Venom Proteomics of Trimeresurus gracilis, a Taiwan-Endemic Pitviper, and Comparison of Its Venom Proteome and VEGF and CRISP Sequences with Those of the Most Related Species

Trimeresurus gracilis is an endemic alpine pitviper in Taiwan with controversial phylogeny, and its venom proteome remains unknown. In this study, we conducted a proteomic analysis of T. gracilis venom using high-performance liquid chromatography-tandem mass spectrometry and identified 155 toxin proteoforms that belong to 13 viperid venom toxin families. By searching the sequences of trypsin-digested peptides of the separated HPLC fractions against the NCBI database, T. gracilis venom was found to contain 40.3% metalloproteases (SVMPs), 15.3% serine proteases, 6.6% phospholipases A2, 5.0% L-amino acid oxidase, 4.6% Cys-rich secretory proteins (CRISPs), 3.2% disintegrins, 2.9% vascular endothelial growth factors (VEGFs), 1.9% C-type lectin-like proteins, and 20.2% of minor toxins, nontoxins, and unidentified peptides or compounds. Sixteen of these proteoforms matched the toxins whose full amino-acid sequences have been deduced from T. gracilis venom gland cDNA sequences. The hemorrhagic venom of T. gracilis appears to be especially rich in PI-class SVMPs and lacks basic phospholipase A2. We also cloned and sequenced the cDNAs encoding two CRISP and three VEGF variants from T. gracilis venom glands. Sequence alignments and comparison revealed that the PI-SVMP, kallikrein-like proteases, CRISPs, and VEGF-F of T. gracilis and Ovophis okinavensis are structurally most similar, consistent with their close phylogenetic relationship. However, the expression levels of some of their toxins were rather different, possibly due to their distinct ecological and prey conditions.


Introduction
Taiwan is a mountainous island, with two thirds of its territory covered by mountain forests.Today, alpine organisms worldwide face various threats such as climate change, air Toxins 2023, 15, 408 2 of 18 pollution, and human development; thus, research on alpine organisms is urgently needed.Among the more than 200 mountains in Taiwan higher than 3000 m, some glacial refuges and relic species have considerably high conservation value [1][2][3].Trimeresurus gracilis Oshima, 1920 [4] is an endemic medium-sized pit viper distributed mainly at altitudes above 2000 m in central Taiwan.Remarkably, the taxonomy and genus name of T. gracilis have been controversial [5].Phylogenetic analyses based on the mitochondrial and nuclear gene sequences of various pitvipers revealed that T. gracilis is phylogenetically close to Ovophis okinavensis in central Ryukyu [6][7][8].However, T. gracilis gives live birth to its offspring, whereas O. okinavensis is among the few egg-laying pit viper species.Moreover, the prey ecology of T. gracilis and O. okinavensis is rather different [9,10], and how this affects their venom proteomes remains to be further explored and clarified.
T. gracilis snakebites mainly elicit hemorrhagic symptoms in patients, including local tissue damage (myonecrosis, dermal necrosis, edema, hemorrhage, and blistering) and systemic coagulopathy [11], although T. gracilis envenoming cases are rare.To date, no specific antivenom is available for T. gracilis envenomation, and the venom proteome has not been resolved.T. gracilis-envenomed patients have been treated with local "bivalent hemotoxic-antivenom" against Viridovipera stejnegeri and Protobothrops mucrosquamatus, but this antivenom failed to relieve the local lesions of the patients before they received surgical intervention [11].
Previously, we cloned and sequenced T. gracilis venom proteins belonging to three major toxin families: an acidic phospholipase A 2 (PLA 2 ) and a Lys49-homolog of PLA 2 [12], ten venom serine proteases (SVSPs) [13], and five metalloproteinases (SVMPs) including PII-class and its disintegrin (DIS) domain [14].We have shown that the amino acid sequences of some representative toxins, including Lys49-PLA 2 , several SVSP variants, and the PI-class metalloprotease of T. gracilis, are highly similar to those of the corresponding venom toxins of O. okinavensis [15,16].The taxonomy of T. gracilis and its relationship with other eastern Asian pitvipers, meanwhile, remains puzzling, and this species and those under the genus Gloydius have been shown to be the most likely Asian sisters of New World pitvipers [7,[17][18][19].Thus, it is not surprising that PIII-SVMP and some SVSP variants expressed in T. gracilis venom bear high structural similarities to the corresponding toxins from New World pitvipers [13,14].
Venomic studies may provide important insights into the pathophysiology of envenoming and the molecular evolution of venom toxin multigene families [20].Our aim was to investigate the venom proteome and full sequences of not only the major but also the secondary or minor toxin families of T. gracilis in order to better understand the composition and evolution of its venom and to treat snakebites effectively.In the present study, T. gracilis venom proteomics were studied using high-performance liquid chromatography (HPLC) and liquid chromatography-tandem mass spectrometry (LC-MS/MS).We also cloned and sequenced the venom gland cDNAs encoding Cys-rich secretory proteins (CRISPs) and vascular endothelial growth factors (VEGFs).These T. gracilis toxins were compared with other pitviper toxins using a BLAST search and sequence alignments.Our results may provide a deeper understanding of the venom composition of T. gracilis and the evolutionary relationships between T. gracilis and other related pitvipers, such as the east Asian Ovophis, Protobothrops, and the New World Crotalus.

