Quantifying venom in African snakes: Insights into protein content, yield and body size associations

TL;DR

This study compares methods to measure protein in African snake venoms and finds that Elapidae snakes have higher protein concentrations than Viperidae.

Contribution

The study identifies the BCA assay as the most accurate method for quantifying African snake venom proteins and reveals higher concentrations in Elapidae compared to Viperidae.

Findings

BCA assay was the most accurate for quantifying venom proteins.

Elapidae venoms had significantly higher protein concentrations than Viperidae venoms.

Venom protein concentrations varied more between Elapidae species than within the same genus.

Abstract

Snake venoms are complex mixtures primarily composed of toxic proteins used during prey capture and defence. There is limited knowledge concerning the protein concentration of snake venom and the biases of different protein determination methods. Here, we assess the ability of the Qubit protein assay, bicinchoninic acid (BCA) assay, Bradford assay and NanoDrop spectrometry (A280 with a mass extinction coefficient of one) to accurately quantify protein concentrations of toxins isolated from venom, including three-finger toxins and phospholipase A2. The Bradford assays severely underestimated three-finger toxin concentrations and NanoDrop spectrometry overestimated phospholipase A2 concentrations, whilst the BCA assay was the most accurate. Venom from six major African venomous snake genera was also assessed: coral cobras (Aspidelaps spp.); mambas (Dendroaspis spp.); cobras (Naja spp.); bush vipers (Atheris sp.); adders (Bitis spp.); and saw-scaled vipers (Echis sp.). Protein concentration results were inconsistent between methods. Protein concentrations were found to be lowest for Bitis spp. venom and highest for Naja spp. venom and did not vary between species of the same genus. However, in general, Elapidae species had venoms with significantly higher protein concentrations than Viperidae species. Moreover, there was greater variability between Elapidae species. We also determined wet venom yields and used this to provide a tentative estimate of the total protein quantity that may be injected during a snake bite. We found snake weight and length influenced wet venom yield for the Atheris squamigera but not for Bitis arietans and Echis romani. Our results aim to improve our understanding of the physical properties of snake venom.

Image 1

•Assessment of the accuracy of protein assays and standards for snake venom.•The Bradford assay is inaccurate for determining protein concentration in Elapidae.•The BCA assay with BSA standard is an accurate method for African snake venoms.•African Elapidae have higher venom protein concentrations than Viperidae.

Assessment of the accuracy of protein assays and standards for snake venom.

The Bradford assay is inaccurate for determining protein concentration in Elapidae.

The BCA assay with BSA standard is an accurate method for African snake venoms.

African Elapidae have higher venom protein concentrations than Viperidae.

Figures

A, Kruskal-Wallis, p > 0.05), but was significantly lower from the genera of small snakes (SVL < 70 cm), including *Aspidelaps spp., Atheris squamigera* and *Echis romani* relative to the genera of larger snakes (SVL > 80 cm) including *Bitis* spp.*, Dendroaspis* spp. and *Naja* spp. (Supplementary Fig. 1B, Non-parametric one-way ANOVA, p < 0.0001)*.* The yields from spitting cobras, *N. mossambica, N. nigricollis, N. pallida* and *N. nigricincta,* were not statistically significantly different from non-spitting cobras, *N. haje, N. subfulva* and *N. annulifera.* Yields between species of the

A–D). The IgG standards showed similar absorbance values to venoms from Viperidae in the Bradford assay and these values were consistently lower than the BSA standard (Supplementary Fig. 2E–H). The BSA standard showed excellent overlap with venom from all species and localities in the BCA assay, whilst the IgG standards consistently produced higher absorbance values than venom standards (Supplementary Fig. 3A–H). Protein concentrations were calculated using all three standards (Supplementary Table 1). However, to allow comparability with the Qubit assay the results utilising BSA standards were

A, p-value <0.001, Mann-Whitney *U* test) and there was no difference between species of the same genus (Supplementary Fig. 4A, p-value >0.05, Non-parametric one-way ANOVA).

Full text

Abbreviations:

BCABicinchoninic acidPBSphosphate-buffered saline3FTxthree-finger toxinSVMPsnake venom metalloproteinasesPLA_2_phospholipase A_2_SVLsnout to vent lengthSVSPsnake venom serine proteasesCTxcytotoxic three-finger toxinsTZATanzaniaNGANigeria

Introduction

1

Snakebite envenoming causes approximately 100,000 deaths and over 400,000 life-altering disabilities annually (Gutiérrez et al., 2017). It disproportionally affects impoverished areas and is recognised by the WHO as a priority neglected tropical disease. In sub-Saharan Africa, it is a significant public health burden (Gutiérrez et al., 2017).

