Proteomics and antivenomics of Echis carinatus carinatus venom: Correlation with pharmacological properties and pathophysiology of envenomation

The proteome composition of Echis carinatus carinatus venom (ECV) from India was studied for the first time by tandem mass spectrometry analysis. A total of 90, 47, and 22 distinct enzymatic and non-enzymatic proteins belonging to 15, 10, and 6 snake venom protein families were identified in ECV by searching the ESI-LC-MS/MS data against non-redundant protein databases of Viperidae (taxid 8689), Echis (taxid 8699) and Echis carinatus (taxid 40353), respectively. However, analysis of MS/MS data against the Transcriptome Shotgun Assembly sequences (87 entries) of conger E. coloratus identified only 14 proteins in ECV. Snake venom metalloproteases and snaclecs, the most abundant enzymatic and non-enzymatic proteins, respectively in ECV account for defibrinogenation and the strong in vitro pro-coagulant activity. Further, glutaminyl cyclase, aspartic protease, aminopeptidase, phospholipase B, vascular endothelial growth factor, and nerve growth factor were reported for the first time in ECV. The proteome composition of ECV was well correlated with its biochemical and pharmacological properties and clinical manifestations observed in Echis envenomed patients. Neutralization of enzymes and pharmacological properties of ECV, and immuno-cross-reactivity studies unequivocally point to the poor recognition of <20 kDa ECV proteins, such as PLA2, subunits of snaclec, and disintegrin by commercial polyvalent antivenom.

Tandem mass spectrometry coupled to protein database search has evolved as a gold standard for snake venom protein identification [11][12][13] . Although, several isotope labeling methods such as SILAC, ICAT, and iTRAQ, are applied for mass spectrometry-based protein quantification 24 , albeit some limitations of these techniques may restrict their use for quantification of venom proteins 24,25 . On the contrary, the label free protein quantification techniques such as spectral count (MS2) and area-based (MS1) methods in addition to circumvent the above disadvantages are also comparable to other approaches of protein quantification including isotope labeling and mass spectral peak intensities 26 . Nevertheless, label free protein quantification methods are not devoid of technical difficulties like protein size, availability of trypsin cleavage sites along the protein, and lack of a comprehensive snake venom protein databases 24,27 . Moreover, the relative abundance of a protein in snake venom invariable depends on the number of peptides identified which in turn is influenced by the degree of sequence similarity between homologous sequences present in the database 27 . Therefore, in this study to avoid unambiguous protein identification as well as to improve the protein quantification, the semi-tryptic sequences were also considered, the spectral count as well as the area under peptides were normalized by the number of theoretical peptides 13,24,27 , and at least one overlapping distinct peptide was considered for ECV protein identification 13 . Further, to enhance the relevance of the venom proteomic analysis we have compared the ECV proteome with the published reports on venom proteome composition of congeneric snakes from other locales 5 .
In the present study, approximately 10% of the collected MS/MS spectra were matched to peptide sequences with high confidence. The collected MS/MS spectra usually show 10-30% high confidence matching to peptide sequences and several factors, for example existence of MS/MS spectra with non-canonical fragment ions that are excluded by database search engines, absence of the peptide sequences in the target database, and unanticipated post-translational modifications can be attributed to this effect 28 . The ESI-LC-MS/MS analyses of tryptic peptides generated from the gel filtration fractions of ECV showed sequence similarities with more than 1500 proteins deposited in Viperidae venom protein databases. Our stringent protein identification process; however, eliminated redundancies and led to a final recognition of 90 distinct enzymatic and non-enzymatic proteins belonging to 15 snake venom protein families of ECV (Table 1 Table S5a,b). Because mass-spectrometry-based proteomic analysis is database-dependent 29 ; therefore, it is quite reasonable to understand that the difference in number of proteins identified in ECV proteome by searching the different protein databases was dependent on the database entry of genus or species-specific snake venom proteins. Consequently, several protein families such as vascular endothelial growth factor (VEGF), Kunitz-type protease inhibitor (KSPI), aminopeptidase (APase), NT, and glutaminyl cyclase (GC) of ECV identified by searching the Viperidae databases were not identified in Echis as well as E. carinatus venom protein databases due to limitations in genus and species-specific database entries.
The proteomic composition of E. c. carinatus venom, when searched against species-specific protein databases ( Fig. 2c) corroborates well with the published transcriptomic data of E. c. sochureki venom gland (Fig. 2e) 5 ; nevertheless, the minor variation in the relative abundances of venom proteins in these two congeneric snakes reflects the sub-species-specific difference and/or geographical variation in snake venom composition. Notably, the LAAO identified in E. c. sochureki venom by transcriptomic approach 5 was not identified by species-specific database analysis of ECV, although this enzyme was identified in the venom under study by biochemical analysis as well as by genus-specific database search (Fig. 2b, Supplementary Table S3a,b). This result suggests the absolute necessity of species-specific transcriptomic database in proteomic study 29,30 . Further, identification of few homologous protein isoforms in ECV by searching the TSA sequences of E. coloratus may presumably be explained on the basis of venom variability between the two species of Echis.
