Proteomic Characterization and Comparison of Malaysian Tropidolaemus wagleri and Cryptelytrops purpureomaculatus Venom Using Shotgun-Proteomics

Tropidolaemus wagleri and Cryptelytrops purpureomaculatus are venomous pit viper species commonly found in Malaysia. Tandem mass spectrometry analysis of the crude venoms has detected different proteins in T. wagleri and C. purpureomaculatus. They were classified into 13 venom protein families consisting of enzymatic and nonenzymatic proteins. Enzymatic families detected in T. wagleri and C. purpureomaculatus venom were snake venom metalloproteinase, phospholipase A2, l-amino acid oxidase, serine proteases, 5′-nucleotidase, phosphodiesterase, and phospholipase B. In addition, glutaminyl cyclotransferase was detected in C. purpureomaculatus. C-type lectin-like proteins were common nonenzymatic components in both species. Waglerin was present and unique to T. wagleri—it was not in C. purpureomaculatus venom. In contrast, cysteine-rich secretory protein, bradykinin-potentiating peptide, and C-type natriuretic peptide were present in C. purpureomaculatus venom. Composition of the venom proteome of T. wagleri and C. purpureomaculatus provides useful information to guide production of effective antivenom and identification of proteins with potential therapeutic applications.


Introduction
Snake venom is a highly complex mixture of proteins and polypeptides with a myriad of biological activities. It functions as an important tool for defense against predators, prey immobilization, and facilitation of prey digestion [1]. There are two families of terrestrial venomous snake that cause the majority of envenoming cases in Malaysia: Elapidae and Viperidae. Among the potentially dangerous snakes that belong to the viperid subfamily Crotalinae (pit vipers) are Tropidolaemus wagleri (Temple or Wagler's pit viper) and Cryptelytrops purpureomaculatus (mangrove pit viper) [1]. Envenomation from these pit vipers is often associated with intense pain, local swelling, necrosis, hemorrhage, and blood pressure disruption [1].
Tropidolaemus wagleri is recognized as a primitive and unique pit viper, morphologically as well as biochemically [2,3]. Enzymatic properties of the venom highlight similarity of venom components to other vipers and include nonlethal enzymes such as phosphodiesterase, thrombin-like enzyme (TLE), L-amino acid oxidase (LAAO), and phospholipase A 2 (PLA 2 ) [4]. Lethal components of T. wagleri venom were identified as low molecular-weight toxins, namely waglerins-1, -2, -3, and -4 [5]. These toxins were classified as neurotoxins due to their competitive antagonism with the nicotinic acetylcholine receptor (nAChR) [6,7]. This feature is considered unusual for snakes of the family Viperidae, as the presence of postsynaptic neurotoxins is well-known in the venoms of elapid snakes [8]. Cryptelytrops purpureomaculatus is known to be very aggressive and is commonly found on mangrove mudflats on the coastal region of west peninsular Malaysia. The color of C. purpureomaculatus is variable, ranging from grayish to olive and brownish purple [1,9]. The venom is known to be nonfatal to humans but causes local swelling and pain [10,11]. The venom was found to have anticoagulant, thrombin-like, hemorrhagic, and other enzymatic activities, and it is more toxic than several other Asian arboreal viper species [12].
Information on the venom proteome is important for understanding and predicting the clinical consequences of envenomation and for formulating an effective antivenom that will target and neutralize venom components common to viper species [13]. At present, online protein database searches using UniProt [14] for "Tropidolaemus wagleri" and "Cryptelytrops purpureomaculatus" listed only five and four proteins, respectively. Hence, much is still unknown about the proteome of T. wagleri and C. purpureomaculatus venoms. Here we performed proteomic profiling and compared the crude venom composition of T. wagleri and C. purpureomaculatus using a shotgun-proteomics approach. Shotgun-proteomics and liquid chromatography tandem mass spectrometry (LC-MS/MS) have been described as a rapid, bottom-up proteomic techniques that allows direct analysis of simple and complex protein samples to generate a global profile of the total protein components [15,16].

