Snake Venomics and Antivenomics of Cape Cobra (Naja nivea) from South Africa: Insights into Venom Toxicity and Cross-Neutralization Activity

Naja nivea (Cape Cobra) is endemic to southern Africa. Envenoming by N. nivea is neurotoxic, resulting in fatal paralysis. Its venom composition, however, has not been studied in depth, and specific antivenoms against it remain limited in supply. Applying a protein decomplexation approach, this study unveiled the venom proteome of N. nivea from South Africa. The major components in the venom are cytotoxins/cardiotoxins (~75.6% of total venom proteins) and alpha-neurotoxins (~7.4%), which belong to the three-finger toxin family. Intriguingly, phospholipase A2 (PLA2) was undetected—this is a unique venom phenotype increasingly recognized in the African cobras of the Uraeus subgenus. The work further showed that VINS African Polyvalent Antivenom (VAPAV) exhibited cross-reactivity toward the venom and immunorecognized its toxin fractions. In mice, VAPAV was moderately efficacious in cross-neutralizing the venom lethality with a potency of 0.51 mg/mL (amount of venom completely neutralized per milliliter of antivenom). In the challenge-rescue model, VAPAV prevented death in 75% of experimentally envenomed mice, with slow recovery from neurotoxicity up to 24 h. The finding suggests the potential para-specific utility of VAPAV for N. nivea envenoming, although a higher dose or repeated administration of the antivenom may be required to fully reverse the neurotoxic effect of the venom.


Introduction
Snakebite envenoming is a priority neglected tropical disease designated by the World Health Organization [1]. Each year, it causes 81,000-138,000 deaths and three times as many chronic complications worldwide [2]. Sub-Saharan Africa is one of the most affected regions, with nearly 500,000 cases resulting in more than 30,000 deaths annually. Worse still, this is accompanied by a countless toll of permanent disabilities and amputations [3,4]. The annual burden of snakebite has been recently estimated at 1.03 million DALYs (disabilityadjusted life years), an astonishing figure that is close to or even higher than the burden of many Neglected Tropical Diseases (NTDs) [5].
Toxins 2022, 14, x FOR PEER REVIEW 3 o 17 were minor proteins eluted in the later course of fractionation (beyond 130 min HPLC), consisting of moderate to high molecular weight proteins (>20 kDa) in the veno

Naja nivea Venom Proteome
The tandem mass spectrometry analysis identified a total of 58 proteoforms from all fractions of N. nivea venom (Table 1). Of these, 43 proteoforms were annotated based on peptide sequences that belonged to cobras (Naja spp.; Table 2; Supplementary File S1). These proteins were sorted and categorized into 11 protein families, along with their relative protein abundances ( Figure 2). Three-finger toxins (3FTx) constitute the main protein family, accounting for 84.62% of total venom proteins in which cytotoxins/cardiotoxins (CTX, 75.65%) and α-neurotoxins (short and long neurotoxins, 7.39%) are abundantly expressed. Other proteins with lower abundances (collectively < 20% of total venom proteins) belong to the toxin families of snake venom metalloproteinase (SVMP), cysteine-rich secretory protein (CRiSP), Kunitz-type serine protease inhibitor (KSPI), nerve growth factor (NGF), phosphodiesterase (PDE), 5 -nucleotidase (5 -NUC), L-amino acid oxidase (LAAO), cobra venom factor (CVF), acetylcholinesterase (AChE) and vespryn (VES; Table 2). Mass spectrometric data, including parameters for spectral ions, protein scores and the amino acid sequences of peptides, are provided in the Supplementary Materials (Supplementary File S1). a Protein codes with the prefixes "CL" and "Unigene" were derived from the in-house transcriptome database. b MS search score refers to the protein score based on peptides matched to databases during the protein identification process by LCMS/MS. A score of ≥20 was set as a filter for protein identification in this study. c Protein relative abundance was determined as the percentage of total venom proteins.

Immunological Binding Activity of VAPAV toward N. nivea Venom
In indirect ELISA, the antivenom VAPAV exhibited concentration-dependent immunoreactivity toward the venoms of N. nivea and N. haje (positive control; Figure 3). The binding activities of VAPAV toward N. nivea and N. haje venoms were comparable, as indicated by the values of its maximal absorbance and half-maximal effective concentration (EC 50 ) for the two venoms. Its EC 50 for N. nivea venom was 4.76 ± 0.92 µg/mL, and for N. haje venom, 6.74 ± 1.26 µg/mL (p > 0.05).

