1. Introduction
Venoms are used for competitor deterrence, defence, and prey capture. These chemical cocktails use myriad protein types as starting substrates for the evolution of their toxic arsenals. Many of these protein families have been convergently recruited into the secretions of various venomous animal classes [1]. Novel expression of a normal body protein in the venom gland is followed by fixation and diversification of the proteins, with duplication and diversification leading to neofunctionalisation. In reptiles, such compounds have been recruited for use at different times in the diverse history of venom in this lineage with relative expression levels rising and falling for a particular toxin type along the evolutionary continuum [2]. As venoms evolve through the birth-and-death model of protein evolution [3], there are many instances of toxins also being lost from within a lineage [4].
Rather than prey death, prey immobilization is the driving selection pressure upon the venom, as there is no functional difference between a prey item that is physically unable to flee or defend itself and one that is dead [5,6]. Prey capture involving chemicals will produce a different form of injury than one involving mechanical subjugation techniques such as claws or cutting teeth. Interactions among toxins are selected for a combined function, which facilitates prey subjugation in a time dependent manner. Evolutionary pressures from other variables such as post-envenomation prey detectability and prey escape potential also shape venoms. The evolutionary pressures acting upon predatory venoms select specific actions that facilitate prey capture via target sites reachable by the blood. Actions upon nerve action potentials and the blood clotting cascade are two particularly important areas. Venoms that act on the nerves are typically used by species that include reptiles as major prey items in their diet. The specific function selected for the case of elapid snakes, which feed on sleeping lizards, is caused by alpha-neurotoxins, which antagonistically bind to the post-synaptic nicotinic acetylcholine receptor to produce flaccid paralysis [7]. Conversely, the long-glanded blue coral snake (Calliophis bivirgatus) and mambas (Dendroaspis species), which feed on active prey items with a high escape potential, have presynaptic-acting venoms that produce excitatory neurotoxicity, resulting in uncontrollable, uncoordinated spastic paralysis [6]. In the case of C. bivirgatus, there is an additional selection pressure for rapid acting venom due to specialisation upon other venomous snakes, which in turn have a high chance of prey retaliation.
Toxins that target the blood coagulation cascade do so in one of two mutually exclusive functional mechanisms; overall net anticoagulation or procoagulation. Anticoagulant venoms, such as those found in most pit-vipers, participate in the formation of haemorrhagic shock through a combined chemical assault, which may include the use of procoagulant pathways as part of the net anticoagulant function. Targets ranging from the cleavage of fibrinogen (whether outright destruction or the use of procoagulant pathways to produce weak, unnatural clots that are readily broken down by the abundant plasmin produced by synergistic toxins, which activate plasminogen) through to platelets (aggregation inhibition or unnatural clumping effects, which disrupt the normal role platelets serve in clotting), with these toxic effects accompanied by other toxins that increase the permeability of the vascular bed [1,2,8,9,10,11,12]. These venoms are often employed by ambush feeding vipers, which specialise upon warm-blooded prey items, and are able to be efficiently tracked post-bite using chemoreception and heat detection sensory arrays [13,14]. In contrast, strong clot forming procoagulant venoms are selected for lineages that feed upon prey with fast moving blood circulation rates. Such prey lineages are vulnerable to toxins that produce blood clots that rapidly subjugate through stroke induction, such as taipans (Oxyuranus species) feeding upon rodents [15].
