Integrative multi-omics analysis reveals the contribution of neoVTX genes to venom diversity of Synanceia verrucosa Authors: Zhiwei Zhang, Qian Li, Hao Li, Shichao Wei, Wen Yu, Zhaojie Peng, Fuwen Wei, Wenliang Zhou Published: 2024 ABSTRACT BackgroundAnimal venom systems are considered as valuable model for investigating the molecular mechanisms underlying phenotypic evolution. Stonefish are the most venomous and dangerous fish because of severe human envenomation and occasionally fatalities, whereas the genomic background of their venom has not been fully explored compared with that in other venomous animals.ResultsIn this study, we followed modern venomic pipelines to decode the Synanceia verrucosa venom components. A catalog of 478 toxin genes was annotated based on our assembled chromosome-level genome. Integrative analysis of the high-quality genome, the transcriptome of the venom gland, and the proteome of crude venom revealed mechanisms underlying the venom complexity in S. verrucosa. Six tandem-duplicated neoVTX subunit genes were identified as the major source for the neoVTX protein production. Further isoform sequencing revealed massive alternative splicing events with a total of 411 isoforms demonstrated by the six genes, which further contributed to the venom diversity. We then characterized 12 dominantly expressed toxin genes in the venom gland, and 11 of which were evidenced to produce the venom protein components, with the neoVTX proteins as the most abundant. Other major venom proteins included a presumed CRVP, Kuntiz-type serine protease inhibitor, calglandulin protein, and hyaluronidase. Besides, a few of highly abundant non-toxin proteins were also characterized and they were hypothesized to function in housekeeping or hemostasis maintaining roles in the venom gland. Notably, gastrotropin like non-toxin proteins were the second highest abundant proteins in the venom, which have not been reported in other venomous animals and contribute to the unique venom properties of S. verrucosa.ConclusionsThe results identified the major venom composition of S. verrucosa, and highlighted the contribution of neoVTX genes to the diversity of venom composition through tandem-duplication and alternative splicing. The diverse neoVTX proteins in the venom as lethal particles are important for understanding the adaptive evolution of S. verrucosa. Further functional studies are encouraged to exploit the venom components of S. verrucosa for pharmaceutical innovation.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12864-024-11149-6. Background Venom is one of the most successful convergent evolutionary novelties across all major animal lineages for defense, predation, and competition [1]. Animal venom systems are considered as valuable models for investigating the molecular mechanisms underlying phenotypic evolution, and are widely used for exploring and engineering biologically active molecules for pharmaceutical innovation [2]. Modern venomics involves genetics, evolution, ecology as well as translational studies, facilitating our understanding of the venomous animals including scorpions, snakes and spiders [3, 4]. Most of the animal venoms are highly complexed mixtures of molecules with modifications at the genomic, transcriptomic, and protein level. Recently, the emerging chromosomal level genomes have greatly facilitated venom toxin screening and gene identification in snake [5–7], spider [8, 9], and other venomous animals [10–13]. Large-scale comparative genomic analyses provide evolutionary views of toxin genes, which include co-option of pre-existing nontoxic genes, massive tandem duplication and domain loss [14–17]. Despite the varying venom composition and apparatus spanning venomous metazoans, a comparative analysis of 20 venom gland transcriptomes from wasps, spiders, fish and snakes revealed that a conserved genetic toolkit was recruited to evolve the ability to secret toxins [18]. On the other hand, full-length transcriptomic analysis revealed that more than half of the venom genes exhibited multiple isoforms within the parasitoid wasp venom gland, suggesting that alternative splicing highly contributes to venom diversity and evolution [19]. In the future, integrative genomic, transcriptomic and proteomic approaches will contribute to novel toxin discovery and the elucidation of venom evolution. Approximately 2500 fish species are venomous across the world, whereas most of their venom are left as under investigation and valuable unexplored resource [20]. The stonefish (Scorpaeniformes, Synanceiidae, Synanceia) are recognized as the most venomous and dangerous fish because of severe human envenomation and occasional fatalities [21, 22]. There are five species of stonefish, including S. verrucosa, S. horrida, S. alula, S. nana, and S. platyrhyncha. Along with the development of venomic approaches, attempts have been devoted to decode the stonefish venom composition and functional activities [23] The inceptive research in stonefish venom was predominantly reported in S. horrida, with the first fish-derived protein toxin characterized as stonustoxin (SNTX), which has inspired the venom research in the teleost [24, 25]. Venomous fish usually use their venom as defensive weapon against native predator by causing excruciating pain. Structure analysis showed that SNTX subunits belongs to the perforin superfamily of pore-forming immune effectors, which accounts for its hemolytic activity [26]. As the most widely distributed stonefish species throughout the world, S. verrucosa demonstrated great potential in understanding venom pathology and toxin gene evolution in fish [27, 28]. The lethal components were purified in the venom in the early 1990s, including a tetramer protein VTX and a dimer neoVTX. Homologous to SNTX, VTX/neoVTX demonstrated similar hemolytic, hypotensive, and cardiotoxic activities [29–31]. Recently, a pathophysiological study showed that the cardiotoxicity of the S. verrucosa venom is in part due to binding to the voltage-gated Ca2+ channel, indicating a multifunctional activity of VTX/neoVTX as fish cytolysin [32, 33]. Besides, two widely abundant components in the animal venom, hyaluronidase [34] and lectin [35], have also been characterized in S. verrucosa venom. Studies to date have been limited to single particle-level purification and molecular cloning, leaving most of