Identification of jellyfish neuropeptides that act directly as oocyte maturation-inducing hormones

Fully-grown oocytes maintained within the female gonad are held at first prophase of meiosis until environmental and/or physiological signals initiate cell cycle resumption and oocyte maturation, culminating in release of fertilisation-competent eggs. This process of oocyte maturation is a key feature of animal biology, and is tightly regulated to optimise reproductive success. It involves biochemical cascades activated within the oocyte that are highly conserved across animal phyla, notably involving the kinases Cdk1 (to achieve entry into first meiotic M phase) and MAP kinase (to orchestrate polar body formation and cytostatic arrest) (Amiel et al., 2009; Von Stetina and Orr-Weaver, 2011; Tachibana et al., 2000; Yamashita et al., 2000). These kinase regulations have been well characterised using biochemically tractable model species, notably frogs and starfish, and knowledge extended using genetic methods to other species, including nematodes, Drosophila and mammals. Nevertheless, information is largely lacking on certain critical steps and, in particular, the initial triggering of these cascades in response to the maturation-inducing hormones (MIHs), which act locally in the gonad on their receptors in the ovarian oocytes; the only known examples identified at the molecular level are 1-methyladenine released in starfish (Kanatani et al., 1969), steroid hormones in amphibians and fish (Haccard et al., 2012; Nagahama and Yamashita, 2008), and a sperm protein in Caenorhabditis (Von Stetina and Orr-Weaver, 2011).

Hydrozoan jellyfish provide excellent models for dissecting the molecular and cellular mechanisms regulating oocyte maturation, which in these animals is triggered by light-dark and/or dark-light transitions. Remarkably, oocyte growth, maturation and release continue to function autonomously in gonads isolated from female jellyfish, implying that all the regulatory components connecting light sensing to spawning are contained within the gonad itself (Amiel et al., 2010; Freeman, 1987; Ikegami et al., 1978). Furthermore, as members of the Cnidaria, a sister clade to the Bilateria, hydrozoan jellyfish can provide insight into spawning regulation in early animal ancestors.

In this study we addressed the molecular nature and the cellular origin of MIH in two hydrozoan jellyfish model species, Clytia hemisphaerica (Fig. 1A) and Cladonema pacificum (Fig. 1B), which are induced to spawn by dark-light and light-dark transitions, respectively (Amiel et al., 2010; Deguchi et al., 2005; Houliston et al., 2010). Starting from the hypothesis that hydrozoan MIHs might be neuropeptides, consistent with size filtration and protease-sensitivity experiments (Ikegami et al., 1978), we screened synthetic candidate peptides predicted from gonad transcriptome data by treatment of isolated gonads (spawning assay) or isolated oocytes (MIH assay). We then raised inhibitory antibodies to confirm the presence and activity of the putative peptides in native MIH secreted from gonads in response to light/dark cues, and to characterise the MIH-producing cells and their response to light. We extended our findings by determining the activity of the identified hydrozoan MIH tetrapeptides on males as well as females, and on a selection of other diverse hydrozoan species. A parallel study of Clytia gonad light detection revealed that the light-mediated MIH release reported here is dependent on an opsin photopigment co-expressed in the same population of cells that secretes MIH (Quiroga Artigas et al., 2018). These specialised cells, which have neural-type morphology and characteristics, thus provide a simple and possibly ancestral mechanism to promote synchronous gamete maturation, release and fertilisation.

First we demonstrated that true MIH activity can be recovered from small drops of seawater containing isolated ovaries of either Clytia or Cladonema following the appropriate light transition, as demonstrated previously using other hydrozoan species (Freeman, 1987; Ikegami et al., 1978). Isolated oocytes recovered using this method and incubated in endogenous MIH efficiently complete the meiotic maturation process, which is manifest visually by germinal vesicle breakdown (GVBD) and extrusion of two polar bodies (Fig. 1C,D).