Immunorecognition of N. nivea Venom Fractions by VAPAV
The RP-HPLC profiles of the whole venom, immunoretained and non-immunoretained venom proteins, are shown in Figure 4 (panels A, B, and C, respectively). The result illustrates the antivenomics for VAPAV against N. nivea venom, where the antivenom was shown to immunorecognize and bind to the 12 protein fractions of venom at different degrees of immunoreactivity. Comparing between the profiles ( Figure 4A-C), proteins in RP-HPLC fractions 1, 2, and 3, as well as the tailing fraction 12, were most markedly immunocaptured by VAPAV, whereas proteins in fractions 4-8 were least immunoretained. The degrees of immunoretention of venom proteins in these fractions were tabulated as percentages in Table 3. Accordingly, the VAPAV affinity column efficiently immunorecognized venom proteins in fractions 1, 2, and 3 (68-70% immunoretention). The binding activity of VAPAV for venom proteins in fraction 12 was moderate (~58% immunoretention) but low for those in fractions 4-11 (~20-25% immunoretention). The result of immunoretention was consistent with the corresponding non-immunoretained fractions, as shown in Figure 4C, where unbound proteins were washed out from the affinity column.

Immunorecognition of N. nivea Venom Fractions by VAPAV
The RP-HPLC profiles of the whole venom, immunoretained and non-immunoretained venom proteins, are shown in Figure 4 (panels A, B, and C, respectively). The result illustrates the antivenomics for VAPAV against N. nivea venom, where the antivenom was shown to immunorecognize and bind to the 12 protein fractions of venom at different degrees of immunoreactivity. Comparing between the profiles ( Figure 4A-C), proteins in RP-HPLC fractions 1, 2, and 3, as well as the tailing fraction 12, were most markedly immunocaptured by VAPAV, whereas proteins in fractions 4-8 were least immunoretained. The degrees of immunoretention of venom proteins in these fractions were tabulated as percentages in Table 3. Accordingly, the VAPAV affinity column efficiently immunorecognized venom proteins in fractions 1, 2, and 3 (68-70% immunoretention). The binding activity of VAPAV for venom proteins in fraction 12 was moderate (~58% immunoretention) but low for those in fractions 4-11 (~20-25% immunoretention). The result of immunoretention was consistent with the corresponding non-immunoretained fractions, as shown in Figure 4C, where unbound proteins were washed out from the affinity column.

Venom Lethality and Neutralization by Antivenom
The intravenous median lethal dose (i.v. LD 50 ) of N. nivea venom was determined to be 1.11 µg/g in mice ( Table 4). Considering that VAPAV exhibited high immunoreactivity toward the whole venom and was able to immunorecognize most of its protein fractions, the in vivo neutralization efficacy of the antivenom was subsequently examined in mice. In the venom-antivenom preincubation test, VAPAV was able to cross-neutralize the lethality of N. nivea venom with a neutralization potency (P) of 0.51 mg venom per ml antivenom, equivalent to a normalized potency (n-P) of 3.90 mg venom per gram of antivenom protein (Table 4). For comparison, Table 4

Experimental Envenoming and Rescue with Antivenom
The subcutaneous LD 50 of N. nivea venom was 2.42 µg/g (95% C.I. 2.10-2.79) in mice, approximately one-fold higher than the intravenous LD 50 (Table 4). In the experimental envenoming model, mice injected subcutaneously with 5 × s.c. LD 50 N. nivea venom began to show weakness in movement within the first 30 min of envenoming. As the neurotoxicity progressed, the mice showed hind limb paralysis, movement difficulty, and labored breathing. In the group of untreated mice (n = 8), complete paralysis and death ensued within 2 h from the time of experimental envenoming ( Figure 5). In the treated group, an intravenous bolus of VAPAV at 250 µL administered upon the onset of hind limb paralysis (approximately 30 min post-envenoming) rescued most of the mice (75% survival) from the lethality of N. nivea venom ( Figure 5). Mice that were dead (25%) despite antivenom treatment had a marginally delayed death (dying by approximately 3-4 h post-envenoming) in comparison to the untreated mice, which succumbed to death between 1-2 h post-envenoming. The treated mice which survived the experimental envenoming (75%) showed a gradual reversal of paralysis with full recovery (regaining the ability to move and eat freely) observed by 24 h post-envenoming. venom treatment had a marginally delayed death (dying by approximately 3-4 h envenoming) in comparison to the untreated mice, which succumbed to death be 1-2 h post-envenoming. The treated mice which survived the experimental enven (75%) showed a gradual reversal of paralysis with full recovery (regaining the ab move and eat freely) observed by 24 h post-envenoming.