Due to lack of species-specific (E. c. carinatus) transcriptomic database we have applied both MS1 and MS2-based quantitative approaches for determining the relative protein abundance in ECV and compared their results with each-other. The relative abundance of each protein class calculated by both MS1 (area based) and MS2 (spectral count)-based methods did not show significant deviation ( Supplementary Fig. S1a-h); therefore, the relative abundance of ECV proteins is presented as an average data determined from MS1 and MS2-based methods ( Fig. 2a- Supplementary Fig. S3. It is noteworthy to mention that among the LC-MS/MS-identified protein families, GC, aspartic protease (ASPro), APase, phospholipase B (PLB), VEGF, and nerve growth factor (NGF) were shown to be present in ECV for the first time. By proteomic analysis, the enzymatic proteins identified in ECV include SVMP, PLA 2 , snake venom serine proteases (SVSP), LAAO, NT, APase, PLB, GC, ASPro, and phosphodiesterase (PDE) (Fig. 2c). Proteases of snake venoms are broadly classified as SVMP and SVSP 31 . The proteomic analysis suggested the presence of enzymes from both groups of proteases in ECV; nevertheless, the species-specific database search demonstrated predominance of SVMP group in ECV (Fig. 2c) which corroborates well with the venom proteins composition of E. c. sochureki determined by transcriptomic analysis (Fig. 2e). Remarkably, the relative abundance of SVMP in ECV (Fig. 2c) was found to be lower as compared to SVMP content determined by proteomic analysis in the venoms of its congeneric species such as E. coloratus, E. ocellatus, E. pyramidum leakeyi and E. c. sochureki 5 . Interestingly, the relative abundance of SVMP in ECV varies significantly depending on the target databases ( Fig. 2a-d) owing to different number of entries of venom proteins in these databases. The biochemical analyses suggested that GF 1, followed by GF 2, exhibited the highest metalloprotease activity (Table 2), which was supported by LC-MS/MS analysis of the gel filtration peaks (Table 1). Further, SVMPs are grouped to PI, PII and PIII classes based on their size and domain structure 32 . Analysis of LC-MS/MS data against Viperidae venoms showed that ECV was predominated by PIII-SVMPs (17.9% of the ECV proteome) ( Table 1; Fig. 2c). Because SVMP is the most abundant component in Echis venom, therefore the significant variation in SVMP content among Echis spp. may lead to differences in severity of clinical manifestations and pharmacological activity post Echis spp. bite.   Significant esterolytic activity exhibited by proteins was eluted in GF 2 to GF 5 (Table 2), which may be correlated with the presence of quantitatively higher amounts of pro-coagulant serine proteases, in these fractions (Table 1). Depending on target database search the relative abundance of SVSPs was found to vary in ECV proteome ( Fig. 2a- Table 1). The relative abundance of SVSP (determined by searching the species-specific protein databases) in ECV was found to be comparable with the relative abundance of this class of protease determined by proteomic analysis in the venom of other species of Echis 5 .
The molecular mass of PLA 2 from snake venoms is reported to be in the range of 10-15 kDa 33 . In accordance with its molecular weight, it was mostly eluted in the GF 7-9 fractions (Table 1). In the case of Russell's viper venom (RVV), PLA 2 s have been shown to interact with high molecular weight components of venom leading to their elution in the initial gel filtration fractions of RVV 11,12 . Interestingly, the present study also found the elution of trace quantities of PLA 2 in GF 3 ( Table 1). The relative abundance of PLA 2 determined by analyzing data against Viperidae and Echis venoms (2a-b) was found to be two-fold then its relative abundance determined by species-specific database search (2c) which may presumably due to less entry of this important class of venom protein in latter database. The relative abundance of PLA 2 in ECV (10.9%) determined by species-specific database search (Fig. 2c) was comparable to the relative abundance of the same enzyme in venoms of E. coloratus, E. ocellatus, and E. c. sochureki albeit PLA 2 content of E. pyramidum leakeyi venom was found to be higher 5 which may have an evolutionary significance 34 .
ATPase, ADPase, AMPase, and PDE are poorly characterized, high molecular mass (>50 kDa) proteins of snake venom 20,35 . Our biochemical assays showed that ECV contains ATPase (6.3 ± 0.006 × 10 4 U/mg), ADPase (3.6 ± 0.300 × 10 4 U/mg), AMPase/nucleotidase (8.2 ± 0.4 × 10 3 U/mg), and PDE (18.6 ± 0.08 U/mg) enzyme activities. Nevertheless, ATPase and ADPase from ECV could not be identified by proteomic analysis (in the present study) owing to a limited protein database deposition 11,12 or by transcriptomic analysis of venom glands of Echis spp 5,30 . Occurrence of PDE in ECV could not be identified when the LC-MS/MS data were searched against the genus and species-specific databases (Fig. 2b,c) or by the transcriptomic analysis of E. c. sochureki venom gland for the reason stated above; however, searching the MS/MS data against Viperidae venom proteins resulted in identification of PDE as a minor component of the ECV (Table 1, Fig. 2a) 11,12,36 .