Further characterisation of MIH using Cladonema showed that isolated gonad ectoderm, but not endoderm, tissue (see Fig. 1B) could produce active MIH. MIH activity from Cladonema gonad ectoderm resisted heat treatment at 100°C for 20 min (95% GVBD, n=41), several freeze/thaw cycles (100% GVBD, n=14) and to filtration through a 3000 MW cut-off membrane (90% GVBD, n=18), consistent with the idea that the active molecule is a small, possibly peptidic, molecule (Ikegami et al., 1978).

Cnidarians, including jellyfish, hydra and sea anemones, express many low molecular weight neuropeptides with various bioactivities (Anctil, 2000; Fujisawa, 2008; Takahashi et al., 2008; Takeda et al., 2013). These are synthesised by cleavage of precursor polypeptides and can produce multiple copies of one or more peptides, which are frequently subject to amidation by conversion of a C-terminal glycine (Grimmelikhuijzen et al., 1996). A previous study found that some synthetic Hydra amidated peptides can stimulate spawning when applied to gonads of the jellyfish Cytaeis uchidae, the most active being members of the GLWamide family (active at 10⁻⁵ M minimum concentration) (Takeda et al., 2013). Crucially, however, these did not induce meiotic maturation when applied to isolated oocytes, i.e. they did not meet the defining criterion of MIHs. These previous results were not conclusive because of the use of species-heterologous peptides, but suggest that although jellyfish GLWamide peptides do not act as MIHs, they might be involved less directly in spawning regulation.

To identify endogenous species-specific neuropeptides as candidates for MIH from our model species, we first retrieved sequences for ten potential amidated peptide precursors from a mixed stage Clytia transcriptome (Fig. S1), and then searched for those specifically expressed in the ectoderm, the source of MIH, by evaluating the number of corresponding Illumina HiSeq reads obtained from manually separated ectoderm, endoderm and oocyte gonad tissues (Fig. 2A). In the ectoderm, only three putative neuropeptide precursor mRNAs were expressed above background levels, as confirmed by quantitative PCR (Fig. S2). One was a GLWamide precursor, Che-pp11, expressed at moderate levels. Much more highly expressed were Che-pp1 and Che-pp4, both predicted to generate multiple related short (3-6 amino acid) amidated peptides with the C-terminal signature (W or R)-PRP, -PRA -PRG or -PRY. Potential precursors for both GLWamide (Cpa-pp3) and PRP/Aamides (Cpa-pp1 and Cpa-pp2) were also present among four sequences identified in a transcriptome assembly from the Cladonema manubrium (which includes the gonad; Fig. 1B). Cpa-pp1 contains one copy of the RPRP motif, while Cpa-pp2 contains multiple copies of RPRA motifs (Fig. S1).

As a first screen to select neuropeptides potentially involved in regulating oocyte maturation, we incubated Clytia female gonads in synthetic tetrapeptides predicted from Che-pp1, Che-pp4 and Che-pp11 precursors at 10⁻⁵ M or 10⁻⁷ M (Fig. 2B). We uncovered preferential and potent activity for the WPRPamide and WPRAamide tetrapeptides, which consistently provoked oocyte maturation and release from the gonad at 10⁻⁷ M, while the related RPRGamide and RPRYamide were also active but only at 10⁻⁵ M. RPRPamide, a predicted product of Cpa-pp1 and also of Che-pp8, a precursor not expressed in the Clytia gonad, was also active in this screen at 10⁻⁵ M. By contrast, PGLWamide, which is potentially generated from Che-pp11, did not affect the gonads at either concentration. This result placed WPRP/Aamide-related peptides as the best candidates for jellyfish MIH.