Discussion
In Africa, N. nivea bites are a cause of neurotoxic envenoming [9,14]. Applying tein decomplexation approach, the present study unveiled the first venom proteom nivea, showing the venom composition is predominantly three-finger toxins compr α-neurotoxins (α-NTX), cytotoxins/cardiotoxins (CTX) and weak neurotoxins (WTX α-NTX, though constituting only ~7.4% of the total venom protein, are the principal responsible for neurotoxicity in N. nivea envenoming [14]. The α-NTX of N. nivea h been shown to be highly lethal in mice with an i.v LD50 of 0.08 μg/g [20], similar to NTX of various cobra species with i.v LD50 of approximately 0.1 μg/g in mice (for in [21][22][23]. The high lethality of N. nivea α-NTX is primarily a sequel of the post-sy blockade of nicotinic acetylcholine receptors (nAChR), which disrupts the neuromu

Discussion
In Africa, N. nivea bites are a cause of neurotoxic envenoming [9,14]. Applying a protein decomplexation approach, the present study unveiled the first venom proteome of N. nivea, showing the venom composition is predominantly three-finger toxins comprised of α-neurotoxins (α-NTX), cytotoxins/cardiotoxins (CTX) and weak neurotoxins (WTX). The α-NTX, though constituting only~7.4% of the total venom protein, are the principal toxins responsible for neurotoxicity in N. nivea envenoming [14]. The α-NTX of N. nivea has also been shown to be highly lethal in mice with an i.v LD 50 of 0.08 µg/g [20], similar to the α-NTX of various cobra species with i.v LD 50 of approximately 0.1 µg/g in mice (for instance: [21][22][23]. The high lethality of N. nivea α-NTX is primarily a sequel of the post-synaptic blockade of nicotinic acetylcholine receptors (nAChR), which disrupts the neuromuscular transmission, resulting in paralysis and respiratory failure [24]. The lethal potency of cobra venom, nonetheless, is also modulated by the amount of these neurotoxins in the venom, as it has been established that the relative abundances of α-NTX in cobra venoms are strongly correlated with the venom's lethality [25,26]. In principle, cobra venoms with a high α-NTX abundance (>20% of total venom proteins) have low LD 50 s below 0.2 µg/g (thus more lethal), while those containing less α-NTX (<10%) tend to have high LD 50 above 1 µg/g (thus less lethal), in mice [27,28]. African cobras such as N. nubiae, N. senegalensis, and N. melanoleuca have venoms containing 12-20% α-NTX, and moderate lethal activity with an intermediate i.v. LD 50 (≤0.6 µg/g) in mice [19,29,30]. On the other hand, N. nivea (present study), as with N. nigricollis, N. katiensis, N. pallida, N. mossambica, and N. annulifera have venoms containing α-NTX below 10% of total venom proteins, and therefore higher i.v. LD 50 s (0.9-2.4 µg/g) in mice [30,31]. Despite the somewhat low lethal potency of venom, N. nivea envenoming causes rapid and profound neurotoxicity clinically, which could be attributed to the large amount of venom injected in a bite (considering its medium-to-large body size of up to 1.8 m in length [32]), and the venom's low hu-man equivalent LD 50 (0.09 mg/kg) upon allometric conversion. The other toxins with known neurological effects were named weak toxins (WTX), previously shown to have weak affinity toward neuronal and muscarinic receptors and low toxicity in mice (LD 50 , 5-80 mg/kg) [33,34]. The low abundance of WTX in N. nivea venom indicates its ancillary role in the pathophysiology of envenoming.
Cardiotoxins, interchangeably known as cobra cytotoxins (CTXs), are the most abundant toxins identified in the N. nivea venom proteome (75% of the total venom proteins). The high expression of CTX is consistent with proteomic findings in other African cobras from various subgenera, including Afronaja [30], Boulengerina [29], and Uraeus [19,31]. Compared with the α-NTX, cobra CTXs are relatively less potent in lethal activity, with higher i.v. LD 50 beyond 1 µg/g in mice [22,23,35]. CTXs are usually implicated in local tissue necrosis and venom ophthalmia due to their cytolytic effect, which can be potentiated by synergistic PLA 2 in snake venom [36,37]. The cytolytic activity of CTX varies between CTX proteoforms depending on the amino acid residue present at the tip of the protein's loop II. Based on this characteristic, CTXs are categorized as P-type (Pro30) or S-type (Ser28). Studies on CTX interaction with model lipid membranes demonstrated that the P-type CTXs are more cytolytic than the S-type [38,39], while the latter (S-type CTX) has also been shown to be cytotoxic toward human cell lines [40]. The current study identified three CTX proteoforms, in agreement with the earlier report of three N. nivea cytotoxins (designated as V 1, V 2 and V 3) whose amino acid sequences have been determined and deposited in the database as CTX 1 (Uniprot KB: P01456, S-type), CTX 2 (P01463, P-type) and CTX 3 (P01458, S-type), respectively [13]. Earlier studies, however, demonstrated that N. nivea venom and its CTXs (V 1, V 2, and V 3) had rather weak hemolytic activity in vitro [13,41], and local necrosis has not been reported in in vivo experiments as well as in clinical envenoming cases [14,41]. The present