Introduction

Snakes have been a source of fascination and symbolism since ancient times, being associated with rituals, good and evil mythology, fertility and healing. The snake has been on the medical emblem for over two millennia, based on the Greek God of healing and medicine Asclepius, who held a staff with a snake coiled around it (Antoniou et al. 2011). The advance of science has overtaken centuries of superstition about snakes, and snake venom research now spans the medical, pharmacological, ecological and molecular evolutionary domains (Chiappinelli 1983; Dutertre and Lewis 2010; Giorgianni et al. 2020; Oliveira et al. 2022a; Perry et al. 2022; Saviola et al. 2014; Whittington et al. 2018), with an exponential increase in publications over the last century (Sofyantoro et al. 2022).

The study of snake venom is a multi-disciplinary endeavour encompassing a diverse range of research areas from the therapeutic potential of toxins and drug discovery (Kalita et al. 2021; Li et al. 2018; Omidi et al. 2021; Urra and Araya-Maturana 2022), antivenom development and the treatment of snakebite (de Silva et al. 2016b, c, d, a; Fry et al. 2003; Gutiérrez et al. 2011; Harrison et al. 2011; Maduwage et al. 2015; Visser et al. 2008), pathophysiology of envenomation (Maduwage et al. 2013; Mamede et al. 2020; Silva et al. 2016b, 2016c), using toxins as physiological probes to better understand physiological mechanisms (Chiappinelli 1983; Harvey and Robertson 2005; Kerns et al. 1999; Wijeyewickrema et al. 2005; Yang et al. 2016), and venom characterisation, which includes venom composition and the sequence, structure and pharmacology of toxins (Bolt 2021; Kleiz-Ferreira et al. 2021; Tasoulis et al. 2022b; Yang et al. 2016). More recently the evolutionary biology of toxins and venom has been investigated, including both inter- and intra-specific venom variation and its evolutionary and ecological significance (Margres et al. 2021; Mora-Obando et al. 2020; Rautsaw et al. 2019; Zancolli et al. 2019). Venom is also being used as a model for understanding gene duplication, expression and regulation, and molecular evolution (Giorgianni et al. 2020; Hargreaves et al. 2014; Perry et al. 2022; Whittington et al. 2018).

The evolutionary history of how some snakes became venomous has only recently begun to be elucidated (Kardong 2002; Vidal 2002; Vonk et al. 2008). Snake venom is produced in a pair of post-orbitally located venom glands, and the venom glands of all snakes are homologous (Jackson et al. 2017). There are several tooth-bearing bones in the upper jaws of snakes, but venom-delivering fangs are only located on the marginal upper jawbone called the maxilla, with the location of the fang on the maxilla (front or rear), varying amongst different clades of snakes (Westeen et al. 2020). There are three families of front-fanged venomous snakes, Atractaspididae (Burrowing asps, 72 species), Elapidae (Elapids 389 species) and Viperidae (Vipers, 373 species) (data taken from www. Reptile-database. org) and also a number of rear-fanged venomous snake families and subfamilies (Colubriformes). The front-fanged elapids and vipers (Fig. 1) are responsible for almost all serious human envenoming and deaths globally (Gutiérrez et al. 2017), and venoms from all these groups have been the focus for most research in this field (Tasoulis and Isbister 2017).

Fig. 1
figure 1

Examples of the two most medically significant and speciose families of the front-fanged snakes. A An Elapid, an Australian coastal taipan (Oxyuranus scutellatus), and B a viper, a Sonoran desert sidewinder (Crotalus cerastes cercobombus). Note the large opening below and in front of the eye of the viper. This is an infra-red sensing pit, neurally integrated with the eye and allowing vision in the infra-red spectrum and distinguishes “pit vipers” Crotalinae, from “true vipers” Viperinae, the two major subfamilies of the family Viperidae. Photos courtesy of Shane Black (A), and Ross McGibbon (B)

Full size image

Vipers are classified into three subfamilies, but 99% of viper species belong to just two of these: true vipers (Viperinae; 101 sp.), and pit vipers (Crotalinae; 271 sp.). Pit vipers are distinguished by the presence of infra-red sensitive facial pits (Fig. 1B), giving pit vipers the ability to see in the infra-red spectrum (Gower et al. 2019; Liu et al. 2016). A universally recognised group of vipers are the American rattlesnakes. Elapids are typically slender-bodied snakes and include many of the well-known species of medical significance, such as cobras, mambas, kraits, coral snakes, taipans, tiger snakes and sea snakes.

Snake venoms are heterogeneous mixtures of different toxic protein families with each protein family potentially containing many different toxins. Toxic proteins produced by different genes are classified as different toxins, whilst toxins with different sequences that are produced by the same gene are called proteoforms (Smith and Kelleher 2013); these can be the result of alternative splicing or post-translational modifications (Ogawa et al. 2019). Toxins have evolved to affect various physiological systems, and can be neurotoxic, haemotoxic, myotoxic or cytotoxic (Gutiérrez et al. 2017), depending on the site of action. Toxins are typically monomeric, but toxins can also oligomerise with themselves or other toxins to become multimeric. If they oligomerise with the same toxin, they are homomeric, and if they oligomerise with different toxins, they are heteromeric. Most snake toxins are in the mass range of 4 to 100 kDa (Olaoba et al. 2020; St Pierre et al. 2006); however, some multimeric toxins may have a total mass of 250 kDa; for example, Pseutarin, the procoagulant serine protease toxin from the Australian brown snake genus Pseudonaja (Rao and Kini 2002). Structurally, monomeric toxins can contain several domains, these are conspicuous parts of the protein that fold independently from the rest. In addition to toxin proteins, there are also believed to be several non-toxin protein modification enzymes present in the venom (Isomoto et al. 2022).