We then performed a direct MIH activity assay, i.e. treatment of isolated oocytes with the candidate peptides. For Clytia oocytes we detected potent MIH activity (as assessed by oocyte GVBD; Fig. 2C, Fig. S3) for W/RPRP/Aamide and RPRY/Gamide tetrapeptides, but not for PGLWamide. RPRAamide was more active in triggering GVBD when added to isolated oocytes than to intact gonads, perhaps because of poor permeability through the gonad ectoderm. For Cladonema oocytes, RPRP/Aamides showed very potent MIH activity, and the RPRG/Yamides were also active at higher concentrations, whereas WPRP/Aamides were not active (Fig. 2C, Fig. S3). We also tested, on oocytes of both species, pentapeptides and hexapeptides that might theoretically be generated from the Che-pp1 and Cpa-pp1 precursors, but these had much lower MIH activity than the tetrapeptides, while the tripeptide PRPamide and tetrapeptides lacking amidation were inactive (Fig. 2D, Fig. S3). The response of Clytia or Cladonema isolated oocytes to synthetic W/RPRP/A/Yamides mirrored very closely that elicited by endogenous MIH, proceeding through the events of oocyte maturation with normal timing following GVBD, and advanced by 15-20 min (Clytia) or 10 min (Cladonema) compared with light/dark-induced spawning of gonads (Fig. S4A-D). The resultant mature eggs could be fertilised and developed into normal planula larvae (Fig. S4E,F).

We demonstrated that W/RPRPamide and/or W/RPRAamide peptides are responsible for endogenous MIH activity in Clytia and Cladonema by use of inhibitory affinity-purified antibodies generated to recognise the PRPamide and PRAamide motifs (as determined by ELISA assay; Fig. 3A,B). These antibodies were able to inhibit specifically the MIH activity of the targeted peptides (Fig. 3C,D). Conclusively, pre-incubation of endogenous MIH obtained from Clytia or Cladonema gonads with anti-PRPamide antibody for 30 min completely blocked its ability to induce GVBD in isolated oocytes. Pre-incubation with the anti-PRAamide antibody slightly reduced MIH activity but not significantly compared with a control IgG (Fig. 3E).

Taken together, these experiments demonstrate that WPRPamide and RPRPamide are the active components of endogenous MIH in Clytia and Cladonema, respectively, responsible for triggering oocyte meiotic maturation. Other related peptides, including RPRYamide, RPRGamide, WPRAamide (Clytia) and RPRAamide (Cladonema) also probably contribute to MIH. These peptides almost certainly act at the oocyte surface rather than intracellularly, since fluorescent (TAMRA-labelled) WPRPamide microinjected into Clytia oocytes, unlike externally applied TAMRA-WPRPamide, did not induce GVBD (Fig. 3F).

Single- and double-fluorescence in situ hybridisation showed that the Clytia MIH precursors Che-pp1 and Che-pp4 are co-expressed in a distinctive population of scattered cells in the gonad ectoderm in males and females (Fig. 4A,B, Fig. 5E). Similarly, in Cladonema the predicted RPRPamide precursor Cpa-pp1 was expressed in scattered cells in the manubrium ectoderm, which covers the female or male germ cells (Fig. 4A, Fig. 5A,E). Immunofluorescence with the anti-PRPamide and anti-PRAamide antibodies in both species revealed that the expressing cells have a morphology typical of cnidarian neural cells, comprising a small cell body and two or more long projections (David, 1973), and characterised by the presence of bundles of stable microtubules (Fig. 4C,F). A parallel study further revealed that in Clytia, these MIH-secreting cells express an opsin photoprotein with an essential function in oocyte maturation and spawning (Quiroga Artigas et al., 2018). Given their neural-type morphology, photosensory function and key role in regulating sexual reproduction via neuropeptide hormone production, we propose that these cells have both a sensory and neurosecretory nature. Scattered endocrine cells with both sensory and neurosecretory characteristics are a feature of cnidarians (Hartenstein, 2006). Furthermore, the distribution and organisation of the gonad MIH-producing cells in Clytia and Cladonema are suggestive of the neural nets that characterise cnidarian nervous systems (Koizumi, 2016; Bosch et al., 2017; Dupre and Yuste, 2017). We could not confirm from our immunofluorescence analyses any direct connections between neighbouring cells. Future electron microscopy or calcium imaging techniques (Gründer and Assmann, 2015; Dupre and Yuste, 2017) to identify synapses or action potential transmission could resolve whether these scattered endocrine cells are integrated within a neural network. In intact Clytia jellyfish, both immunofluorescence and in situ hybridisation (Fig. 5B-D, Fig. S5A) revealed the presence of MIH peptides and their precursors at additional sites associated with neural systems: the manubrium (mouth), tentacles and the nerve ring that runs around the bell rim (Koizumi et al., 2015), as well as along the radial canals. This suggests that PRPamide family neuropeptides have other functions in the jellyfish in addition to regulating spawning. Neuropeptides can have both neural and endocrine functions in cnidarians, and are thought to have functioned in epithelial cells in primitive metazoans prior to nervous system evolution (Hartenstein, 2006; Bosch et al., 2017).

