Venomics of Tropidolaemus wagleri, the sexually dimorphic temple pit viper: Unveiling a deeply conserved atypical toxin arsenal

Tropidolaemus wagleri (temple pit viper) is a medically important snake in Southeast Asia. It displays distinct sexual dimorphism and prey specificity, however its venomics and inter-sex venom variation have not been thoroughly investigated. Applying reverse-phase HPLC, we demonstrated that the venom profiles were not significantly affected by sex and geographical locality (Peninsular Malaya, insular Penang, insular Sumatra) of the snakes. Essentially, venoms of both sexes share comparable intravenous median lethal dose (LD50) (0.56–0.63 μg/g) and cause neurotoxic envenomation in mice. LCMS/MS identified six waglerin forms as the predominant lethal principles, comprising 38.2% of total venom proteins. Fourteen other toxin-protein families identified include phospholipase A2, serine proteinase, snaclec and metalloproteinase. In mice, HPLC fractions containing these proteins showed insignificant contribution to the overall venom lethality. Besides, the unique elution pattern of approximately 34.5% of non-lethal, low molecular mass proteins (3–5 kDa) on HPLC could be potential biomarker for this primitive crotalid species. Together, the study unveiled the venom proteome of T. wagleri that is atypical among many pit vipers as it comprises abundant neurotoxic peptides (waglerins) but little hemotoxic proteinases. The findings also revealed that the venom is relatively well conserved intraspecifically despite the drastic morphological differences between sexes.

were likely the two basic and lethal polypeptides subsequently isolated from the venom, termed waglerin I and waglerin II 27 . Waglerins had been shown to elicit neurotoxic signs in laboratory animals [28][29][30] .
The uniqueness of T. wagleri venom was further demonstrated by the lack of immunological recognition of its toxins by antisera or antivenoms raised against several Asiatic viperid species 31,32 . The remarkably different venom properties of this species among Trimeresurus sensu lato is congruent with the revised taxonomy where it diverged earlier from the other paraphyletic arboreal pit vipers. Further characterization of its venom is hence important to understand better the biological and medical importance of pit vipers of this unique lineage. Although several in-depth studies on isolated or synthetic waglerins have been reported [27][28][29][30]33 , to date, the compositional details of the global venom profile of T. wagleri remains unavailable. Confounding this is the lack of insights into venom variability that may accompany the extreme sexual dimorphism of this species. Therefore, this study aimed to investigate the venom proteome of T. wagleri by detailing the subtypes and relative abundances of proteins in the venom. Potential divergence in the venom compositions caused by sex factor was also examined. In addition, the proteomic findings were corroborated by toxicity tests for further insights into the toxicovenomic profile of this species.

Results
Comparative profiling of T. wagleri venoms on reverse-phase HPLC. Figure 2 shows the chromatographic profiles of T. wagleri venom according to male and female specimens and their geographical locales. The findings revealed an overall similar elution pattern under the same condition of high performance liquid chromatography for T. wagleri venom ( Fig. 2A-E), with the exception of the fraction eluted between 38-42 min where the female samples ( Fig. 2A,C,E) showed 2-3 neighboring peaks within the eluted fraction, while the male samples (Fig. 2B,D) showed a better resolved single chromatographic peak (indicated by arrows). Another difference was noted in Fig. 2D (Penang, male sample) where the fraction eluted at 120 min appeared higher than that of the other samples. On the other hand, the venom of Tropidolaemus subannulatus (included as a hetero-specific sample for profile comparison) exhibited a very different elution profile altogether, as shown in Fig. 2F.

Venomics of T. wagleri.
