The seahorse genome and the evolution of its specialized morphology

Seahorses have a specialized morphology that includes a toothless tubular mouth, a body covered with bony plates, a male brood pouch, and the absence of caudal and pelvic fins. Here we report the sequencing and de novo assembly of the genome of the tiger tail seahorse, Hippocampus comes. Comparative genomic analysis identifies higher protein and nucleotide evolutionary rates in H. comes compared with other teleost fish genomes. We identified an astacin metalloprotease gene family that has undergone expansion and is highly expressed in the male brood pouch. We also find that the H. comes genome lacks enamel matrix protein-coding proline/glutamine-rich secretory calcium-binding phosphoprotein genes, which might have led to the loss of mineralized teeth. tbx4, a regulator of hindlimb development, is also not found in H. comes genome. Knockout of tbx4 in zebrafish showed a ‘pelvic fin-loss’ phenotype similar to that of seahorses. Supplementary information The online version of this article (doi:10.1038/nature20595) contains supplementary material, which is available to authorized users.

Members of the teleost family Syngnathidae (seahorses, pipefishes and seadragons) (Extended Data Fig. 1), comprising approximately 300 species, display a complex array of morphological innovations and reproductive behaviours. This includes specialized morphological phenotypes such as an elongated snout with a small terminal mouth, fused jaws, absent pelvic and caudal fins, and an extended body covered with an armour of bony plates instead of scales 1 (Fig. 1a). Syngnathids are also unique among vertebrates due to their 'male pregnancy' , whereby males nourish developing embryos in a brood pouch until hatching and parturition occurs 2,3 . In addition, members of the subfamily Hippocampinae (seahorses) exhibit other derived features such as the lack of a caudal fin, a characteristic prehensile tail, and a vertical body axis 4 (Fig. 1a). To understand the genetic basis of the specialized morphology and reproductive system of seahorses, we sequenced the genome of the tiger tail seahorse, H. comes, and carried out comparative genomic analyses with the genome sequences of other ray-finned fishes (Actinopterygii).

Genome assembly and annotation
The genome of a male H. comes individual was sequenced using the Illumina HiSeq 2000 platform. After filtering low-quality and duplicate reads, 132.13 Gb (approximately 190-fold coverage of the estimated 695 Mb genome) of reads from libraries with insert sizes ranging from 170 bp to 20 kb were retained for assembly. The filtered reads were assembled using SOAPdenovo (version 2.04) to yield a 501.6 Mb assembly with an N50 contig size and N50 scaffold size of 34.7 kb and 1.8 Mb, respectively. Total RNA from combined soft tissues of H. comes was sequenced using RNA-sequencing (RNA-seq) and assembled de novo. The H. comes genome assembly is of high quality, as > 99% of the de novo assembled transcripts (76,757 out of 77,040) could be mapped to the assembly; and 243 out of 248 core eukaryotic genes mapping approach (CEGMA) genes are complete in the assembly.
We predicted 23,458 genes in the genome of H. comes based on homology and by mapping the RNA-seq data of H. comes and a closely related species, the lined seahorse, Hippocampus erectus, to the genome assembly (see Methods and Supplementary Information). More than 97% of the predicted genes (22,941 genes) either have homologues in public databases (Swissprot, Trembl and the Kyoto Encyclopedia of Genes and Genomes (KEGG)) or are supported by assembled RNAseq transcripts. Analysis of gene family evolution using a maximum likelihood framework identified an expansion of 25 gene families (261 genes; 1.11%) and contraction of 54 families (96 genes; 0.41%) in the H. comes lineage (Extended Data Fig. 2 and Supplementary Tables 4.1, 4.2). Transposable elements comprise around 24.8% (124.5 Mb) of the H. comes genome, with class II DNA transposons being the most abundant class (9%; 45 Mb). Only one wave of transposable element expansion was identified, with no evidence for a recent transposable element burst (Kimura divergence ≤ 5) (Extended Data Fig. 3).