The Venom Proteome of T. wagleri and C. purpureomaculatus
The venom proteomes of T. wagleri and C. purpureomaculatus have never been completely reported using proteomic techniques such as shotgun-proteomic and LC-MS/MS. Shotgun-proteomics and LC-MS/MS approaches have been used to characterize many other snake venom proteomes, including kraits and cobras [17,18].

T. wagleri Venom
Earlier studies on T. wagleri venom have demonstrated the presence of enzymatic proteins commonly occurring in pit viper venom such as phosphodiesterase, phosphomonoesterase, hydrolase, TLE, LAAO, and PLA 2 [4]. The venom lethality was comparable to other pit viper species venom but distinct due to the lack of hemorrhagic activity [4,5]. The lethal toxins were then identified as low molecular-weight peptides; waglerins [4,5]. In the public protein database (UniProtKB), only five proteins were reviewed and listed under the heading of Tropidolaemus wagleri: waglerin-3 (WAG13_TROWA), SNACLEC trowaglerix subunit alpha (SLA_TROWA), waglerin-4 (WAG24_TROWA), SNACLEC trowaglerix subunit beta (SLB_TROWA), and NADH-ubiquinone oxidoreductase (NU4M_TROWA). In our study, mass spectrometry analysis of T. wagleri venom detected 13 different proteins that shared similar sequences with proteins in the database (SwissProt, Serpentes) ( Table 1). Protein families that were not reported previously in the venom, (e.g., snake venom metalloproteinase (SVMP)) were detected in the present study (Table 2).

Venom Composition of T. wagleri and C. purpureomaculatus
The venom composition of T. wagleri and C. purpureomaculatus was divided into enzymatic and nonenzymatic protein families (Tables 3 and 4). Enzymatic protein families were found to be the major venom protein family in T. wagleri and C. purpureomaculatus venom with 62% and 73% of the total proteins, respectively (Figures 1 and 2). Nonenzymatic protein families of T. wagleri and C. purpureomaculatus venom were detected at 38% and 27% of the total proteins, respectively.    Table 3. Enzymatic and nonenzymatic proteins from T. wagleri crude venom.