Reverse-phase chromatography resolved the venom into approximately 25 peaks, the majority of which were eluted before 75 min of retention time (comprising 73% of the total peaks area) (Fig. 3a). All eluted proteins were collected manually into 18 fractions designated as Fractions 1-18, respectively. SDS-PAGE of these major fractions revealed the presence of proteins as homogenous bands in the range of 3-7 kDa. Proteins eluted after 110 min showed an increase in heterogeneity and molecular sizes, with masses ranging from 14 kDa to 40 kDa and above (Fig. 3b). Liquid chromatography tandem mass spectrometry (LCMS/MS) detected 76 distinct protein forms from Fractions 6-18 of the venom chromatography ( Table 1). The proteomic information on mass charges, sequences, validity parameters of the peptide spectra are provided in Supplementary File S1. The in-house transcriptomic data matched by the tryptic peptides were provided in Supplementary File S2. In total, 67 proteins were identified as toxins and assigned into 15 different families (Table 2; Fig. 4). Among these, waglerins were found to be the most abundant (38.2% of total venom proteins), followed by phospholipases A 2 (7.3%), serine proteinases (5.5%), snaclecs (3.5%), snake venom metalloproteinases (1.7%), L-amino acid oxidases (1.7%), 5′nucleotidases (1.6%), phosphodiesterases (1.0%) and several toxins of lower abundance (< 1.0% each) comprising cysteine-rich secretory protein, phospholipase B, hyaluronidase, aminopeptidase, phospholipase A 2 inhibitor, cytotoxin and cobra venom factor. Together, these toxins constitute 62.3% of the total venom proteins. A small amount of proteins in the venom (2.5%) were identified as non-toxins; these are mainly physiological proteins or cellular enzymes. Meanwhile, venom proteins eluted in the initial course of the chromatography (Fractions 1-5, totaling 34.5% of venom proteins) (Fig. 3a), were assigned as unspecified low molecular mass proteins as they were unidentified by the current LCMS/MS and data mining approach, even with the use of two different protease enzymes (trypsin and chymotrypsin) for optimal sequence matching. Proteins in Fractions 6-11 which share the same low molecular mass as those in Fractions 1-5, nonetheless, were successfully identified as various forms of waglerin.
Lethality and toxicovenomics of T. wagleri venom. The intravenous median lethal doses of the venom in mice were determined to be 0.56 (95% C.I.: 0.37-0.84) μ g/g and 0.63 (95% C.I.: 0.59-0.66) μ g, respectively, for female and male venom samples (Table 3). Mice injected with a lethal dose of the venom showed prostration and impaired movement with laboured breathing while paralysis set in progressively. Furthermore, the toxicity of the major protein fractions of the venom (Fig. 3a) was evaluated in a mouse model through intravenous injection (Table 4). Of note, Fractions 1-5, unspecified low molecular mass proteins, were not lethal in mice at doses of 1-4 μ g/g. Fractions 6 and 8, both being the major waglerin forms, exhibited potent lethality with an LD 50 value of 0.063 (0.059-0.066) μ g/g and 0.20 (0.16-0.25) μ g/g, respectively, in mice. On the other hand, Fractions 12-18 which contain proteins of medium to high molecular masses were found to be non-lethal in mice even at high doses (from 2-4 μ g/g). For the venom of male specimens, the HPLC fractions eluted between 38-42 min and at 120 min (which showed slight variation from that of the female samples) were also tested in a mouse model. Similarly, these fractions were found to be non-lethal in mice up to 4 μ g/g. When tested in frogs, the male and female T. wagleri venoms showed comparable neurotoxicity (frogs showed inability to move four limbs) and lethality. The LD 50 values of female and male venoms are 38.31 (95% C.I.: 29.99-48.94) μ g/g and 41.77 (95% C.I.: 32.78-53.23) μ g/g, respectively, when injected via lymphatic route into the frogs (Table 3).  Discussion Snake venom variability within the same species has been widely appreciated for its implication on fundamental research, management of envenomation and study of venom evolution. Intraspecific venom variability is often associated with geographical origin, age (stage of development), sex, diet etc. of the snake 7,10,11,34 . Recent proteomic approach revealed that there are at least subtle sex-based differences in the venom compositions of some New World pit vipers notably Bothrops jararaca 12,35,36 ; however, literature on this comparative topic for Asiatic pit vipers remains scarce. Sex-based venom variation was once reported previously in the Malayan pit viper (Calloselasma rhodostoma), although the female-specific protein band (from isoelectric focusing study) was never identified 11 . In most of the studies, the biological significance of sex-based venom variation has not been