LAAOs are high molecular weight (60-150 kDa) snake venom enzymes 19,37 . ECV demonstrated an LAAO specific activity of 0.3 ± 0.01 U/mg. Depending upon the venom protein / transcriptomic database search the relative abundance of LAAO isoenzymes in ECV proteome demonstrated significant variation (2.1-6%) (Fig. 2a,b,d, Table 1, Supplementary Tables 2-5). However, due to limited database entry LAAO could not be identified in ECV proteome by species-specific database search (Fig. 2c, Supplementary Table S4a,b). The LAAO was also reported to be a minor component in venom of other species of Echis 5,30 .
Hyaluronidase activity, at the tested dose of 3 µg/ml, was not found by the biochemical assay or by the LC-MS/ MS analysis of ECV or by proteomic and transcriptomic analyses of venoms of Echis spp 5 . Nevertheless, Urs et al. 38 and Girish et al. 39 demonstrated the presence of hyaluronidase in ECV, albeit they used a much higher concentration of venom (200 µg/ml) in their enzyme assay.
C-type lectin (snaclec), the Ca 2+ -dependent non-enzymatic proteins 9,40 of snake venom, are found to be the most abundant non-enzymatic group of proteins in the ECV proteome; however, depending on non-redundant protein / TSA sequence database search their relative occurrence in ECV was demonstrated to vary between 24-34% ( Fig. 2a- (Fig. 2c) was 3-4 folds higher as compared to the snaclec content in other species of Echis 5 . Because snaclec are one of the major components of Echis venom responsible for adverse pharmacological effects in bite victims (see below) therefore, this species-specific variation in snaclec content in different species of Echis may have a profound clinical significance. The disintegrins are cysteine rich non-enzymatic proteins characterized with a conserved arginine-glycine-aspartic acid (RGD) motif 41 . They were the second most abundant non-enzymatic proteins in ECV (Fig. 2c) and among the identified disintegrins, 5 isoforms (gi|82194569, gi|544584743, gi|182705265, gi|82203514, and gi|182705262) belong to dimeric disintegrin sub-family. The disintegrin content of ECV (14%), determined by species-specific protein database search (Fig. 2c), was found to be slightly higher than the disintegrin content determined by proteomic as well as transcriptomic analysis of E. c. sochureki venom 5 .
Apart from the above mentioned proteins, single isoforms of actin and myosin were also identified by proteomic analysis against Viperidae family of venom proteins (data not shown). Nevertheless, they are not venom components and may be contamination in the ECV from the venom extraction. Considering their likely origin and low abundance (0.3%), they were not considered in calculating the relative abundance of proteins in ECV. The yield of venom per milking from an adult saw-scaled viper (0.8-1.0 feet in length) is approximately 12 mg 43 . Therefore, the concentration of ECV in the blood of an adult human after a full bite would be expected to be in the range of 2.5-3.0 µg/ml. The assays of blood coagulation activity suggest that crude ECV (3.0 µg/ ml) and GF 1-6 fractions were strongly pro-coagulant, whereas the proteins from GF 8 and 9 demonstrated anticoagulant activity ( Table 2). The pro-coagulant activity of ECV is also more pronounced than the activity displayed by crude RVV 12,43 . SVMPs and SVSPs have been shown to influence the hemostatic system by activation of blood coagulation factor Xa, factor V, prothrombin, and by virtue of thrombin-like activity 11,46 . The abundance of SVMPs and SVSPs in the ECV proteome (Fig. 2c), compared to the RVV proteome 12 , makes this venom more pro-coagulant. The SVMPs and SVSPs are usually mid-and high-molecular mass proteins (>33 kDa) of Viperidae venoms 18,47,48 , and they were separated in the GF 1-6 fractions ( Table 2), which is supported by previous reports 11,12 . Subsequently, the low molecular mass proteins (<30-10 kDa) that are eluted in GF 7-10 gel filtration fractions, exhibited anticoagulant activity (Table 2). These fractions are predominated by PLA 2 s, KSPIs, and disintegrins (Table 1), and their anticoagulant properties have been well characterized 23,49,50 .

The proteome composition of ECV is well correlated with its in vitro
SVMPs and SVSPs primarily interfere with the blood coagulation cascade of victim/prey 42,46 that ultimately leads to consumption coagulopathy and the prolongation of whole blood clotting time of E. carinatus bite patients in southern India 45 (personal communication from Dr. A. Zachariah, CMC, Vellore). Interestingly, SVMPs were detected in all gel filtration fractions by proteomic analysis (Table 1) and by biochemical assay (Table 2). Carinactivase 18 and Ecarin 50 possessing prothrombin activation property have been well characterized as high molecular weight (>50 kDa) SVMPs from ECV. Although our stringent protein identification process did not identify the above proteases in ECV, the remarkable prothrombin activation exhibited by ECV and its gel filtration fractions, GF 1-3 (Fig. 3a,b), may have been due to the presence of several pro-coagulant SVMPs and RVV-X activator-like proteins in ECV (Tables 1 and 2).