In Clytia gonad ectoderm, the anti-PRPamide and PRAamide antibodies decorated a single cell population, whereas in Cladonema the two peptides were detected in distinct cell populations (Fig. S5B,C), presumably being generated from the Cpa-pp1 and Cpa-pp2 precursors, respectively (Fig. S1). Immunofluorescence analysis of Cladonema gonads revealed a reduction in the anti-PRPamide signal within 20 min after darkness, whereas the anti-PRAamide signal was relatively unaffected (Fig. 4D). Clytia gonads showed a moderate reduction of staining with both antibodies 45 and 120 min after light stimulus (Fig. 4E). More detailed examination indicated that in each stained cell the numbers of antibody-positive dots, presumably representing peptide-filled vesicles, decreased following light exposure in both the cell body and in the microtubule-rich projections (Fig. 4F). It would be interesting to determine, for instance by live imaging techniques, whether MIH-containing vesicles are secreted from particular sites or exhibit any trafficking following the light/dark cues, or whether vesicle release occurs throughout the cell, as would be typical of a neurosecretory cell type (Hartenstein, 2006).

The similar distribution of MIH-producing cells in female and male gonads (Fig. 4A, Fig. 5E) suggests that these neuropeptides might play a general role in regulating gamete release, and not only in the initiation of oocyte maturation in female medusae. We found using male jellyfish of both Clytia and Cladonema that synthetic MIH peptides at 10⁻⁷ M provoked release of active sperm from the gonads (Table 1). This confirms that the oocyte maturation-stimulating effect of MIH is just one component of a wider role in reproductive regulation. It also raises the intriguing possibility that MIH neuropeptides released into the seawater from males and females gathered together at the ocean surface during spawning might facilitate precise synchronisation of gamete release during the periods of dawn and dusk.

Our experiments revealed some selectivity in the MIH activity of different peptides between Clytia and Cladonema. The most potent MIH peptides for Clytia oocytes were the main Che-pp1/Che-pp4-derived tetrapeptides WPRPamide, WPRAamide and RPRYamide, which were clearly active even at 10⁻⁸ M (Fig. S3). The best candidate for Cladonema MIH is RPRPamide (from Cpa-pp1), while RPRAamide (from Cpa-pp2) was slightly less active (Fig. S3). Correspondingly, the RPRP sequence is not found in precursors expressed in the Clytia gonad, while WPRP/Aamides are not predicted from any Cladonema precursors (see Fig. 2A, Fig. S1).

Further testing on oocytes from eight other hydrozoan jellyfish species revealed responsiveness with different sensitivities to W/RPRP/A/G/Yamide-type tetrapeptides in Obelia, Aequorea, Bouillonactinia and Sarsia, but not Eutonina, Nemopsis, Rathkea or Cytaeis (Fig. 6). The responsive and non-responsive species included members of two main hydrozoan groups, namely the Leptomedusae and Anthomedusae. These comparisons suggest that W/RPRXamide-type peptides functioned as MIHs in ancestral hydrozoan jellyfish. We can speculate that variation in the peptide sequences that are active between related species might reduce stimulation of spawning between species in mixed wild populations.