fully elucidated although it could be sensibly related to dietary differences between male and female snakes. In the present work, the venom variability driven by sexual dimorphism was expected to be much drastic in T. wagleri when considering the extreme differences in their body sizes (male-to-female body ratio can be as high as 1:20). Hypothetically, the venom of female T. wagleri would be one that is streamlined for the predation of larger animals such as warm-blooded rodents or birds, while the smaller male adults and juveniles feed mainly on amphibians and reptiles. Surprisingly, the current study revealed highly similar chromatographic profiles of both the male and female venoms, in particular the lethal components i.e. waglerins. The findings imply that the venom contents of T. wagleri are consistent regardless of the sex and body size of the snake, and that predation is accomplished with the venom neurotoxic activity mediated through waglerins. Functionally, this is further supported by the comparable lethal potency and paralytic effect of male and female venoms in each prey type (mice and frogs). In this study, the mice appeared to be more susceptible than frogs to the lethal effect of both male and female venoms, possibly because waglerin has a higher affinity toward muscle cholinergic receptors of mammalian species. Essentially, the present work suggested that both male and female T. wagleri share the similar composition of lethal principles in their venoms and both induce comparable toxicity. This finding is in agreement with earlier studies which indicated that composition variations in some snake venoms occur at negligible to minor scale between sexes of the same species 37 . On the HPLC profiles, the minor difference noted between 38-42 min of elution time (where splitting of fraction was noted in the female profile) may be considered for the presence of protein isoforms in the pooled female venoms. Proteins eluted in this fraction have low molecular mass similar to waglerins but are likely insignificant medically, since these proteins are non-lethal and did not induce any remarkable sign when injected into the animals at doses more than 5 times of the venom LD 50 . This also applies to the corresponding protein fractions (eluted between 38-42 min and 120 min) of the male specimens where no toxicity was observed at high doses. The majority of these non-lethal, high molecular mass proteins are hydrolytic    enzymes that may serve digestive and antibacterial purposes instead of foraging 9 . It should be noted that the interpretation was limited to the individual fractions tested; potential toxicity resulting from synergistic interactions of the different non-lethal fractions remain to be further explored.

Tropidolaemus wagleri venom LD 50 (μg/g) in mice via intravenous route LD 50 (μg/g) in frogs via intra-lymphatic route
The current quantitative proteome of T. wagleri venom verified the abundance of waglerins in the venom (close to 40% total venom proteins), which has not been clearly elucidated hitherto in a global profile, although based on toxin-isolation studies the estimated abundance could be, rather inconsistent, ranging from less than 1% 27 to 8-13% 26 of total venom proteins. The inconsistency was presumably due to the different methods adopted for protein purification. This present study also demonstrated that waglerins predominate in at least 6 protein forms based on their different degrees of hydrophobic interaction with the C18 column. The two most abundant forms of waglerins isolated from Fraction 6 and Fraction 8 demonstrated high lethal activity with the former being more potent by approximately 4-fold. Our finding is hence in agreement with Weinstein et al. 27 where two lethal, low molecular mass peptides (approximately 3 kDa, termed Lethal I and Lethal II) were isolated from the venom of this species. The lack of lethal effect in the other proteins supported waglerins as the principal lethal toxins of the venom. From the evolutionary and biological aspects, waglerin has been compared to azemiophin, a unique short neurotoxic peptide originated from a primitive viper, Azemiophs feas, where both toxins share a homologous C-terminal hexapeptide and block nicotinic receptors, but azemiophin is distinct by not possessing disulfide bridges 38 . Waglerins are also proline-rich short peptides akin to azemiophin and other snake venom metalloproteinase propeptides originated from the colubrid Psammophis mossambicus 39 . Of note, the proline-rich property creates a "kinky" peptide orientation which probably resisted tryptic digestion of waglerin initially in this study. This was overcome by the subsequent use of chymotrypsin and the corresponding enzyme-specific database search engine that enabled the identification of waglerins in the proteome.