Fibrinogenolytic activity was prominent in ECV, and especially, the GF 1-5 fractions due to elution of proteases (SVMPs and SVSPs) in these fractions (Fig. 3c,d). The occurrence of these enzymes was also evident from biochemical and proteomic analyses (Tables 1 and 2). Furthermore, the above fractions are rich in esterase activities, predominantly BAEE-esterase activity ( Table 2), which is often exhibited by SVSPs that possess fibrinogenolytic activity 47,48 . The crude venom (or GF 1-5) proteins degraded the Aα chain of fibrinogen leaving β and γ chains intact (Fig. 3c), which suggests that ECV is predominated by α-fibrinogenase. Further, it has been well documented that prothrombin-activating SVMPs and procoagulant SVSPs together play a predominant role in inducing coagulation and in in vivo defibrinogenating activities 42,47,48 . Interestingly, tryptic peptides VIGGDECNIN and TSTYIAPLSLPSSPPR of Factor V activator pro-coagulant serine protease isoenzymes 47 and a thrombin-like serine protease (Russelobin) 48 , detected in this ECV proteome, have also been shown to cause in vivo defibrinogenation in mice 47,48 . Further, crude ECV and its gel filtration fractions (GF 2-5) exhibited remarkable fibrinogen clotting activity leading to the formation of tenuous fibrin clots ( Table 2). This pharmacological property can be correlated to the presence of Russelobin-like serine proteases (gi|311223824) 48 in these fractions (Table 1). Taken together, defibrinogenation (consumption of fibrinogen) and prothrombin as well as factor V (FV) activation, induced by the abundant SVMPs and SVSPs in ECV, may be responsible for the prolongation of blood coagulation in bite victims, which is a major clinical symptom of E. c. carinatus envenomation 46 .
SVMPs are often associated with additional cysteine-rich, and disintegrin domains that are cleaved off by proteolysis or autolysis during venom secretion 32 . Further, in some cases, C-type lectin domains (snaclec) are found to be linked with PIII-SVMPs by disulfide linkage 32 . Among them, the CRISP and snaclec domains are reported to trigger inflammation by recruiting inflammatory cells at the bite site 51 . Further, Ecarpholin (accession no. gi|163311140), a basic PLA 2 isolated and characterized from ECV was predicted to be myotoxic in nature 21 . It was also detected by LC-MS/MS analysis in the current study (Table 1). Since this basic myotoxic PLA 2 can induce edema in the footpad of experimental Swiss albino mice 52 , the persistent local swelling and edema observed in Echis bite patients 43,44 can be correlated with the presence of significant amounts of CRISPs, snaclecs and PLA 2 s in the ECV proteome (Fig. 2a-c).
Further, other symptoms of envenomation, such as hemoptysis, haematemesis, and haematuria, observed in a few patients bitten by E. carinatus 43,44 (personal communication from Dr. A. Zachariah, CMC, Vellore) may be correlated to the existence of PLA 2 s, SVMPs, and SVSPs in ECV (Table 1, Fig. 2c). Hemorrhage and local tissue damage is yet another important clinical symptom observed in Echis bite patients 43,44,53 . Among the SVMP classes, PII and PIII SVMPs are associated with non-metalloproteinase domains (disintegrin and cysteine-rich) that are reported to target extracellular matrix (ECM) and cause hemorrhage 54 . Therefore, this clinical feature can be correlated with the abundance of both the sub-classes of SVMPs in southern India ECV proteome. Hypotension and shock, the two other prominent clinical symptoms of Echis envenomation 43 are probably attributed to venom ATPase and VEGF (Fig. 2c) 55 as well as coagulopathy and bleeding.
A decrease in the circulatory platelet count (thrombocytopenia) after Echis bite is a very common clinical finding; 43,44 and ECV and its gel filtration fractions induced loss of platelet integrity in vitro, albeit this effect was more pronounced by GF 1 and 2 proteins (Table 2) containing abundant SVMPs and snaclecs (Tables 1 and 2). Because the above two proteins collectively account for snake venom-induced thrombocytopenia 12,56,57 , it could be concluded that a high proportion of the SVMPs and snaclecs in ECV (Fig. 2c) are liable for the observed thrombocytopenia induced by E. carinatus venom in human (Table 2). Although ECV did not exhibit significant aggregation or deaggregation of washed platelets (data not shown), it rapidly (<60 s) induced agglutination of platelet rich plasma (PRP). A heterodimeric snaclec Echicetin, purified from venom of E. carinatus from unknown regions of India and identified in southern India ECV (accession nos. gi|802148, gi|32452854, gi|40889261, gi|2829697) ( Table 1)   the only choice for the treatment of snakebite. Nevertheless, the safety and efficacy of antivenoms are major concerns for efficient antivenom treatment. Therefore, the immuno-recognition of different commercial polyvalent antivenoms towards ECV and its gel filtration fractions was studied by ELISA, Western blot, and immuno-chromatographic (second generation antivenomics) analyses. Results from the ELISA experiment indicated that immuno-recognition of high (>45 kDa) and mid molecular weight (20-45 kDa) proteins was significantly higher (p < 0.05) than the recognition of low molecular mass proteins (<20 kDa) of ECV for all tested PAVs (Fig. 4a). This finding essentially reflects the poor immunogenicity of low molecular weight ECV proteins, or alternatively, the geographical source of ECV used for raising antivenom was different than the locale of the ECV used in this study. The corroboration of results from ELISA and Western blotting analysis also showed that ECV proteins ranging in mass from 10-20 kDa were less recognized by the tested antivenoms (Fig. 4b).