The most medically important snakes of sub-Saharan Africa are front-fanged snakes from the Elapidae and Viperidae family, which possess enlarged, hollow anterior maxillary teeth (fangs) used for venom injection (Mackessy, 2022). Elapidae, including cobras (Naja spp.), coral cobras (Aspidelaps spp.) and mambas (Dendroaspis spp.), have short hollow fangs at the anterior of the maxilla. Elapidae are generally neurotoxic with their venom inducing ptosis, neuromuscular paralysis, respiratory failure, and death. The venom gland structure of Viperidae, such as adders (Bitis spp.) and saw-scaled vipers (Echis spp.), differs from Elapidae, with a larger central venom gland lumen and long hollow fangs that can be rotated and folded (Mackessy, 2022). Venoms from snakes of the Viperidae family are primarily haemotoxic and can cause haemorrhage, coagulopathy and localised tissue necrosis (Gutiérrez et al., 2017).

Snake venom is a highly complex mixture composed of proteins, peptides, small molecules and salts used primarily for predation and defence (Mackessy, 2009). Approximately 90% is made up of proteins, and these toxins have evolved through neofunctionalization (Mackessy, 2009). These include, but are not limited to: enzymes, such as snake venom metalloproteinases (SVMPs) and snake venom serine proteases (SVSPs); and non-enzymatic proteins such as three-finger toxins (3FTX); and phospholipase A2s (PLA_2_) that exist in both enzymatic and non-enzymatic isoforms (Offor et al., 2022; Tasoulis and Isbister, 2017).

As a generalization, venoms from snakes of the Elapidae family are primarily composed of 3FTXs and PLA_2_s, whilst those from the Viperidae family are generally dominated by SVMPs, SVSPs and PLA_2_s (Offor et al., 2022). However, venom composition varies interspecifically, intra-specifically, and with snake age, locality, diet and sex (Mackessy, 2009). Adding to this variability, is the differing quantities of venom expelled by snakes both between species and between bites (Young et al., 2002), and the differing protein concentrations within the venom (Avella et al., 2021; Marsh and Glatson, 1974).

In this study, we assess the biases of different protein quantification methods for crude venom and isolated toxins, alongside measurements of wet venom yield and venom protein concentrations for a broad range of medically relevant African snakes. This data are used to gain a tentative estimate of the total toxin quantity that – assuming milking is an appropriate model for snakebite – could be delivered during envenomation.

Materials and methods

2

Venom collection

2.1

Samples were sourced from the on-site herpetarium at the Liverpool School of Tropical Medicine (LSTM). The facility and its protocols for the husbandry of snakes are approved by the UK Home Office and the LSTM Animal Welfare and Ethical Review Boards. Snake venoms used in this study came from the following viper (Viperidae) and elapid (Elapidae) species: Aspidelaps lubricus lubricus (captive-bred, n = 3), Aspidelaps lubricus cowlesi (captive-bred, n = 3), Atheris squamigera (Cameroon, n = 15), Bitis arietans (captive-bred, Eswatini, Nigeria, Tanzania; n = 24), Bitis gabonica (Cameroon, Tanzania, Uganda; n = 6), Bitis rhinoceros (captive-bred, n = 3), Dendroaspis angusticeps (Tanzania, n = 3), Dendroaspis viridis (Togo, n = 2), Echis romani (Nigeria, n = 12), Naja annulifera (Eswatini, n = 2), Naja mossambica (Eswatini, n = 2), Naja nigricollis (Nigeria, Tanzania; n = 6), Naja pallida (Kenya, Tanzania; n = 3), Naja nigricincta (captive-bred, n = 2), Naja haje (captive-bred, Uganda, n = 3), and Naja subfulva (Cameroon, Uganda, n = 6). Venom was collected by allowing snakes to voluntarily bite onto a sterile collection vessel, followed by light glandular massage to promote complete venom release. All venoms were collected either prior to or in accordance with the Nagoya protocol. The wet weight of the resulting venom was recorded immediately by recording the mass of the collection dish before and after extraction. Snout to vent length (SVL) and snake weight were also recorded. Venom samples were immediately stored at −80 °C prior to protein determination.