Different toxin protein families are mostly located on different chromosomes (Suryamohan et al. 2020), and multi-gene protein families occur as tandemly arrayed gene clusters (Giorgianni et al. 2020). These have expanded from single ancestral genes through repeated gene duplication events (Dowell et al. 2016; Shibata et al. 2018). The mechanisms for how these protein families may be integrated into regulatory networks for coordinated expression is an exciting new area of research (Perry et al. 2022).

Since 2004, there has been a rapid increase in the number of publications characterising snake venom proteomes, partly facilitated by the advent of transcriptomics, which has enabled far more rapid identification of toxins. A review in 2017 identified studies of the venom composition of 132 snake species (Tasoulis and Isbister 2017) and showed that snake venom protein families could be grouped using a combination of occurrence and abundance into a hierarchy of categories. There were four dominant protein families—phospholipase A2 (PLA2), snake venom metalloprotease (SVMP), snake venom serine protease (SVSP) and three-finger toxin (3FTx), six secondary protein families—L-amino acid oxidase (LAAO), cysteine-rich secretory protein (CRiSP), kunitz peptides (KUN), natriuretic peptides (NP), C-type lectins (CTL) and disintegrins (DIS) and nine minor protein families–nerve growth factor (NGF), nucleotidase (NUC), vascular endothelial growth factor (VEGF), phosphodiesterase (PDE), phospholipase B (PLB), hyaluronidase (HYAL), vespryn/ohanin, (VES/OHA) snake venom metalloprotease inhibitor (svMPi), acetylcholinesterase AChE). More recent analysis using a larger dataset suggests that there are in fact 14 minor families although some of these may not possess toxic activity. The remaining protein families were classified as rare.

Analytical techniques used to characterise snake venoms

The first aim in characterising a snake’s venom proteome is toxin identification and quantification. This usually involves an initial venom separation (fractionation) step known as venom decomplexation. However, whole venom proteome characterisation is also often employed by researchers. Prior to the turn of the millennium, toxin identification was a laborious and time-consuming process involving Edman degradation (protein N-terminal sequencing). However, modern analytical techniques utilising workflows combining reverse-phase high-performance liquid chromatography (RP-HPLC), sodium dodecyl sulphate polyacrylamide gel electrophoresis (SDS-PAGE), mass spectrometry (MS) and transcriptomics, have revolutionised the study of venom composition (Tan 2022).

There have been several recent excellent reviews comprehensively covering the methodologies of characterising snake venom proteomes (Sahyoun et al. 2022; Slagboom et al. 2022; Tan 2022), so they will only be briefly discussed here. All studies of snake venom investigations start with the collection of venom from accurately identified snakes either from licensed companies of snake venoms, or with clear information on the snake identification and location where the snake came from. After venom extraction, the venom from multiple individuals of the same species is usually pooled and then lyophilised by freeze drying. Lyophilised venom can then retain viability for at least eight decades (Jesupret et al. 2014). More recently, the importance of also studying the venom from individual snakes has been understood, and such investigations require identification and collection of data on all snake individuals involved. The characterisation (identification and quantification of constituent protein families and toxins) of snake venom proteomes can be achieved by a variety of workflows, all of which have different advantages and disadvantages. Several studies have indicated that combining different methodological approaches appears to maximise proteomic coverage (Chanda and Mukherjee 2020; Choudhury et al. 2017). A recent review which compared all the different workflows used over a 4-year period (2017 to 2021) (Tasoulis et al. 2022a), found that the strategy of choice was bottom-up proteomics, in particular involving a preliminary two-dimensional decomplexation of the venom, consisting of RP-HPLC, followed by SDS-PAGE, prior to in-gel trypsin digestion and MS analysis (Fig. 2). Other methods which are commonly used include shotgun proteomics, which uses digested crude venom without prior fractionation, top-down proteomics (TDP), which performs MS analysis directly on undigested crude venom or venom fractions, and bottom-up proteomics (BUP), which involves a sample preparation step of trypsin digestion, which can be either in-solution or in-gel, prior to MS. Several comparative studies have indicated that preliminary decomplexation of the venom by RP-HPLC followed by SDS-PAGE increases proteomic coverage, as the second decomplexation step of SDS-PAGE is useful for separating co-eluting toxins with similar hydrophobicity but often belonging to different protein families with different masses.

Fig. 2
figure 2

Simplified summary of published workflows used from 2017 to 2021 to characterise the venom proteomes of vipers and elapids, showing how almost half of the total studies used bottom-up proteomics (thick blue arrows), consisting of a two-dimensional decomplexation of the crude venom by reverse-phase high-performance liquid chromatography (RP-HPLC), followed by sodium dodecyl sulphate polyacrylamide gel electrophoresis (SDS-PAGE). The venom fractions are then subjected to in-gel trypsin digestion prior to mass spectrometric analysis. Diagram modified from Tasoulis et al. 2022a (colour figure online)

Full size image

Snake venom composition

Snake venom has been shown to mostly consist of a relatively small number of protein families. We have previously attempted to classify these protein families in a hierarchical system based on a combination of occurrence across species, combined with relative abundance of these protein families as a percentage of the total venom. We graphed these values from 179 species of front-fanged snakes, 68 species of elapids, 26 species of true vipers, and 84 species of pit vipers, plus 1 species of viper from the monogeneric subfamily Azemiopinae (Fig. 3). These categories are discussed in more detail below.