The immuno-chromatographic approach also showed that mostly the low molecular weight (<20 kDa) venom proteins were not efficiently captured by PAV (Fig. 4c,d). Subsequently, by LC-MS/MS analysis (antivenomics study) these low molecular mass ECV proteins (Fig. 4c) were identified as PLA 2 s, αand β-subunits of snaclecs, disintegrins, and NGFs (Table 3), and the percent of these proteins did not bind to PAV-immuno-affinity column is shown in Fig. 4e. Because PLA 2 s as well as snaclecs play a significant role in ECV-induced toxicity and adverse pharmacological effects in bite victim; 44,58,59 therefore, least neutralization and/or recognition of these components of ECV by commercial PAV is a serious concern for effective antivenom therapy.
The major challenge in managing Echis bite patients is the efficient treatment of local swelling, which does not subside even after several vials of antivenom treatment, thus leaving the victims partially crippled (personal communication from R. Whitaker, Chennai). PLA 2 s are responsible for edema induction or local swelling 44 , whereas snaclecs can induce thrombocytopenia 12,56 . Therefore, the poor immunogenicity of these ECV components may be one of the reasons that explain the persistent local swelling in bite victims even after several vials of antivenom therapy. Henceforth, strategies must be designed either by inclusion of antibodies specifically raised against these low molecular weight toxic venom components or improving their antigenicity so as to mitigate the deleterious effect of the proteins for better hospital management of Echis bite patients. Further, presence of hemorrhagic SVMPs in ECV contributes to the formation of stable neutrophil extracellular trap (NET) at the bite site which forms a barrier for the free flow of blood and prevents antivenom from reaching the damaged site. Therefore, although high molecular weight SVMPs are well recognized and neutralized, antivenom treatment fails in efficient reversal of ECV induce local toxicity and swelling 60 .

Neutralization of enzymatic activity and pharmacological properties of ECV by commercial polyvalent antivenom highlights the requirement of a well-designed immunological protocol for antivenom production.
Our previous studies have demonstrated a good correlation between enzyme function and the pathophysiology of snake envenomation [11][12][13] . The efficient treatment of ECV bite will depend on effective neutralization of enzymatic activities and pharmacological properties of this venom. Commercial PAVs showed variable inhibitions of the enzymatic and pharmacological properties of ECV, which again, may be correlated to the source of ECV used for antivenom production. The enzymatic activities exhibited by high molecular mass proteins (>50 kDa), such as LAAO, PDE, SVMP, ATPase, ADPase, AMPase were efficiently neutralized by the antivenom (Fig. 5a); however, the enzymatic activities of mid and low molecular weight proteins such as PLA 2, SVSP, Nα-p-Tosyl-L-arginine methyl ester hydrochloride (TAME), and Nα-Benzoyl-L-arginine ethyl ester hydrochloride (BAEE) exhibited mostly by SVSPs were least neutralized by antivenom (Fig. 5a). Interestingly, these proteins (except PLA 2 ) were well recognized by PAV (Fig. 4b,c,d), suggesting that the catalytic sites of the enzymes may be poor immunogens. Further, the negligible neutralization of PLA 2 activity suggests that the low molecular mass (~14-15 kDa) pharmacologically active proteins of ECV also serve as poor antigens (Fig. 5a).
Interference with the hemostatic system is the major pharmacological effect of ECV (Table 2); therefore, the potency of commercial antivenoms to neutralize this property of crude ECV was also investigated. All of the PAVs at 1:10 (venom:antivenom) efficiently neutralized the pro-coagulant activity and loss of platelet integrity induced  Table 3. List of proteins identified by in-gel trypsin digestion and subsequent ESI-LC-MS/MS analysis of the ECV proteins poorly recognized by commercial PAVs. To enhance the ECV protein coverage, the data was searched against Viperidae family of proteins. by ECV (Fig. 5b). This may be correlated to the efficient neutralization of pro-coagulant SVMPs (Fig. 5b), indicating a good correlation between metalloprotease activity of Viperidae venom and the above pharmacological properties 12 .