Protein determination of whole venom

2.2

Standards used in protein determination assays consisted of an eight-point calibration curve ranging from 0 to 1.2 mg/ml serially diluted in PBS. Quick Start Bovine gamma globulin was purchased prediluted at 2 mg/ml (Bio-Rad Laboratories) and subsequently diluted. BSA (Millipore) standards were prepared by weighing and subsequently diluted in PBS. Venom standards were prepared identically using lyophilised venom. Wet venom samples were diluted 500x in PBS and analysis performed in triplicate. The Pierce BCA Protein Assay (Thermo Fisher Scientific) was performed according to manufacturer instructions. Briefly, 10 μl diluted venom sample or calibration standards and 200 μl working reagent were incubated at 37ᵒC for 30 min in 96-well plate setup then absorbance read at 562 nM. The Bradford assay was performed by incubating 5 μl diluted venom sample or calibration standards and 250 μl Bradford reagent (Sigma) at room temperature for 10 min in 96-well plate setup. Absorbance was then read at 595 nM. The three different standards (BSA, IgG and sample matched venom) were then used to calculate the protein concentrations of wet venom.

The Qubit™ protein assay (Thermo Fisher Scientific) was undertaken following manufacturer instructions. Briefly, 10 μl diluted venom sample or 10 μl Qubit™ BSA standards were combined with 190 μl Qubit™ working solution, then incubated at room temperature for 15 min. Fluorescence was then measured on the Qubit 3.0 Fluorometer.

NanoDrop One (Thermo Fisher Scientific) was used to directly measure absorbance at 280 nm. Protein concentrations were then calculated using two methods:

• A)Assuming a mass extinction coefficient of 1 at 280 nm for a 1 mg/ml solution ε0.1% (Wilson and Walker, 2000):

$$[eqn]$$

• B)Following Beer-Lambert law and using an experimentally determined mass extinction coefficient (Pace et al., 1995):

$$[eqn]$$

where ε0.1% is the mass extinction coefficient (Supplementary Table 1) and L is the cuvette's light path in cm. ε0.1% was determined experimentally by measuring absorption at 280 nm of each venom sample at a known total concentration of 1 mg/ml. The samples of known concentration were prepared by resuspending a known mass of lyophilised venom in PBS.

Total concentration, including non-protein components, was additionally determined using the dry weight of lyophilised samples. After protein determination, the remaining sample volume was lyophilised using the Scanvac Cool Safe -55C 4L Basic Freezer Dryer (Labogene). Concentration was then calculated by dividing the mass by the corresponding sample volume.

Toxin isolation and protein determination

2.3

All chromatography was carried out using an AKTA Pure M LC system (Cytiva). Chromatography buffers were freshly prepared, and vacuum filtered (0.1 μm) immediately prior to use. Cytotoxin 1 (CTx, Uniprot ID P01468) and basic phospholipase A_2_ (PLA_2_, Uniprot ID P00605) were isolated from Naja nigricollis (Tanzania) venom according to Bartlett et al. (2024). Short neurotoxin-1 (sNTx-1, Uniprot ID P01426) was purified from the venom of Naja pallida (Tanzania) venom using two cation exchange chromatography steps (Bartlett et al., 2024). The first step was carried out using 50 mM sodium phosphate pH 6.0 and 10 CV gradient a gradient of 0-0.5 M NaCl, and the second on a 1 mL HiRes Capto S column using a 25 CV 0-0.3 M NaCl gradient in 25 mM sodium phosphate pH 7.2. Alpha-neurotoxicity was confirmed using an established nicotinic acetylcholine receptor inhibition assay (Patel et al., 2023). Bitis arietans (Nigeria and Tanzania) SVMP PI (Accession numbers PP950451 and PP50538, respectively) were isolated from venoms. Both isolation and confirmation of SVMP activity were performed as described in Wilkinson et al. (Wilkinson et al., 2024, Preprint). Snake venom serine protease (Uniprot ID Q6T6S7) was purified from Bitis gabonica (Tanzania) venom using size-exclusion chromatography in 25 mM sodium phosphate pH 7.2 on Superdex 200HR. This was then further purified by cation exchange chromatography on a 1 mL HiRes Capto S column with a 25 CV 0-0.4 M gradient of NaCl in 50 mM sodium phosphate, pH 6.0. Serine protease activity was confirmed by inhibition of insulin B chain digestion in the presence of 2 mM phenylmethylsulfonyl fluoride. Prior to the protein concentration measurements, all proteins were dialysed against 25 mM sodium phosphate, 0.15 M NaCl pH 7.2 and then concentrated using centrifugal concentrators (3 kDa molecular weight cut-off for the 3FTx and PLA_2_ proteins, 10 kDa molecular weight cut-off for the proteases).