Chromatographic and Electrophoretic Profiling of T. gracilis Venom
Using a C 18 reverse-phase column, crude T. gracilis venom was separated into 39 peptide/protein fractions by HPLC (Figure 1A).These fractions were collected and analyzed using sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) under reducing conditions (Figure 1B).The first 13 fractions (eluted within the first 55 min) failed to show any bands on the gel.Notably, most medium-to high-MW venom proteins (5000-260,000) were eluted between 55 and 130 min and collected in fractions 14-39 (Figure 1A).As expected, most of the small peptides or proteins eluted earlier than large proteins, and the basic variants of the venom toxins eluted earlier than the acidic variants of the same toxin family.However, SDS-PAGE results revealed that most of the HPLC peaks contained multiple proteins or subunits rather than a single purified protein (Figure 1B).
Toxins 2023, 15, x FOR PEER REVIEW 3 of 21 lyzed using sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) under reducing conditions (Figure 1B).The first 13 fractions (eluted within the first 55 min) failed to show any bands on the gel.Notably, most medium-to high-MW venom proteins (5000-260,000) were eluted between 55 and 130 min and collected in fractions 14-39 (Figure 1A).As expected, most of the small peptides or proteins eluted earlier than large proteins, and the basic variants of the venom toxins eluted earlier than the acidic variants of the same toxin family.However, SDS-PAGE results revealed that most of the HPLC peaks contained multiple proteins or subunits rather than a single purified protein (Figure 1B).

T. gracilis Venom Proteomic Analysis
Mass spectrometric analyses of the proteins in the HPLC peaks (particularly the broader ones) revealed that they may contain several proteins from different toxin families (Table 1 and Supplementary Table S1).Overall, 155 toxin proteoforms were identified in the HPLC fractions, except for peaks 1-5 and 7. Sixteen of these proteoforms were T. gracilis venom proteins whose full amino-acid sequences have been deduced from venom gland transcriptome, including three SVMPs, six SVSPs, two PLA 2 s, two CRISPs, one DIS, and two VEGFs (Supplementary Table S2).By measuring the peak area under the curve of the HPLC chromatogram, the relative abundances of low-molecular-weight peptides, nontoxins (e.g., keratins), and unidentified compounds in the venom together accounted for 20.1%, which were mainly from the first 13 fractions.The identified peptides/proteins were categorized into 13 toxin families and sorted quantitatively (Supplementary Table S2).* Updated toxin names are indicated in square brackets.** The sequence matches gracilisin [14].*** The sequence matches Tgc-MP [14] and okinalysin [16].