Fractionation of ECV through gel filtration chromatography and SDS-PAGE analysis of crude ECV.
2 mg of ECV (dry weight) was dissolved in 20 mM Tris-Cl, pH-7.4 buffer containing 150 mM NaCl (buffer A) and centrifuged at 10,000 rpm for 10 min. The supernatant was filtered through 0.2 µ membrane syringe filter and protein content was estimated by the Lowry method 61 . Then 0.1 ml filtrate containing 1.5 mg of protein was fractionated on a Shodex KW-803 gel filtration column (8 × 300 mm, 5 µm) equipped with Dionex Ultimate 3000 UHPLC system (Thermo Fisher Scientific, USA). The column was pre-equilibrated with the above buffer and the flow rate was adjusted to 15 ml/h at room temperature (~23 °C). The elution of protein was monitored at 280 nm and fractions of 0.25 ml were collected. The peaks were pooled for the enzymatic activity assay, pharmacological characterization, and LC-MS/MS analysis. Protein content of crude ECV and fractions was determined 61 and from a standard curve of BSA, the concentration of unknown protein was determined at 660 nm.
The crude ECV was analyzed on 12.5% SDS-PAGE in both reduced and non-reduced conditions. Protein bands were visualized by PhastGel Blue R stain (GE Healthcare, Sweden). The molecular masses of the ECV proteins were also determined by MALDI-TOF-MS analysis (4800 MALDI TOF/TOF ™ Analyzer, Applied Biosystems) as described earlier 12,48 . The masses were determined in the m/z ranges of 5-20, 21-40, 41-100, and >100 kDa.

Proteomic identification of ECV proteome by LC-MS/MS analysis of GF fractions.
To identify the venom proteome, 80 µg of each gel filtration fraction was subjected to ESI-LC-MS/MS analysis [11][12][13] . The venom proteins were digested with sequencing grade trypsin (13 ng/μL in 10 mM ammonium bicarbonate containing 10% acetonitrile) at an enzyme substrate ratio of 1:30 at 37 °C for 18 h. The tryptic peptides were desalted, concentrated using ZipTip C 18 (Merck, USA) and reconstituted in 0.1% formic acid. They were separated on a Zorbax 300SB-C 18 analytical column (75 μm × 150 mm, 3.5 μm, Agilent) at a flow rate of 300 nL/min applying the following mobile phase gradient: from 11% B for 5 min, 11 to 25% B in 20 min, 25 to 53% B in 16 min, 53 to 100% B in 5 min, 100% B for 4 min, and then 11% B for 4 min. Solvent A and B were 0.1% formic acid and 80% acetonitrile containing 0.1% formic acid, respectively. The eluted peptides were then analysed on an LTQ Orbitrap Discovery hybrid mass spectrometer (ThermoFisher Scientific, Bremen, Germany) interfaced to an Agilent 1200 HPLC via a Nanomate Triversa (Advion BioSciences, Ithaca, NY). The ionization voltage was set at 1.7 kV.
The raw data was acquired and processed by Xcalibur software (ThermoFisher Scientific, Bremen, Germany) in a data-dependent acquisition (DDA) mode with 1 MS survey scan followed by 5 MS/MS scans. The full-scan MS spectra were acquired in the FT mode in the scan range of m/z 300−2000 (lock mass was set to 445.12 corresponding to polysiloxane) with a resolution of 30, 000 (full width at half-maximum). MS/MS fragmentation was collision-induced dissociation (CID) in linear mode with the following triggering conditions: minimum signal intensity, 10,000; charge state, +2, +3; maximum injection time for MS/MS, 500 ms; and isolation width, 2 amu.
The LC-MS/MS data were searched independently against the entries in non-redundant NCBI database with taxonomy set to (a) Viperidae (taxid: 8689 with 56,902 protein entries), (b) Echis (taxid: 8699 with 788 protein entries) and (c) Echis carinatus (taxid: 40353 with 162 protein entries). The data were analyzed by PEAKS 8.5 software (Bioinformatics Solutions Inc., Ontario, Canada). Carbamidomethylation of cysteine, oxidation of methionine, deamidation of asparagine and glutamine and pyro-glu of N-terminal glutamine residues were set as variable modifications. Precursor and fragment mass tolerances were set to 10 ppm and 0.8 Da, respectively, up to two missed cleavages were allowed and non-specific cleavage at one end (semi-tryptic) was considered. The false discovery rate (FDR) was kept very stringent (0.1%). To exclude the contaminating proteins from identification, the contaminant database with 115 protein entries (ftp://ftp.thegpm.org/fasta/cRAP/crap.fasta) was included in the database search process. All the redundant peptides were removed from the data set and thereafter each protein entry was manually verified. The following criteria were set for the purposes of protein identification: (a) only matching proteins and peptides showing a −10 log P value ≥ 30 and 20, respectively, and (b) occurrence of at least one overlapping distinct peptide.