study re-investigated the local tissue-damaging effect of N. nivea venom by inoculating it intradermally in C57BL/6 mice (n = 4 per dose), and found no dermonecrosis even at the highest sublethal dose tested (50 µg venom per mouse; data not shown). The finding suggests a lesser role of N. nivea CTX in the pathogenesis of local tissue necrosis or that the CTXs possibly exhibit other toxic activities, such as cardiotoxicity. Cardiotoxicity is considered a rare complication of cobra bite through disrupting the calcium regulation or mitochondria fragmentation in cardiac muscle cells, resulting in cardiac arrhythmia [42,43]. Earlier studies showed that N. nivea venom caused an unusual form of cardiac failure (systolic arrest of the ventricle) ex vivo in isolated perfused mammalian heart preparation [24,41], but this toxicity has not been well documented in human envenoming cases. It is possible that the abundant CTXs interact in synergism with other toxins present in the venom, thereby contributing to the overall toxicity of the venom. Further studies are warranted to elucidate the biological activity of N. nivea CTX and its correlation with the pathophysiology of envenoming.
In addition to 3FTx, other components detected in N. nivea venom were of low abundances (3-7%). These include snake venom metalloproteinases (SVMP) and cysteine-rich secretory proteins (CRiSP). Cobra SVMPs are zinc-dependent multidomain enzymes that may be instrumental in the activation of the complement system and inflammatory response to envenoming, contributing to the venom's toxicity [44]. CRiSP is a common component in snake venoms with diverse biological activities such as inhibition of smooth muscle contraction, blockade of various cation channels, and induction of hypothermia in prey animals [45,46]. Other minor venom components (<2%) include Kunitz-type serine protease inhibitor (KSPI), nerve growth factor (NGF), phosphodiesterase (PDE), 5 -nucleotidase (5 -NUC), L-amino acid oxidase (LAAO), cobra venom factor (CVF), acetylcholinesterase (AChE), and vespryn (VES). KSPI are small proteins (5-7 kDa) that may be involved in diverse biological activities such as blood coagulation, fibrinolysis, inflammation, and blockade of ion channels [47,48]. NGF possibly acts as an inhibitor of metalloproteinase, preventing SVMP autodigestion [49]. Snake venom PDE and 5 -NUC can act on ATP and AMP molecules to release adenosine, thus facilitating venom spread in envenomed subjects [50,51]. CVF is also known to facilitate venom spread through the activation of the complement system, thus increasing inflammation and vascular permeability [52,53]. LAAO and AChE are both large enzymatic proteins in cobra venoms. Snake venom LAAO has recently been shown to stimulate the production of inflammatory mediators, contributing to local inflammatory effects in snakebite envenoming [54]. AChE may potentiate venom-induced neurotoxicity through the degradation of the neurotransmitter acetylcholine at the neuromuscular junction [55]. VES, first discovered in King Cobra venom, has been shown to induce hyperalgesia and hypolocomotion in prey animals [56]. Although these proteins are classified as putative venom toxins, their exact roles in specific cobra envenoming await further elucidation.
Intriguingly, the present study revealed a lack of phospholipase A 2 (PLA 2 ) in the venom proteome of N. nivea. Although PLA 2 has long been regarded as a ubiquitous venom component across all snake lineages, proteomic and biochemical studies increasingly showed African non-spitting cobras from the Uraeus subgenus have venoms that contain little to no PLA 2 [19,31,57,58]. This is a unique venom phenotype not shared by other African and Asiatic cobras whose venoms contain a significant amount of PLA 2 in general (12-20% of total venom proteins) [25,29,30,59,60]. It is speculated that the loss of PLA 2 in its venom evolution partly results in the lack of cytotoxic and necrotizing properties of N. nivea venom despite containing abundant CTX, since PLA 2 and cobra CTX (known as direct lytic factor previously) have been known for a long time to act in synergism [36]. Further studies are needed to elucidate the cause and mechanism (probably through pseudogenization) as well as the evolutionary impact of PLA 2 loss in the venom.
In sub-Saharan Africa, resource allocation and subsequent antivenom supply, access, affordability, and availability are inadequate [61]. Domestic antivenoms are high-priced and limited in supply. Antivenoms produced outside Africa (for instance, products procured from India targeting the African snake species) are increasingly imported into Africa since their supply is consistent and the price is low. There is, however, limited data to support the clinical effectiveness of their use in Africa [15]. In this study, we scrutinized the immunoreactivity and neutralization efficacy of VAPAV, a polyvalent antivenom developed in India for use in Africa, as a para-specific antivenom against N. nivea venom. The immunological binding activity of VAPAV toward N. nivea venom was comparable to the homologous N. haje venom (the venom used in hyperimmunizing