The African boomslang (Dispholidus typus) and the twig snakes (Thelotornis species) are closely related iconic African snakes belonging to the family Colubridae, and both inhabit arboreal, predatory niches that include prey items with a high potential for escape [16,17]. The limited work that has been undertaken upon them has revealed that both have clot-forming, procoagulant venoms [17,18,19,20,21,22]. While the venoms appear to be functionally similar in their potent action upon blood chemistry, the relative complexity of the venoms is unknown. While Dispholidus has been shown to have a venom dominated by P-III snake venom metalloprotease (SVMP) [22,23,24], other toxin types are present in the venom in lower amounts [3,23]. In vipers, certain procoagulant P-III SVMPs have been shown to activate prothrombin [25,26,27,28] or Factor X [27,28,29]. Strong clot forming procoagulant function has convergently evolved in the venom of D. typus and similar activities are suggested for T. capensis [30]. The venom composition for any Thelotornis species is unknown so the toxin type responsible for procoagulation can only be inferred as likely being a P-III SVMP like its related genera Dispholidus. From a functional perspective, these two related but very divergent snakes share a coagulotoxic action with their last common ancestor [3,23]. These snakes have evolved procoagulant venoms convergently in relation to vipers in the Daboia genus [31,32] or elapids in the Oxyuranus or Pseudonaja genera [33,34]. Daboia use serine proteases to activate factor X, while Oxyuranus and Pseudonaja use a mutated factor Xa:factor Va complex to escape hemostatic control [35].
Dispholidus is genetically close to the genus Thrasops. Dispholidus and Thrasops are in turn sister to Thelotornis. The two genera studied here differ in their occupied niches and morphology. Dispholidus is a monotypic genus consisting of the boomslang (D. typus), which is an active foraging snake with very large eyes with circular pupils and acute vision, a large round head, and variable colour, ranging from uniform brown to green with darkened scale edges [17,20]. The large head is quite similar to the tree cobras (e.g., Naja goldii), including enlarged eyes. In contrast, Thelotornis is a genus of ambush feeding, highly cryptically and intricately patterned snakes with large eyes, horizontal pupils, and near binocular vision [17,20,36]. Their elongate heads are convergently similar to other Colubridae, particularly the Ahaetulla genus from Asia and the Oxybelis genus of the southern U.S. and Central America.
Birds account for the majority of Dispholidus prey items as it hunts in the canopy. The fast-moving, endothermic blood circulatory system of avians is similar to mammals in being vulnerable to rapid prey subjugation by stroke formation, thus neutralising the high escape potential posed by prey items with flight ability. While lizards such as chameleons form a percentage of the Dispholidus diet [16,17], they are non-dangerous and slow moving and are also taken during their diurnal activity period when blood circulation is at its maximum. Thelotornis may have a more complex diet than D. typus as it occupies a niche located near or on the forest floor and snakes of this genus are perhaps also more likely to be sensitive to prey escape potentials due to their slower movements relative to the fast moving Dispholidus [16,17,36].
Neither genus of snakes was thought to be of significant medical consequence until each killed an eminent herpetologist (Robert Mertens and Karl Patterson Schmidt, respectively) [24]. However, the ability for D. typus to cause serious harm to humans, and even death, had previously been reported by F.W. Fitzsimons as early as 1909 [17,20,21,37,38,39,40]. Strikingly, T. capensis has been shown to produce nearly identical human clinical symptoms as reported for D. typus, with death the result of internal bleeding such as cerebral haemorrhage [19,30,41]. However, the existing antivenom manufactured to treat bites caused by D. typus is said to be ineffective in cases of T. capensis envenomation [17,21,30]. While prey injected with significant concentrations of procoagulant venoms items succumb to stroke [42], human deaths from these venoms are due to internal haemorrhage due to the dilution of the procoagulant venom into a volume of blood much larger than the concentrated effects in prey animals, resulting in defibrinogenation due to venom induced consumptive coagulopathy [2,15,28,41,43].
Despite a monovalent antivenom existing and routinely administered to bite victims [17,22], limited research has been conducted on the function of D. typus venom, with the literature being dominated by bite case reports or proteomics in the absence of bioactivity testing. Even less is known about Thelotornis venoms, with limited previous publications having mainly dealt only with T. capensis, while the rest of the genus is unknown in its venom composition. We aimed to investigate the key similarities and differences between the two genera utilising a range of proteomic and bioactivity approaches to reconstruct their functional evolution in response to prey preference and ecological niche occupied. It is hypothesised that T. mossambicanus has a greater diversity of venom toxin components, reflective of its diverse diet and their associated heighted potential for prey escape, than the more specialised feeder D. typus.