  1. 1.

    Dominant protein families (4): mean abundance > 10% of the whole venom and occur in > 33% of species.

  2. 2.

    Secondary protein families (6): mean abundance > 3% but < 10% of the whole venom and occur in > 33% of species.

  3. 3.

    Minor protein families (14): occur in > 5% of species with a mean abundance of < 3% (12) or are key toxins in one or a small number of species with an abundance > 3% (Fig. 3).

  4. 4.

    Rare protein families (18): any protein family with likely toxic activity with a mean abundance < 5% and occur in < 5% of species. Many of the listed rare protein families may be non-toxin protein modification enzymes.

Fig. 3
figure 3

XY graph showing the criteria for our hierarchical classification system of the 42 protein families recorded from the venoms of 179 species of vipers and elapids (23.5% of all species), based on a combination of mean abundance as a percentage of the total venom (x-axis), and the number of species in which it is present (occurrence) (y-axis). Categories are: dominant protein families—present in > 33% of species, and with a mean abundance of > 10% of the venom (square icons above and to the right of dashed red lines). Secondary protein families—present in > 33% of species, and with a mean abundance of > 3% and < 10% of the venom (filled circles between the two vertical red dashed lines). Minor protein families—present in > 5% of species with a mean abundance of < 3% of the venom (triangles between the Y-axis and the vertical dashed red line line), with two exceptions; waglerin and crotamine. Rare protein families—present in < 5% of species and with a mean abundance of < 5% of the venom (empty black circles in lower left corner of graph). Details of the dominant protein family are given in Table 1 (colour figure online)

Full size image
Table 1 The four dominant protein families as based on a combination of mean abundance as a percentage of the total venom protein, and the number of species in which it is present (occurrence)
Full size table

Phospholipase A 2 (PLA 2 )

PLA2s are one of the most important protein families in snake venoms globally, occurring in the majority of elapids and vipers. It has been recorded in all species of vipers except for some species of Palm vipers (Bothriechis), and in all species of elapids except for Senagalese cobra (Naja senegalensis), and east African green mamba (Dendroaspis angusticeps), although it is expressed in very low amounts in all mambas. The type of PLA2 differs between the two snake families, with elapids possessing pancreatic type (group I) PLA2 and vipers having synovial type (group II) PLA2 (Suranse et al. 2022). When averaged across all species that contain PLA2, the mean abundance of PLA2 is 26%, (31% in elapids and 23% in vipers). They are more abundant in the Australasian elapid clade (Hydrophiinae) and can make up 90% of the venoms of some of the species in this clade (e.g. Pseudechis papuanus (Pla et al. 2017)) and are likely the reason for systemic myotoxicity being an important clinical manifestation of Australian snake envenomation (Johnston and Isbister 2021). They have a diverse pharmacological profile and are responsible for pre-synaptic neurotoxicity from many elapids (Johnston et al. 2017a; Kuruppu et al. 2005), myotoxicity reported in Australasian elapids (Hart et al. 2014; Johnston et al. 2013), cytotoxicity, inflammation and anticoagulant effects (Johnston et al. 2013; Lane et al. 2011). They have a monomeric mass of 13 to 18.5 kDa (Cendron et al. 2012; Kang et al. 2011), and may oligomerise with other PLA2s (e.g. trimeric taipoxin) or with other toxin protein families such as kunitz peptides—(e.g. β-bungarotoxin in krait venoms) (Cendron et al. 2012; Rowan 2001). The most potent PLA2 neurotoxins are reported to be oligomers (e.g. taipoxin) (Montecucco and Rossetto 2008). In the viperid group II PLA2s, the 49th amino acid residue is aspartate (ASP49 or D49) and is in the active site (Suranse et al. 2022). This amino acid appears to be responsible for the catalytic activity of these PLA2. Interestingly, there is poor correlation between enzymatic activity and toxicity (Montecucco and Rossetto 2008; Tasoulis et al. 2020a). The presence of PLA2 in almost all venomous snakes means the detection of PLA2 activity may be a useful way to detect systemic envenoming in human snakebite patients. Preliminary studies have demonstrated that the detection of PLA2 activity in serum of snakebite patients distinguishes between patients with and without systemic envenoming (Isbister et al. 2020; Maduwage et al. 2014).

Three-finger toxins (3FTx)

3FTxs are one of the two most abundant protein families present in elapid venoms, with PLA2. The mean abundance of 3FTxs in all elapid species is 52%. They are extremely rare in viper venoms being only recorded in five species. In elapids, they have been recorded in all species and can make up 95% of the venom (e.g. Micrurus tschudii, M. surinamensis). These non-enzymatic toxins have a mass of between 6 and 9 kDa (Clarke et al. 2006), and are named after their characteristic spatial structure, as all have a similar highly conserved folding pattern consisting of three β-stranded loops (fingers), extending from a globular central core which is stabilised by four conserved disulfide bridges (Kini and Doley 2010). An important sub-group of this protein family are the α-neurotoxins, post-synaptic neurotoxins that act at the nicotinic acetylcholine receptors (nAChRs) (Nys et al. 2022). Alpha-neurotoxins competitively antagonise the ligand-gated ion channels at the acetylcholine binding site and prevent the opening of the channel. Alpha-neurotoxins are further sub-divided into two groups: 1) short-chain neurotoxins 60–62 amino acid residues and four disulfide bonds, and 2) long-chain 66–75 amino acid residues and five disulfide bonds, having an extra disulfide bond in the central finger (Kini 2002). It is important to note that human nAChR have an exceptionally low affinity for short-chain α-neurotoxins compared to long-chain α-neurotoxins (Ishikawa et al. 1985), and long-chain neurotoxins are likely to be more important in human envenomation (Silva et al. 2018). Other 3FTxs target other types of ion channels, such as muscarinic 3FTxs (Kini 2002). African mamba (Dendroaspis) venoms contain a further two types of specialised 3FTx neurotoxins—calciseptines, which selectively block L-type calcium channels, and fasciculins, which inhibit acetylcholinesterase (Casewell et al. 2020). The Malayan Blue Coral Snake (Calliophis bivirgata) contains the 3FTx neurotoxin, calliotoxin, which activates the voltage-gated sodium channel, Nav 1.4 (Yang et al. 2016). In some cobra venoms, particularly spitting cobras, cytotoxic 3FTxs are thought to be responsible for serious local dermonecrosis (Hiu and Yap 2022; Lin et al. 2022).

