopressin/oxytocin (VP/OT)-759 type neuropeptides and precursors. An example of the latter are neuropeptide-S 760 (NPS)/crustacean cardioactive peptide (CCAP)-type neuropeptides and precursors, which are 761 paralogs of VP/OT-type neuropeptides and precursors (Semmens et al. 2015 with neuropeptides in protostomes and deuterostomes exhibiting modest sequence similarity. 766 The discovery of sNPF-type signalling in echinoderms has provided a unique 767 opportunity to speculate on the ancestral characteristics of this signalling system in 768 Urbilateria. It is noteworthy that, by comparison with the protostome sNPF-type peptides, the 769 echinoderm sNPF-type peptides have more features in common with the paralogous 770 NPY/NPF-type peptides. The echinoderm sNPF-type peptides are not as long as NPY/NPF-771 type peptides but they are nevertheless much longer than protostome sNPF-type peptides. 772 Furthermore, it was the sequence similarity that echinoderm peptides share with NPY/NPF-773 type peptides that originally facilitated their discovery (Zandawala et al., 2017). Additionally, 774 the structure of the echinoderm sNPF-type precursors is similar to NPY/NPF-type precursors 775 because the neuropeptide is located immediately after the signal peptide, whereas this is not a 776 feature of protostome sNPF-type precursors. Based on these observations, we propose that 777 echinoderm sNPF-type peptides and precursors may more closely resemble the ancestral 778 characteristics of this signalling system in Urbilateria. Furthermore, we speculate that the 779 common ancestor of the paralogous NPY/NPF-type and sNPF-type neuropeptide precursors 780 may have been similar to NPY/NPF-type precursors with respect peptide, precursor and gene 781 structure. Then, following gene duplication, these ancestral characteristics were retained in 782 the paralog that gave rise to the bilaterian NPY/NPF-type peptides/precursors. In contrast, the 783 paralog that gave rise to sNPF-type signalling diverged from the ancestral condition. 784 However, the extent of divergence varies in the echinoderm and protostome lineages. In 785 echinoderms, the sNPF-type peptides/precursors have many NPY/NPF-type characteristics 786 and we conclude that this reflects less divergence from the proposed ancestral condition. 787 Conversely, in the protostomes, the sNPF-type peptides/precursors exhibit little similarity 788 with NPY/NPF-type peptides/precursors and we conclude that this reflects more divergence 789 from the proposed ancestral condition. 790 Lastly, we need to consider more broadly the evolutionary history of sNPF-type 791 signalling in deuterostomes, and in particular the non-echinoderm phyla -the hemichordates 792 and hemichordate PrRP-type receptors do not clade with sNPF-type receptors. A limitation of 805 cluster analysis of receptor relationships is that it is based on pairwise comparisons that 806 cannot resolve paralogy/orthology relationships because speciation/duplication nodes are 807 not retrieved (Gabaldón 2008;Kim et al. 2008). Thus, the determination of deep homology 808 relationships is normally accomplished by generating phylogenetic trees (Gabaldón 2008 have the common characteristic of an intron that interrupts the coding sequence at a position 821 corresponding to their N-terminal or central regions. Furthermore, the intron consistently 822 interrupts the coding sequence in the same frame, at a position between the first and second 823 nucleotide of the interrupted codon, as denoted by +1 in Figure 7B. This contrasts with 824 NPY/NPF-type genes that have a highly conserved intron interrupting the coding sequence at 825 a position corresponding to the C-terminal region of the mature peptides, with the intron 826 located between the second and third nucleotide of the codon for a conserved arginine 827 residue, as denoted by -1 in Figure 2. Collectively, these findings are supportive of the 828 hypothesis that echinoderm sNPF-type neuropeptides are orthologs of vertebrate PrRP-type 829 neuropeptides and the novel PrRP-like peptides that we have identified here in the 830 hemichordate S. kowalevskii and the cephalochordate B. floridae. Furthermore, it is 831 noteworthy that orthologs of vertebrate PrRP-type receptors have been identified in 832 cephalochordates and hemichordates (Mirabeau and Joly 2013) (see also Figure 3). Thus, 833 there are mutually exclusive patterns in the phylogenetic distribution of sNPF-type receptors 834 and PrRP-type receptors, with the former found only in protostomes and echinoderms and the 835 latter found only in vertebrates, cephalochordates and hemichordates (Figure 8)