Widespread convergent evolution of alpha-neurotoxin resistance in African mammals

Abstract

Convergent evolution is central to the study of adaptation and has been used to understand both the limits of evolution and the diverse patterns and processes which result in adaptive change. Resistance to snake venom alpha-neurotoxins (αNTXs) is a case of widespread convergence having evolved several times in snakes, lizards and mammals. Despite extreme toxicity of αNTXs, substitutions in its target, the nicotinic acetylcholine receptor (nAChR), prevent αNTX binding and render species resistant. Recently, the published meerkat (Herpestidae) genome revealed that meerkats have the same substitutions in nAChR as the venom-resistant Egyptian mongoose (Herpestidae), suggesting that venom-resistant nAChRs may be ancestral to Herpestids. Like the mongoose, many other species of feliform carnivores prey on venomous snakes, though their venom resistance has never been explored. To evaluate the prevalence and ancestry of αNTX resistance in mammals, we generate a dataset of mammalian nAChR using museum specimens and public datasets. We find five instances of convergent evolution within feliform carnivores, and an additional eight instances across all mammals sampled. Tests of selection show that these substitutions are evolving under positive selection. Repeated convergence suggests that this adaptation played an important role in the evolution of mammalian physiology and potentially venom evolution.

1. Introduction

Convergent evolution has offered insight into the ways in which diverse organisms evolved to cope with similar selective pressures [1,2]. Within mammals, venom resistance has convergently evolved in marsupials, rodents, carnivores, eulypotyphlans and artiodactyls [3–9]. Old world mammals which prey upon and sustain bites from venomous snakes in the family Elapidae face envenomation with deadly alpha-neurotoxins (αNTXs, specifically three-finger toxins), which bind to the α1 domain of the muscular nicotinic acetylcholine receptor (nAChR), blocking nerve–muscle communication and causing rapid muscular paralysis and death. Despite this, the Egyptian mongoose, honey badger, domestic pig and hedgehogs regularly prey on Elapids and have convergently evolved mutations in their nAChRs which confer resistance to αNTXs and experience strong positive selection [3,10].

Mutations associated with αNTX resistance in mammals occur at four sites in the nAChR epitope, and function via three distinct biophysical mechanisms [9]. The first two sites 187 and 189, are aromatic amino acids (W¹⁸⁷, F¹⁸⁹) in most mammals, and directly interact with αNTX. Resistance at these sites is present either via steric hindrance mediated by a glycosylation (N¹⁸⁷, T¹⁸⁹) in mongooses or via arginine-mediated electrostatic repulsion (R¹⁸⁷) in honey badgers, hedgehogs and pigs. The second two sites involved in mammalian resistance, 194 and 197 are prolines in most mammals, are necessary for αNTX binding, and replacement of either (e.g. P¹⁹⁴ to L¹⁹⁴, P¹⁹⁷ to H¹⁹⁷ in the Egyptian mongoose) results in the loss of αNTX binding [9]. Extensive experimental work has validated the function of these substitutions in diverse genetic backgrounds [10–20]. Hereafter, we refer to these mechanisms as electrostatic repulsion resistance, glycosylation resistance and proline resistance, respectively.

Other species of mongooses and meerkats in the same family (Herpestidae) are well-known to exhibit lack-of-avoidance, cooperative mobbing and predation upon venomous snakes (figure 1, electronic supplementary material, table S3) [7,9,10,22]. The newly published meerkat genome (GCA_006229205.1) reveals that meerkats share the same changes as the Egyptian mongoose (i.e. N¹⁸⁷, T¹⁸⁹ and L¹⁹⁴) and likely enjoy the same protective glycosylation and partial proline-mediated resistance. αNTX resistance in related species of African feliform carnivores remains unexplored, including in the families Eupleridae (Malagassy carnivores) and Viverridae (civet cats), the latter of which contains many species that exhibit predatory and/or aggressive behaviour towards venomous snakes (figure 1; electronic supplementary material, table S3) [23]. While their ecology, biogeography and relatedness to resistant taxa suggest that snake venom may be an important selective pressure for many species in Herpestidae, Eupleridae and Viverridae, it is unknown if they are resistant to αNTX, and whether resistance is the result of an ancient adaptation at the base of these three clades or has evolved convergently in response to repeated selection pressure for venom resistance.

Figure 1.

The evolutionary tree of mammals showing the relationships between species for which the nAChR α1 subunit epitope was available. The nAChR epitope alignment from sites 175–198 is displayed for each species with dots denoting amino acids that do not differ from the consensus. Designated foreground clades (Herpestidae, Viverridae and Eupleridae) are marked with black rectangles. Sites highlighted in grey rectangles indicate signatures of positive selection in these clades (table 2). Mutations known to confer resistance are reconstructed onto their corresponding branches in colour based on the mechanism of resistance: electrostatic repulsion via a positively charged replacement (R or K) at site 187 is marked in red, steric hindrance via a glycosylation indicated by an N-T replacement at sites 187/189 is marked in green and replacement of prolines 194 and/or 197 (proline resistance) is marked in blue. Changes at functional and selected sites are mapped onto branches of the tree using a maximum-likelihood-based codon model ancestral state reconstruction [21]. Species known to prey on venomous snakes are denoted with a snake icon (citations and summary of interactions in electronic supplementary material, table S3).

To examine the evolution of αNTX resistance across these clades, we leveraged an exhaustive search of museum tissue samples from all species available from North American museums for Eupleridae, Viverridae and Herpestidae and sequenced the muscular nAChR. We also used bioinformatic searches to identify muscular nAChR sequences from additional species. Using a comparative phylogenetic approach, we leveraged maximum-likelihood tests of selection, as well as ancestral sequence reconstruction, to examine whether the species exhibited resistant nAChR, what mechanisms were present, and whether resistance mutations arose ancestrally or convergently across clades.