the horses), implying conserved venom protein antigenicity between the two sister species which share close phylogenetic relatedness within the subgenus of Uraeus. On antivenomics, VAPAV was also found to have considerably good immunorecognition capacity for the venom protein fractions that were eluted in the initial course of RP-HPLC (fractions 1, 2 and 3, approximately 70% of protein binding capacity). These fractions represent alpha-neurotoxins as profiled by C18 reverephase HPLC in the proteomic study, which is also consistent with the decomplexed venom profiles of other cobra species [21,25,29,30,60], including N. annulifera and N. senegalensis from within the N. haje complex itself [19,31]. The immunoretention of these neurotoxins indicated the cross-neutralization potential of VAPAV against N. nivea venom-induced neurotoxicity. As such, the in vivo neutralization activity of VAPAV against N. nivea venom was investigated, applying the venom and antivenom preincubation protocol for immunocomplexation as per the WHO guideline [62]. Consistent with the immunoreactivity finding, VAPAV was moderately effective in cross-neutralizing the venom's lethal effect in mice. The parameter "normalized potency" was then used to compare its neutralization efficacy with the reported efficacy of other antivenom products, i.e., SAIMR polyvalent antivenom (South African Institute for Medical Research, South Africa), Antivipmyn-Africa (Bioclon Institute, Mexico), and VACSERA (The Egyptian Company for Biological Products and Vaccines, Giza, Egypt) [16][17][18]. Of note, SAIMR polyvalent antivenom appeared to be most potent (in neutralizing N. nivea venom lethality per gram of antivenom protein), and the logical interpretation is that N. nivea venom was used as part of the immunogen cocktail for hyperimmunization during the antivenom manufacturing, in contrast to the other products that lack species-specificity for N. nivea. Between Antivipmyn and VAPAV, the former was approximately twice as potent. The immunogens used in the production of both antivenoms include cobra venoms from N. haje, N. melanoleuca and N. nigricollis, whereas Antivipmyn has an additional component from Naja pallida. Another antivenom product, VACSERA, was found to be ineffective against N. nivea venom, although N. haje and N. nigricollis venoms were included as the immunogen in its production [18]. The discrepancy in the neutralization efficacy of these antivenoms against N. nivea venom could be related to the composition of venom immunogen used by the respective manufacturers or technical factors, e.g., horses' health conditions in mounting an immune response, immunoglobulin purification method, product purity, and experimental conditions such as the lethal dose (challenge dose) used in the neutralization assay.
Although the commonly adopted venom-antivenom preincubation technique is preferred as a standard method, it should be noted that the in vivo interaction between venom and antivenom can be complex. This can be affected by the toxicokinetics of various toxins in the venom, and the pharmacokinetics of different antivenom preparations, thus contributing to variations between studies. Thus, to further verify the in vivo cross-neutralization potential of VAPAV, the experimental envenoming and rescue mouse experiment was conducted, mimicking the actual treatment of snakebite where the antivenom is administered as indicated by the sign of neurotoxicity. The in vivo neutralization effect of VAPAV, when given as a bolus upon the onset of paralysis in the envenomed mice, was moderate. The antivenom was able to reduce fatality (rescue with VAPAV resulted in 75% survival of the envenomed mice), although the reversal of neurotoxicity was prolonged (close to 24 h) for the envenomed mice to regain normal limb functions. The finding of slow recovery from neurological paralysis echoes the clinical observation of difficulty in reversing the already developed paralysis in patients envenomed by N. nivea even with the use of the specific antivenom SAIMR [14]. In contrast, using the same experimental model of envenoming and rescue, previous works from the same laboratory showed that effective antivenoms were able to reverse venom-induced neurotoxicity rapidly, and fully rescue all mice envenomed with N. naja and N. kaouthia venoms (100% survival) [22,28]. A slow recovery from neurotoxicity indicates that a higher dose of antivenom may be required, or the antivenom may have to be administered repeatedly in order to expedite the clearance of residual toxins and promote recovery. The finding, therefore, revealed the limitation of the heterologous antivenom in cross-neutralizing N. nivea venom, and underscored the need for a poly-specific antivenom of high potency that can neutralize the venom effectively. This may be achieved by enriching the venom immunogen formulation with N. nivea venom or, more ideally, its fractionated principal neurotoxins. This can be seen in the production of an experimental pan-region, poly-specific antivenom that utilized a diverse toxin repertoire containing venoms and toxins from multiple elapid species [63,64].