the venom components remained uncharacterized, which limits our understanding of how venom exerts its poisoning action and how the different components contribute to the physiological effects of envenomation. Two criteria have been widely adopted to identify major functional venom components in animals: (1) genes should exhibit highly reliable expression in the venom gland; (2) highly expressed genes should be further characterized in the venom proteome. In this study, we assembled the chromosomal level genome of S. verrucosa with a set of toxin genes preliminarily annotated. The major venom protein components were then deciphered and characterized through a comprehensive analysis of genomic, transcriptomic, and proteomic data. Notably, we revealed a mechanism of tandem-duplicated neoVTX genes that underwent massive alternative splicing, which underlines the venom diversity and evolution in S. verrucosa. Methods Animals Adult S. verrucosa fish were purchased from a local fishery market and maintained in the laboratory aquarium system for at least 2 weeks before sampling. The fish were anesthetized using MS222 (General reagent, China) before dissection. Muscle, skin, and venom glands were dissected and rinsed with autoclaved artificial seawater, followed by flash-freezing in liquid nitrogen. The samples were then stored at − 80 °C. All the experiments were conducted in accordance with guidelines approved by the Animal Ethical and Welfare Committee of Southern Marine Science and Engineering Guangdong Laboratory (Guangzhou). Genomic DNA extraction and sequencing Genomic DNA was extracted from the muscle using the DNeasy Blood and Tissue Kit (Qiagen, USA). The integrity, purity, and concentration of the extracted DNA were analyzed by gel electrophoresis and NanoDrop™ (Thermo Fisher Scientific, USA). The DNA concentration was further quantified using a Qubit® 4.0 Fluorometer (Invitrogen, USA). A total of 15 µg DNA was then fragmented for HiFi sequencing library construction and the library was then sequenced using the PacBio Sequel II series sequencer (Pacific Biosciences, CA, USA). Whole genome sequencing (WGS) was included to facilitate genome assembly. Briefly, a library with a 300-bp inserted size was prepared using genomic DNA isolated from muscle following the protocol of the MGIEasy FS DNA Library Prep Kit (MGI Tech, China) and sequenced on the DNBSEQ-T7 platform (MGI Tech, China). For Hi-C sequencing, the fresh muscle tissue was fixed with 2% formaldehyde to enable cross-linking between genomic DNA and chromatin. The crosslinked DNA was then extracted, and a library with a 150-bp inserted size was constructed. Sequencing was performed using the DNBSEQ T7 platform (MGI Tech, China). RNA extraction and sequencing Total RNA was isolated from the muscle, skin, and venom glands (n = 7) using TRIzol reagent (Thermo Fisher Scientific, USA). The concentration of the RNA was analyzed by NanoDrop™ and a Bioanalyzer 2000 (Agilent, USA) was used to assess integrity. For next-generation sequencing (NGS), the cDNA libraries were constructed following the manufacturer’s instructions and sequenced using the MGI DNBSEQ-T7 sequencing platform. Venom gland full-length transcriptome sequencing was performed using the ISO-Seq PacBio platform (Pacific Biosciences, CA, USA). Genome assembly and annotation To generate a chromosomal level genome of S. verrucosa, a high-quality contig genome assembly was generated using Hifiasm v0.16.1 [36], followed by redundant sequence removal using Purge_dups v1.2.5 [37]. The assembled contigs were then aligned to the Hi-C sequencing data of S. verrucosa with default parameters using chromap [38], followed by genome assembly with default parameters using the yahs pipeline [39]. Finally, the completeness of the genome assembly was assessed using BUSCO v5.4.7 with genome mode and referring to the database of actinopterygii_odb10 [40]. Repetitive sequences in the genome were screened using RepeatModeler v2.0 [41]. The S. verrucosa genome sequences were then mapped against de novo repeat libraries and the Repbase TE library v16.02 to identify transposable elements (TEs) using RepeatMasker v4.1.1 [42], and TE-related proteins were further examined using RepeatProteinMask v.4.0.6 [43]. Besides, Tandem Repeat Finder (TRF) v.4.07 was used to search for tandem repeat sequences [44]. Homology-based, transcriptome-based, and de novo prediction were used to annotate the genes in the genome, with the combined use of EvidenceModeler (EVM) v1.1.1 and PASA v2.4.0 [45]. For homology-based predictions, the protein sequences of Danio rerio, Bufo gargarizans, Cyclopterus lumpus, Pseudoliparis swirei, Sebastes umbrosus, Gasterosteus aculeatus, Scatophagus argus, Thalassophryne amazonica, Siniperca chuatsi, Oncorhynchus mykiss, Oryzias latipes, Salmo salar, Oreochromis niloticus, Electrophorus electricus, Ostracion cubicus, and Acanthochromis polyacanthus were obtained from the NCBI and aligned to the assembled S. verrucosa genome [46]. For transcriptome-based prediction, RNA-seq data were aligned to the genome with default parameters using Hisat v2.1.0 [47]. Subsequently, transcripts were reconstructed with default parameters using StringTie v2.0 [48]. For de novo gene predictions, Augustus v3.4.0, Genscan v3.1, and GlimmerHMM v3.0.1 were used to analyze the repeat-masked genome [49–51]. Functional annotation of the genes was performed using Blastp and Blastn (-e 1e-5) to blast protein or nucleotide sequences against various databases including the NCBI nonredundant protein (NR), the NCBI nonredundant nucleotide sequence (NT), as well as the SwissProt databases. Gene Ontology (GO) terms were then retrieved and assigned to the S. verrucosa query sequences, and enzyme codes (EC) corresponding to GO terms were mapped to KEGG pathway annotations for further functional insights. Long-read RNA alignment and isoform identification The raw reads obtained from the full-length transcriptome sequence were analyzed following the Iso-Seq3 pipeline (v3.1). Briefly, circular consensus sequences were created from the raw subreads, and the cDNA primers were removed using lima. Full-length reads were then filtered, clustered, and polished (Iso-Seq3, v3.1). The polished sequences were aligned to the S. verrucosa genome using pbmm2 (v1.13.0). Isoform identification was done using TAMA [52]. Redundant isoforms were removed by