We have demonstrated that in the hydrozoan jellyfish Clytia and Cladonema, short amidated peptides with the prototype sequence W/RPRPamide are responsible for inducing oocyte maturation, resulting ultimately in release from the gonad of active mature gametes (Fig. 7A). These peptides act as bona fide MIHs, i.e. they interact directly with the surface of resting ovarian oocytes to initiate maturation. Related W/RPRXamide peptides act as MIHs also in other hydrozoan jellyfish species. Regarding the later events of the spawning process, we hypothesise that egg release through the gonad ectoderm in female medusae is not triggered directly by MIH but via a secondary signal emitted by the oocyte towards the end of the meiotic maturation process. More specifically, it could depend on Mos-MAP kinase activation in the oocyte, since maturation in the absence of spawning can occur when Mos translation is inhibited in ovarian oocytes (Amiel et al., 2009). In male gonads, MIH peptides presumably act on inactive, postmeiotic spermatozoids to initiate the spawning response, but the mechanisms involved are not yet known.

Some GLWamide family peptides are also able to provoke oocyte maturation and spawning, albeit at higher concentrations than the PRPamides, but do not induce maturation of isolated oocytes (this study; Takeda et al., 2013), suggesting that the role of these peptides in regulating spawning is indirect. We can imagine that inhibitory or sensitising factors, possibly including GLWamides, could act either in the gonad MIH-secreting cells or in other ectodermal cells to modulate the light response, and account for species-specific dawn or dusk spawning. It remains to be seen whether regulation of spawning by MIH neuropeptides related to those in Clytia and Cladonema extends beyond hydrozoan jellyfish to other cnidarians. If so, further layers of regulation could allow the integration of seasonal cues and lunar cycles to account for the well-known mass annual spawning events seen in tropical reef corals (Harrison et al., 1984).

The identification of MIH in Clytia and Cladonema is a significant step forward in the oocyte maturation field because the molecular nature of the hormones that trigger oocyte maturation is known in surprisingly few animal species, notably 1-methyladenine in starfish and steroid hormones in teleost fish and amphibians (Kanatani et al., 1969; Nagahama and Yamashita, 2008; Yamashita et al., 2000; Haccard et al., 2012). The very different molecular natures of these known MIH examples from across the (bilaterian plus cnidarian) clade could be explained by an evolutionary scenario in which secretion of neuropeptide MIHs from cells positioned near to the oocyte was the ancestral condition, with intermediate regulatory tissues, such as endocrine organs and ovarian follicle cells, evolving in the deuterostome lineage to separate neuropeptide-based regulation from the final response of the oocyte (Fig. 7B). Such interpolation of additional layers of regulation is a common feature of endocrine system evolution (Hartenstein, 2006). During the evolution of reproductive regulation, various neuropeptides, including vertebrate gonadotropin-releasing hormones (GnRHs) (Roch et al., 2011) as well as modulatory RFamide peptides such as Kisspeptins and gonadotropin-inhibitory hormone (GnIH) (Parhar et al., 2012), regulate various aspects of reproduction including gamete release in both males and females. Chordate GnRHs are PGamide decapeptides, which stimulate the release of peptidic gonadotropic hormones (GTHs), such as vertebrate luteinizing hormone (LH), from the pituitary. Similarly, starfish gonad-stimulating substance (GSS/Relaxin) (Mita et al., 2009) is a GTH produced at a distant ‘neuroendocrine' site, the radial nerve. In both cases, these peptidic GTHs in turn cause oocyte maturation by inducing MIH release from the surrounding follicle cells, or, in the case of mammals, GAP junction-mediated exchange of cyclic nucleotides between these cells (Shuhaibar et al., 2015). Regulation of reproduction by GnRHs probably predated the divergence of deuterostomes and protostomes (Roch et al., 2011; Tsai, 2006), the best evidence coming from mollusc species in which peptides structurally related to GnRH, and synthesised at various neuroendocrine sites, regulate various reproductive processes (Osada and Treen, 2013).