2. Results
2.1. Skull and Venom Gland Anatomical Comparisons
Both species displayed enlarged rear-fangs, consistent with their demonstrated efficient venom delivery in predation and also defensive bites. Reflective of their difference in head shape, the T. mossambicanus venom gland was more elongate than that of D. typus due to contributions by both hard and soft tissue (Figure 1).
2.2. Proteomics
Shotgun mass spectrometry analysis recovered the same toxin types from each venom; CRiSP, PLA2 Type IIE, 3FTx, and SVMP. However, 1D and 2D gels showed significantly different relative abundances and complexity levels. While both 1D and 2D gels indicate that both species are dominated by P-III SVMP (Figure 2 and Figure 3, Supplementary Data), T. mossambicanus also contained phospholipase A2 (PLA2 Type IIE) in high amounts. The presence of PLA2 Type IIE is a significant discovery as it is the third type of PLA2 to be characterised in snake venoms [44]. The molecular weights of identified P-III SVMP correspond to previous transcriptome and proteome data for D. typus [24].
2.3. Enzymatic Assays
Fluorescent Determination of Matrix Metalloprotease and Kallikrein Activity
Despite T. mossambicanus having a more complex venom profile than D. typus, and thus lower relative amounts of P-III SVMP (Supplementary Data), it actually displayed a higher rate of cleavage of a matrix metalloprotease specific substrate (Figure 4). A two-way (RM) ANOVA (α = 0.05) indicates that there is a significant difference in Column Factor (species and concentration), accounting for 78.6% of the total variance (after adjusting for matching); F(3, 8) = 45.2, p < 0.001. Tukey’s post-hoc test indicated that there is a significant difference in activity between D. typus and T. mossambicanus at a concentration of 0.5 µg/100 µL (p ≤ 0.04). The levels of T. mossambicanus were of a high level, similar to that of the viperid snake Gloydius saxitilis included for comparison.
Contrastingly, D. typus venom demonstrated a much higher kallikrein activity compared to that of T. mossambicanus. D. typus had an active concentration of 0.1 µg/100 µL and 0.5 µg/100 µL, compared to T. mossambicanus, which was only slightly active at 0.5 µg/100 µL (Figure 5). A two-way (RM) ANOVA (α = 0.05) indicates that there is a significant difference within the test, with Column Factor (species and concentration) accounting for 80.1% of the total variance (after adjusting for matching); F(3, 7) = 136.6, p ≤ 0.001. Tukey’s post-hoc test indicated that there is a highly significant difference in activity between D. typus and T. mossambicanus at both concentrations (p ≤ 0.001). This is indicated by the increased presence of kallikrein-like serine proteases in D. typus’ venom profile (Figure 2). However, even the proportionally higher D. typus activity levels were insignificant in relation to the representative viperid Trimeresurus vogeli included for comparison, thus indicating that this function is a trivial aspect of the venom, consistent with the non-detection in our proteomics and low-levels in the published D. typus venom gland transcriptome [24].
2.4. Fluorescent Determination of sPLA<sub>2</sub> Activity
We tested for continuous secretory phospholipase A2 (sPLA2) enzymatic activity over 100 measurement cycles using and following the EnzChek® Phospholipase A2 Assay Kit (Cat#E10217). T. mossambicanus showed increased levels of sPLA2 activity compared to D. typus at a concentration of 50 ng/µL (Figure 6). This increase in activity by T. mossambicanus is also mimicked in the venom profiling of both venoms via 1D SDS PAGE (Figure 2). A two-way RM ANOVA (α = 0.05) indicates that there is a significant difference within the assay, with Column Factor (species and concentration) accounting for 63.41% of total variance (after adjusting for matching); F(2, 6) = 78.75, p ≤ 0.001. Tukey’s post-hoc test indicated there is a highly significant difference in activity between D. typus from the positive control and between D. typus and T. mossambicanus (p ≤ 0.001). Tukey’s post-hoc test also indicates a significant difference of T. mossambicanus from the positive control (p ≤ 0.01). The T. mossambicanus levels were similar to that of the representative viperid snake Azemiops feae included for comparison.