Phylogenomics and evolutionary rate
The phylogenetic relationships between H. comes and other teleosts were determined using a genome-wide set of 4,122 one-to-one orthologous genes (Supplementary Note 4.2). The phylogenetic analysis (Fig. 1b) showed that H. comes is a sister group to other percomorph fishes analysed (stickleback, Gasterosteus aculeatus; medaka, Oryzias latipes; Nile tilapia, Oreochromis niloticus; fugu, Takifugu rubripes; and Seahorses have a specialized morphology that includes a toothless tubular mouth, a body covered with bony plates, a male brood pouch, and the absence of caudal and pelvic fins. Here we report the sequencing and de novo assembly of the genome of the tiger tail seahorse, Hippocampus comes. Comparative genomic analysis identifies higher protein and nucleotide evolutionary rates in H. comes compared with other teleost fish genomes. We identified an astacin metalloprotease gene family that has undergone expansion and is highly expressed in the male brood pouch. We also find that the H. comes genome lacks enamel matrix protein-coding proline/glutamine-rich secretory calcium-binding phosphoprotein genes, which might have led to the loss of mineralized teeth. tbx4, a regulator of hindlimb development, is also not found in H. comes genome. Knockout of tbx4 in zebrafish showed a 'pelvic fin-loss' phenotype similar to that of seahorses. platyfish, Xiphophorus maculatus) with the exception of blue-spotted mudskipper (Boleophthalmus pectinirostris), a member of the family Gobiidae. Our inference, which placed the mudskipper as the outgroup, differs from that of a previous phylogenetic analysis based on fewer protein-coding genes that had placed syngnathids as an outgroup 5 . Estimated divergence times of H. comes and other teleosts calculated using MCMCTree suggest that H. comes diverged from the other percomorphs approximately 103.8 million years ago, during the Cretaceous period (Extended Data Fig. 2). Interestingly, the branch length of H. comes is longer than that of other teleosts, suggesting a higher protein evolutionary rate compared to other teleosts analysed in this study (Fig. 1b). This result was found to be statistically significant by both relative rate test 6 and two cluster analysis 7 (Supplementary Tables  4.3 and 4.4). To determine whether the neutral nucleotide substitution rate of H. comes is also higher, we generated a neutral tree on the basis of fourfold degenerate sites and calculated the pairwise distance of each teleost to the spotted gar (an outgroup) (Supplementary Fig. 4.4). The pairwise distance of H. comes was again higher compared with other teleosts, indicating that the neutral evolutionary rate of H. comes is also higher than that of other teleosts. The reasons for this higher molecular evolutionary rate in H. comes are unclear.

Gene loss
Gene loss or loss of function can contribute to evolutionary novelties and can be positively selected for 8,9 . We identified several genes that are not found in the H. comes genome but are found in other sequenced teleost genomes.
Secretory calcium-binding phosphoprotein (SCPP) genes encode extracellular matrix proteins that are involved in the formation of mineralized tissues such as bone, dentin, enamel and enameloid. Bony vertebrate genomes encode multiple SCPP genes that can be divided into two groups, the acidic and the proline/glutamine (P/Q)rich SCPP genes. Acidic SCPPs regulate the mineralization of collagen scaffolds in bone and dentin whereas the P/Q-rich SCPPs are primarily involved in enamel or enameloid formation 10 . Analysis of the H. comes genome and the transcriptomes of H. comes and H. erectus showed that both contain two acidic SCPP genes, scpp1 and spp1 (Extended Data Fig. 4). However, no intact P/Q-rich gene could be identified. The only P/Q-rich gene present in the H. comes genome assembly, scpp5, is represented by only three out of ten exons, indicating that it has become a pseudogene. Seahorses and pipefish (family Syngnathidae) are toothless, a phenomenon known as edentulism. Besides syngnathids, edentulism has occurred convergently in several other vertebrate lineages 11 , the most notable ones being birds 12 , turtles, and some mammals such as baleen whales, pangolins and anteaters 13 . The loss of teeth in birds, turtles and mammals has been attributed to inactivating mutations in one or more P/Q-rich enamel-specific SCPP genes such as Enam, Amel, Ambn and Amtn, and the dentin-specific gene, Dspp 12,14 . In the case of H. comes, the complete loss of functional P/Q-rich SCPP genes may explain the loss of mineralized teeth.
Animals use their sense of smell, or olfaction, for finding food, mates and avoiding predators. Olfaction is mediated by olfactory receptors (ORs), which constitute the largest family of G-proteincoupled receptors. We were able to identify in the H. comes genome a significantly smaller repertoire of OR genes than in other teleosts (P value < 0.05, Wilcoxon rank-sum test). Our sensitive search pipeline (based on TblastN and Genewise) and manual inspection identified only 26 OR genes in the H. comes genome-the smallest OR repertoire identified in any ray-finned fish genome analysed so far (60 to 169 OR genes) ( Fig. 2 and Extended Data Fig. 5).