The abundance and protein type of phospholipases A 2 revealed in the current proteomic study are consistent with previous toxin isolation studies by Weinstein et al. 27 and Tsai et al. 26 . Clinically, the effects of T. wagleri PLA 2 on platelet dysfunction (acidic PLA 2 ) or myonecrosis (basic K49-PLA 2 ) have not been reported. In laboratory mice, the present study confirmed that these PLA 2 were non-lethal, supporting the observation reported by Weinstein et al. 27 . Instead, the role of the PLA 2 may be related to ancillary function for the digestion of prey tissues. In addition, the current study also reported the presence of three snaclec subtypes (C-type lectins) in the venom with species-specific sequences matched to T. wagleri transcriptome database, implying species uniqueness in the peptide sequences of these snaclecs. Snaclecs are toxins responsible for platelet aggregation or disintegration; however, thrombocytopaenia has not been reported clinically in humans, thus suggesting that the action of T. wagleri snaclecs may be species-specific or prey-restricted.
The composition of snake venom metalloproteinase (SVMP) is considerably small in T. wagleri venom as shown in its proteome. In contrast, the venom proteomes of most Asiatic pit vipers typically charted 20-40% SVMPs that function as hemorrhagins or anticoagulant toxins 10,[40][41][42] . Indeed, the venom of T. wagleri lacks hemorrhagic effect which is otherwise commonly seen in envenomation by vipers and pit vipers 24,25 . On the other hand, the composition of snake venom serine proteinases (SVSPs) including procoagulant enzymes detected in the current study is comparable to that reported for the basal, terrestrial pit vipers in Asia (Calloselasma rhodostoma, Hypnale hypnale and Deinagkistrodon acutus) 40,[42][43][44] . However, the SVSPs and the venoms of these terrestrial pit vipers were known to exhibit much potent procoagulant activity 24,45 . There is considerably overlapping of prey between the arboreal T. wagleri and these terrestrial pit vipers; nonetheless, in the latter the potent procoagulant effect of venom is a strategy needed for prey subjugation and killing, while T. wagleri relies on the neurotoxic action of waglerin for hunting. Again, the feeble procoagulant activity of T. wagleri venom is more parallel to the venom effect of Fea's viper, Azemiops feas, which is also non-hemorrhaging but neurotoxic 38,46 .
Earlier, the enzymatic activities of phosphodiesterase, phosphomonoesterase, arginine ester hydrolase and L-amino acid oxidase (LAAO) had been detected in T. wagleri venom but the abundance of these enzymes were uncharacterized 24 . The current study successfully revealed the composition of these venom enzymes in addition to several other minor components (phospholipase B, hyaluronidase and aminopeptidase) previously not well reported from this venom. These venom enzymes are likely involved in tissue digestion and venom spread, while LAAO may be responsible for anti-microbial effect of the venom 47 . The current proteomic study was also able to detect the minute phospholipase A 2 inhibitor, the function of which is unknown in venom but be related to the stability of venom storage in the glands. It is also noteworthy that cobra venom factor (a C3 complement factor) and a cytotoxin homologue conventionally found in elapid venoms 7,48 were also detected proteomically as expressed proteins in T. wagleri venom, albeit at a small quantity. Further protein isolation and characterization study is hence warranted to fully validate and elucidate the phenomenon.
Meanwhile, it should be noted that the venom samples were collected from wild-caught specimens and each sexual sample (male, female) was matched to the locality. The main sexual samples collected within Malaysia, in total, were from 8 males and 12 females before stratification into the locales of Penang (6 females, 4 males) and Perak (6 females, 4 males). In addition, the Sumatran sample was a pool from 6 female specimens. We acknowledge that the sample size (n = 8-12 per sex group) could be a limitation in this study; however importantly, the conclusion was not generated from bias based on a single T. wagleri specimen.

Conclusions
The current study unveiled the unique composition of T. wagleri venom in a global profile, addressing the qualitative and quantitative details of the venom proteins. Of note, venoms from both sexes of the snake (paired peninsular and insular collections) demonstrated comparable lethality and protein profile, suggesting that the venom composition is relatively well conserved in the species despite its drastic sexual dimorphism. C 18 reverse-phase high-performance liquid chromatography (HPLC) and SDS-polyacrylamide gel electrophoresis (SDS-PAGE). The amount of venom sample subjected to HPLC was 1 mg each for comparative chromatographic profiling and 2 mg (Perak, female specimens) for mass spectrometry study of the protein fractions. Crude venom samples were first reconstituted in ultrapure water and centrifuged at 10,000 g for 5 min. The supernatants were subjected to LiChrospher ® WP 300 C 18 reverse-phase column (5 μ m) using a Shimadzu LC-20AD HPLC system (Japan). The venom components were eluted at 1 ml/min with a linear gradient of 0.1% trifluoroacetic acid (TFA) in water as Solvent A and 0.1% TFA in 100% acetonitrile (ACN) as Solvent B, as follows: 5% B for 10 min, 5-15% B for 20 min, followed by 15-45% B for 120 min and 45-70% B for 20 min. In preparation for mass spectrometric analysis, the chromatographic fractions were collected manually at absorbance 215 nm and the lyophilized fractions were further electrophoresed on glycine SDS-PAGE (18%, reducing condition). The protein bands were visualized using Coomassie blue staining. Low range protein marker was used as molecular mass standard (1.7-40 kDa).