Two Cysteine-Rich Secretory Proteins (CRISPs) of T. gracilis Venom
The cDNAs encoding the two novel CRISPs (Tgc-CRa and Tgc-CRb) were cloned and fully sequenced from venom glands and submitted to GenBank under the accession numbers ACE73569.1 (gi|190195325) and ACE73570.1 (gi|190195327), respectively.Tgc-CRa and Tgc-CRb appeared to be paralogs, with only 67% sequence similarity.They were aligned with possible orthologous snake venom CRISPs retrieved using BLASTp (Figures 3A and 3B, respectively).We are not able to find any venom-CRISP sequences of other Trimeresurus species in databanks.All the pitviper venom CRISPs contain 221 amino acid residues (Figure 3A,B), whereas those from true vipers may contain 220 residues [21].The 16 Cys residues and sequences in their N-terminal half, including the pathogenesis-related protein-1 (PR-1) domain [22], are highly conserved.Tgc-CRa is acidic, and its sequence is 95% identical to that of Ook-CR [15] and >99% similar to the venom CRISPs of Bothriechis schlegelii and Protobothrops species (Figure 3A).In contrast, Tgc-CRb is basic and most similar to serotriflin from the blood of P. flavoviridis [23] and a serotriflin-like protein (i.e., Pmu-CRL) from P. mucrosquamatus; it is also similar to some basic CRISPs of elapid venom (Figure 3B), and these CRISPs are possibly also expressed in tissues other than venom glands.As shown in Table 1, Tables S1 and S2, Tgc-CRa and Tgc-CRb were eluted in the HPLC fractions 14 and 16, respectively (Table 1, Figure 2), and the content of Tgc-CRa was higher than that of Tgc-CRb.

Three VEGFs Are Expressed in the T. gracilis Venom Gland
Both the tissue and the venom-types of VEGFs have distinct biochemical properties and are common components of most viperid venom [24,25].Here, we cloned and sequenced the cDNAs encoding three distinct VEGFs from T. gracilis venom glands, which were deposited to GenBank with accession numbers OQ614863-OQ614865, for Tgc-VGFa, Tgc-VGFb, and Tgc-VGFc, respectively.Their sequences were aligned with possible orthologous snake venom VEGFs retrieved from a BLAST search, respectively (Figure 4A,B).Thus far, NCBI databases do not contain any venom VEGF sequences from other Trimeresurus species.Apparently, the 154-residue Tgc-VGFa is identical to or >99% similar to the VEGFs expressed in the venom of O. okinavensis, Protobothrops, and some New World pitvipers, and those present in the venom of true vipers (subfamily Viperinae) (Figure 4A).Both Tgc-VGFb and Tgc-VGFc contain 122 residues and are approximately 98% similar to each other; both are 93% identical to the sequence of the Ook-VGF protein (Figure 4B) and eluted in fractions 10 and 15, respectively (Table 1, Figure 2).Both types of VEGFs contain conserved receptor-binding residues, and their C-terminal residues contain potentially basic regions responsible for binding heparin (Figure 4A,B).