Snake venom metalloprotease (SVMP)

SVMPs are the most abundant toxin proteins in viper venoms, with a mean abundance of 34% across all viper species. SVMPs have been recorded in every viper species except Bothriechis nigroviridis. Some Bothrops spp. contain as much as 74% SVMPs in their venom. In elapids, SVMPs are far less abundant (mean abundance 5% across all species), although a recent study indicates that it may be more abundant in the currently understudied Australian elapid clades with Hoplocephalus stephensii containing 37% SVMPs (Tasoulis et al. 2022b). Metalloproteases are classified into three major classes, P-I, P-II and P-III, based on the domain structure (Olaoba et al. 2020). The P-III are the largest and believed to be the ancestral form, with P-II and P-I having lost some of the domains. P-III consist of a metalloprotease domain, a disintegrin domain, and a cysteine-rich domain (type P-III also contains a C-type lectin domain). P-II are only made up of a metalloprotease domain and a disintegrin domain, whilst P-I consist simply of a metalloprotease domain. Toxins belonging to this family typically, but not always, possess haemorrhagic activity, although some are procoagulants, such as the potent prothrombin activator Ecarin in Echis venom (Ainsworth et al. 2018). Ecarin is arguably one of the most medically important snake toxins in human envenoming, being responsible for venom-induced consumption coagulopathy (VICC) for the thousands of Echis envenoming cases (Mion et al. 2013).

Snake venom serine protease (SVSP)

This is the least abundant of the dominant protein families with a mean abundance of 14% in vipers but present in every species. SVSPs can make up as much as 53% of the venom in the pit viper Ovophis convictus and 26% in the true viper Bitis gabonica. SVSPs appear to be far less compositionally important in elapids with a mean abundance of only 1.3% and present in just over a third of the species. They appear to be a more important component in the venoms of the Australian clade, and the few published studies of Australian elapid venom proteomes have found higher than average values for elapids (6% for both Notechis scutatus and H. stephensii). This is consistent with venom-induced consumption coagulopathy VICC being the most important clinical manifestation in Australian snake envenomation (Johnston et al. 2017b). Pharmacologically SVSPs are procoagulants, disrupting haemostasis and triggering the clotting pathway by partial cleavage of the inactive clotting factor proteins (zymogens) to the active or partially active enzymes (Ainsworth et al. 2018; St Pierre et al. 2005). These toxins include factor X activators, factor V activators, prothrombin activators and thrombin-like enzymes (TLEs) (Isbister 2010). These toxins are medically important in some Australian elapids causing VICC (Isbister et al. 2010b), and responsible for significant coagulopathy in many vipers e.g. Calloselasma rhodostoma, and likely also Bothrops asper (Otero-Patiño 2009; Wongtongkam et al. 2005). This protein family also includes the anticoagulant protein C activator which are mainly found in the venoms of the pit viper Agkistrodon complex (Gempeler-Messina et al. 2001). Another class of toxins in this family are the kallikreins, which in addition to having fibrinogenolytic activity can also cleave kininogen to release bradykinin which can result in hypotensive effects (Felicori et al. 2003; Fry 2015).

L-Amino acid oxidase (LAAO)

This is the most ubiquitous of the secondary protein families, present in 93% of vipers and 69% of elapids (83% combined) (Table 2). LAAOs have a mean abundance of 4% and a maximum of 17% in vipers and a maximum of 11% in elapids. They are homodimers with monomeric masses of 50 to 70 kDa (Fox 2013). They have become a focus of research for their antimicrobial and anticancer potential (Cecchini et al. 2005; Ciscotto et al. 2009; Costa et al. 2014; Tan et al. 2018). Despite them being so widespread and extremely well characterised, their functional role in human envenomation remains unclear (Izidoro et al. 2014; Pawelek et al. 2000; Tasoulis et al. 2022b).

Table 2 The six secondary protein families as based on a combination of mean abundance as a percentage of the total venom protein, and the number of species in which it is present (occurrence)
Full size table

Cysteine-rich secretory protein (CRiSP)

This is the second most ubiquitous of the secondary protein families, present in 74% of species, with a mean abundance of 3% and a maximum of 14%. CRiSPs are more commonly recorded in vipers than elapids (86% versus 56%, respectively). They are monomeric with a molecular mass of approximately 25 kDa (Suntravat et al. 2019). They belong to the catabolite activator protein (CAP) superfamily (Tadokoro et al. 2020). CRiSPs from Australian elapids are the only known protein blockers of cyclic nucleotide-gated (CNG) ion channels which play pivotal roles in sensory transduction by retinal photoreceptors and olfactory neurons (Suzuki et al. 2008). Their role in the pathophysiology of human envenomation is still poorly understood (Boldrini-França et al. 2017), but a change or loss of smell has been reported for a number of Australian elapid bites (Churchman et al. 2010; Johnston et al. 2013).