Cnidarians use neuropeptides to regulate multiple processes including muscle contraction, neural differentiation and metamorphosis from larva to polyp (Anctil, 2009; Fujisawa, 2008; Takahashi and Hatta, 2011). Transcript sequences predicted to produce many copies of short neuropeptides have also been found in ctenophore and placozoan genomes (Moroz et al., 2014; Nikitin, 2015), and neuropeptides are thought to have been the predominant neurotransmitters in the ancient common ancestor of these groups (Grimmelikhuijzen and Hauser, 2012). Although independent evolution of neuropeptide regulation or reproduction between animal clades cannot be ruled out, the identification of the MIH neuropeptides in Clytia and Cladonema, along with other evidence from cnidarians (Takeda et al., 2013; Tremblay et al., 2004) as well as bilaterians (see above), suggests that neuropeptide signalling played a central role in coordinating sexual reproduction in the bilaterian-cnidarian ancestor, and might have been involved in coordinating spawning events in the marine environment. In Clytia medusae we found cells producing PRPamide family peptides not only in the gonad but also in the manubrium, tentacles and bell margin (Fig. 5C,D), so these presumably have wider functions beyond orchestrating gamete release. It will be of great interest to investigate the activities of related peptides across a wide range of species in order to track the evolutionary history of the neuroendocrine regulation of reproduction.

Laboratory strains of Clytia hemisphaerica (‘Z colonies'), Cladonema pacificum (6W, NON5, UN2) and Cytaeis uchidae (17) were maintained throughout the year (Deguchi et al., 2005; Houliston et al., 2010; Takeda et al., 2006). Wild specimens of Cladonema as well as Eutonina indicans, Nemopsis dofleini, Obelia sp., Rathkea octopunctata and Sarsia tubulosa were collected from Sendai Bay, Miyagi Prefecture, and Aequorea coerulescens and Bouillonactinia misakiensis from Mutsu Bay, Aomori Prefecture. The brand of artificial seawater (ASW) used for culture and for functional assays in Japan was SEA LIFE (Marine Tech, Tokyo), and for Clytia hemisphaerica culture, transcriptomics and microscopy in France was Red Sea Salt.

Fully-grown oocytes were obtained from ovaries of intact jellyfish or pre-isolated ovaries placed under constant illumination for 20-24 h following the previous spawning. Ovarian oocytes were aspirated using a mouth pipette or detached using fine tungsten needles. During oocyte isolation, jellyfish were in some cases anaesthetised in excess Mg²⁺ ASW (a 1:1 mix of 0.53 M MgCl₂ and ASW). Pre-isolated ovaries of Clytia, Aequorea and Eutonina were bathed in ASW containing 1 mM sodium citrate to facilitate the detachment of oocytes from ovarian tissues.

Active MIH was recovered from cultured ovaries of Clytia and Cladonema by a similar approach to that used previously (Freeman, 1987). A chamber formed between a plastic dish and a coverslip separated by two pieces of 400 or 500 µm-thick double-sided sticky tape was filled with silicon oil (10 cSt; TSF451-10, Momentive Performance Materials), and a drop of ASW (0.5-1 µl) containing several ovaries separated from two Clytia jellyfish or several ovarian epithelium fragments stripped from three to five Cladonema jellyfish was inserted into the oil space (Fig. 1A,B). The oil chambers were subjected to light-dark changes (light after dark for Clytia, and dark after light for Cladonema) and the ASW with MIH activity was collected 60 min later. Prior to MIH assays, isolated oocytes were cultured in seawater for at least 30 min and any oocytes showing damage or GVBD were discarded. MIH assays were performed at 14-16°C for Eutonina, Nemopsis, Obelia, Sarsia and Rathkea or at 18-21°C for Clytia, Cladonema, Cytaeis and Bouillonactinia.