comparison with the raw data of the NGS transcriptome data. An in-depth characterization of the isoforms and artifact removal was done using SQANTI3 [53]. Briefly, isoform characterization was performed using sqanti_qc.py, and filtering was done using sqanti_filter.py. The filtered data set was rerun through sqanti_qc.py for the final confirmation. The SQANTI-filtered isoforms were used for the remaining analyses. Expression analysis of the genes in the venom gland HISAT2 v2.1.0 was used to align the quality-controlled RNA-seq reads to the S. verrucosa genome. The aligned reads were assembled into transcripts, and expression was quantified using feature Counts [54]. Differential expression analysis was performed using DESeq2 v1.26.0 to identify differentially expressed genes in each tissue type [55]. Genes with an adjusted p-value < 0.05 and a log2(fold change) > 1 were considered to be differentially expressed. Visualization of differentially expressed toxin genes across all samples was performed using the pheatmap package v1.0.12 in R. Transcript per million (TPM) values was used to assess the expression level among the venom glands, muscles, and skin of S. verrucosa. Toxin genes annotation Because of the propensity of toxin gene families to be organized in extensive tandem arrays, the number of paralogs in particular gene families may be compromised, thereby impeding the accuracy of gene annotation [56]. Consequently, additional annotation steps were included in the present study. Combining the ToxProt-based annotation with the isoforms generated from the third-generation sequencing transcriptome of the venom gland, we performed empirical annotation using FGENESH (http://www.softberry.com) as well as manual annotation using IGV-GSAman (https://gitee.com/CJchen/IGV-sRNA). Finally, each re-annotated gene was searched against the ToxProt database to confirm its identity and further confirmed its homology through a phylogenetic analysis. We also downloaded HMM models from the Pfam database v35.0 [57], and then blasted the annotated toxin genes set using hmmsearch in HMMER v3.1b1 (http://hmmer.org/). Sequences with E-values below the default inclusion threshold (E-value < 1e-05) were retrieved and identified as toxin homologs. Proteomic analysis of the venom Lyophilized venom (100 µg, n = 4) was dissolved in a triethylammonium bicarbonate buffer. Next, 10 mM of Tris (2-carboxyethyl) phosphine was added and the venom solution was incubated at 37 °C for 60 min to reduce the disulfide bonds in the proteins. Subsequently, 40 mM of iodoacetamide was added to block the sulfhydryl groups of the cysteine residues to prevent the reformation of disulfide bond reformation. Followed by the addition of pancreatic trypsin at an enzyme-to-protein ratio of 1:50, the venom was digested overnight at 37 °C. The resulting peptides were then desalted using HLB and quantified (Thermo Fisher Scientific). The samples were dissolved in sample loading buffer (2% acetonitrile, 0.1% formic acid) at a concentration of 0.25 µg/µL for mass spectrometry analysis. The samples were separated by high-performance liquid chromatography and analyzed using a timsTOF Pro2 mass spectrometer (Bruker, Germany) in data-dependent acquisition (DDA) mode. The ionization mode was set to positive, with an ion source voltage of 1.5 kV, and both MS and MS/MS data were acquired using time-of-flight analysis. The mass spectrometry scanning range was set to 100–1700 m/z. The data acquisition mode used parallel accumulation serial fragmentation (PASEF), in which 10 rounds of PASEF were performed after each full MS scan. Second-level spectra with charge states ranging from 0 to 5 were acquired and a dynamic exclusion time of 24 s was set for tandem mass spectrometry scans to avoid repeated ion selection. A local Scorpaeniformes venom protein database was established. This database included toxin gene set from S. verrucosa we annotated and the toxin protein of S. horrida in the Tox-Prot database. The venom proteomics data was searched against this local database using MaxQuant v2.4.2.0, and the results were filtered based on a parameter Peptide FDR ≤ 0.01. Protein abundance was analyzed using intensity-based absolute quantification (iBAQ) method. iBAQ value is calculated by dividing the sum of peak intensities of all peptides matching to a specific protein by the number of theoretically observable peptides [58]. For native PAGE electrophoresis, the lyophilized venom was dissolved in a phosphate buffer solution (10 mg/mL) and centrifuged at 5,000 g for 15 min at room temperature to remove insoluble material and primarily cellular debris. The venom (50 µg) was then subjected to preparative native-PAGE (6%) gels after diluting with Tris-glycine buffer (0.05 M, pH 8.0). The gels were run at a constant voltage of 120 V at 4 °C. Gel strips were stained with Coomassie brilliant blue R-250 for 30 min and subsequently destained using water. The protein bands of interest were cut and sent for mass spectrometry analysis (Sangon Biotech, China). Results Chromosome-scale de novo assembly and toxin genes annotation A high quality genome greatly facilitates the genetic decoding of the components of animal venom. In this study, a total of 19.80 Gb HiFi and 154.59 Gb Hi-C library sequencing data were generated respectively, resulting in an assembled genome size of 677.90 Mb for S. verrucose (Fig. 1A). The assembled genome was further evaluated and the results indicated that a high-quality of S. verrucose genome was obtained in this study (Fig. 1B and Supplementary Fig. 1). Fig. 1Toxin genes in the genome of S. verrucosa a) Visualized circle plot of chromosomal level genome assembly of S. verrucosa, with N ratio, GC ratio, repeat content and gene density shown from outside to inside, corresponding to each pseudochromosomes. b) Summary of the genome assembly evaluation. c) Protein families of all the annotated toxin genes. d) Tandem-duplicated SNTX/VTX genes clustered together with sequence divergence Toxin genes in the genome of S. verrucosa a) Visualized circle plot of chromosomal level genome assembly of S. verrucosa, with N ratio, GC ratio, repeat content and gene density shown from outside to inside, corresponding to each pseudochromosomes. b) Summary of the genome assembly evaluation. c) Protein families of all the annotated toxin genes. d) Tandem-duplicated SNTX/VTX genes