T. gracilis Venom Proteome
The major toxin families expressed in the venom of most pit vipers are metalloproteases, phospholipase A 2 , serine proteases, and snaclecs [26,27], which are also present in the T. gracilis venom.At least 12 different protein families have been identified in T. gracilis venom, with eight major families (SVMPs, SVSPs, PLA 2 s, LAAOs, CRISPs, DISs, VEGFs, and snaclecs) comprising 79.8% of all the venom components (Figure 1C).Additionally, low levels of NGFs, PLBs, HYAs, 5 NTs, and cystatins were identified (Supplementary Table S2).Other unidentified components of T. gracilis venom may mainly include low-molecular-weight peptide families, such as bradykinin-potentiating peptides, C-type natriuretic peptides, and tripeptidyl SVMP inhibitors [28].To further verify the presence of these peptides in the T. gracilis venom, we searched for both trypsin-digested and non-trypsin-digested sequences by mass spectrometry using the non-redundant NCBI database.We detected two proteoforms, bradykinin-potentiating and natriuretic peptides, in fractions 4-6 by partially matching previously published sequences of O. okinavensis and Bothrops atrox (see Supplementary Table S3).Other minor venom enzymes, such as glutaminyl cyclase, aminopeptidase, and phosphodiesterase [26], are expected to be present in T. gracilis venom, but this has not been confirmed.
Among the 10 Tgc-SVSP variants [13], relatively high levels of Tgc-KN1, Tgc-KN4, Tgc-PAH1/2, and Tgc-PA3 were detected by proteomic analysis (Table 1; Supplementary Table S2).Among the five reported Tgc-SVMPs [14], Tgc-MP (PI class), Tgc-PIIc, and Tgc-PIII were clearly present in the venom (Table 1).In addition, two CRISP variants (Tgc-CRa and Tgc-CRb) and two venom-type VEGF variants (Tgc-VGFb and Tgc-VGFc) were also identified (Table 1).Venom VEGFs usually comprise 2-5% of the pit viper venom proteome [29], and Tgc-VGFb and VGFc contribute 2.9% of the venom proteome (Figure 1C).We previously cloned and isolated an acidic PLA 2 (Tgc-E6W30) from Tgc venom (collected much earlier, not from Mt. Daxue) with a total yield of approximately 6% (w/w) [12], which is consistent with the relative abundance of acidic PLA 2 s in the T. gracilis venom proteome (6.6%; Supplementary Table S2).Other minor acidic PLA 2 variants, or possibly an E6A30-PLA 2 , may also be present in the Tgc-venom analyzed in the present study, which could be highly similar to the acidic PLA 2 s isolated from O. okinavensis and G. brevicaudus (formerly G. halys or G. blomhoffii) (Table 1 and Table S1).A proteoform eluted by HPLC in fraction 26 (Supplementary Table S1) was assigned as Tgc-K49 by searching on the NCBI database, but it is more likely to be an acidic PLA 2 variant for three reasons.(1) Basic K49-PLA 2 homolog usually eluted earlier than acidic PLA 2 s from the RP-HPLC column in 0.1% TFA, but this proteoform was eluted in fraction 26 like other acidic PLA 2 s (eluted in fractions 22-28).( 2) Venom content of K49-PLA 2 homologs is usually higher than those of the enzymatically active PLA 2 s, but this proteoform was detected only once and based on two peptides which match a mutated region (residues 70-100) in the Tgc-K49 sequence, and the region happens to be highly similar to the corresponding regions in Tgc-E6W30 and Ook-E6A30 PLA 2 s [12].(3) The high number of acidic PLA 2 s proteoforms detected (Table 1 and Supplementary Table S2) strongly suggests the presence of more than one acidic PLA 2 isoforms in T. gracilis venom and this proteoform is likely a E6A30-PLA 2 .