A derived phenotype of seahorse and other syngnathids is the complete lack of pelvic fins 15,16 . Pelvic fins are homologous to tetrapod hindlimbs and primarily serve a role in body trim and subtle swimming manoeuvres during teleost locomotion 17-19 . In addition, pelvic spines have an important role in protection against predators 15 . Pelvic fin loss has occurred independently in several teleost lineages, including Tetraodontidae (for example, pufferfishes), Anguillidae (eels) and Gasterosteidae (some populations of sticklebacks), and is frequently associated with a reduced pressure from predators and/or the evolution of an elongated body plan 15 . In pufferfish (fugu), pelvic fin loss is associated with a change in the expression pattern of hoxd9a 20 . In freshwater populations of stickleback, the loss of pelvic fins has been demonstrated to be due to deletions in the pelvic fin-specific enhancer of pitx1 (ref. 21).
Analysis of the H. comes genome and the transcriptomes of H. comes and H. erectus (see Supplementary Information, section 2), suggested that tbx4, a transcription factor conserved in jawed vertebrates, is not present in the seahorse genome ( Fig. 3a) (Supplementary Information, section 9). To verify this, we carried out degenerate polymerase chain reaction (PCR) using genomic DNA from H. comes and several other species of syngnathids and some non-syngnathids. While the degenerate primers amplified a fragment of tbx4 from non-syngnathids, they failed to amplify a tbx4 fragment from syngnathid fishes (see Supplementary Information, section 9). Tbx4 is a T-box DNAbinding domain-containing transcription factor that acts as a regulator of hindlimb formation in mammals 22-24 . Loss of function of this gene in mouse leads to a failure of hindlimb formation 22,23 as well as strong pleiotropic defects in lung 25 and placental development 22 . Expression of zebrafish tbx4 specifically in pelvic fins suggests a similar role in appendage patterning in fishes 24 . Given the major role of tbx4 in   hindlimb formation in mammals, we hypothesized that its absence in H. comes might be associated with the loss of pelvic fins. To test this hypothesis, we generated a CRISPR-Cas9 tbx4-knockout mutant zebrafish line. Interestingly, unlike homozygous mouse Tbx4 mutants, which fail to develop a functional allantois 22 , the homozygous zebrafish mutants are viable but completely lack pelvic fins without exhibiting any other gross morphological abnormalities in pectoral or median fins ( Fig. 3c and Extended Data Fig. 6; see also Supplementary Information, section 9.3, in particular Supplementary Fig. 9.6 for additional phenotype analysis). This finding is consistent with the results of a recent study that showed that mutations in tbx4 are associated with the loss of pelvic fins in a naturally occurring zebrafish strain called pelvic finless 26 (see also Supplementary Information, section 9.3). These results show that tbx4 has a role in pelvic fin formation in teleosts and suggests that the loss of pelvic fins in H. comes may be related to the loss of tbx4.

Expansion of the patristacin gene family
Male pregnancy is an evolutionary innovation unique to syngnathids. In teleosts, the C6AST subfamily of astacin metalloproteases-such as high choriolytic enzyme (HCE) and low choriolytic enzyme (LCE)-are involved in lysing the chorion surrounding the egg, leading to hatching of embryos 27 . A member of this subfamily, patristacin (pastn), was found to be highly expressed in the brood pouch of pregnant males of the Gulf pipefish, Syngnathus scovelli, leading to the suggestion that this gene may have a role in the evolution of male pregnancy 28 . A pastn gene was also found to be highly expressed in the brood pouch of the male big belly seahorse, H. abdominalis, during mid-and late pregnancy 29 , suggesting a shared role for this gene in male pregnancy in syngnathids.
The H. comes genome contains six pastn genes (pastn1 to pastn6; Fig. 4a) organized in a cluster. To examine their expression patterns in the brood pouch, we carried out RNA-seq analysis at different stages of brood pouch development (see Supplementary Information Fig. 2). Quantitative reverse transcription PCR (qRT-PCR) analysis of these genes showed that some of them are expressed at significantly higher levels in early-and late-pregnant stages (Fig. 4c). For example, pastn2 is expressed at significantly higher levels in early-and late-pregnant stages compared to the non-pregnant stage, whereas pastn1 and pastn3 are expressed at significantly higher levels during the late-pregnant stage compared to non-pregnant stage (Fig. 4c). This expression pattern suggests a role for these pastn genes in brood pouch development and/ or hatching of embryos within the brood pouch prior to parturition.