C-type lectin (CTL)

CTLs are present in 69% of species, with  a mean abundance of 6% and a maximum of 38%. They are more highly expressed in vipers than elapids (mean abundance 6.9% versus 0.8%) and are present in more species of vipers than elapids (89% versus 38%). They are haemotoxic, and can be anticoagulant, platelet aggregation inhibitors, or platelet aggregation agonists (Ogawa et al. 2005). There are two major groups; CTLs which are homodimeric (or higher multimers), and snaclecs (snake C-type lectins), which are heterodimeric (Clemetson et al. 2009).

Disintegrin (DIS)

This protein family is present in 47% of species with a mean abundance of 5% and a maximum of 20% for vipers, and 6% for elapids. Disintegrins are problematic to quantify, as in many species they are only present as a domain of SVMP and are quantified as part of that protein family, so they may be underrepresented. They are classified into five different groups based on their polypeptide length and number of disulphide bonds (Calvete et al. 2003). In addition to occurring as a domain of SVMP, they may also occur as monomers or dimers released in the venom by proteolytic processing of P-II and P-III SVMPs or translated from short-coding mRNA (Arruda Macêdo et al. 2015; Calvete et al. 2005). They selectively block integrin receptors and are potent inhibitors of integrin–ligand interactions (Calvete et al. 2005). Disintegrins are being investigated as novel therapeutic leads for cancer treatments because of their ability to target specific integrins (Akhtar et al. 2021). They could conceivably interfere in important processes involved in carcinogenesis, tumour growth, invasion and migration (Arruda Macêdo et al. 2015).

Natriuretic peptide (NP)

This protein family has been recorded in 36% of species. They are more commonly occurring and more highly expressed in vipers than elapids and are most widespread and highly expressed in pit vipers. The mean abundance values in pit vipers are 12% in the 51% of species that they occur, and in true vipers are 7% in the 46% of species that they occur. They are far less common in elapids with a mean abundance of only 1% in the 21% of species that they occur. The maximum abundance of NPs is 39% (Bothriechis supraciliaris) and it is a prominent protein family in the venoms of Bushmasters (Lachesis), and palm vipers (Bothriechis), but is absent in the genus Agkistrodon. They have a molecular mass of approximately 4 kDa (St Pierre et al. 2006). NPs also contain both bradykinin-potentiating peptides (BPP), and bradykinin-inhibiting peptides (BIP) (Fry 2015). In some studies NPs are recorded as a different protein family, “snake venom metalloprotease inhibitors” (svMPi), but the svMPis reported in this review either lacked sufficient sequence information to confirm they belonged to a different protein family or they matched directly to a NP (Giribaldi et al. 2020; Neri-Castro et al. 2020). Tripeptide svMPis have previously been shown to be encoded by precursor NPs (Boldrini-França et al. 2017), so we have merged svMPi and NP in Fig. 3. The drug captopril which is widely used for the treatment of hypertension and heart disease was developed from a BPP toxin isolated from the venom of the South American pit viper Bothrops jararaca (Bryan 2009; Vonk et al. 2011). It inhibits angiotensin-converting enzyme (ACE), and numerous drugs in the same family have been subsequently developed.

Kunitz peptide (KUN)

This protein family is present in 36% of species but is unequally distributed across snake families as it has been recorded in 65% of elapid and true viper species, but only in 4% of pit vipers. KUN are also more highly expressed in elapids than vipers, with a mean abundance of 6.4% in elapids versus 3.3% in true vipers, and only 0.06% in pit vipers. In elapids, they are highly expressed in African mambas (Dendroaspis), and desert cobras (Walterinnesia), and also occurs in Asian kraits (Bungarus), and some Australian elapids. In black mamba Dendroaspis polylepis, a KUN makes up 50% of the venom (dendrotoxin), a neurotoxin which blocks neuronal Kv1 ion channels (Harvey and Robertson 2005; Owen et al. 1997). Dendrotoxin appears to preferentially block inactivating forms of K + current, and can induce repetitive firing in neurons and facilitate neurotransmitter release (Harvey 1997).

Vascular endothelial growth factor (VEGF)

VEGFs are most common in true vipers, being present in 77% of species (mean abundance 3%), then pit vipers, 42% of species (mean abundance 1.5%), and only 13% of elapids (mean abundance 0.06%) (Table 3). They have been recorded as being 25 kDa homodimeric heparin-binding proteins (Ferreira et al. 2021), but one toxin isolated from a pit viper was heterodimeric with a molecular mass of 37 kDa (monomeric masses of 13 and 26 kDa) (Nakamura et al. 2014).

Fig. 4
figure 4

Family and subfamily comparisons of the abundance of the 22 most ubiquitous protein families in snake venoms. Y-axis is % of whole venom and X-axis is the different protein families. Coloured boxes show median and interquartile range Please move Figure 4 to the section on Variation in snake families (colour figure online)

Full size image
Table 3 The 14 minor protein families as based on a combination of mean abundance as a percentage of the total venom protein, and the number of species in which it is present (occurrence)
Full size table

Nucleotidase (NUC)

This protein family contains 5’ nucleotidase, ATPase and ADPase (Boldrini-França et al. 2017). They are one of the two most widespread of the minor protein families along with NGF, occurring in 39% of species, and are more commonly reported in vipers than elapids. They have a mean abundance of 0.7% and a maximum of 4.6%. There are few studies concerning the isolation and characterisation of snake venom NUCs, but they have been shown to inhibit platelet aggregation in both true and pit vipers (Ouyang and Huang 1983; Trummal et al. 2005).