Comparison of Venom Composition and Toxicity among Closely Related Species
Tgc-MP (a PI-SVMP) is probably as hemorrhagic as okinalysin because of their 95% sequence similarity [14,16].By acting synergistically with other venom components, abundant Tgc-MP may play a crucial role in the pathophysiology of T. gracilis envenoming.One of the prominent pathologies associated with the hemorrhagic PI-SVMPs is the development of extensive blistering [30], which may become a reservoir of venom toxins that can continuously damage the local tissues.Another distinct function of a number of PI-SVMPs is their ability to activate potent inflammatory responses directly [31].Our results reveal high structural similarities between the venom proteins of T. gracilis and O. okinavensis; not only are the full sequences of their venom PLA 2 s, PI-SVMPs, CRISPs, and VEGFs most similar, but also the tryptic peptide sequences of their LAAO, PLB, and HYA match each other (Supplementary Table S2).Although T. gracilis is phylogenetically closely related with O. okinavensis, the toxicity of O. okinavensis venom (LD 50 11 µg/g mouse, via intravenous injection; [32]) is much weaker than that of T. gracilis venom (LD 50 3 µg/g mouse, via intraperitoneal injection; Tsai et al., unpublished data).The difference in their lethality could be partly explained by the differential expression of their venom toxins, i.e., SVSPs are dominant in O. okinavensis venom [15,33], whereas SVMPs are dominant in T. gracilis venom.T. gracilis venom promotes hemorrhage, hypotension, and impaired blood coagulation, which is consistent with mammalian predation by adult T. gracilis.O. okinavensis venom is comprised of overwhelmingly abundant SVSPs and fewer SVMPs (Figure 5), apparently representing a hybrid strategy optimized mainly for frogs [16], in addition to small mammals.potension, and impaired blood coagulation, which is consistent with mammalian predation by adult T. gracilis.O. okinavensis venom is comprised of overwhelmingly abundant SVSPs and fewer SVMPs (Figure 5), apparently representing a hybrid strategy optimized mainly for frogs [16], in addition to small mammals.and O. tonkinensis [33], Gloydius brevicaudus [34], Protobothrops mucrosquamatus and Viridovipera stejnegeri [35], Crotalus atrox [36] and C. lannomi [37], T. albolabris, and T. purpureomaculatus [38].SVMP, snake venom metalloproteinase; SVSP, snake venom serine protease; PLA2, phospholipase A2; LAAO, L-amino acid oxidase; CRISP, cysteine-rich secretory protein; DIS, disintegrin; VEGF, vascular endothelial growth factor; snaclec, C-type lectin-like protein.
The venom proteomes of many Asian and New World pit viper species have recently been reported [27,37,38].We are able to compare the venom proteome of T. gracilis with those of hemorrhagic and phylogeographically related pit viper species [7,8], as shown in Figure 5. SVMPs are the most prominent toxin family in the venom of T. gracilis, G. brevicaudus, P. mucrosquamatus, C. atrox, C. lannomi, and V. stejnegeri; however, the proportions of their PI-, PII-, and PIII-SVMPs are rather diverse (Supplementary Figure S1).Snaclecs are dominant in the venom of T. albolabris and T. purpureomaculatus compared to other species in Figure 5, and the venom proteome of both arboreal species are not similar to that of T. gracilis.Of note, both T. gracilis and C. atrox venom are most abundant in SVMPs and SVSPs, followed by acidic PLA2s, while SVSPs are the most abundant family in Ovophis venom [15,33] that appears to lack DISs (Figure 5).It is also recognized that T. gracilis is the Asian sister of the New World pitvipers [6][7][8] and T. gracilis and C. atrox share high sequence similarities in their PIII-SVMPs and some SVSP variants [13,14].Possibly because of similar diet ecology in adults, T. gracilis and C. atrox share the venom proteome with grossly similar proportions of the major toxin families (Figure 5), and their lethalities to mice are close, as LD50 of C. atrox venom is 5.0 μg/g mouse for intraperitoneal injection or 2.72 μg/g mouse for intravenous injection [39,40].Comparison of the venom proteome of Trimeresurus gracilis to those of other related pitvipers.
The venom proteomes of many Asian and New World pit viper species have recently been reported [27,37,38].We are able to compare the venom proteome of T. gracilis with those of hemorrhagic and phylogeographically related pit viper species [7,8], as shown in Figure 5. SVMPs are the most prominent toxin family in the venom of T. gracilis, G. brevicaudus, P. mucrosquamatus, C. atrox, C. lannomi, and V. stejnegeri; however, the proportions of their PI-, PII-, and PIII-SVMPs are rather diverse (Supplementary Figure S1).Snaclecs are dominant in the venom of T. albolabris and T. purpureomaculatus compared to other species in Figure 5, and the venom proteome of both arboreal species are not similar to that of T. gracilis.Of note, both T. gracilis and C. atrox venom are most abundant in SVMPs and SVSPs, followed by acidic PLA 2 s, while SVSPs are the most abundant family in Ovophis venom [15,33] that appears to lack DISs (Figure 5).It is also recognized that T. gracilis is the Asian sister of the New World pitvipers [6][7][8] and T. gracilis and C. atrox share high sequence similarities in their PIII-SVMPs and some SVSP variants [13,14].Possibly because of similar diet ecology in adults, T. gracilis and C. atrox share the venom proteome with grossly similar proportions of the major toxin families (Figure 5), and their lethalities to mice are close, as LD 50 of C. atrox venom is 5.0 µg/g mouse for intraperitoneal injection or 2.72 µg/g mouse for intravenous injection [39,40].Nevertheless, the results of comparing the proteomic data from different studies may be confounded by the variations in the protein detection method, ages of the snakes, or other factors (summarized in Supplementary Table S4), and need to be explained with caution.For example, conditions of pre-treatment by trypsin could be inconsistent in the proteomic studies, in-solution tryptic digestion provided a higher number of proteins identified, and a larger sequence coverage for bottom-up proteomic studies, as compared to using in-gel digestion [41].