Absorbance of isolated toxins was measured at 280 nm using a NanoDrop One Spectrophotometer (Thermo Fisher Scientific). The mass extinction coefficient for a 1 mg/ml solution (ε0.1%) for each toxin was predicted computationally from the amino acid sequence (Supplementary List 1) using the following equation on Expasy ProtParam (web.expasy.org/protparam):

$$[eqn]$$

Exact protein concentrations were then calculated using the modification of Beer-Lambert law as stated above. Toxins were then diluted to 0.4 mg/ml in PBS.

Statistical analysis

2.4

Normality of data and equality of variance was assessed in R (version 4) by the Shapiro-Wilk test and the Brown-Forsythe test, respectively, and non-parametric or parametric tests undertaken as appropriate. Linear regression followed by Bonferroni correction was applied to assess correlations. Correlations with an R^2^ > 0.3 and an adjusted p-value < 0.05 were deemed significant. GraphPad Prism (version 10) was employed for statistical analysis and data visualization. ∗, p < 0.05; ∗∗p < 0.01, ∗∗∗p < 0.001.

Results

3

Venom wet mass

3.1

Snakes showed a large level of variation in wet venom yields between individuals of the same species (Table 1, Supplementary Fig. 1). Yields were not significantly different between Elapidae and Viperidae (Supplementary Fig. 1A, Kruskal-Wallis, p > 0.05), but was significantly lower from the genera of small snakes (SVL < 70 cm), including Aspidelaps spp., Atheris squamigera and Echis romani relative to the genera of larger snakes (SVL > 80 cm) including Bitis spp., Dendroaspis spp. and Naja spp. (Supplementary Fig. 1B, Non-parametric one-way ANOVA, p < 0.0001). The yields from spitting cobras, N. mossambica, N. nigricollis, N. pallida and N. nigricincta, were not statistically significantly different from non-spitting cobras, N. haje, N. subfulva and N. annulifera. Yields between species of the same genus were not significantly different. (Supplementary Fig. 1C, Non-parametric one-way ANOVA, p > 0.05). Data for individual snakes can be viewed in Supplementary Table 1.Table 1. Wet venom yields, snout to vent length and weight of African Elapidae and Viperidae*.Table 1. Wet venom yield (mg)Snout to vent length (cm)Snake weight (g)nAverageRangeAverageRangeAverageRangeAspidelaps lubricus cowlesi4321 - 705846 - 6611681 - 1333Aspidelaps lubricus lubricus3924 - 575846 - 66141101 - 1893Atheris squamigera9128 - 2054739 - 585126 - 14515Bitis arietans443199 - 9358272 - 100824520 - 154524Bitis gabonica943760 - 12949490 - 9712761030 - 17236Bitis rhinoceros443628 - 12748580 - 9111411025 - 12893Dendroaspis angusticeps423258 - 554130126 - 138448375 - 5503Dendroaspis viridis327246 - 409149149 - 149455430 - 4792Echis romani7721 - 1334439 - 497660 - 9212Naja annulifera445386 - 505127119 - 134958950 - 9652Naja haje459221 - 817127120 - 135583440 - 7253Naja mossambica430327 - 534106106 - 106678600 - 7552Naja nigricincta406297 - 5158382 - 84238220 - 2552Naja nigricollis1336755 - 2000128124 - 133693600 - 7756Naja pallida902608 - 1060105103 - 108430425 - 4403Naja subfulva*197108 - 275136125 - 152586515 - 7406n represents number of snakes.

Effect of weight and SVL on venom yield

3.2

Hayes et al. have shown that venom expenditure varies dependent on intrinsic and extrinsic factors with snake size being a predominant factor (Hayes and Herbert, 2002). We therefore assessed whether the extrinsic factors of weight and SVL affected venom yield for snake species with sufficient statistical power (N > 10), specifically Atheris squamigera, Bitis arietans and Echis romani. A. squameriga venom yield showed a significant and positive correlation with SVL and weight (Fig. 1A and B, Simple linear regression, R^2^ = 0.41 and 0.51, adjusted p-value = 0.03 and 0.03, respectively). However, venom yields from Bitis arietans and Echis romani did not correlate with either snake weight or SVL (Fig. 1C, D, E, F, Simple linear regression, R^2^ ≤ 0.2, adjusted p-value >0.05).Fig. 1Relationship of snake weight and SVL with venom yield at the species level. Wet venom yield, snake weight and SVL were determined for individual milked snakes. Correlations are shown between venom yield and weight (A, C, E) and between venom yield and SVL (B, D, F) for Atheris squamigera (A, B), Bitis arietans (C, D), Echis romani (E, F). Data are shown as values for individual snakes (black solid circles). R^2^ values were calculated by performing linear regression.Fig. 1