Potential amidated peptide precursor sequences were recovered from a Clytia reference transcriptome derived from mixed larva, polyp and jellyfish samples. Open reading frames (ORFs) and protein sequences were predicted using an R script (Lapébie et al., 2014). Potential secreted proteins were identified by the presence of signal peptide using SignalP 4.0 (Petersen et al., 2011). Then, sequences rich in the amidated pro-peptide cleavage motifs GR/K and lacking domains recognised by InterProScan-5.14-53.0 (EMBL-EBI) were selected. Finally, sequences containing repetitive motifs of fewer than 20 amino acids were identified using TRUST (Szklarczyk and Heringa, 2004). Among this final set of putative peptide precursors, some known secreted proteins with repetitive structures were identified by BLAST (NCBI) and removed.

To prepare a Cladonema transcriptome, more than 10 µg total RNA was isolated from the manubrium of female jellyfish (6W strain) using the NucleoSpin RNA purification kit (MACHEREY-NAGEL). RNA-seq library preparation and sequencing (Illumina HiSeq 2000) were carried out by BGI (Hong Kong, China). Using an assembled dataset containing 74,711 contigs and 35,957 unigenes, local BLAST searches were performed to find peptide precursors using published cnidarian neuropeptide sequences or the Clytia pp1 and pp4 sequences as bait.

The ORFs of putative candidate Clytia and Cladonema peptide precursors were cloned by PCR into the pGEM-T Easy vector (Promega), or retrieved from our Clytia EST collection cDNA library prior to probe synthesis. Sequences and accession numbers are given in Fig S1.

For Clytia gonad tissue transcriptome comparisons, Illumina HiSeq 50 nt reads were generated from mRNA isolated using the RNAqueous Micro Kit (Ambion Life Technologies) from ectoderm, endoderm and oocytes manually dissected from ∼150 Clytia female gonads. Quantitative PCR was performed to check for contamination between samples using endogenous GFP genes expressed in oocyte, ectoderm and bell tissue (Fourrage et al., 2014), and to quantify the expression of selected peptide precursors (see Fig. S2B for primers). The reads were mapped against a Clytia reference transcriptome using Bowtie 2 (Langmead and Salzberg, 2012). The counts for each contig were normalised per total reads of each sample and per sequence length and visualised using the heatmap.2 function in the gplots R package.

WPRP-NH₂, WPRA-NH₂, RPRP-NH₂, RPRA-NH₂, RPRG-NH₂, RPRY-NH₂, PGLW-NH₂, DAWPRP-NH₂, AWPRP-NH₂, FNIRPRP-NH₂, NIRPRP-NH₂, IRPRP-NH₂, PRP-NH₂, WPRP-OH and RPRP-OH were synthesised by GenScript or Life Technologies. These peptides were dissolved in deionised water at 10⁻² M or 2×10⁻³ M, stored at −20°C, and diluted in ASW at 10⁻⁵-10⁻¹⁰ M prior to use. TAMRA-WPRPamide (TAMRA-LEKRNWPRP-NH₂) was synthesised by Sigma and a 5×10⁻⁴ M solution in H₂O was injected at 2-17% of the oocyte volume, to give an estimated final oocyte concentration of 1-9×10⁻⁵ M (Deguchi et al., 2005).

Polyclonal antibodies against XPRPamide and XPRAamide were raised in rabbits using keyhole limpet hemocyanin (KLH)-conjugated CPRA-NH₂ and CPRP-NH₂ as antigens, and antigen-specific affinity purified (Sigma-Aldrich). For MIH inhibition experiments, antibodies or control normal rabbit IgG (MBL, MP035) were concentrated using a 30,000 MW cut-off membrane (Millipore), giving a final protein concentration of 10⁻⁶ M, and the buffers were replaced with seawater through repeated centrifugations.