Enzymatic Protein Families of T. wagleri and C. purpureomaculatus Venom
SVMP was identified to be 7% of the total proteins detected in T. wagleri (Table 5). In contrast, it is the major venom protein family representing 35% of the total proteins in C. purpureomaculatus, 5 times more compared to T. wagleri ( Table 5). The high abundance of SVMP in C. purpureomaculatus is consistent with the biochemical studies on the venom [19,20]. SVMPs are one of the most important components of the viperid venoms, causing hemorrhage by affecting blood coagulation and/or integrity of extracellular matrix components such as collagen, laminin, and fibronectin [21][22][23]. SVMPs are classified into three classes, P-I to P-III, based on their domain structure [24]. The majority  SVMP was identified to be 7% of the total proteins detected in T. wagleri (Table 5). In contrast, it is the major venom protein family representing 35% of the total proteins in C. purpureomaculatus, 5 times more compared to T. wagleri ( Table 5). The high abundance of SVMP in C. purpureomaculatus is consistent with the biochemical studies on the venom [19,20]. SVMPs are one of the most important components of the viperid venoms, causing hemorrhage by affecting blood coagulation and/or integrity of extracellular matrix components such as collagen, laminin, and fibronectin [21][22][23]. SVMPs are classified into three classes, P-I to P-III, based on their domain structure [24]. The majority of SVMPs that were detected in both T. wagleri and C. purpureomaculatus belong to P-II and P-III classes (Tables 1 and 2). P-II class contains metalloproteinase and disintegrin domains while P-III class contains metalloproteinase, disintegrin-like, and cysteine-rich domains [24]. Biological activities of P-II SVMPs include proteolytic induction [25], platelet aggregation inhibition [26], and hemorrhagic [27]. Most P-III SVMPs were identified as hemorrhagic and have the most potent biological activity among the three SVMP classes [24]. Additionally, some P-III SVMPs can induce apoptosis in vascular endothelial cells [28][29][30]. The detection of SVMP in T. wagleri is intriguing, as several earlier reports have shown the absence of hemorrhage activity from the venom [4,31]. Higher sensitivity and resolution of LC-MS/MS compared to biochemical methods could be the reason for the detection of SVMPs in the T. wagleri venom. SVMPs from T. wagleri have never been isolated and studied, therefore there is a need to study these SVMPs and elucidate their biological activity and function. It is possible that intraspecific snake venom variations and regional variations could contribute toward the presence of SVMPs in the venom [32,33]. More extensive study using samples from various regions where T. wagleri is endemic is required to confirm this possibility. LAAO and PLA 2 were other major components in the venoms (Table 5). LAAO and PLA 2 activities of T. wagleri and C. purpureomaculatus venoms have been described in other studies [4,19]. LAAO is a ubiquitous component of snake venom, but its abundance varies between species [34]. This was demonstrated by our present findings where the percentage of LAAO was found to be lower in T. wagleri (8%), compared to C. purpureomaculatus (10%) ( Table 5). The LAAOs that were detected in this study showed sequence similarity with LAAO from several genera including Calloselasma, Gloydius, Viridovipera, Bothrops, Daboia, and Cerastes (Tables 1 and 2). LAAOs in T. wagleri and C. purpureomaculatus could be partially or synergistically responsible for various biological effects upon envenomation, including hemorrhage [35], edema [36], and platelet aggregation [37,38]. Venom PLA 2 may interrupt normal physiological processes, causing various pharmacological effects such as neurotoxicity, myotoxicity, and cardiotoxicity [39][40][41]. The PLA 2 isoforms that were detected in T. wagleri and C. purpureomaculatus venom showed sequence similarity with PLA 2 that were described in the other viper species venoms including Bothrops, Trimeresurus, and Ovophis venoms (Tables 1 and 2). PLA 2 detected from both species were identified as basic and acidic isoforms of PLA 2 (Tables 1 and 2). The presence of PLA 2 isoforms in T. wagleri was consistent with the discovery of acidic and basic variants from Sulawesi and Sumatran origin Tropidolaemus genus [42].
Other families of enzymatic proteins detected in T. wagleri and C. purpureomaculatus include snake venom 5 -nucleotidase, venom phosphodiesterase (vPDE), and PLB. Each of these proteins was detected at 8% in T. wagleri and 2% in C. purpureomaculatus (Table 5). Snake venom serine protease (SVSP) was determined at 8% in T. wagleri and 12% in C. purpureomaculatus (Table 5). Glutaminyl-peptide cyclotransferase (QPCT) was demonstrated in C. purpureomaculatus at 2%, but was not found in T. wagleri (Table 5). Snake venom QPCT has been suggested to have an indirect contribution to venom toxicity via posttranslational modification of venom proteins [43,44]. QPCT is important in the N-terminal glutamine cyclization that induces toxin maturation, protects from exopeptidase degradation, and/or assists proper protein conformation [44].