Comparing methods of determining venom protein concentrations

3.3

To determine the most suitable standard for the BCA and Bradford assays, BSA and IgG standards were compared to venom standards (Supplementary Figs. 2 and 3). The BSA standard and IgG standards showed no overlap with venom from Elapidae in the Bradford assay, with venoms producing much lower absorbance values than BSA and IgG (Supplementary Fig. 2A–D). The IgG standards showed similar absorbance values to venoms from Viperidae in the Bradford assay and these values were consistently lower than the BSA standard (Supplementary Fig. 2E–H). The BSA standard showed excellent overlap with venom from all species and localities in the BCA assay, whilst the IgG standards consistently produced higher absorbance values than venom standards (Supplementary Fig. 3A–H). Protein concentrations were calculated using all three standards (Supplementary Table 1). However, to allow comparability with the Qubit assay the results utilising BSA standards were subsequently taken forward.

The different methods yielded protein concentrations of several hundred milligrams per millilitre, with large variation between some approaches (Fig. 2A–G). The Bradford method consistently produced the lowest predicted protein concentration for venom from all Elapidae species whilst the Qubit assay produced the second lowest (Fig. 2A–D). Venom samples from Viperidae showed more consistency between methods. Results from the BCA and Bradford were similar for Bitis spp. venom samples. Echis romani and Atheris squamigera venom protein concentrations were similar between the Bradford, BCA and Qubit assay. Nanodrop spectrometry with an assumed ε0.1% of one produced the highest predicted protein concentration for most venoms and this effect was most pronounced in Bitis spp.Fig. 2Protein concentrations of snake venom and isolated venom toxins determined with multiple methods. The protein concentration of wet venom was determined for individual milked snakes using Bradford assay (black), BCA assay (white), Qubit protein assay (light grey) and by NanoDrop spectrometry assuming a $[eqn]$ of one (A280, dark grey). Results are shown for Atheris squamigera (A), Aspidelaps spp. (B), Dendroaspis spp. (C), Bitis spp., (D), Echis romani (E), Naja spp. (spitters) (F), Naja spp. (non-spitters) (G). Isolated venom toxins at a concentration of 0.4 mg/ml (dotted line) were also analysed, specifically N. pallida sNTX-1 (P01426), N. nigricollis CTx (P01468), B. gabonica SVSP (Q6T6S7), N. nigricollis PLA2 (P00605), B. arietans TZA SVMP PI (∗PP950538), B. arietans NGA SVMP PI (∗PP950451) (H). Total concentration was also determined, by lyophilisation (purple) and Nanodrop spectrophotometry using experimentally determined $[eqn]$ (green, A280/ $[eqn]$ for wet venoms. Data are shown as the mean ± SEM of individual snakes (A-G) and three technical replicates (H). (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)Fig. 2

To further understand these differences and deduce the most accurate methods, total venom concentrations were estimated by lyophilisation and Nanodrop spectrophotometry with an experimentally determined $[eqn]$ (Fig. 2). These results were most comparable to the BCA assay for both Elapidae and Viperidae venoms. However, lyophilisation and experimentally determined $[eqn]$ rely on total venom mass including non-protein components; therefore, isolated venom toxins of a known protein concentration (without non-protein components) were also analysed (Fig. 2H). The Qubit assay underestimated most toxins but was accurate for N. nigricollis CTx and B. gabonica SVSP. The Bradford assay severely underestimated concentrations of 3FTxs – specifically N. pallida sNTX-1 and N. nigricollis CTx by 5- and 10-fold, respectively – but was accurate for N. nigricollis PLA_2_ and B. arietans SVMP concentrations The BCA assay correctly estimated the protein concentration for the majority of toxins but overestimated the concentration of N. pallida sNTX-1 and N. nigricollis PLA_2_. The BCA assay was therefore deemed most accurate.