Sequence Comparison of T. gracilis Venom CRISPs
CRISPs are generally not abundant in snake venom, but are widely distributed taxonomically.The presence of CRISP toxins with high degrees of sequence similarity in all snakes suggests earlier diversification of CRISPs before the divergence between Viperidae and the remaining Colubroids [42,43].The 19-residue signal peptides of CRISPs are highly conserved and favorable for cDNA cloning and sequencing using PCR.In the present study, two venom CRISPs, Tgc-CRa and Tgc-CRb, were fully sequenced for the first time, and a single CRISP transcript was identified in the O. okinavensis transcriptome (Figure 3).Tgc-CRa is highly similar to CRISPs identified in the venom of O. okinavensis, Gloydius and Protobothrops (Figure 3A).Their C-terminal Cys-rich domain (CRD) contained three highly conserved disulfide bridges and a proline bracket [44] (Figure 3A,B).Triflin (from P. flavoviridis) and ablomin (from G. blomhoffi) are L-type Ca 2+ -channel antagonists of arterial smooth muscle contractions that promote vasodilation and hypotension.CRISPs purified from Bothrops venom species may induce inflammatory responses and interfere with complement pathways, generating bioactive fragments (C3a, C4a, and C5a) and anaphylatoxins [45].Similar to triflin, both Ook-CRa and Tgc-CRa contain hydrophobic residues at Phe 189 , Met 195 , Tyr 205 , and Phe 215 , which were shown by crystallographic studies to obstruct the target ion channels, and the highly conserved Glu 186 and Phe 189 are the most likely functional residues [46].In contrast to most of the known venom CRISP sequences, an N-glycosylation site was present at N 48 in serotriflin and N 44 in Pgu-CRX2 (Figure 3B).In both Tgc-CRb and serotriflin, Phe 189 is replaced by Tyr 189 , and the "Pro-bracket" regions 84-90 show low similarities to those of Tgc-CRa and triflin; thus, they are unlikely to bind identical ion-channels.

Sequence Comparison of T. gracilis Venom VEGFs
We deduced that the protein sequences of three Tgc-VEGF variants, Tgc-VGFa of 154 amino acid residues, appeared to be tissue type-specific variants and similar to human VEGF-A (Figure 4A), whereas Tgc-VGFb and Tgc-VGFc of 122 residues were snake venom types (Figure 4B), which are strongly hypotensive toxins [29].As pointed out previously [25], the structures of tissue-type VEGFs (or VEGF-A) are highly conserved among venomous snakes and even among all vertebrates (Figure 4A), whereas those of venom-type VEGFs (also annotated as VEGF-F) are highly diversified in the regions around the receptor-binding loops and C-terminal putative coreceptor-binding regions (Figure 4B) and show different affinities to heparin [29].Ook-VFa (AB852007.1),154 residues long, is the most highly expressed VEGF in O. okinavensis venom [15].In contrast, T. gracilis venom contains mainly Tgc-VGFb and Tgc-VGFc but not Tgc-VGFa (Table 1), and Tgc-VGFb is 93% identical to Ook-VFb (Figure 4B).As expected, Tgc-VGFb and Tgc-VGFc may increase vascular permeability, cause hypotension, and facilitate the spread and transport of toxin molecules, particularly when synergized with the KN subtype of SVSP [24,29].

Conclusions
Our proteomic profiling of T. gracilis venom was facilitated by using both comprehensive and more specific or restricted databases resulting from extensive cDNA sequencing of the three major toxin families (PLA 2 , SVSP, and SVMP) and the resolution of full sequences of T. gracilis venom CRISPs and VEGFs in the present study.Our results demonstrate that T. gracilis venom toxins qualitatively resemble those of O. okinavensis rather than other Trimeresurus and Protobothrops species, which is consistent with the close phylogeographic linking of T. gracilis and O. okinavensis [7,19].The differential expression of venom proteins of T. gracilis and O. okinavensis (Figure 5) can be explained by the adaptation of both species to different environments and prey ecologies.Being an Asian sister of the New World pit vipers, T. gracilis retains some ancient venom genes (e.g., PIII-SVMP) that bear high sequence similarity to the corresponding toxin genes of some hemorrhagic Crotalus species [14].In contrast to most Crotalus venom, T. gracilis venom lacks crotaminelike myotoxins and highly diversified PII-SVMPs, but is rich in hemorrhagic PI-SVMPs and VEGF-F.The relatively abundant Tgc-MP, Tgc-KN1 and Tgc-KN4, and Tgc-VGFb and Tgc-VGFc could explain the tissue damage, hypotension, and coagulopathy observed in T. gracilis envenoming.It has been demonstrated that using pan-specific effective antivenoms immunized with venoms from only a few species of pitvipers could treat the envenoming by other pitvipers if the immunizing venoms contain toxin families that are representative of the species to which the antivenom is targeted [47].Our results suggest that antivenom prepared with stronger antigenicity against pitvipers' PI-SVMPs could be a better candidate to treat T. gracilis envenoming.It is also possible to test whether the antivenom against hemorrhagic Crotalus venom (or adding it to the Taiwan bivalent hemotoxic-antivenom) could effectively treat T. gracilis envenoming.Further studies on the genome and taxonomy of T. gracilis and the pharmacology of its venom toxins are required to clarify its conservation status as well as its venom pathophysiology.