The presence and activity of snake venom 5 -nucleotidase and vPDE in T. wagleri and C. purpureomaculatus have been demonstrated by biochemical assays, which agrees with our findings [4,19]. LC-MS/MS data showed that these proteins shared sequence similarity to proteins found in Gloydius brevicaudus and Crotalus adamanteus (Tables 1 and 2). These proteins were thought to inhibit platelet aggregation through the liberation of purines using endogenous source from their envenomed victims [45]. We have also detected the presence of PLB in T. wagleri and C. purpureomaculatus crude venoms, similar to those found in Crotalus adamanteus (Tables 1 and 2). The role of PLB in snake venom was not well understood, but its hemolytic activity has been demonstrated in vitro [46]. SVSP has several biological activities such as platelet aggregation, coagulation, and fibrinolysis [47,48]. In this study, different classes of SVSP were detected from T. wagleri and C. purpureomaculatus including TLE, fibrinogenase, and vPA (Tables 1 and 2). TLE plays a significant role in coagulation process [49]. TLE has been purified and characterized from C. purpureomaculatus, but not from T. wagleri [50]. The antibody developed from purpurase, a TLE isolated from C. purpureomaculatus, was found to strongly react with Trimeresurus complex venom, suggesting sequence homology of TLEs within the complex [50]. This finding is consistent with our LC-MS/MS data that detected TLEs that shared similar sequences with TLEs found in Cryptelytrops albolabris, and Viridovipera stejnegeri ( Table 2). Fibrinogenase affects the blood clotting mechanism through fibrinolytic action [51,52]. vPA is another class of serine protease that activates plasminogen to plasmin, promoting fibrinolysis upon envenomation of its victim [53].
Waglerin was found exclusively in T. wagleri and it is the second largest protein family, which constitutes 15% of the crude venom ( Figure 1, Table 5). Waglerin has been well characterized and considered as T. wagleri's most unique and most lethal protein [5,7,58,59]. To date, four types of waglerins have been described [5,58]. The peptides waglerins-1 and -2 differ by one amino acid at position 10 (waglerin-1: histidine, waglerin-2; tyrosine). Waglerins-3 and -4 are almost homologous to waglerins-1 and -2, respectively, except for two additional amino acids (serine and leucine) at their N-terminal [5,58]. Waglerins identified in this study were noted as waglerins-3 and -4 because the database used in this study indicated that waglerins-1 and -2 corresponded with waglerins-3 and -4, respectively (Table 1). Waglerin exerts its toxicity through competitive antagonism with the nicotinic acetylcholine receptor (nAChR) at nanomolar concentrations [7,58]. Earlier studies found that waglerin-1 selectivity binds to the α-ε nAChR interface with 2000-fold higher affinity compared to other binding sites [6]. The finding of high abundance of waglerin in our study (15%) agrees with several reports that identified the protein as the major toxic component of the venom [4,5,7]. In addition, the percentage of peptide coverage identified by MS spectra correlates with the protein abundance [60]. This data is supported by LC-MS/MS, which showed 100% waglerin peptide detection coverage ( Table 1, Supplementary File 1).
Cysteine-rich venom protein (CRVP), also known as snake venom cysteine-rich secretory protein (CRISP) and BCNP were found specifically in C. purpureomaculatus at 2% of the total venom ( Table 5). The amino acid sequence of CRVP from various snake species has been well characterized, however, the biological activity has not been fully understood [61]. Nonetheless, CRVP isolated from several viper species may exhibit neurotoxicity through the blocking of several ion channels [62,63]. Three different CRVPs were detected from C. purpureomaculatus venom using LC-MS/MS. These CRVPs shared sequence similarity with CRVP from Viridovipera stejnegeri, Protobothrops flavoviridis and a CRVP from C. purpureomaculatus, tripurin (Table 2). BCNP is a unique snake venom protein, and its cDNA encodes both bradykinin-potentiating peptide (BPP) and C-type natriuretic peptides (CNP), thereby integrating two different vasodilating molecules into one [64]. Biological effects of BCNP upon envenomation include hypotension and loss of consciousness [65,66]. In this study, BCNP was found specifically in C. purpureomaculatus at 2% of the total venom (Table 5). BPPs can inhibit angiotensin-converting enzyme activity and, along with CNP, have been investigated for the treatment of human hypertension and several other cardiovascular diseases, including congestive heart failure [65][66][67]. We believe this is the first study to identify the presence of BCNP in C. purpureomaculatus venom.