Phosphodiesterase (PDE)

This protein family is present in just over one third of species and had a mean abundance of 0.7% and a maximum of 3.6%. They are mono or multimeric with a molecular mass of 90 to 160 kDa (Oliveira et al. 2022b). They may facilitate venom diffusion through the prey’s tissues (Oliveira et al. 2022b) and PDEs isolated from rattlesnake venoms (Crotalus spp.) were found to produce an immediate profound hypotensive crisis but did not interfere with neurotransmission (Russell et al. 1963).

Nerve growth factor (NGF)

This protein family is one of the two most widespread of the minor protein families along with NUC, occurring in 39% of species. They have a mean abundance of 0.6% and a maximum of 3.6% and belong to the class of structurally related proteins known as neurotrophins (Sunagar et al. 2013). They have a molecular mass of 12 to 37 kDa (Islam et al. 2021).

Phospholipase B (PLB)

This protein family was present in 30% of species and had a mean abundance of 0.5% and a maximum of 3%. They have a molecular mass of 35 to 55 kDa (Bernheimer et al. 1987; Ullah and Masood 2020). There have been limited investigations into this protein family, and their function and role in human envenomation is unclear.

Cobra venom factor (CVF)

This protein family has only been recorded from cobra venoms with two exceptions, the sea snake Hydrophis cyanocinctus (0.2%), and the pit viper Trimeresurus insularis (1%) (Jones et al. 2019; Wang et al. 2020). The largest amounts recorded were in king cobra Ophiophagus hannah (2.8 to 5.5%) (Tan et al. 2015; Vonk et al. 2013). It has a molecular mass of 149 kDa and is composed of three chains (Liu et al. 2018; Vogel and Fritzinger 2010). It continuously activates the complement system resulting in a depletion of complement activity (Van den Berg et al. 1991). As such it has attracted much experimental research to study the role of the complement system in host defence and immune response, as well as the pathogenesis of diseases (Vogel and Fritzinger 2010).

Vespryn/ohanin (VES/OHA)

Vespryns were most common in elapid venoms, being recorded in 25% of species and with a mean abundance of 1.6%. They were recorded in 7% of pit vipers with a mean abundance of 0.6%, but there were no records from true vipers. It is most abundantly expressed in the venom of the Malayan Blue Coral Snake Calliophis bivirgata flaviceps (15%) (Tan et al. 2016).

Acetylcholinesterase (AChE)

AChEs is present in almost a quarter of elapid species, but no vipers except the true viper Echis (Ghezellou et al. 2021). They are particularly prevalent in krait (Bungarus spp.) venoms, and almost completely absent from South American Coral Snakes (Micrurus). They are monomeric with a molecular mass of 59 kDa (Tan and Tan 1988). Although the function of these enzymes in hydrolysing acetylcholine is well described, it is unclear what their role is in snake venom or in human envenoming. There are several nomenclatural taxonomies used for AChE, but we are here treating it as belonging to the cholinesterase subfamily of type-B carboxylesterases (Wheelock et al. 2005).

Hyaluronidase (HYAL)

Hyaluronidases are recorded in 16% of species and have a mean abundance of 0.3% and a maximum abundance of 1.9%. They are less common in the venoms of pit viper venoms. They have a molecular mass of 28 to 70 kDa (Boldrini-França et al. 2017). They are commonly regarded as a “spreading factor”, and a non-toxic enzyme that degrades hyaluronic acid, the main component of extracellular matrix, so potentially allows the spread of other toxins (Bala et al. 2018). It has been shown to increase the diffusion of crotoxin and PLA2 through mouse tissue in Crotalus durissus, thus potentiating crotoxin (Bordon et al. 2012).

Cystatin (CYS)

Cystatins have been recorded in 15% of true vipers and 12% of elapids, but there are no records in pit vipers. A study has shown that they inhibit tumour angiogenesis and that they may have therapeutic potential as antiangiogenic and anti-metastatic agents (Xie et al. 2013). Their role in envenomation is currently unknown (Boldrini-França et al. 2017).

Glutaminyl cyclase

Glutaminyl cyclases have only been recorded in viper venoms and are one of the least understood protein families in snake venoms (Wang et al. 2014). They may not possess toxic activity, but instead may play a role in N-terminal modifications of other toxins (Pawlak and Manjunatha Kini 2006). These enzymes belong to the family of transferases, sometimes referred to as aminotransferases or acyltransferases (Huang et al. 2005).

Aminopeptidase

Aminopeptidases were one of the least commonly reported of the minor protein families, reported in 27% of true vipers with a mean abundance of 0.3%, in 6% of pit vipers with a mean abundance of 0.4%, and in 1.5% of elapids with a mean abundance of 0.004%.

Crotamine

This toxin belongs to the protein family β-defensins but is typically referred to as crotamine in venom proteome studies (Fry 2015). It has only been recorded from the venom of some species of rattlesnakes (Crotalus), and is a 42 residue peptide with a molecular mass of 4.9 kDa (Mancin et al. 1998). Crotamine makes up 37% of the venom of Crotalus viridis (Saviola et al. 2015). It is a myoneurotoxin with multiple activities and has been reported to be a cell-penetrating peptide, antitumoral agent and a potent Kv channel inhibitor (Kerkis et al. 2014; Peigneur et al. 2012).