Interestingly, the platyfish (X. maculatus), in which fertilization and hatching of eggs occur within the maternal body (ovoviviparity), contains a cluster of six c6ast genes (Fig. 4a), with potential hatching enzyme-like activity 30 . Phylogenetic analysis of c6ast family genes in H. comes, platyfish and other fishes showed that H. comes pastn genes and platyfish c6ast genes form separate clades (Fig. 4b), indicating that they have expanded independently in the two lineages. Thus, this is an interesting instance of a gene family (C6AST subfamily of astacin metalloproteases) that has undergone expansion independently in different teleost lineages and shows new expression patterns and functions associated with similar evolutionary innovations (that is, ovoviviparity in female platyfish and male pregnancy in seahorse).

Loss of conserved noncoding elements
Vertebrate genomes contain thousands of noncoding elements that are under purifying selection 31-33 . Many of these conserved noncoding elements (CNEs) function as cis-regulatory elements such as enhancers, repressors and insulators 34,35 . Evolutionary loss of CNEs has important roles in phenotypic differences and morphological innovations 21,36,37 .
To determine the extent of loss of CNEs in seahorse, we predicted genome-wide CNEs in H. comes and four other percomorph fishes (stickleback, fugu, medaka and Nile tilapia) using zebrafish as the reference genome (see Supplementary Information). We identified 239,976 CNEs (average size of 168 bp) that are conserved in zebrafish and at least one of the five percomorph fishes (Supplementary Table 6.1).
To determine the extent to which CNEs are lost in H. comes, we searched for CNEs that are uniquely lost in each of the percomorph fishes. We restricted our analyses to a high-confidence set of CNEs situated in gap-free syntenic intervals (Supplementary  Analysis of zebrafish CNEs that are lost in H. comes indicated that they are present in the neighbourhood of 728 genes enriched in functions such as regulation of transcription, regulation of the fibroblast growth factor receptor signalling pathway, embryonic pectoral fin morphogenesis, steroid hormone receptor activity and O-acetyltransferase activity (Supplementary Tables 6.8 and 6.9). The top 20 genes adjacent to regions with the highest number of CNEs lost in H. comes include sall1a, shox and irx5a (Supplementary Tables 6.10 and 6.11), which are involved in the development of the limbs, nervous system, kidney, heart and skeletal system. Altered expression patterns of these genes can potentially lead to altered morphological phenotypes. For example, loss of regulatory regions of the human SHOX gene is the cause of Leri-Weill dyschondrosteosis, a dominantly inherited skeletal dysplasia that is characterized by moderate short stature caused by short mesomelic limb segments 38,39 .
To verify the potential cis-regulatory functions of CNEs that were absent in H. comes but present in other teleost genomes, we assayed the function of seven selected zebrafish CNEs that were uniquely absent in H. comes. Of the seven CNEs assayed in transgenic zebrafish, four CNEs drove reproducible patterns of reporter gene expression in F1 embryos (Extended Data Fig. 7 and Supplementary Table 6.12). Thus, our transgenic assay indicates that some of the CNEs absent in H. comes may function as cis-regulatory elements in other teleosts. Further studies are required to examine whether the loss of CNEs may have played a role in the evolution of seahorse morphology.

Summary
Seahorses possess one of the most highly specialized morphologies and reproductive behaviours. We sequenced the genome of the tiger tail seahorse and performed comparative analysis with other teleost fishes. Our genome-wide analysis highlights several aspects that may have contributed to the highly specialized body plan and male pregnancy of seahorses. These include a higher protein and nucleotide evolutionary rate, loss of genes and expansion of gene families, with duplicated genes exhibiting new expression patterns, and loss of a selection of potential cis-regulatory elements. It is becoming recognized that evolutionary changes in cis-regulatory elements, particularly the loss and gain of enhancers, might play a major part in the evolution of morphological innovations and phenotypic changes across species 21,36,37,40 .