2.5. Procoagulation Analysis
While both venoms were potently procoagulant, we find that the venom of T. mossambicanus coagulates plasma much faster that that of D. typus (Figure 7 and Figure 8). At a 20 µg/mL concentration, T. mossambicanus venom clots the plasma on average in 7.5 s (SD 0.15 s), compared to D. typus, which clots the plasma on average in 11.2 s (SD 0.06 s) (Table 1). The T. mossambicanus venom is amongst the most potently coagulotoxic of any we have tested, being on par with Oxyuranus and Pseudonaja venoms tested under identical conditions (unpublished results). At a 0.05 µg/mL concentration, T. mossambicanus venom clots the plasma on average in 80.4 s (SD 0.52 s), compared to D. typus, which clots the plasma on average in 146.6 s (SD 5.23 s) (Table 1). Figure 6 demonstrates the relative potency of T. mossambicanus venom against human plasma, compared to D. typus, in the reduced variation in clotting times between dilutions. The distance between the first vertical line and the second vertical line is longer in T. mossambicanus than in D. typus, illustrating that there is a longer rise in clotting times, depicting that T. mossambicanus venom holds its potency between dilutions more so than D. typus (Figure 7, Table 2). This is also demonstrated by the point where the curve crosses the x axis, which is higher in D. typus than in T. mossambicanus (R0 ϕ2 = 37.83 and 23.17 s respectively) (Table 2). The additional investigation with and without cofactors CaCl2 and phospholipid had no impact on the clotting times of either species, with clotting times remaining the same at all dilutions (data not shown). This is indicative of both venoms being calcium and phospholipid independent in exerting their procoagulant coagulotoxic actions.
With the addition of diluted boomslang monovalent antivenom to the dilution series, it is evident that the antivenom neutralises D. typus remarkably well, in contrast to having little neutralising effect on T. mossambicanus (Figure 8). At 20 µg/mL venom concentration with the antivenom at a final concentration in the 250 µL cuvette volume of 1% of that of the original vial concentration, T. mossambicanus clotted the plasma in 9.6 s (SD 0.17 s) instead of the 7.5 s without antivenom, compared to D. typus, which clotted the plasma at 20.7 s (SD 1.07 s) instead of the 11.2 s without antivenom (Table 1). At a 0.05 µg/mL venom concentration, the added antivenom had some effect on T. mossambicanus, which clotted in 381.2 s (SD 4.71 s) instead of 80.4 s without antivenom (Table 1). D. typus however reached maximum clotting time of 999.9 s (Table 1) at a venom concentration of 0.05 µg/mL with the addition of antivenom, demonstrating a significantly greater effective neutralising ability against this species.
Comparing EC50 outputs, it is evident that, despite the variation in clotting times between the species (Figure 8), half-maximal concentration is reached at a similar point by both species. This is due to the dilution series following the same trajectory regardless of time (Figure 8). When transformed, this trajectory becomes more evident, as there is little x-axis difference between T. mossambicanus with and without antivenom (0.24 µg/mL: 95% CI 0.22–0.26 µg/mL, and 0.23 µg/mL: 95% CI 0.20–0.26 µg/mL respectively) and T. mossambicanus and D. typus without antivenom (0.24 µg/mL and 0.20 µg/mL: 95% CI 0.18–0.23 µg/mL respectively) (Figure 9). However, when antivenom is introduced with D. typus, the EC50 x-axis shifts significantly to the right (1.62 µg/mL: 95% CI 1.49–1.76 µg/mL) (Figure 9). Due to each sub data set being transformed, first by log concentration and then the normalisation of clotting time, each final data point reaches 100% (Figure 9). Taking into account the y-axis shift in addition to the x-axis shift (Figure 8), the T. mossambicanus venom has an antivenom induced relative shift in the clotting curve of 4.69, while the D. typus venom has an antivenom induced relative shift in the clotting curve of 53.12. Thus the antivenom is 11.3 times more effective at neutralizing D. typus venom than T. mossambicanus venom. Thus if the antivenom had no significant effect upon T. mossambicanus under such ideal circumstances as conducted in this study, then there is little chance of it having an therapeutic effect in a clinical scenario without the use of extreme amounts of antivenom.