clustered together with sequence divergence The toxin gene identification was performed using a combined analysis with the assembled genome and venom gland transcriptomic data. A total of 478 potential toxin genes were preliminarily identified at the whole-genome level. Most of the genes were clustered into 56 gene families, with the peptidase S1 (102), venom metalloproteinase (36), the SNTX/VTX toxin (32), the type-B carboxylesterase/lipase (19), the ficolin lectin (18), the true venom lectin (18), peptidase A1 (17), and venom Kunitz-type (17) as the most abundant toxin gene families (Fig. 1C, Supplementary Data 1). Further GO annotation revealed that two-thirds (321 out of 476) of the genes exhibit explicit toxin activity (GO:0090729). The other 155 annotated genes were assigned to other functional categories, including calcium channel regulator activity (GO:0005246), killing of cells of another organism (GO:0031640), hyalurononglucosaminidase activity (GO:0004415), and serine-type endopeptidase inhibitor activity (GO:0004867) (Supplementary Fig. 2). Toxin genes usually demonstrated remarkable tandem duplication events during evolution. In the present study, a total of 13 toxin genes were characterized as tandem repeated genes in S. verrucose genome, including the SNTX/VTX genes, true venom lectin genes, peptidase S1 genes, glycosyl hydrolase 56 genes and etc. In particular, eight annotated SNTX/VTX genes were clustered on chromosome 14 (Fig. 1D). The front 6 SNTX/VTX genes showed high sequence homology (> 85%) with neovtxa (GenBank: AB262392.1) and neovtxb (GenBank: AB262393.1) (Fig. 1D and Supplementary Fig. 3). The six genes were arranged in the following order: neovtx-b1, neovtx-b2, neovtx-a1, neovtx-a2, neovtx-b3, neovtx-a3. Interestingly, the other two SNTX/VTX genes (neovtx-like1 and neovtx-like2) clustered in this region shared a much lower sequence identity (< 50%) and clustered into a separate branch of the phylogenetic tree, whereas an InterPro scan revealed that they all contained similar conserved domains, including a Stonustoxin helical domain (IPR048997), Thioredoxin_11 domain (IPR040581), PRY domain (IPR006574) and SPRY domain (IPR003877) (Fig. 1D). Toxin genes expression in the venom gland NGS of transcriptome was performed using venom gland (n = 7) to further characterize the toxin genes. A total of 14, 380 genes were considered as actively expressed in the venom gland with a mean TPM > 1. Further analysis revealed that 240 genes corresponded to the previous annotated toxin genes set (Supplementary Data 1). We characterized a number of dominantly expressed toxin genes (DETGs) with a TPM > 500 in the venom gland. Twelve toxin genes were designated DETGs, including one cysteine rich venom protein (CRVP) gene (sv_crvp1), the six tandem-repeated neoVTX subunit genes on chromosome 14, one hyaluronidase gene (sv_hyal1), one translationally-controlled tumor protein gene (sv_tpt1), two calglandulin protein genes (sv_caglp1 and sv_caglp2), and one Kunitz-type serine protease inhibitor gene (sv_kspi1) (Fig. 2A, B). The 12 DETGs accounted for 93.70% of the totally expressed toxin genes. Moreover, we also examined the highly expressed non-toxin genes in the venom gland, which may be pivotal to regulate the toxin gene expression and/or produce important non-toxic components in the venom. The top ten non-toxin genes expressed in the venom gland were shown in Fig. 2B, including sv_fabp6, sv_tmsb4, sv_eef1a, sv_gpx1, sv_rpl39, sv_clu, sv_rpl26, sv_cstb, sv_ybx1 and sv_krt14. Interestingly, four unannotated genes with a TPM > 500 were identified in the venom gland, suggesting the presence of unique genes in S. verrucosa not previously found in other venomous animals (Fig. 2B). Fig. 2Transcriptomic analysis of the venom gland a) A heatmap of the highest 50 toxin genes transcription with hierarchical clustering in the venom gland among all the 7 samples. The color represented the expression level based on TPM value. b) Categorizes of the dominantly expressed toxin genes (DETGs) with TPM > 500, the highest 10 non-toxin genes, and 5 unannotated genes with TPM > 500. The value was shown as mean ± SD, n = 7. c) Volcano plots showing the differentially expressed genes (DEGs) between the venom gland and the skin tissue, with log2 (fold change) thresholds of -1 and 1, and P < 0.05. d) GO enrichment bubble plot analysis of the DEGs from non-toxin genes set (with TPM > 500 in VG) between the VG and SK. The bubble size indicated the gene number, the color represented the adjusted p value. BP, biological process; CC, cellular components; MF, molecular function Transcriptomic analysis of the venom gland a) A heatmap of the highest 50 toxin genes transcription with hierarchical clustering in the venom gland among all the 7 samples. The color represented the expression level based on TPM value. b) Categorizes of the dominantly expressed toxin genes (DETGs) with TPM > 500, the highest 10 non-toxin genes, and 5 unannotated genes with TPM > 500. The value was shown as mean ± SD, n = 7. c) Volcano plots showing the differentially expressed genes (DEGs) between the venom gland and the skin tissue, with log2 (fold change) thresholds of -1 and 1, and P < 0.05. d) GO enrichment bubble plot analysis of the DEGs from non-toxin genes set (with TPM > 500 in VG) between the VG and SK. The bubble size indicated the gene number, the color represented the adjusted p value. BP, biological process; CC, cellular components; MF, molecular function A comparative analysis was further performed between the venom gland (VG) and its closely related anatomic tissues, including skin (SK) and muscle (MU). The results indicated that the VG exhibited distinct expression profiles, and a higher transcriptome similarity with the SK compared with MU (Supplementary Fig. 4). The results indicated that skin may serve as the potential evolutionary origin of VG in S. verrucosa. Differentially expressed genes (DEGs) in the VG compared with other tissues generally indicated their preferred contributing roles to the venom. A total of 2,700 genes showed significantly higher expression in the VG (log2 Fold Change > 1 and padj < 0.05) compared with the SK. By blasting with the actively expressed 240 toxin genes, we obtained a total of 87 VG preferentially expressed toxin genes including ten of the 12 previously characterized DETGs. The six neoVTX subunit