Waglerin

This protein family has only been recorded from the Oriental pit viper genus Tropidolaemus (Tan et al. 2017). It is 22 residues in length with a proline-rich sequence and a molecular mass of 2.5 kDa (Lin et al. 1995; Weinstein et al. 1991). It is a neurotoxin that competitively antagonises muscle nicotinic acetylcholine receptors (nAChRs) (Molles et al. 2002). They are the only characterised toxin from viper venoms that target nicotinic receptors (Molles et al. 2002). It makes up 38% of the venom.

Rare protein families

This category contains 18 protein families (Table 4). The classification criteria for inclusion in this group were: present in < 5% of species, and with a mean abundance of < 5% (Fig. 3). The number of protein families in this category has been reduced from the 36 listed in the 2017 review, as many of the protein families included therein are now known to be either impurities, non-toxins, synonyms, or protein subfamilies (Supplementary file 1. Tables 3, 4).

Table 4 Rare protein families
Full size table

Venom variation

Types and levels of venom variation

Snake venom can vary in composition at all taxonomic levels; families, clades, genera, species (inter-specific variation) and within species (intra-specific or inter-population variation (Saviola et al. 2017)). The expression of different protein families can also vary within populations (intra-population variation) (Tasoulis et al. 2020b). Venom composition can also vary in a number of ways, including different relative amounts of each protein family, different numbers of toxins within each protein family and different types of toxins with each protein family (amino acid sequence differences).

Variation in snake families

Elapid venoms are overwhelmingly dominated by just two protein families, 3FT and PLA2 (Fig. 4A), making up 82% of the venom when averaged across all elapid species. The remainder of their venoms are mostly comprised of small amounts of SVMP, and three of the six secondary protein families, LAAO, CRiSP and KUN. Viper venoms are mainly comprised of three of the dominant protein families, SVMP, PLA2 and SVSP (Figs. 4B, C), which make up 69% of their venoms when averaged across all viper species. Viper venoms contain more of the secondary protein families than elapids (combined average of 23% versus 8%). The differences between the venoms of true and pit vipers are minor, with true vipers containing less SVMP (median 22% versus 35%), and much more of the secondary protein family KUN and the minor protein family VEGF.

Intra-specific venom variation

There has been little uniformity in the results of studies investigating intra-specific venom variation in snakes. Inter-population expression variation has been found in eastern diamondback rattlesnake Crotalus adamanteus (Margres et al. 2015), Malayan pit viper Calloselasma rhodostoma (Daltry et al. 1997), jararaca Bothrops jararaca (Goncalves-Machado et al. 2016), and Lancehead Bothrops asper (Mora-Obando et al. 2020). Conversely, it was not found in the elapid, eastern coral snake Micrurus fulvius (Margres et al. 2015), and only minimally in the coastal taipan Oxyuranus scutellatus (Tasoulis et al. 2020b). A study on the sidewinder rattlesnake Crotalus cerastes (Rautsaw et al. 2019), found little evidence for toxin coding sequence differentiation across populations, suggesting that their venoms are not under strong positive selection pressures. A study on strongly dichotomous venom expression in populations of the Mojave rattlesnake Crotalus scutulatus found that the venom variation could not be explained by diet (Zancolli et al. 2019). Likewise, the combined results of three studies on the trophic generalist sea snake Hydrophis curtus from three localities spanning a distance of almost 5000 km, suggest a general cline in venom variation rather than positive selection shaping venom composition for local differences in diet (Neale et al. 2017; Tan et al. 2019; Wang et al. 2020). The venom of this species varied from PLA2 dominant at the southern end of its distribution (QLD, northern Australia), to almost equal parts PLA2 and 3FTx in the centre of its distribution (Penang Malaysia), to 3FTx dominant in the northern part of its range (Hainan, China). Interestingly, studies on insular populations of the Australian tiger snakes have shown that there has been strong selective pressure on snake body size imposed by the size of available prey items, so venom may not always be the trait under strongest selection pressure (Aubret 2015; Keogh et al. 2005). This raises the question about whether selection can simultaneously optimise venom and non-venom traits.

Although numerous studies have found an association between venom variation and diet, others have not found such an association. One possibility is that geographical differences in diet generally drive venom variation, but there are circumstances in which non-venom traits are under stronger selection pressure than venom. Perhaps circumstances like these mean that venom expression may alter in ways not associated with diet. Such geographic differences in venom expression could then easily be misinterpreted as being the result of positive selection for diet. Another possibility is that the dominant protein families are sufficiently efficacious at prey subjugation that selection cannot detect differences in venom composition below a certain toxicity range. This would allow for initial random venom variation between fragmented or newly founded populations to be retained in some circumstances. Such initial random variation could then potentially reset the venom composition of these populations onto radically new evolutionary trajectories. Like all traits, venom evolution would therefore be the result of the interplay between pre-existing composition, chance and ecological drivers.