Conclusions
The present study unveiled the venom proteome of N. nivea and showed it is dominated by three-finger toxins (3FTx, >80% of total venom proteins), especially cytotoxins/cardiotoxins (~75.6%) and alpha-neurotoxins (~7.4%). The venom proteome lacks PLA 2 -a unique venom phenotype increasingly recognized in the African non-spitting cobras of the Uraeus subgenus. The hetero-specific antivenom, VAPAV (an Indian antivenom product manufactured for African use) showed immunoreactivity toward the venoms of N. nivea and the phylogenetically related N. haje (whose venom was used in VAPAV production), while antivenomics indicated that the venom's alpha-neurotoxins were immunorecognized by the antivenom with a binding capacity of~70%. In mice, the antivenom was moderately effective in cross-neutralizing the venom's lethality. The finding implies the need for further refinement of antivenom production, so that a pan-region antivenom with a higher potency and a broader coverage of snake species can be developed for snakebite envenoming in southern Africa where N. nivea envenoming is endemic and prevalent.

Animal Supply
ICR albino mice (20-25 g) used in the preincubation neutralization study were supplied by the Animal Experimental Unit, Faculty of Medicine, University of Malaya. Mice of the same strain and body weight were sourced from the BioLASCO Taiwan Co. Ltd. for the envenoming-rescue study conducted in the National Tsing Hua University, Taiwan. All animal study protocols were referenced to the guidelines provided by the Council for International Organizations of Medical Sciences (CIOMS) [65].

Reverse-Phase High-Performance Liquid Chromatography (RP-HPLC)
Three milligrams of N. nivea venom were reconstituted in 200 µL ultrapure water, and the supernatant was subjected to LiChrospher ® WP300 C-18 reverse-phase fractionation using Shimadzu LC-20 AD HPLC system (Shimadzu, Kyoto, Japan). The C18 column was pre-equilibrated with 0.1% TFA in water (Buffer A), and the sample was separated with 0.1% TFA in ACN (Buffer B) using a linear gradient of 5% B for 10 min, 5-15% B over 20 min, 15-45% B over 120 min and 45-70% B over 20 min at a flow rate of 1 mL/min. The elution of the proteins was monitored at wavelength 215 nm, and fractions were collected manually, lyophilized, and stored at −20 • C until use.

Sodium Dodecyl Sulfate-Polyacrylamide Gel Electrophoresis (SDS-PAGE)
Naja nivea venom and the RP-HPLC-collected fractions were reconstituted in ultrapure water and separated by 15% SDS-PAGE under reducing conditions at 100 V for 2 h. Spectra™ Multicolor Broad Range Protein Ladder (10-260 kDa) was used for molecular mass calibration. Gels were stained with Coomassie Brilliant Blue R-250 and scanned using Image ScannerIII Labscan 6.0 (GE Healthcare, Uppsala, Sweden).