Venom protein concentrations

3.4

Average venom protein concentrations were determined as the mean of the BCA assay and Qubit™ assay results. Spitting cobras displayed the highest average venom protein concentration, and Bitis spp. displayed the lowest (Table 2). Overall, venoms from snakes of the Elapidae family had statistically significantly higher concentrations than snakes of the Viperidae family with a mean protein concentration of 337 mg/ml and 197 mg/ml, respectively (Fig. 3A, p-value <0.001, Mann-Whitney U test). Within Viperidae family, Echis spp. venom had a statistically significant higher protein concentration than Bitis spp. Within Elapidae, spitting cobras had a statistically significant higher protein concentration than Aspidelaps spp*.* There were no significant differences in protein concentration between species of the same genus (Fig. 3C, Non-parametric one-way ANOVA). However, the mean protein concentration of all the non-spitting cobras species were lower than that of the spitting cobras.Table 2. Protein concentrations of venom extracted from African Elapidae and Viperidae.Table 2. Protein concentration (mg/ml)nAverageRangeAspidelaps lubricus lubricus253211 - 3143Aspidelaps lubricus cowlesi298251 - 3353Dendroaspis angusticeps360282 - 5023Dendroaspis viridis361356 - 3672Bitis arietans188140 - 24824Bitis gabonica187112 - 2576Bitis rhinoceros207166 - 2403Echis romani219175 - 30912Naja annulifera288266 - 3102Naja mossambica359353 - 3663Naja nigricollis360342 - 3585Naja pallida364343 - 3795Naja nigricincta388343 - 43216Naja haje333249 - 38026Naja subfulva335217 - 38613Atheris squamigera195118 - 2427n represents number of snakes.Fig. 3Protein concentration of venom*.* Protein concentrations calculated from BCA assay with BSA standards with individuals separated based on family (A), genus (B) or species (C). Species of the same genera are shown by the same colour bars: Aspidelaps spp. (solid white), Dendroaspis spp. (solid light grey), spitting cobra (solid dark grey), non-spitting cobras (solid black), Atheris squamigera (dashed white), Bitis spp. (dashed light grey), and Echis romani (dashed dark grey). Data are shown as individual snakes (circles) and the mean ± SEM. P-value were calculated by performing Mann-Whitney U test (A) and Non-parametric one-way ANOVA (B, C). ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.Fig. 3

To ensure differences were not due to biases in the protein determination assays, these findings were compared to total concentration determined by averaging results from lyophilisation and Nanodrop spectrometry using experimentally determined ε0.1% (Supplementary Fig. 4). Consistent with protein concentration results, Bitis spp*.* displayed the lowest total concentration and cobras the highest. Again, venoms from snakes of the Elapidae family had statistically significantly higher concentrations than snakes of the Viperidae family (Supplementary Fig. 3A, p-value <0.001, Mann-Whitney U test) and there was no difference between species of the same genus (Supplementary Fig. 4A, p-value >0.05, Non-parametric one-way ANOVA).

It was assessed whether wet venom yield, SVL or snake weight influenced protein concentration for snake species for which there was sufficient statistical power (N > 10). For all three individual species assessed, Atheris squamigera, Bitis arietans and Echis romani, venom yield did not correlate significantly with protein concentration (Fig. 4A–C, adjusted p-value > 0.05, R^2^ < 0.2). This was also true for snake weight (Fig. 4D–F, adjusted p-value > 0.05, R^2^ < 0.1) and SVL (Fig. 4G–I, adjusted p-value > 0.05, R^2^ = 0.00).Fig. 4Relationship of snake weight, SVL and venom yield with protein concentration at the species level. Wet venom yield, snake weight, SVL and protein concentration were determined for individual milked snakes. Correlations are shown between venom yield and protein concentration (A-C), snake weight and protein concentration (D-F) and, SVL and protein concentration (G-I) for Atheris squameriga (A, D, G), Bitis arietans (B, E, H), Echis romani (C, F, I). Data are shown as values for individual snakes (black solid circles), R^2^ values were calculated by performing linear regression.Fig. 4