For single or double anti-PRPamide/anti-PRAamide staining, specimens were anesthetised using excess Mg²⁺ ASW and fixed overnight at 4°C in 10% formalin-containing ASW, then rinsed three times for 10 min each in phosphate buffered saline (PBS) containing 0.25% Triton X-100 (PBS-Triton). They were incubated in anti-PRPamide or anti-PRAamide antibody diluted 1/1000-1/10,000 in PBS-Triton overnight at 4°C. After washes in PBS-Triton, the specimens were incubated with secondary antibody (Alexa Fluor 488 or 568 goat anti-rabbit IgG, both 1/1000; Invitrogen, A-11034, A-11036) for 2 h at room temperature and nuclei stained using 50 µM Hoechst 33258 or 33342 (Invitrogen) for 5-20 min. Zenon antibody labelling kits (Molecular Probes, Z-25313, Z-25302, Z-25306) were used for double peptide staining. In control experiments, PBS-Triton alone or normal rabbit IgG (3 mg/ml; Zymed, 100500C) diluted 1/1000 in PBS-Triton replaced the anti-PRPamide or anti-PRAamide antibodies. Images were acquired using a laser scanning confocal system (C1, Nikon).

For co-staining of neuropeptides and microtubules (Fig. 4C,D), dissected Clytia gonads were fixed overnight at 18°C in HEM buffer (0.1 M HEPES pH 6.9, 50 mM EGTA, 10 mM MgSO₄) containing 3.7% formaldehyde, then washed five times in PBS containing 0.1% Tween 20 (PBS-T). Treatment on ice with 50% methanol in PBS-T, then 100% methanol, plus storage in methanol at −20°C improved visualisation of microtubules in the MIH-producing cells. Samples were rehydrated, washed in PBS containing 0.02% Triton X-100, blocked in PBS with 3% BSA overnight at 4°C, then incubated in anti-PRPamide antibody and anti-alpha tubulin (YL1/2, Sigma-Aldrich, 92092402; 1/500) in PBS with 3% BSA at room temperature for 2 h. After washes, the specimens were incubated with secondary antibodies (Rhodamine goat anti-rabbit and Cy5 donkey anti-rat IgG, both 1/100; Jackson ImmunoResearch, 111-025-003, 712-175-153) overnight in PBS at 4°C and nuclei stained using Hoechst 33258 for 20 min.

For in situ hybridisation, isolated gonads or whole jellyfish were processed as described previously (Fourrage et al., 2014) except that 4 M urea was used instead of 50% formamide in the hybridisation buffer (Sinigaglia et al., 2017). For double fluorescent in situ hybridisation, female Clytia gonads were fixed overnight at 18°C in HEM buffer containing 3.7% formaldehyde, washed five times in PBS-T, then dehydrated on ice using 50% methanol/PBS-T then 100% methanol. In situ hybridisation (Lapébie et al., 2014; Sinigaglia et al., 2017) was performed using a DIG-labelled probe for Che-pp1 and a fluorescein-labelled probe for Che-pp4. A 3 h incubation with a peroxidase-labelled anti-DIG antibody was followed by washes in MABT (100 mM maleic acid pH 7.5, 150 mM NaCl, 0.1% Triton X-100). For Che-pp1, the fluorescence signal was developed using the TSA Plus Fluorescence Amplification Kit (PerkinElmer) and Cy3 fluorophore [1/400 in TSA buffer (PBS with 0.0015% H₂O₂)] at room temperature for 30 min. After three washes in PBS-T, fluorescence was quenched with 0.01 M HCl for 10 min at room temperature, and washed again several times in PBS-T. Overnight incubation with a peroxidase-labelled anti-fluorescein antibody was followed by washes in MABT. The anti-Che-pp4 fluorescence signal was developed using the TSA kit with Cy5 fluorophore. Nuclei were stained using Hoechst 33258. Images were acquired using a Leica SP5 confocal microscope and maximum intensity projections of z-stacks prepared using ImageJ software (NIH).

We thank P. Dru, S. Chevalier and L. Leclère for generating and assembling the Clytia reference transcriptome; A. Ruggiero and C. Sinigaglia for sharing in situ hybridisation protocols; S. Yaguchi for advice on immunofluorescence; J. Uveira for technical assistance; and our group members, ‘Neptune' network colleagues, Clare Hudson and Hitoyoshi Yasuo for useful discussions.