3. Discussion
Both venoms tested here were found to be highly procoagulant without requiring calcium or phospholipid as co-factors. However T. mossambicanus is notably more potent than D. typus, as well as being extremely poorly neutralised by SAIMR boomslang antivenom (Table 2, Figure 7, Figure 8 and Figure 9). The decoupling from co-factors is a notable discovery as this lack of shift between tests with or without co-factors had only been well-documented for Echis carinatus. This is the first extensive antivenom comparison of both species marking the relative effectiveness of the available boomslang monovalent antivenom.
Previous reports have described these two species as exhibiting strikingly similar lethal envenomations [17,19,20,21,30,37,38,39,40] that can be attributed to their venoms being dominated by P-III SVMPs (Figure 2 and Figure 3). It is notable that, despite T. mossambicanus having a lower concentration of SVMP due to its greater venom complexity, it displayed a higher relative rate of metalloprotease activity in addition to being more potently coagulotoxic (Figure 4, Figure 7 and Figure 8). Further, T. mossambicanus possesses two distinct molecular weight classes of SVMP as opposed to the single band present in D. typus venom (Figure 2). Thus the differential proteomics profile is mirrored by differential coagulotoxic activity. This key difference between the species’ venom profiles may also shed light as to why the available antivenom for D. typus has only a small neutralising effect on T. mossambicanus venom (Figure 8 and Figure 9, Table 1).
Calcium-independent prothrombin activation is a rare feature in snake venoms, with most venoms typically requiring calcium for such a coagulotoxic activity. A previously documented notable exception is the ecarin-type P-III SVMP from Echis carinatus, which does not have a marked shift between tests with or without co-factors [25,45]. Thus, the presence of such an activity in the venoms of these two colubrid snakes points towards a remarkable case of functional convergence within the same toxin class.
Procoagulation accompanied by plasmin inhibition is an effective way to potentiate the coagulotoxic effects as the blood clots formed would have a longer half-life since the role of plasmin is to break blood clots down. This synergistic activity has been documented in viper venoms (e.g., Daboia) and elapid venoms (e.g., Pseudonaja), where the procoagulation is accomplished by very different toxin types (SVMP and fXa:fVa, respectively), but the plasmin inhibition is exerted by the same toxin type (kunitz peptide) [46,47]. However, neither D. typus or T. mossambicanus venoms displayed this synergistic action.
Despite having the same procoagulant mechanism, it appears evident that additional external evolutionary pressures are driving the variation between these two species. In addition to differential complexity, the venoms were functionally variable relative to each other in non-procoagulant activities, with D. typus having serine protease activity (Figure 5), while conversely T. mossambicanus had strong PLA2 activity (Figure 6). It is known that evolution among arboreal specialists further influences venom composition of toxins’ families [48]. This is usually in the form of prey ecology, dependent on niche occupation. Even though both species occupy an arboreal habitat, their differential morphology and behaviour allow for varying prey interactions, thus potentially further contributing to venom variation and punctuated evolution. As D. typus is an agile and fast-moving snake, it is able to pursue prey, thus minimising escape potential. T. mossambicanus however is a slower moving cryptic snake that ambush feeds, and thus the prey escape potential may be significantly higher. Thus, there may be a stronger selection pressure operating upon T. mossambicanus for a venom which rapidly immobilises prey. This is consistent with what has been observed in other lineages of venomous animals in which prey escape potential is a significant shaping factor and thus the venom is under extreme selection pressure for rapid immobilising action [2,5,6,15,49,50,51,52,53,54,55,56,57,58,59].
This investigation not only has revealed the differential evolution of D. typus and T. mossambicanus venoms and the poor performance of boomslang antivenom against T. mossambicanus venom, but it has also reinforced that there are commonalities to venom evolution such that there may be levels of predictability in regards to niche occupation and prey escape potential shaping venom relative rates of action. The differences in ecological niches occupied were also reflected in differential skull morphology (Figure 1). This reinforces that, in addition to the applied application of venom research in regards to biodiscovery and clinical effects, research such as this contributes to the growing body of venom evolutionary theory.