genes exhibited the highest differential expression level (Fig. 2. C and Supplementary Data 2). The other two DTEGs, the sv_tpt1 and sv_caglp2 (Sver.16G.00747), exhibited high expression levels in VG, as well as in SK, indicating their potential non-toxic roles outside the VG. We further examined the DEGs of the non-toxin genes set (with TPM > 500 in VG) between the VG and SK. A total of 84 genes were characterized and predominantly showed the molecular function of enzyme inhibitor, antioxidant and peroxidase activities, and involved in the biological processes of detoxification, metabolism and biosynthesis (Fig. 2D and Supplementary Data 3). They may to be partially involved in the resistance of S. verrucosa to its own venom. Proteomic evidence of the toxin genes in venom As the major components of the venom, the proteins encoded by the previously identified genes were further characterized by examining the venom proteome. A 4D-label free proteomic approach was employed to analyze the venom protein components of S. verrucosa. A total of 662 proteins were obtained, and 33 matched the previously annotated toxin genes (Supplementary Data 4). We characterized the major venom proteins into two categories: toxin proteins and non-toxin proteins. A quantitative analysis showed that the toxin proteins accounted for 43.19% of the total venom proteins (Fig. 3A). Next, we analyzed the proteomic data for the 12 DETGs. As shown in Fig. 3B, the most abundant toxin proteins were helothermine like protein (SvCRVP) from sv_crvp1 (22.35%). The neoVTX proteins accounted for approximately 17.21% of the total venom protein. Consistent with the VG transcriptomic data, almost all of the neoVTX proteins were derived from the six tandem-repeated neoVTX subunit genes. Of these, neoVTX a2 (50.55%) and neoVTX b3 (47.04%) accounted for most of the neoVTX proteins, and only a small proportion was contributed by the other four genes, indicating their dominant roles in protein production (Fig. 3C). Interestingly, native-PAGE analysis showed that two major protein bands with molecular weights of approximately 160 and 420 kDa, and both contained neoVTX proteins by mass spectrometry (Fig. 3D). Proteins from the other DETGs, including Kunitz-type serine protease inhibitors (SvKSPI) from sv_kspi1 (1.39%), the calglandulin-like protein (SvCAGL, 0.45%), and hyaluronidase (0.44%) were also evident in the venom proteome (Fig. 3B). The above 11 DTEGs accounted for 96.91% of all the designated toxin proteins and served as the major toxin genes in S. verrucose (Fig. 3E). The only DETGs with no proteomic evidence in the venom was tpt1, suggesting a major role in the VG rather than as venom components. Fig. 3Proteomic analysis of the crude venom a) Pie charts illustrating the composition of toxin and non-toxin proteins. b) Protein abundance of the major toxins and non-toxins, n = 7. c) Pie charts showing the contribution of each neoVTX subunit genes at transcriptomic and proteomic level. d) Native SDS-PAGE of lyophilized crude venom. Lane 1, protein ladders; lane 2, lyophilized crude venom. The asterisk indicated neoVTX proteins with different molecular weight. e) Illustration of the major protein toxin genes (11) deciphering in S. verrucosa. The contribution (proportion in total) of the 11 major protein toxin genes to the whole expressed toxin genes (240) in the VG and toxin proteins (33) was mapped to the corresponding column with diagonal lines Proteomic analysis of the crude venom a) Pie charts illustrating the composition of toxin and non-toxin proteins. b) Protein abundance of the major toxins and non-toxins, n = 7. c) Pie charts showing the contribution of each neoVTX subunit genes at transcriptomic and proteomic level. d) Native SDS-PAGE of lyophilized crude venom. Lane 1, protein ladders; lane 2, lyophilized crude venom. The asterisk indicated neoVTX proteins with different molecular weight. e) Illustration of the major protein toxin genes (11) deciphering in S. verrucosa. The contribution (proportion in total) of the 11 major protein toxin genes to the whole expressed toxin genes (240) in the VG and toxin proteins (33) was mapped to the corresponding column with diagonal lines Other notable toxin proteins in the venom included homologs of putative endothelial lipase (Uniprot ID: J3RZ81), lactotransferrin (K9IMD0), snaclec coagulation factor IX/X binding protein subunit A (SLXA, Q9DG39), snaclec B1 (B4XT00), peroxiredoxin-4 (P0CV91), cobra venom factor (Q91132) and cystatin-2 (J3SE80), whose abundance were higher compared with the lowest expressed DETG (neoVTX-b2) in the VG. As for the non-toxin proteins, gastrotropin protein (25.13%) ranked the highest in abundance in the venom of S. verrucose. The other highly expressed (abundance > 1%) non-toxin proteins included GPx1, PTGS1, cystatin-B, and TCAF1 (Fig. 3B). Alternative splicing contributes to venom diversity Alternative splicing (AS) of toxin genes was proposed as an important molecular mechanism underlying venom diversity and evolution. To further examine the complexity of the venom components at the transcriptional level, an integrative analysis pipeline of ISO-Seq and NGS and was performed with the same RNA samples from the VG (Fig. 4A). Fig. 4Massive alternative isoforms from the neovtx genes in the venom gland a) The work flow for sequencing and analysis for full length transcriptome construction. b) Distribution of the number of alternative splicing isoforms per annotated toxin gene. c) Percent of the AS isoforms from each DETGs in the total isoforms (856) mapped to the annotated toxin genes (476). d) The 6 tandem-repeated neovtx genes demonstrated the most isoforms. The isoforms were mapped to each gene with the exon usage shown. 5 trans-splicing isoforms were listed at the lower panel, showing a single isoform embraced exons from different genes. e) Expression level of 27 cis-splicing and 5 trans-splicing isoforms from the 6 neovtx genes, the value was illustrated as mean ± SD, n = 7 Massive alternative isoforms from the neovtx genes in the venom gland a) The work flow for sequencing and analysis for full length transcriptome construction. b) Distribution of the number of alternative splicing isoforms per annotated toxin gene. c) Percent of the AS isoforms from each DETGs in the total isoforms (856) mapped to the annotated toxin genes (476). d) The 6 tandem-repeated neovtx genes demonstrated the most isoforms. The isoforms were mapped to each gene with the exon usage shown. 