Venom protein quantity

3.5

The total protein content delivered during extraction was then estimated using an approximation of the density of venom (1.05 g/cm^3^) (Achee et al., 2019; Triep et al., 2013) and wet venom yield (Table 3). Overall, the protein quantity expelled from Elapidae family was significantly higher than the Viperidae family (Fig. 5A, p-value <0.001, Mann-Whitney U test). Within Viperidae family, Echis romani and Atheris squameriga expelled very similar protein quantities and these were both significantly lower than Bitis spp. Within the Elapidae family, Aspidelaps squameriga expelled statistically lower venom protein quantities than Dendroaspis spp., spitting cobras and non-spitting cobras. The protein quantities delivered during extraction were not statistically significantly different between spitting cobras, non-spitting cobras and Dendroaspis spp*.* There were no significant differences in total protein quantity between species of the same genus (Fig. 5C, Non-parametric one-way ANOVA).Table 3. Estimation of protein venom quantity delivered during extraction from African Elapidae and Viperidae.Table 3. Protein quantity (mg)AverageRangeAspidelaps lubricus lubricus9.25.2 - 11.5Aspidelaps lubricus cowlesi12.46.2 - 22.2Dendroaspis angusticeps133.8122.4 - 155.9Dendroaspis viridis113.183.4 - 142.7Bitis arietans79.737.3 - 198.3Bitis gabonica169.981.2 - 258.2Bitis rhinoceros179.7143.8 - 201.9Echis romani15.46.2 - 24.3Naja annulifera120.9114.1 - 127.7Naja mossambica147.9109.7 - 186.2Naja nigricollis453.3254.1 - 681.5Naja pallida315.9198.7 - 374.8Naja nigricincta154.697.1 - 212.2Naja haje154.452.5 - 288.0Naja subfulva64.927.5 - 101.2Atheris squamigera16.96.3 - 34.9Protein quantity represents an estimation of the total protein content delivered during extraction. It was estimated using the wet venom yields reported in Table 1, assuming the equivalence of 1.05 mg vet venom to 1 μl. Number of snakes as stated previously.Fig. 5Total protein quantities of venom*.* Total protein quantity was calculated using the protein concentrations from BCA assay and wet mass, with an assumed venom density of 1050 kg/m^3^. Results are shown separated based on family (A), genus (B) and species (C). Species of the same genera are shown by the same colour bars: Aspidelaps spp. (solid white), Dendroaspis spp. (solid light grey), spitting cobra (solid dark grey), non-spitting cobras (solid black), Atheris squamigera (dashed white), Bitis spp. (dashed light grey), and Echis romani (dashed dark grey). Data are shown as individual snakes (circles) and the mean ± SEM. p-values were calculated by performing Mann-Whitney U test (A) and Non-parametric one-way ANOVA (B, C). ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.Fig. 5

Discussion

4

There is very limited research on the protein concentrations of snake venom. This is in part due to limited sample availability and difficulties associated with accurately determining protein concentrations for mixtures of protein and peptides. Here, we assess the ability of four protein determination assays previously used on snake venom samples (Bocian et al., 2020) – NanoDrop spectrometry with an assumed ε0.1% of one, Bradford assay, BCA assay and Qubit Protein assay – to accurately quantify snake venom.

The Bradford assay is based on the ability of Coomassie Brilliant Blue to bind to amino acids lysine, arginine and histidine (Bradford, 1976). The BCA assay detects amino acids which reduce Cu^2+^ to Cu^+^, primarily tryptophan, cysteine, methionine and tyrosine (Freeman et al., 1961). The Qubit protein assay quantifies protein using a fluorophore that binds to hydrophobic regions of proteins (Bell and Karuso, 2003). Differences in protein size and amino acid composition therefore results in unpredictable biases.

For spectrophotometric and colorimetric assays, the most accurate and precise results would be obtained by preparing a protein calibration curve with the same sample as the measured protein. However, this is often not used in practise, especially for venoms which have non-protein components, which can be difficult to solubilise (Mark Wilkinson, personal communication) and could have limited availability. In such cases, a generic protein standard, such as BSA or IgG, is used as an approximation. We therefore compared the closeness of these standards to venom standards. BSA was found to be a more appropriate standards than IgG for the BCA assay, whilst for the Bradford assay IgG was appropriate standard for venom from Viperidae. Neither standard was appropriate for the use of venom from Elapidae in the Bradford assay. In accordance with this, the Bradford assay was found to underestimate the protein concentrations of venoms dominated by 3FTXs, such as Naja spp*. and Aspidelaps* spp. This is consistent with previous reports on Naja ashei venom (Bocian et al., 2020) and the limited sensitivity of the Bradford assay to smaller peptides (Wilson and Walker, 2000). NanoDrop spectrometry (with an assumed ε0.1% of one) overestimated the protein concentration of Bitis spp. venoms, and this aligns with reports of interfering compounds, such as adenosine, in the venom of this genus (Aird, 2005).

In an attempt to determine the most accurate method, results were compared to venom concentrations estimated by total mass after lyophilisation, and Nanodrop spectrophotometry (applying a modification of Beer-Lambert law with experimentally determined $[eqn]$ . However, it is important to note that these two methods have their own limitations. Both assume total venom mass is protein, which is incorrect due to the presence of non-protein component, such as adenosine in Bitis spp. (Aird, 2005). We therefore also analysed specific toxins at a known concentration. The BCA assay showed greater accuracy for specific toxins than the Bradford assay, Qubit and NanoDrop spectrometry, the BCA assay was therefore utilized in subsequent analysis.