In-Solution Tryptic Digestion and Liquid Chromatography-Tandem Mass Spectrometry
RP-HPLC-collected protein fractions (10 µg each) were reduced by dithiothreitol (DTT) and alkylated using iodoacetamide (IAA). MS-grade trypsin (Thermo Fisher Scientific, Waltham, MA, USA) was subsequently used to digest the proteins according to the manufacturer's protocol. The peptides were then desalted and concentrated using Millipore ZipTip ® C18 Pipette Tips. The digested peptides were then reconstituted in 7 µL of 0.1% formic acid in the water and analyzed using nano-electrospray ionization liquid chromatography-tandem mass spectrometry (nano-ESI-LCMS/MS). Agilent 1260 Infinity Nanoflow LC system, coupled with Agilent 6550 Accurate-Mass Q-TOF LC/MS system (Agilent, Santa Clara, CA, USA), was used for the analysis. The samples were loaded to HPLC Large-Capacity Chip Column Zorbax 300-SB-C18 (160 nL enrichment column, 75 µm × 150 mm analytical column and 5 µm particles; Agilent part no. G4240-62010; Agilent, Santa Clara, CA, USA) for peptide separation. The injection volume was 1 µL per sample, and the flow rate was set to 0.4 µL/min, with a linear gradient of 5-70% of elution solvent (0.1% formic acid in 100% acetonitrile). The flow of drying gas was delivered at 11 L/min at 290 • C. The fragmentor voltage was set to 175 V, and the capillary voltage was 1800 V. The mass spectra were obtained using Mass Hunter acquisition software (Agilent, Santa Clara, CA, USA) in MS/MS mode with an MS scan range of 200-3000 m/z and MS/MS scan range of 50-3200 m/z. Data were extracted based on MH+ mass range between 50 and 3200 Da and processed with Agilent Spectrum Mill MS Proteomics Workbench software packages version B.04.00. The merged database incorporating non-redundant NCBI databases of Serpentes (taxid: 8570) and an in-house transcripts database was used for peptide matching and protein identification purposes. Carbamidomethylation was specified as a fixed modification and oxidized methionine as a variable modification. The following filters were used for the validation of peptides and proteins: protein score >20 and peptide score >10. Results of protein identification based on ≥2 "distinct peptides" were considered significant.

Estimation of Protein Relative Abundance
The relative abundance of proteins (%) was estimated based on the chromatographic peak area of protein eluted and the mean spectral intensity (MSI) of peptides previously described [25]. The following calculation was adopted: Relative abundance of protein A in fraction Z (%) = Mean spectral intensity of protein A in fraction Z Total mean spectral intensity in HPLC fraction Z × % AUC of HPLC fraction Z

Immunoreactivity of Antivenom
The immunoreactivity of VAPAV toward the venom antigens of N. nivea and N. haje were studied with an indirect enzyme-linked immunosorbent assay (indirect ELISA) as previously described [66]. The wells of the 96-well immunoplate were pre-coated with 10 ng of venom antigen at 4 • C overnight. The immunoplate was flicked dry and rinsed with phosphate-buffered saline with 0.5% Tween ® 20 (PBST) 4 times to remove the excess antigens. The antivenom was appropriately diluted at 1:100, 1:300, 1:900, 1:2700, 1:8100, and 1:24,300 from a stock concentration of 20 mg/mL. A total of 100 µL of diluted antivenom was then added to each of the antigen-coated wells for incubation at room temperature for 1 h. The incubates were then removed by flicking the immunoplates and washing them with PBST. 100 µL of horseradish peroxidase-conjugated antihorse-IgG (Jackson ImmunoResearch Inc., West Grove, PA, USA) pre-diluted in PBST (1:10,000) was then added into each well, and the incubation took place for 1 h at room temperature. The excess unbound conjugated antibodies were then removed by flicking the immunoplates and washing them 4 times with PBST. Fifty microliters of freshly prepared 3,3 5,5 -tetramethylbenzidine (TMB) substrate solution was subsequently added, and the enzymatic reaction took place in the dark for 10 min at room temperature. The reaction was terminated by adding 50 µL of 12.5% sulfuric acid, and the absorbance was measured at 450 nm using a Tecan i-control™ infinite M1000Pro microplate reader (Tecan, Männedorf, Switzerland). All values were means ± S.E.M. of triplicate experiments. The immunoreactivity was expressed as the half-maximal effective concentrations (EC 50 ), interpreted as the antivenom concentration that results in 50% of the venom-antivenom binding reaction.