The relative abundance of the venom proteins was determined by MS1 (area under peptide) as well as MS2 (spectral count) based label-free methods. For both the methods, the sum of areas or the spectral count was normalized by number of theoretical peptides and the normalized values were calculated using equation 1. The number of theoretical peptides for each identified protein was determined using MassSorter v3.1 software. Thereafter, the relative abundance of a protein in a particular GF fraction was calculated using equation 2:

= ×
Relative abundance of X in GF fraction Y mean area or spectral count of X in Y total mean areas or spectral count of all proteins in Y protein yield of Y (%) (2) The average of the relative protein abundance of ECV determined by two different methods (MS1 areas-based and MS2 spectral count) was considered to represent the relative abundance of ECV venom proteome.
In addition, the MS/MS data were searched against Transcriptome Shotgun Assembly (TSA) sequences of Echis coloratus venom, the only translated Echis protein sequence database in NCBI (BioProject: PRJEB2884, 87 translated protein entries) and analyzed using PEAKS 8.5 software. The search parameters were the same as stated under LC-MS/MS data analysis. After removing the redundant peptides the translated protein entries were manually verified. The matching proteins and peptides showing a −10 log P value ≥ 30 and 20, respectively and occurrence of at least one overlapping distinct peptide were the pre-requisite for protein identification 12,13 . The relative abundances of ECV proteins were determined by MS2 mean spectral count and MS1 area-based methods (equations 1 and 2). The data was presented as an average of the relative protein abundance of ECV determined by two different methods.
Biochemical characterization. Esterolytic activity of crude ECV and its GF fractions were checked by the spectrophotometric method using Nα-Benzoyl-L-arginine ethyl ester hydrochloride (BAEE) and Nα-p-Tosyl-L-arginine methyl ester hydrochloride (TAME) as substrate 48 . One unit of TAME and BAEE-esterase activity is defined as an increase in absorbance of 0.01 at 254 nm and 244 nm, respectively, during the first 5 min of the reaction at 37 °C. PDE activity was assayed by the spectrophotometric method using bis-p-nitrophenyl phosphate as a substrate 62 . One unit of PDE activity is defined as micromoles of p-nitrophenol released per min. The protease activity of crude ECV and its GF fractions was determined by incubating 3.0 µg/ml of crude ECV or 0.5 µg/ml of GF fraction or 1X PBS (control) with fibrinogen (2.5 mg/ml) for 3 hours at 37 °C. The fibrinogen degradation products were separated by 12.5% SDS-PAGE (reduced) and percent fibrinogenolytic activity was calculated by measuring the degradation of the Aα chain of fibrinogen. The band intensity of Aα chain of fibrinogen molecule after crude ECV treatment was considered as 100% fibrinogenolytic activity and other values were compared to that. The protein band intensities were measured by Image Quant TL 8.1 software (GE Healthcare, Sweden). PLA 2 activity of crude ECV (3 µg/ml) and GF fractions (0.5 µg/ml) was assayed by the turbidometric method 63 . One unit of PLA 2 activity is defined as a decrease in 0.01 absorbance at 740 nm after 10 min of incubation. L-kynurenine was used as a substrate for screening of L-amino acid oxidase (LAAO) activity of crude ECV (3 µg/ml) and GF peaks (0.5 µg/ml). The unit of LAAO activity was defined as nmol kynurenic acid produced/min under the assay conditions and specific LAAO activity was expressed as unit/mg protein 11,12 . ATPase and ADPase activity of crude ECV (3 µg/ml) and GF fractions (0.5 µg/ml) was assayed by the method of Williams and Esnouf 64 with slight modifications as described by Mukherjee et al. 11 The 5′-nucleotidase (AMPase) activity was determined according to the protocol of Sinsheimer and Koerner 65 with slight modifications described in our previous publication 11 . One unit of ATPase/ADPase/AMPase activity was defined as μM of Pi released per min at 37 °C.
Hyaluronidase activity of crude ECV (3 µg/ml) was assayed as described previously 12 . One unit of enzyme activity was defined as a decrease in turbidity by 1% as compared to the control and activity was expressed as U/ mg protein 66 . SVMP activity was assayed by using azocasein as a substrate. Crude ECV (3 µg/ml) or GF peaks (0.5 µg/ml) were added to reaction mixture and incubated at 37 °C for 10 min. The activity was checked by the spectrophotometric method and specific activity was expressed as ΔA 450nm /min/mg protein 11 . Pharmacological characterization. Goat blood obtained from slaughter house was collected in 3.8% tri sodium citrate and the blood was centrifuged at 4300 rpm for 10 min at 4 °C 11,63 . The pellet was discarded and the yellowish supernatant was termed as platelet poor plasma (PPP) and it was used within 4 hours after its collection. The anticoagulant activity of crude venom or its fraction on PPP was determined by Ca-clotting time 11,63 . For the control, 1X PBS instead of venom was added to PPP. One unit of coagulant or anticoagulant activity was defined as a decrease or an increase of 1 second of clotting time of PPP incubated with crude ECV or GF fractions, compared to control PPP 63 . Prothrombin time (PT) and activated partial thromboplastin time (APTT) of crude ECV (3 µg/ml) and GF peaks (0.5 µg/ml) were determined using commercial diagnostic kits following the instructions of the manufacturer 63 .