Male pregnancy is a unique developmental feature of seahorses and pipefishes (family Syngnathidae, comprising 57 genera and approximately 300 species). In the seahorse genome, the astacin subfamily of c6ast metalloprotease genes has undergone tandem duplications giving rise to six genes. This subfamily of metalloprotease includes the hatching enzyme (also known as choriolysin), HCE-like and HCE2-like enzymes that are responsible for hatching of embryos in fishes 27 . Of the six duplicated genes in seahorse, five are highly expressed in the male brood pouch, suggesting that they may be involved in male pregnancy, possibly through rewiring of their regulatory network. The loss of pelvic fins in seahorse is associated with the evolution of an armour-like covering of its body and gain of an  elongated, flexible, substrate-gripping tail. By combining comparative genomics and gene-knockout experiments in zebrafish, we suggest that loss of tbx4 may have a role in this phenotype in seahorse. The loss of mineralized teeth in seahorse is associated with the fusion of the jaws into a tube-like snout and a small mouth, which is extremely efficient in sucking small food items that are abundant in the benthic environment. In teleosts, P/Q-rich SCPP genes are involved in the mineralization of enameloid, which is the equivalent of enamel in tetrapods 10 . The seahorse genome does not contain any intact P/Q-rich SCPP genes that code for enamel matrix proteins, suggesting that the loss of these genes could have played a part in the loss of its mineralized teeth. Our analyses of the H. comes genome sequence and comparative genomics with other teleosts highlighted several genetic changes that may be involved in the evolution of the unique morphology of seahorses.
Online Content Methods, along with any additional Extended Data display items and Source Data, are available in the online version of the paper; references unique to these sections appear only in the online paper. (Promega, Madison, WI) and sequenced to identify carrier fish transmitting a frameshift mutation. These carrier fish were crossed again to AB wild type and the resulting F1 fish were raised to maturity. The F1 were genotyped using fin clipping, DNA isolation, PCR, T7 endonuclease to identify heterozygous mutant fish followed by cloning and sequencing of the mutant PCR products to validate presence of the frameshift allele. The CRISPR-Cas9 mutation strategy is schematically shown in Extended Data Fig. 5.
In the F0 mutant tbx4 fish we observed pelvic fin loss at low frequency. gRNA#1 gave 3/42 fish with either double-or single-sided pelvic fin loss whereas 1/34 had single-sided pelvic fin loss for gRNA#2 (Extended Data Fig. 5). We observed mutant allele transmission for both gRNA#1 and gRNA#2 but failed to identify a deletion leading to a frameshift mutation for gRNA#2 so no stable line was generated for this CRISPR. For gRNA#1 we identified several frameshift mutants, one of which was further analysed. This mutant has a deletion/replacement mutation in which eight nucleotides are replaced by three nucleotides, leading to an effective 5 bp deletion and the introduction of a frameshift mutation (Extended Data Fig. 5). This mutation introduces a downstream STOP codon leading to a severely truncated protein lacking the DNA binding domain ( Supplementary  Figs 9.4 and 9.5). The mutant line is maintained on an AB wild-type background. Loss of CNEs. Using zebrafish as the reference genome, whole-genome alignments of six teleost fishes were generated. The soft-masked genome sequence for zebrafish (Zv9, April 2010) was downloaded from the Ensembl release-75 FTP site. The following soft-masked genome sequences were downloaded from the UCSC Only chromosome sequences of zebrafish were aligned while unplaced scaffolds were excluded. The reference (zebrafish) genome was split into 21 Mb sequences with 10-kb overlap, while the percomorph fish genomes (H. comes, stickleback, fugu, medaka and Nile tilapia) were split into 10 Mb sequences with no overlap. Pairwise alignments were carried out using Lastz v.1.03. 54 (ref. 55) with the following parameters: -strand = both-seed = 12of19-notransition-chain-gappedgap = 400,30-hspthresh = 3000-gappedthresh = 3000-inner = 2000-masking = 50ydrop = 9400-scores = HoxD55.q-format = axt. Coordinates of split sequences were restored to genome coordinates using an in-house Perl script. The alignments were reduced to single coverage with respect to the reference genome using UCSC Genome Browser tools 'axtChain' and 'chainNet' . Multiple alignments were generated using Multiz.v11.2/roast.v3 (ref. 56) with the tree topology "(Zv9 (hipCom0 ((fr3 gasAcu1) (oryLat2 oreNil2))))".