Animals and Venom
We collected four adult T. gracilis (3 females and 1 male) samples from Mount Daxue, Central Taiwan, from September 2020 to December 2022.The corresponding author of this study is responsible for the taxonomic identification of snakes.According to Lin and Tu [9], snakes with a snout-vent length (SVL) larger than 22 cm were considered adults.The mean ± SEM (range) of the SVL and body mass of each collected snake was 50.3 ± 9.46 (44.0-64.0)cm and 64.0 ± 14.1 (48.0-79.0)g, respectively.Venom samples were collected manually at intervals of 14 d or more after venom collection or snake feeding.The wet venom yield per first collection of each snake was 47.0 ± 28.9 (10.7-81.2) mg.Crude venoms from the four T. gracilis specimens were pooled in equal proportion, lyophilized in an FD-series freeze-dryer and CES-series centrifugal evaporator (Panchum Scientific Corp., Kaohsiung, Taiwan), and stored at −80 • C until analysis.

Determination of Venom Protein Concentration
The lyophilized venom samples were dissolved with ultrapure water and centrifuged 10,000× g at 4 • C for 10 min, and the protein concentration of the supernatant was determined in triplicate using a protein assay dye (Bio-Rad, Hercules, CA, USA) with bovine serum albumin as calibration standard.

Reverse-Phase High-Performance Liquid Chromatography
T. gracilis venom containing 1.0 mg venom proteins was reconstituted in 20 µL ultrapure water and subjected to C 18 reverse-phase fractionation using an HPLC system (Chromaster 5160 Pump and Chromaster 5410 UV detector, Hitachi, Tokyo, Japan).The C 18 column was pre-equilibrated with 0.1% TFA in water (Buffer A) and eluted with 0.1% TFA in ACN (Buffer B) at a flow rate of 1.0 mL/min, using a linear gradient of 5% B for 5 min, 5-10% B for 5 min, 10-20% B for 30 min, 20-30% B for 5 min, 30-60% B for 90 min, and 60-70% B for 5 min.The protein elution was monitored at 215 nm and fractions were collected manually, lyophilized, and stored at −80 • C until use.

Sodium Dodecyl Sulfate-Polyacrylamide Gel Electrophoresis
Protein fractions collected during RP-HPLC were further analyzed by SDS-PAGE according to the method described by Laemmli [48].The ExcelBand™ 3-color broad-range protein marker (5-245 kDa) was used as a calibration standard.Approximately 5 µg of protein from each fraction was loaded to the 12.5% polyacrylamide gel under reducing conditions and electrophorized at 110 V for 2 h.The gels were stained with Coomassie Brilliant Blue R-250 (Bio-Rad) and de-stained for visualization.