The prothrombin activation (FXa-like activity) property of crude ECV and GF peaks, if any, was analyzed by 12.5% SDS-PAGE of human prothrombin incubated with ECV or fractions or FXa (control) 42 . Formation of thrombin from prothrombin was quantified by measuring the density/intensity of thrombin (36 kDa) band in Image Quant TL software 8.1 (GE Healthcare, Sweden) 49 . Presence of thrombin-like enzyme activity was ascertained by measuring the fibrinogen clotting activity of crude ECV (3 μg/ml) and the GF peaks (0.5 μg/ml) 48 . Loss of platelet integrity induced by crude ECV or GF peaks was determined by incubation of crude ECV (3 µg/ml) or GF peaks (0.5 µg/ml) with washed platelet or Tyrode's solution (control) at 37 °C for 6 hour in a CO 2 incubator.
The platelets were stained with trypan blue and counted in the hemocytometer using Motic Images plus 3.0 ML software 12 .
Immunological cross-reactivity, antivenomics, neutralization of enzymatic activities, and tested pharmacological properties of crude ECV by commercial polyvalent antivenom. The immunological cross-reactivity of commercial PAVs towards ECV and its GF fractions was studied by ELISA and western blot analysis, as described previously 11,12 . Briefly, for ELISA experiment, 100 ng protein of ECV or GF fractions was coated in 96 well microtiter ELISA plate (in triplicate) for overnight at 4 °C and washed three times with 1X PBS buffer containing 0.05% tween-20 (wash buffer). Two hundred ng of PAV was used as primary antibody and incubated for 2 h at room temperature followed by washing with wash buffer. Anti-horse IgG HRP conjugated secondary antibody (1:2000 dilution) was incubated for 2 h at room temperature to detect the primary antibody. Color was developed by adding substrate (1X TMB/H 2 O 2 ) to the well for 30 minutes in dark condition and reaction was stopped by adding 50 µl of 2 M H 2 SO 4. The absorbance was taken at 492 nm against blanks in Multiskan GO (Thermoscientific, USA) microplate reader.
Western blotting experiment was performed by running 100 µg protein of ECV (reduced) in 12.5% SDS-PAGE and transferred to PVDF membranes 12 . Non-specific bindings were blocked by 5% fat free skimmed milk for overnight at 4 °C. After that the membranes were washed with 1X TBS with 0.05% tween-20 (washing buffer). Primary antibodies (15 mg/ml PAVs) at a dilution of 1:1000 ratios were incubated for 1 h at room temperature. Thereafter the membranes were washed and ALP conjugated secondary antibody (1:15000 dilution) was incubated for 2 h at room temperature. Blots were developed using BCIP/NBT substrate kit (Sigma Aldrich, USA) and densitometry scanning (Epson America, Inc) was done.
Immunological cross-reactivity of crude ECV against PAV (PSVPL) was assessed by an immunoaffinity chromatographic approach 67 . Briefly, after coupling the NHS-activated Sepharose 4 Fast Flow column (GE Healthcare) with 15 mg PAV (PSVPL), the excess antivenom was washed with 1X PBS pH 7.4 buffer 67 . Then, 500 µg of crude ECV was loaded onto the column and incubated at 37 °C for 2 h in a shaker orbiter. Unbound venom proteins were eluted with 5 column volumes of 1X PBS, pH 7.4, desalted in PD 10 column (GE Healthcare, Sweden) and vacuum dried (Labconco, Model: 7670061, USA). Bound venom proteins were eluted by washing the column with 5 column volumes of 0.1 M glycine, pH 2.0. The pH of the eluted proteins was immediately neutralized by 1 M Tris-Cl, pH 9.0. The bound and unbound fractions, as well as 500 µg of crude ECV were separated by 12.5% SDS-PAGE. The poorly immunogenic, low molecular mass unbound proteins were excised from the gel and subsequently identified by LC-MS/MS analysis against Viperidae protein databases as described above. Percentage of unbound proteins was quantified by spectral count method; the total mean spectral count of PAV unbound proteins was compared with the total mean spectral count of corresponding ECV proteins (100%).
The neutralization potency of polyvalent antivenom (PAV) towards the pro-coagulant activity of crude ECV was determined by pre-incubating PAVs with crude ECV at 1:10 (protein: protein) for 30 min at 37 °C prior to the assay of Ca 2+ clotting time of PPP, and enzymatic activities of venom 11,12,68 . The percent inhibition was calculated by comparing the pharmacological/enzymatic activity of crude ECV in the absence of PAVs (100% activity).
All the experiments were performed according to Tezpur University ethical committee and bio-safety committee guidelines.
Data accessibility. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE 69 partner repository with Project Name "Echis carinatus carinatus (India) venom proteomics" and the dataset identifier PXD007980.