Antivenomics: Immunorecogniztion of N. nivea Venom Fractions by Antivenom
The ability of VAPAV to immunorecognize N. nivea venom fractions was examined using an affinity chromatography approach. Briefly, 1 mL of NHS-activated Sepharose 4 fast flow (GE Healthcare, Danderyd, Sweden) was packed in a column and washed with five matrix volumes (or column volume, CV) of 1 mM ice-cold HCl followed by two CV of coupling buffer (0.2 M NaHCO 3 , 0.5 M NaCl, pH 8.3). The matrix was then incubated with 50 mg of antivenom dissolved in 1 CV of coupling buffer at room temperature for 3 h. After the immobilization of antivenom to the column matrix, the non-reactive NHS matrix groups were blocked with 1 mM ethanolamine at room temperature for 30 min. The column matrix was repeatedly washed six times with 5 CV of low (0.1 M sodium acetate, 0.5 M NaCl, pH 4.0) and high pH buffer (0.5 M ethanolamine, 0.5 M NaCl, pH 8.3) before equilibration with 5 CV of binding buffer (PBS). One milligram of N. nivea venom dissolved in 1 mL of binding buffer was loaded to the antivenom-immobilized column matrix and incubated at room temperature for 30 min. Non-retained fractions were collected with 5 CV of PBS, and the immunocaptured venom fractions were eluted with 5 CV of elution buffer (0.1 mM glycine, pH 1.5, neutralized immediately with 1M Tris-HCl, pH 9.0). Both retained, and non-retained fractions were concentrated and subjected to reverse-phase high-performance liquid chromatography (RP-HPLC).
RP-HPLC was performed with a C18 column (300 Å, 250 mm × 4.6 mm particle size, 5 µm pore size; Phenomenex, CA Torrance) using a Shimadzu LC-20AD HPLC system (Shimadzu, Kyoto, Japan). The column was pre-equilibrated with 0.1 % TFA in ultrapure water, and eluted with a linear gradient of 0.1% TFA in acetonitrile (Buffer B) at 2% B for 5 min, 2-10% B over 2 min, 10-16% B over 6 min, 16-28% B over 2 min, 28-65% B over 37 min and 65-80% B over 5 min, with a flow rate of 0.3 mL/min. The venom fractions were monitored at a wavelength of 280 nm. The degree of VAPAV immunorecognition toward a venom fraction was estimated by the chromatographic peak area under the curve (AUC) using OriginLab 8.0 software (OriginLab Corporation, Northampton, UK). The immunorecognition was expressed as the percentage of proteins bound by the antivenomcoated affinity column (regarded as antivenom-treated venom) in comparison to that of the untreated venom profiled by the RP-HPLC under the same conditions.

Venom Lethality and Neutralization by Antivenom
A hundred microliters of the venom in various doses were injected intravenously into ICR mice through the caudal vein (20-25 g, n = 4 per dose). The survival ratio at each venom dose was recorded at 24 h post-injection. In the lethality neutralization assay, a challenge dose of the venom (5 × LD 50 ) was pre-incubated with various dilutions of antivenom in a total volume of 200 µL at 37 • C for 30 min, followed by intravenous injection into the mice through the caudal vein (20-25 g, n = 4 per dose). The survival ratio of mice was recorded at 24 h post-injection. The venom intravenous median lethal dose (LD 50 ), antivenom median effective dose (ED 50 ) and the 95% confidence intervals (C.I.) were calculated by the Probit analysis method [67], using BioStat 2009 analysis software (AnalystSoft Inc., Walnut, CA, USA). The neutralizing capacity was also expressed as potency (P), defined as the amount of venom completely neutralized by one mL of antivenom (mg/mL), calculated as previously described [68,69]. For comparison purposes, the potency (P) was further divided by the antivenom protein concentration to obtain normalized potency (n-P), defined as the amount of venom neutralized by one gram of antivenom protein (mg/g).

Experimental Envenoming and Rescue Experiment
The experiment mimics the natural setting of snakebite and antivenom treatment. It was conducted in an envenoming and rescue model in mice as previously described [22,70]. The subcutaneous LD 50 of N. nivea venom was first established by injecting the venom at various doses into the loose skin over the neck of the mice (20-25 g, n = 4 per dose). The number of deaths of the mice at each dose was recorded at 24 h. In the envenoming and rescue model, a challenge dose of venom (5 × s.c. LD 50 ) dissolved in 50 µL saline was inoculated subcutaneously into the mice (20-25 g, n = 8). The mice were monitored closely for the development of neurological signs. In the rescue group, the antivenom (VAPAV) was administered as a bolus of 250 µL intravenously (via the caudal vein) upon the onset of neurotoxicity indicated by posterior limb paralysis. Mice in the control (untreated) group (20-25 g, n = 8) received 250 µL of saline instead upon the onset of neurotoxicity. The progress of neurotoxicity (deterioration or reversal) in the mice was closely monitored over 24 h. All mice were allowed free access to food and water ad libitum throughout the experiment.