Conclusions
We have successfully characterized and compared the venom proteome of Malaysian T. wagleri and C. purpureomaculatus by using shotgun-proteomics, LC-MS/MS, and protein de novo sequencing. The present data have revealed the complex composition of the crude venom from both species. The venom proteome of both snakes consists of enzymatic and nonenzymatic protein families. The proteins detected in both T. wagleri and C. purpureomaculatus were SVMP, SVSP, LAAO, PLA 2 , vPDE, snake venom 5 -nucleotidase, and SNACLEC. Neurotoxin waglerin was unique in T. wagleri venom, whereas BCNP and QPCT were unique in C. purpureomaculatus venom. This information may be useful to predict the clinical prognosis after envenoming and provide better guidance for the production of effective antivenom. Moreover, proteins detected in T. wagleri and C. purpureomaculatus venoms could be selectively investigated for therapeutic potentials.

Materials
Snake Venom Crude T. wagleri and C. purpureomaculatus venoms were obtained from Mr. Zainuddin Ismail, a private snake enthusiast from Perlis in Peninsula Malaysia. All snakes originated from the west of Peninsular Malaysia. Snake venoms were collected in a sterile container covered with parafilm. The venoms were transported back to Monash Malaysia campus on ice and frozen at −80 • C before subjecting them to the freeze-drying process. Freeze dried venom was weighed, labeled, and stored at −20 • C until use. Venom samples from three different sampling sessions were used for the experiment.

In-Solution Tryptic Digestion
Approximately 0.5 mg of crude venom was added into 1.5 mL tube in triplicates and mixed with 25 µL of 100 mM ammonium bicarbonate (ABC), 25 µL of trifluoroethanol, and 1 µL of 200 mM 1,4-dithiothreitol (DTT). The mixture was then briefly vortexed, and incubated at 60 • C for 1 h. Next, the protein in the tube was alkylated by adding 4.0 µL of 200 mM iodoacetamide, briefly vortexed, and incubated at room temperature in the dark (covered with aluminum foil) for 1 h. Subsequently, 1 µL of 200 mM DTT was added to the tube and incubated at room temperature in the dark for another 1 h. Double-distilled water and 100 mM ABC were then added to the sample mixture to dilute the protein denaturant and raised the pH to 7-9. Trypsin solution was added to the tubes at the weight ratio of 1:50, briefly vortexed and incubated overnight at 37 • C. On the next day, 1 µL of formic acid was added to stop the trypsin digestion, briefly vortexed, and left in a vacuum concentrator overnight to concentrate the digested proteins. Samples were kept at −20 • C prior to LC-MS/MS analysis.

Nanoflow Liquid Chromatography Electrospray-Ionization Coupled with Tandem Mass Spectrometry (Nanoflow-ESI-LC-MS/MS)
The digested peptides were loaded into an Agilent C18 300 Å Large Capacity Chip (Agilent, Santa Clara, CA, USA) column that was equilibrated with 0.1% formic acid in water (solution A). The peptides were eluted from the column with 90% acetonitrile in 0.1% formic acid in water (solution B) using the following gradient; 3%-50% solution B over 0-30 min, 50%-95% solution B over 2 min, 95% solution B for 7 min, and 95%-3% solution B over 39-47 min. Quadrupole-time of flight (Q-TOF) polarity was set at positive with capillary and fragmenter voltage being set at 2050 V and 300 V, respectively, and 5 L/min of gas flow with a temperature of 300 • C. The peptide spectrum was analyzed in auto MS mode ranging from 110-3000 m/z for MS scan and 50-3000 m/z for MS/MS scan. The spectrum was then analyzed with Agilent MassHunter (Agilent Technologies, Santa Clara, CA, USA) data acquisition software and then PEAKS 7.0 software (Bioinformatics Solutions Inc., Waterloo, ON, Canada).

Venom Protein Identification by Automated de Novo Sequencing (PEAKS Studio 7.0)
Protein identification by automated de novo sequencing was performed with PEAKS Studio 7.0 (Bioinformatics Solution Inc., Waterloo, ON, Canada). SwissProt.Serpentes (May 2015) database was used for protein identification and homology search by comparing the de novo sequence tag. Carbamidomethylation was set as fixed modification with maximum mixed cleavages at 3. Parent mass and fragment mass error tolerance were both set 0.1 Da with monoisotopic as the precursor mass search type. Trypsin was selected as the enzyme used for digestion. False discovery rate (FDR) of 1% and unique peptide ≥2 were used for filtering out inaccurate proteins. A −10lgP score of greater than 20 indicates detected proteins are relatively high in confidence as it targets very few decoy matches above that threshold [68]. The percentage of the venom protein family in the crude venom was calculated using the following formula: no. of proteins (protein family) total proteins detected using LC − MS/MS × 100