5 trans-splicing isoforms were listed at the lower panel, showing a single isoform embraced exons from different genes. e) Expression level of 27 cis-splicing and 5 trans-splicing isoforms from the 6 neovtx genes, the value was illustrated as mean ± SD, n = 7 A total of 379,860 ISO-Seq reads were obtained, with an N50 length of 2,408 bp. All the reads were then aligned to the above assembled genome and 374,821 of them matched the corresponding target genome block. A total of 27,600 unique isoforms were retained, corresponding to 10,945 gene loci. These unique isoforms were further blasted with the 476 annotated toxin genes, and a total of 825 cis-splicing isoforms were mapped to 191 of these genes (Fig. 4B, Supplementary Data 5). Further analysis revealed that the 12 DETGs accounted for 52.48% of the isoforms among all the toxin genes (Fig. 4C). The results indicated that AS predominantly occurred in the highly expressed toxin genes. The six neoVTX genes accounted for 411 AS isoforms, which was approximately half of the total number of toxin gene isoforms (Fig. 4C). Mapping the isoforms to the six neoVTX genes revealed that neovtx-a2 (154) and neovtx-b3 (101) had the highest number of AS isoforms, followed by neovtx-a3 (71), neovtx-a1 (35), neovtx-b1 (17), and neovtx-b2 (4) (Fig. 4D). Interestingly, 29 trans-splicing isoforms were found among the six neoVTX genes, which utilized exons across these genes (Fig. 4D). The expression of these AS isoforms was further analyzed referring to the NGS data for quantitation analysis. A total of 27 isoforms accounted for about 90% expression of the six neoVTX genes predominantly associated with the neovtx-a2, neovtx-b3 and neovtx-a3 genes. In addition, the heterogenous expression pattern of the AS isoforms among all the seven samples revealed the diversity of venom genetic regulation at an individual level (Supplementary Fig. 5). In contrast, all the trans-splicing isoforms exhibited much lower expression (Fig. 4E and Supplementary Data 5). These results indicated that the alternative splicing of the neoVTX genes contributes to the venom diversity. Discussion High-throughput sequencing and rapid development of bioinformatic analyses have greatly facilitated the modern venomic research. We examined the venom components of S. verrucosa through an integrative analysis of genomic, transcriptomic and proteomic data. The chromosome-level genome provides a fundamental genetic resource for venom components characterization and in-depth biomedical studies. We first assembled a high-quality S. verrucosa genome, which enabled us to preliminary annotate 476 toxin genes. Further transcriptomic analysis of the VG characterized 12 DTEGs as the predominantly expressed toxic genes and 11 of these were validated by analyzing the venom proteome. The unique venom composition of S. verrucosa underlies its evolution adaptation A total of 467 toxin genes were annotated in the S. verrucosa genome, with the peptidase S1 family, venom metalloproteinase, and SNTX/VTX toxin families as the most abundant (Fig. 1C). Peptidase S1 family genes encode serine proteases, which are responsible for coordinating various physiological functions, including digestion, the immune response, blood coagulation, and reproduction. Snake venom serine protease (SVSP) was characterized as one of the four dominant protein families, together with phospholipase A2 (PLA2), snake venom metalloprotease (SVMP), and three-finger toxins (3FXs) [59]. In the S. verrucosa genome, peptidase S1 family genes (102) were listed as the most abundant toxin genes in S. verrucosa genome (Fig. 1C). Interestingly, only two of them showed expression levels with TPM > 100, and the peptidase S1 genes as a whole accounted for only a small proportion of the VG transcripts. This was similar with the cases in scorpion, spider and centipede [60–62]. Correspondingly, only around 0.13% of peptidase S1 proteins were found in the crude venom (Supplementary Data 3). A total of 36 venom metalloproteinase family genes were annotated in the S. verrucosa genome. All of these showed low-level expression with TPM < 50 and no metalloproteinase protein was identified in the crude venom. Similarly, six PLA2 genes (TPM < 1) were barely detected in the VG transcriptome. Instead, neoVTX proteins from the six tandem-repeated genes were the most abundant toxin proteins in the venom (Fig. 3B). Although we characterized 12 DETGs in the VG transcripts, only three toxin proteins accounted for more than 1% in the crude venom, including a pair of neoVTX subunits, a CRVP like protein, and a Kuntiz-type serine protease inhibitor. The results indicated that the venom is relatively less complex compared with other venomous animals, which is consistent with a similar study in S. horrida [25]. The last but not the least, we found 4 unannotated genes with high expression (TPM > 500) in the VG but no proteins in the venom (Fig. 2B), which suggests their potential roles as unique genes involved in the VG function of S. verrucosa. Taken together, the above results suggest that the unique venom composition underlined the venom phenotype evolution of S. verrucosa. Protein toxins offer invaluable source for the development of active drugs. Pore-forming proteins have shown great potential in tumor therapy due to its ability to deliver antibodies into the cancer cells [63]. SNTX/neoVTX toxins demonstrate high pore-forming activities, and the purified SNTX proteins have been proved to inhibit MCF-7 breast cancer cell growth by triggering apoptosis [64]. SNTX/neoVTX toxins also show ability to bind to voltage gated Ca2+ channel mimotopes to exert cardiotoxicity. Molecular modified SNTX/neoVTX toxins toward a voltage gated Ca2+ channel blocker could also be used in cardiovascular therapy [65]. Integrative analysis reveals a contribution of neoVTX genes to the venom diversity of S. verrucosa The major lethal molecules of S. verrucosa venom have been purified and molecular characterized as early as the 1990s. The VTX and neoVTX were purified by two different research groups and characterized as different proteins and remained unclarified thus far. The VTX was first characterized as a tetrameric protein consisting of VTXa and VTXb subunits [30, 31], whereas the neoVTX was a dimeric protein