Fourfold degenerate (4D) sites of zebrafish genes (Ensembl release-75) were extracted from the multiple alignments. These 4D sites were used to build a neutral model using PhyloFit in the rphast v.1.5 package 57 (general reversible "REV" substitution model). PhastCons was then run in rho-estimation mode on each of the zebrafish chromosomal alignments to obtain a conserved model for each chromosome. These conserved models were averaged into one model using PhyloBoot. Subsequently, conserved elements were predicted in the multiple alignments using PhastCons with the following inputs and parameters: the neutral and conserved models, target coverage of input alignments = 0.3 and average length of conserved sequence = 45 bp. To assess the sensitivity of this approach in identifying functional elements, the PhastCons elements were compared against zebrafish protein-coding genes. Eighty per cent of protein-coding exons (197,508/245,556 exons) were overlapped by a conserved element (minimum coverage 10%), indicating that the identification method was fairly sensitive.
A CNE was considered present in a percomorph genome if it showed coverage of at least 30% with a zebrafish CNE in Multiz alignment. To identify CNEs that could have been missed in the Multiz alignments due to rearrangements in the genomes, or due to partitioning of the CNEs among teleost fish duplicate genes, we searched the zebrafish CNEs against the genome of the percomorph using BLASTN (E < 1 × 10 −10 ; ≥ 80% identity; ≥ 30% coverage). Those CNEs that had no significant match in a percomorph genome were considered as missing in that genome. To account for CNEs that might have been missed due to sequencing gaps, we identified gap-free syntenic intervals in zebrafish and the percomorph genomes, and generated a set of CNEs that were missing from these intervals. These CNEs represent a high-confidence set of CNEs missing in the percomorph fishes and thus were used for further analysis. Functional enrichment of genes associated with CNEs was carried out using the GREAT software 58 with each CNE assigned to the genes with the nearest transcription start site and within 1 Mb in the zebrafish genome, and significantly enriched functional categories identified based on a hypergeometric test of genomic regions (false discovery rate (FDR) q value < 0.05). We identified the statistically significant gene ontology biological process terms, molecular function terms and zebrafish phenotype descriptions of the genes that are associated with CNEs.
We also predicted CNEs in the Hox clusters of H. comes and other representative teleost fishes using the global alignment program MLAGAN. Orthologous Hox clusters were aligned using MLAGAN with zebrafish as the reference sequence and CNEs were predicted using VISTA. Functional assay of CNEs. Seven representative zebrafish CNEs that have been lost in H. comes (the largest among the lost CNEs) were assayed for enhancer activity in transgenic zebrafish using GFP as the reporter gene. The CNEs were amplified by PCR using zebrafish genomic DNA as template. The products were cloned into a miniTol2 transposon donor plasmid linked to the mouse cFos (McFos) basal promoter and the coding sequence of GFP. Transposase mRNA was generated by transcribing cDNA in vitro using the mMESSAGE mMACHINE T7 kit (Ambion; Life Technologies). The CNE-containing McFos-miniTol2 construct and transposase mRNA were co-injected into the yolk of zebrafish embryos at the one to two-cell stage. Each CNE construct was injected into 250-350 embryos and the injections were repeated on two days. The embryos were reared at 28 °C, and GFP was observed at 24, 48 and 72 h post-fertilization (hpf). The survival rate of the embryos post-injection was 70-80%. Consistent GFP expression in at least 20% of F0 embryos was considered as specific expression driven by a CNE. Such embryos were reared to maturity and mated with wild type zebrafish to produce F1 lines. The expression of GFP in F1 embryos was observed under a compound microscope fitted for epifluorescence (Axio imager M2; Carl Zeiss, Germany) and photographed using an attached digital microscope camera (Axiocam; Carl Zeiss, Germany). Pigmentation was inhibited by maintaining zebrafish embryos in 0.003% N-phenylthiourea (Sigma-Aldrich, Sweden) from 8 hpf onwards. Consistent GFP expression observed in at least three lines of F1 fishes was considered as the specific expression driven by a CNE.
All animals were cared for in strict accordance with National Institutes of Health (USA) guidelines. The zebrafish gene knockout protocol was approved by the Institutional Animal Care and Use Committee of Sun Yat-Sen University. The zebrafish transgenic assay protocol was approved by the Institutional Animal Care and Use Committee of Biological Resource Centre, A* STAR, Singapore. Data availability statement. The tiger tail seahorse (H. comes) whole-genome sequence has been deposited in the DDBJ/EMBL/GenBank database under accession number LVHJ00000000. RNA-seq reads for H. erectus and H. comes have been deposited in the NCBI Sequence Read Archive under accession numbers SRA392578 and SRA392580, respectively.