In-solution tryptic/chymotryptic protein digestion and protein identification by tandem mass spectrometry (nano-ESI-LCMS/MS). The protein fractions from reverse-phase HPLC were subjected to reduction with DTT, alkylation with iodoacetamide, and in-solution digestion with mass-spectrometry grade trypsin or chymotrypsin proteases as described previously 50 . The protease-digested peptides were desalted with Millipore ZipTip ® C 18 Pipette Tips (Merck, USA) according to the manufacturer's protocol to enhance the performance of mass spectrometry. The peptide eluates were then subjected to nano-electrospray ionization (ESI) MS/MS experiment, respectively. The experiment was performed on an Agilent 1200 HPLC-Chip/MS Interface, coupled with Agilent 6520 Accurate-Mass Q-TOF LC/MS system. Samples were loaded in a large capacity chip 300 Å, C18, 160 nL enrichment column and 75 μ m × 150 mm analytical column (Agilent part No. G4240-62010) with a flow rate of 4 μ l/min from a capillary pump and 0.3 μ l/min from a Nano pump of Agilent 1200 series. Injection volume was adjusted to 1 μ l per sample and the mobile phases were 0.1% formic acid in water (A) and 90% acetonitrile in water with 0.1% formic acid (B). The gradient applied was: 3-50% solution B for 30 min, 50-95% solution B for 2 min, and 95% solution B for 5 min, using Agilent 1200 series nano-flow LC pump. Ion polarity was set to positive ionization mode. Drying gas flow rate was 5 L/min and the drying gas temperature was 325 °C. Fragmentor voltage was 175 V and the capillary voltage was set to 1995 V. Spectra were acquired in MS/MS mode with MS scan range of 110-3000 m/z and MS/MS scan range of 50-3000 m/z. Precursor charge selection was set as doubly, triply or up to triply charged state with the exclusion of precursors 922.0098 m/z (z = 1) and 121.0509 (z = 1) set as reference ions. Data was extracted with MH + mass range between 600-4000 Da and processed with Agilent Spectrum Mill MS Proteomics Workbench software packages. Carbamidomethylation of cysteine was set as a single modification. The peptide finger mapping was modified to specifically search against an in-house database that has merged non-redundant NCBI protein sequences of Serpentes (taxid: 8570) with dataset derived the venom-gland transcriptome of Tropidolaemus wagleri of a Malaysian origin. Protein identifications were validated with the following filters: protein score > 11, peptides score > 6 and scored peak intensity (SPI) > 60%. The proteins identified were classified as toxins or non-toxins according to their reported putative functions or known toxicity 9,48 . The abundance of individual venom toxin was estimated based on its mean spectral intensity (MSI) relative to the total MSI of all proteins identified through the in-solution mass spectrometry, as reported previously 51,52 . (CIOMS) guidelines on animal experimentation, and was approved by the Institutional Animal Care and Use Committee of the University of Malaya (Ethics clearance numbers: 2014-09-11/PHAR/R/TCH and 2016-190607/ TCH/R/PHARM). The median lethal doses (LD 50 ) of the venoms and toxin fractions were determined from a serial dose-response study, where the venoms or toxin components were injected intravenously into the caudal veins of mice, or injected into the dorsal lymph sac of frogs. The number of animals tested for each compound was 16-20 (n = 4 per dose, 4-5 doses in total). The survival ratio was recorded after 48 h and LD 50 was calculated using the Probit analysis method 53 .