In-Solution Tryptic Digestion and Peptide Identification by Mass Spectrometry
After lyophilization, 10 µg of protein from each fraction was reduced with DTT, alkylated with IAM, and hydrolyzed with trypsin at an enzyme:substrate ratio of 1:25.The resulting peptides were desalted using MIcroSpin™ columns according to the manufacturer's protocol (Cytiva, Amersham, UK).The samples were lyophilized, reconstituted in 5% ACN/0.1% FA in water, and subjected to nanoscale electrospray ionization liquid chromatography-tandem mass spectrometry (nano-ESI-LCMS/MS) using a Dionex Ultimate 3000 RSLC system (Thermo Scientific, Waltham, MA, USA) coupled with a Q Exactive mass spectrometer (Thermo Scientific).Samples were loaded in a C 18 column (75 µm × 150 mm, 2 µm, 100 Å) (Thermo Scientific Acclaim™ PepMap™, Waltham, MA, USA) at a flow rate of 0.25 µL/min.The injection volume was 5 µL per sample and the mobile phase was 0.1% FA in water (Solution A) and 0.1% FA in 95% ACN (Solution B).The gradient applied was: 1% B for 5.5 min, 1-30% B for 39.5 min, 30-60% B for 3 min, 60-80% B for 2 min, 80% B for 10 min, 80-1% B for 5 min, and 1% B for 5 min.The ion polarity was set to positive ionization mode.Spectra were acquired in MS/MS mode with an MS scan range of 200-3000 m/z and an MS/MS scan range of 50-3000 m/z.The 10 most intense ions from the MS scan were subjected to fragmentation for MS/MS spectra.Data were analyzed with PEAKS Studio 10.5 (Bioinformatics Solutions Inc., Waterloo, ON, Canada), and the peptide-mass-finger-printing results were searched based on the non-redundant NCBI database of Serpentes (taxid:8570).Carbamidomethyl was used for static modification, and oxidation was used for dynamic modification.Protein/peptide identification was validated using the following filters: protein false discovery rate (FDR) ≥1% and unique peptides ≥1; the protein/peptide found was based on the identity of partial sequences.Keratin peptides were eliminated from further analyses.The relative abundance (%) of individual proteins in each chromatographic fraction was determined following previous methods [49,50]: Relative abundance of protein Q (%) = (mean spectral intensity of protein Q in fraction R/total mean spectral intensity in fraction R) × AUC of fraction R from HPLC (%) The area under the curve (AUC) for each collected fraction was automatically integrated and determined from the HPLC chromatogram using Chromaster software (Hitachi, Tokyo, Japan).For each protein identified in the individual fraction, the number of spectra, the number of unique peptides, the "mean spectral intensity of protein Q in fraction R", and "total mean spectral intensity in fraction R", as well as other detailed data, are provided in Supplementary Table S1.

Molecular Cloning and Sequence Determination of CRISPs and VEGFs
T. gracilis venom cDNAs were prepared from venom gland mRNAs as described previously [12].To amplify and clone the cDNAs encoding venom CRISPs, PCR was conducted using SuperTaq DNA polymerase with a pair of mixed-base oligonucleotide primers [51] designed according to the highly conserved cDNA regions in the nucleotide database.Primer 1 was designed in the sense direction: TTCA(A/C)AACA(A/G)(C/T)AGAAATG, and primer 2 was designed in the antisense direction: GATGCTACA(T/C)AG(T/G)CTTGTG [52].DNA fragments of approximately 1.0 kb were amplified by PCR, as shown by electrophoresis of the products on a 1% agarose gel, and harvested.
The abbreviation of T. gracilis (Tgc) was used to name novel peptides.cDNAs encoding both the venom and serum types of Tgc-VEGFs were cloned using different sets of degenerate primers for PCR amplification.Specific primer pairs were designed based on the conserved nucleotide sequences previously used for venom VEGFs [25,53].The PCR-amplified DNA products were analyzed on a 1% agarose gel and harvested.After treatment with polynucleotide kinase, the amplified cDNAs were inserted into the pGEM-T easy vector (Promega Corp., Madison, WI, USA) and transformed in Escherichia coli strain JM109.White transformants and cDNA clones were selected.The DNA Sequencing System (model 373A) and TaqDye-Deoxy terminator-cycle sequencing kit (PE Applied Biosystems, Waltham, MA, USA) were used to determine the nucleotide sequences.The protein sequences of T. gracilis venom CRISPs and VEGFs were deduced from their nucleotide sequences.

BLAST Analyses and Sequence Alignments
Protein-to-protein BLAST (BLASTp) was used to retrieve the most similar sequences for each of the novel Tgc-CRISP and Tgc-VEGF variants, using the non-redundant NCBI database (http://www.ncbi.nlm.nih.gov(accessed on 21 March 2023)).The retrieved toxin homologs were from a broad selection of pitviper genera, and preferably those have been purified and characterized.The sequences were aligned using Clustal X2 [54] and MUSCLE [55] in MEGA X [56]; gaps were introduced to optimize the comparison, and % identities or % similarities of the sequences were calculated using the sequence manipulation suite [57].