Medical aspects of venom composition and variation

Until recently, a large proportion of research on snake venom composition has focussed on medically important toxins, either as potential drug leads or for their functional role in human envenomation. What is interesting is that the four dominant protein families appear to be the most important toxin families in human envenomation, being responsible for VICC, pre- and post-synaptic neurotoxicity, myotoxicity and cytotoxicity. Potentially this is because these protein families likely possess broad spectrum toxicity and lethality across all classes of vertebrate prey, and therefore have similar toxicity in humans (Barua and Mikheyev 2019). This would explain why toxins in sea snake venom, that may have been selected to kill fish, are also highly toxic in humans. Clearly there are differences in the effects of different toxins in humans compared to animals, such as the relative resistance of humans to short-chain neurotoxins (Silva et al. 2018), or differences in the effects of snake procoagulants on different animal toxins. (Maduwage et al. 2016) Possibly, the actual lethality of the toxins in these four dominant families across vertebrates is more important than the specific toxic effect.

There have been numerous reviews on the potential value of toxins for the development of novel drugs (Oliveira et al. 2022a; Saviola et al. 2014), following the success of captopril being developed from the discovery of the bradykinin-potentiating peptides (C-type natriuretic peptides) in B. jararaca venom (Ferreira 1965). However, despite a multitude of studies attempting to isolate novel toxins over the last few decades, there have been very few drugs developed from toxins that are now used in clinical practice. An example is ancrod, the purified thrombin-like enzyme from Calloselasma rhodostoma (Malayan Pit Viper), which was trialled in the treatment of acute ischaemic stroke (Hennerici et al. 2006). However, it was shown to not improve outcomes and increased bleeding in patients (Levy et al. 2009).

The role of snake venom variation in human envenomation is somewhat contentious. There are clearly differences in the clinical effects of snake venoms, based on the occurrence and predominance of the different toxin protein families, as discussed above. The effects of toxins from the four dominant toxin families are well described in humans, and different relative abundances of these toxin families in different snakes correlate well with the range of different clinical effects. Vipers have a predominance of SVSPs and SVMPs consistent with haemotoxic and local effects being more common in human envenomation (Warrell 2010). Conversely, neurotoxicity is far more common in elapids, such as kraits, cobras and coral snakes (Silva et al. 2017), in which PLA2s and 3FTx are the most abundant toxin families. The difference in venom composition for different genera also appears to be associated with different clinical effects, for example in the medically important Australasian elapids (Johnston et al. 2017b). It is less clear whether variation in the individual toxins within toxin families is important in human envenomation.

An understanding of venom composition is also central to developing efficacious antivenoms. Antivenoms are mixtures of antibodies raised against multiple toxins in a snake venom in large mammals (e.g. sheep, horse) (Silva and Isbister 2020). For an antivenom to be efficacious, it needs to contain antibodies to all medically important toxins in a snake venom. In the past, this has been difficult, without a complete understanding of venom composition. More recently, proteomic techniques have been used to investigate the toxicological profile of a snake venom and hence identify which protein families are most important for antivenom immunorecognition (Gutiérrez et al. 2009; Laustsen 2018).

Currently antivenoms are only effective against snakes in certain geographical regions, because only venoms from the specific geographical region are used in producing the antivenom (Silva and Isbister 2020). Worldwide this has meant that numerous antivenoms are required, and an antivenom produced in one country is highly unlikely to be useful in another country, unless a close neighbour with similar snake fauna. This has contributed to an ongoing crisis in antivenom supply to resource poor countries, many which cannot produce their own antivenoms (Lalloo et al. 2002).

A different approach to antivenom development is to consider the similarity in venom composition between snakes worldwide, rather than the differences. It appears that most medically important effects in humans are due to the four dominant toxin families, with a few important exceptions, such as natriuretic peptides in vipers (Péterfi et al. 2019; Souza et al. 2007), crotamine in South American pit vipers and dendrotoxins in mambas (Erulu et al. 2018; Harvey 2001; Quarch et al. 2017). A universal antivenom could be developed that contained antibodies against the four dominant toxin families and other key toxin groups. Such an antivenom would then be efficacious across a much larger geographical range, potentially covering continents or multiple continents.

There is considerable support for a universal antivenom in the many studies of antivenom cross-neutralisation (Huynh et al. 2022; Isbister et al. 2014, 2010a; Kornhauser et al. 2013; Silva et al. 2016a). For example, commercial antivenoms developed in Asia cross-neutralise the neurotoxic effects of Australasian elapids, and conversely those developed in Australia cross-neutralise the effects of Asian elapids (Silva et al. 2016a). In the case of pit vipers, an antivenom developed against American pit vipers appeared to be efficacious for an Asian pit viper envenomation (Isbister et al. 2014). To develop a universal antivenom requires designing toxin-specific antivenoms using bioinformatics and multi-epitope DNA immunisation. One study showed that antiserum raised against a single synthetic multi-epitope DNA immunogen (epitope string) from SVMPs in E. ocellatus, cross neutralised haemorrhage induced by both E. ocellatus and Cerastes cerastes venoms (Wagstaff et al. 2006).

Conclusion

There has been a rapid expansion in our knowledge of snake venom composition with modern proteomics. Studies of hundreds of snake species have shown that the venoms of vipers and elapids are made up of 42 different protein families, but 24 of these make up 96% of the venom on average. There are 4 dominant, 6 secondary, 14 minor and 18 rare families. From the four dominant protein families, 3FTx and PLA2 make up 82% of elapid venoms on average, and SVMP, PLA2 and SVSP make up 69% of viper venoms on average. Venom variation occurs at all taxonomic levels including within populations, and venom variation cannot always be explained as being the result of positive selection for local differences in diet. With only a few exceptions, human envenomation syndromes appear to result from the toxic effects of the four dominant protein families, suggesting broad ranging toxic effects across all vertebrates. The use of proteomic techniques will also help to improve antivenom therapy worldwide.