consisting of neoVTXa and neoVTXb subunits [66]. They were considered different proteins primarily based on the difference in their deduced amino acid sequences. neoVTXb exhibited 90% identity compared with the reported VTXb, lower than that with the SNTXb from another species, S. horrida. Further analysis found that VTXb contains an additional nucleotide at the 3’ end of the CDS, leading to a frame-shift, which contributes to the major difference between them. It was suggested that the geographical variation of the fish in the two studies contributed to major loci difference between the two toxin genes [66]. In our assembled genome, the discriminative loci in the three tandem-repeated beta subunit genes were all consistent with that of the neoVTXb gene (GenBank: AB262393.1) (Supplementary Fig. 6). Interestingly, we identified two protein bands with different molecular weights close to the reported dimeric and tetrameric forms of the neoVTX proteins by native PAGE electrophoresis. The two proteins were further confirmed by mass spectrometry sequencing (Fig. 3D). The results indicate that neoVTX proteins are the major lethal toxins in the fish strain that we used and they may be present in either a dimeric or tetrameric form. Gene duplication plays an important role in the evolution and diversification of toxin genes [67]. The major toxins in the snake venom, including metalloproteinases, phospholipases A2, and three-finger toxins, all experienced gene duplication events during evolution [15, 16]. A total of 32 SNTX/VTX family genes were annotated in our assembled S. verrucosa genome. The integrative analysis with the VG transcriptome and venom proteome revealed that most of the neoVTX proteins in the venom were derived from the six tandem-repeated neoVTX subunit genes, predominantly a pair of a2 and b3 genes (Fig. 3C). Two other SNTX-like genes were also located adjacent to this region. They all contained four conserved domains, but with much lower sequence identity and were clustered into a separate branch of the phylogenetic tree (Fig. 1D). In addition, the transcriptomic data indicated that these two SNTX-like genes were minimally expressed in the VG, and no protein was detected in the venom proteome. The results indicate that they may share the same origin as the neoVTX genes but show no functional roles in S. verrucosa. It is hypothesized that SNTX/neoVTX genes were generated by gene duplication throughout the evolutionary history of Percomorpha fish [26]. The identification of gene tandem duplication extremely relies on high quality genome. There are only two published chromosomal level genomes of Scorpaeniformes, Synanceia verrucosa and Sebastes schlegelii up to now, making it difficult to perform evolutionary analysis. Recently published work has provided solid evidence that a clustered SNTX genes could only be identified in the two venomous Scorpaeniformes species, namely S. verrucosa and S. schlegelii, leading to the generation of authentic toxic SNTX/neoVTX genes by neofunctionalization [28]. The understanding of evolution pathway of the SNTX/neoVTX and their tandem duplication events requires additional genomic resource, especially of the closely related species. Alternative splicing of venom genes is considered as a complementary mechanism for the generation of venom complexity across animal lineages [19, 68, 69]. We identified 856 isoforms expressed in the VG from 198 toxin genes, and 411 of them including 29 trans-splicing isoforms were mapped to the six tandem-repeated neoVTX subunit genes, which markedly increased the potential to produce diverse neoVTX proteins. Expression analysis of these AS isoforms revealed that a pair of neoVTX a2 and b3 genes transcribed the several highest expressed isoforms (Fig. 4E). Unfortunately, our proteomic sequencing approach failed to distinguish these proteins with various sizes and charges. A 2D gel electrophoresis of the crude venom from the closely related species S. horrida revealed 15 SNTXa isomers and 26 SNTXb isomers with varied length of ORF, partially supported the alternative splicing of neoVTX genes in this study. The results indicated that the AS of the SNTX/neoVTX genes were conserved mechanism in stonefish underlined venom diversity [25]. Interestingly, although we identified a total of 29 trans-splicing isoforms among the six tandem-duplicated neoVTX genes, they all showed much lower expression level (Fig. 4D and E), leaving their functional roles and biological significance unknown. The alternative splicing could contribute to the venom complexity, which is considered as an important adaptive strategy of S. verrucosa facing physiological changes and geographic location, as well as against the native predators. Further work is required to investigate the biological significance of these neoVTX isomers. The other major toxin proteins characterized in the venom of S. verrucosa The most abundant toxin protein other than the neoVTX proteins was characterized as CRVP. Its homologue in S. horrida is defined as Golgi-associated plant pathogenesis related protein (ShGAPR) instead. Although ShGAPR showed a higher sequence similarity with the human GAPR1, it may act as a venom toxin similar to CVRPs [25]. The transcriptomic analysis of the S. verrucosa venom gland indicated that sv_crvp1 was found to be the highest expressed toxin gene in the VG and considered a DEG compared with the SK (Fig. 2B and C), which further supports its potential role as a toxin gene in S. verrucosa. The most homologous gene to SvCRVP annotated in ToxProt was helothermine, which was first isolated from the venom of the Mexican beaded lizard (Heloderma horridum horridum) [70]. Lizard helothermine is a peptide toxin that blocks ryanodine receptors, which are responsible for Ca2+ release within skeletal, cardiac, and neuronal cells [71, 72]. Further studies are encouraged to illustrate the potential role of SvCRVP as the most abundant venom component besides the neoVTX proteins in S. verrucosa. Venomous snakes use protease inhibitors in the venom as weapons to disrupt the homeostasis of the prey’s physical and biochemical reactions, such as the blood coagulation and blood pressure regulation, which results in immobilization or death of the prey [73]. Protease inhibitors may protect peptide/protein toxins in the venom from degradation by proteases from the prey or its predators [74]. Kuntiz-type serine protease inhibitors, C-type lectins and phospholipase A2 are the most extensively distributed protease inhibitors in the snake venom [75]. We annotated 17 Kunitz-type family genes in the S. verrucosa genome and only one of them (sv_kspi1/SvKSPI) was characterized as a DETG in the VG (Figs. 2B and 3B) and accounted for 1.84% in total toxin proteins. Homologs of Snaclec B1 and SLXA belonging to C-type lectins were also identified in the crude venom (Fig. 3B), and together with SvKSPI, served as the major protease inhibitors in S. verrucosa. Calglandulin was previously identified from the VG of Bothrops insularis and is characterized as a putative Ca2+ binding protein with four EF-hand motifs [76]. Using antiserum against recombinant calglandulin, it was only detectable in the VG of Bothrops insularis, but not in the crude venom or other tissues, indicating that calglandulin may be primarily involved in the cellular control mechanism of the secretion of toxins from the gland into the venom [76, 77]. Whereas, calglandulin was identified as a major component in the crude venom of the fire ant Solenopsis Invicta [78] and stingray [79], indicating its unknown functions beyond the toxins secretion. In S. verrucosa venom, a small proportion of calglandulin proteins (0.60%) were also evidenced, which indistinguishably corresponded to sv_caglp1 and sv_caglp2 due to their same deduced amino acid sequence (Supplementary Fig. 7). Notably, transcriptomic analysis revealed that only sv_caglp1 gene was characterized as DEGs between the VG and the skin, indicating a transcriptional regulatory mechanism driving sv_caglp1 as the major gene involved in toxins secretion (Fig. 2B, C). Hyaluronidase is an enzyme commonly found in the venom of snake, scorpions, spiders and leeches [80]. Hyaluronidase activity is considered critical for the spreading of toxins and interrupting the integrity of the target’s extracellular matrix by degrading hyaluronic acid, which is considered a therapeutic target against snake venom [81, 82]. The cDNA of S. verrucosa hyaluronidase has been previously cloned and characterized with a conserved active site consisting of one catalytic residue and four substrate positioning residues [34]. In the present study, we characterized the hyaluronidase gene as a DETG in the VG and further supported by the proteomic data (Figs. 2B and 3B). Among all the 12 DETGs defined in the S. verrucosa VG, only the sv_tpt1 gene exhibited no proteomic evidence in the crude venom (Fig. 3B). The Sv_tpt1 gene encodes a translationally controlled tumor protein (TCTP), which is predominantly described as a venom toxins in spiders, that contributes to the allergic and inflammatory activity of the venom [83, 84]. There is few studies of the TCTP toxins in other venomous animals and its biological functions under investigation. The expression of the TCTP gene was reported in the VG from three scorpionfishes (Scorpaenidae.spp), indicating its wide distribution in the venomous teleost [85]. In the S. verrucosa, the sv_tpt1 gene showed comparable high expression levels in the VG and SK (Fig. 2C), along with its absence in the crude venom, which indicates other potential roles beyond as toxins in the S. verrucosa. A group of toxin proteins beyond the 12 DETGs were also identified based on the abundance in the crude venom (Fig. 3B). A putative endothelial lipase first identified in the VG transcriptome of the eastern diamondback rattlesnake was suggested to exhibit phospholipase and triglyceride lipase activity [86]. Lactotransferrin protein is a major iron-binding protein usually found in exocrine fluids, including mucosal secretions and mammal breast milk. Lactotransferrin is annotated as a toxin because of its presence in the proteome of venom accessory glands in the vampire bat (Desmodus rotundus) and antimicrobials activity [87]. Similarly, peroxiredoxin-4 [88], Snaclec B1 [89], and cobra venom factor-like protein [90] are also evidenced as venom proteins in other venomous animals. The presence of these proteins in the crude venom of S. verrucosa further supports their characterization as a toxin protein, whereas their underlying molecular mechanism need to be further investigated. A notable non-toxin protein in the venom A set of non-toxin proteins are shown in Fig. 3B. Although some of these have not been reported to be present in the crude venom, they may have housekeeping or hemostasis maintaining roles in the VG. Of these, a gastrotropin-like protein was the second most enriched following the neoVTX proteins. It is encoded by the FABP6 gene, which also showed high expression levels in the VG (Figs. 2B and 3B). Gastrotropin/FABP6 is a fatty acid-binding protein and therapeutic target associated with immune infiltration [91]. Gastrotropin was first characterized as venom protein in S. horrida, but not in other venomous animals. It may contribute to the unique activities of the venom by maintaining venom hemostasis and protecting venom components from degradation [25]. We found that the extracted crude venom of S. verrucosa exhibits high activity at room temperature after hours; however, further studies are encouraged to investigate the biological significance of the gastrotropin-like protein in S. verrucosa venom. Pharmaceutical innovation. Conclusions In this study, the major venom components of S. verrucosa were thoroughly investigated by integrative multi-omics approaches in this work, for the first time in a venomous teleost. Six tandem-duplicated neoVTX subunit genes were identified in the genome, which contained massive AS isoforms and diverse neoVTX proteins. The diverse neoVTX proteins in the venom as lethal particles are important to understand the adaptive evolution of S. verrucosa, particularly when engaging nature predators. Further functional studies are needed to elucidate the mechanism of these alternative isoforms in venom. Additional biomedical studies are also encouraged to exploit the venom components of S. verrucosa for pharmaceutical innovation. Electronic Supplementary Material Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1 Supplementary Material 2 Supplementary Material 2 Supplementary Material 3 Supplementary Material 3 Supplementary Material 4 Supplementary Material 4 Supplementary Material 5 Supplementary Material 5 Supplementary Material 6 Supplementary Material 6