Protein concentrations from Bitis spp. and Naja spp. venom were 112 - 257 mg/ml and 249 - 432 mg/ml, respectively. This is similar to previous reports of 131 - 282 mg/ml for Bitis spp., specifically B.arietans and B.gabonica, but higher than previous reports of 116 – 155 for Naja spp., specifically N. haje, N. nigricollis, N. pallida, N. subfulva and N. mossambica (Avella et al., 2021; Marsh and Glatson, 1974). Our results appear to represent a first in determining the total protein concentrations of venoms from Aspidelaps spp., Dendroaspis spp., Echis sp. and Atheris sp., which were found to be 211 - 335 mg/ml, 282 - 502 mg/ml, 175 - 309 mg/ml, and 118 - 242 mg/ml, respectively.

Viperidae had a significantly lower concentration of protein in their venom compared to Elapidae. However, no differences were seen between species of the same genus. It is important to note that interpretation of differences in protein concentration should be treated with caution. The apparent differences in protein concentration could be argued to be due to the different toxins present in the venoms which show differential sensitivity to the assays. However, the total concentration was also significantly lower for Viperidae venom relative to Elapidae, giving us confidence in these findings. We found no evidence that protein concentration was affected by SVL, snake weight or venom yield intra-specifically for the snakes assessed, specifically B. arietans, E. romani and A.squamigera.

Historically, dry mass has been used to assess venom yield (de Roodt et al., 1998; Fix, 1980; Gao et al., 2019; Healy et al., 2019; Marsh and Glatson, 1974; Salafranca, 1973; Tare et al., 1986; Tumbare and Khadilkar, 2004) and very few studies have reported yield as wet mass (Avella et al., 2021; Kochva et al., 1982; Mirtschin et al., 2006). Here we assessed yield by measuring wet mass and evaluated links with snake morphology for B. arietans, E. romani, and A. squamigera. We found no evidence that snake weight or SVL influenced venom yields for B. arietans and E. romani, however, a significant positive correlation was seen for A. squamigera. Positive correlations have been seen previously for pit vipers, specifically Bothrops alternatus, B. ammodytoides, B. moojeni, B. neuwiedi and Crotalus durissus terrificus (de Roodt et al., 1998, 2016). This suggests that the influence of snake size on venom yield may vary between species. To test this hypothesis, it will be important to undertake extractions with additional snakes to increase statistical power, as well as with additional species.

By collecting wet mass, we were able to estimate both volume and total protein quantity expelled during extraction, by assuming a density of approximately 1050 kg/m^3^. This density has been reported for venom from Naja pallida (1084 ± 25 kg/m^3^) and Daboia russelii (1030 – 1070 kg/m^3^) (Bücherl et al., 1968; Kalita et al., 2018; Triep et al., 2013) and, it is likely to be true for other snakes because venom is a water-based mixture. To our knowledge, this is the first study to determine protein quantity delivered during extraction.

Overall, this study provides valuable insight into the wet venom yields and protein concentrations for several medically relevant African snakes, and an estimation of the total protein quantity delivered during milking. Assuming milking is an appropriate model for envenomation, this data could provide a tentative estimation of total protein toxin content delivered during a bite and could provide useful in improving snakebite models. This includes models for local envenomation, such as ex vivo human skin grafts (Alsolaiss et al., 2024) and organotypic models (Ahmadi et al., 2022), as well as in silico pharmacokinetic models of systemic pathology (Sanhajariya et al., 2018). This study also provides a basis for selecting the most appropriate methods for venom standardization in bioassays; a process highly relevant to quality control laboratories involved in the manufacture of antivenom. The physical properties of venom have received very little research focus, and the data presented here helps to address this knowledge gap.

CRediT authorship contribution statement

Stephanie French: Writing – review & editing, Writing – original draft, Visualization, Validation, Supervision, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Rachael Da Silva: Writing – review & editing, Resources, Investigation, Conceptualization. Martijn ten Have: Writing – review & editing, Resources, Investigation. Edouard Crittenden: Writing – review & editing, Resources, Investigation, Data curation. Paul Rowley: Writing – review & editing, Resources, Investigation, Data curation. India C. Cullen: Resources, Investigation, Data curation. Zachary Holland: Writing – review & editing, Data curation. Mark C. Wilkinson: Writing – review & editing, Resources, Methodology, Investigation. Cassandra M. Modahl: Writing – review & editing, Supervision, Resources, Funding acquisition.

Funding

CMM is supported by the Royal Society (ICAO\R1\231068), 10.13039/501100000691Academy of Medical Sciences (SBF008\1127) and 10.13039/501100000265Medical Research Council Impact Accelerator Account (MR/X502911/1 to the Tropical Infectious Disease Consortium).

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.