2. Methods

(a) . Tissues and DNA extraction

Thirty-nine tissue samples across 33 different species from Eupleridae, Herpestidae and Viverridae were obtained through specimen loans (electronic supplementary material, table S1). Genomic DNA was extracted as previously described [3] with a skin wash step for recalcitrant samples [24]. Previously designed primers were used to amplify 850 bp of the alpha subunit of the muscular nAChR gene (chrna1) that included the ligand-binding (and αNTX) epitope corresponding to residues 122–205 of the protein sequence [3] (see electronic supplementary material, Methods). Amplified PCR products were treated with ExoSAP-IT and Sanger sequenced by the University of Minnesota Genomics Center on an ABI 3730XL DNA Analyzer using BigDye Terminator v3.1 chemistry (Applied Biosystems, USA). Resulting DNA sequences were assembled, edited and aligned using Geneious v8.1 [3,25].

(b) . Bioinformatic retrieval of sequence

All mammalian sequences included in the NCBI orthologue database (accessed 24 May 2022) for the gene containing the muscular nAChR sequence, chrna1, were included in this dataset. Additional blastP searches were conducted using Viverrid, Herpestid and Euplerid sequences generated from this project. Results were filtered for duplicate species and a final dataset of 199 sequences was aligned using MUSCLE [25] and edited in Geneious v8.1. A 90 bp fragment (175–198aa) covering the acetylcholine/αNTX-binding epitope was used for analyses, as this was the longest fragment with complete data for most species (electronic supplementary material, table S3). Timetree was used to generate a phylogenetic tree, and missing species were subsequently added in Mesquite using existing topologies [26–29].

(c) . Tests of selection

We used a suite of likelihood-based codon models which use the ratio of non-silent substitutions to silent substitutions (dN/dS) to detect positive selection [21,30–35]. Because we were interested in selection specifically on branches of the tree leading to Viverridae, and Herpestidae + Eupleridae (figure 1), we used a branch-site test for positive selection identifying these lineages and all subtending branches as the foreground. Similarly, site tests were used to identify sites which exhibit signatures of selection [21,35]. Lastly, to discern signals of convergent versus pervasive selection, we employed a ‘drop-out’ site test in which we removed all species in Viverridae, Herpestidae and Eupleridae as well as honey badger (Mellivora capensis), pig (Sus scrofa) and hedgehog (Erinaceous europea) from the topology in Mesquite and re-run site tests [27,36]. Ancestrally reconstructed sequences were generated from the best-fit site model (M8) in PAML v4.6 (table 1) and aligned using MUSCLE v3.5 [21,26]. For complete details, see electronic supplementary material, methods.

Table 1.

Results of branch-site tests for positive selection on the CHRNA1 gene.

model and log-likelihood	site classa	proportion of sites	w background	w foregroundb
w2 = 1	0	0.746	0.0139	0.0139
lnL = −1395.62	1	0.152	1.0	1.0
	2a	0.085	0.0139	1.0
	2b	0.017	1.0	1.0
w2 > 1	site class	proportion of sites	w background	w foregroundb
lnL = −1384.06	0	0.705	0.013	0.013
	1	0.150	1.0	1.0
	2a	0.121	0.013	10.20
	2b	0.026	1.0	10.20

^aSite class 0 and 1 apply to foreground and background lineages and include sites under purifying selection (0 < w < 1) and neutral sites (w = 1), respectively. Site class 2 allows a proportion of positively selected sites (w > 1) in the foreground lineages, where 2a includes sites under purifying selection (0 < w < 1) in the background lineages and 2b includes neutral sites (w = 1) in the background lineages. 2Δln = 20, d.f. = 1, p < 0.0001. When Mellivora capensis, Sus scrofa and Erinaceous euopus are included in foreground branches 2Δln = 30, d.f. = 1, p < 0.0001.

^bHerpestidae, Eupleridae and Viverridae were included in the foreground class.

3. Results and discussion

(a) . Evolutionary history of resistance in Herpestidae, Eupleridae and Viverridae

To assess αNTX resistance, we used aligned sequence data for the αNTX-binding epitope of nAChR (figure 1) and interpreted sequence data using extensive prior literature that has experimentally demonstrated the effects of specific amino acid substitutions on αNTX binding. We found that all newly sequenced nAChRs from species belonging to the family Herpestidae have N¹⁸⁷, T¹⁸⁹ and L¹⁹⁴, H¹⁹⁷ which confer strong glycosylation and proline resistance, respectively (figure 1) [10–20]. All Malagasy carnivores (Eupleridae) as well as two civets (Viverridae) have substitution R¹⁸⁷ and I/L¹⁸⁹ which confer electrostatic repulsion resistance (figure 1) [3,13,18,19]. Within civets (Viverridae), we found novel mutations (K¹⁸⁷ and E¹⁸⁹) which arose at the base of the genus Genetta (small African carnivores including the genet). Lysine (K) is one of only three positively charged amino acids and may mimic electrostatic repulsion resistance seen with R¹⁸⁷; however, biophysical testing is needed to definitively assess the impact of K¹⁸⁷ on αNTX binding and how this translates to whole-organism resistance.

To examine whether αNTX evolved convergently, we employed a codon model (CODEML) maximum-likelihood ancestral reconstruction of sequences implemented in PAML v4.6 [6,21,37]. Our ancestral reconstruction supports that electrostatic repulsion resistance via R¹⁸⁷ arose once at the base of Eupleridae + Herpestidae and twice within Viverridae. Subsequent glycosylation resistance (N¹⁸⁷, T¹⁸⁹) as well as proline resistance (L¹⁹⁴, H¹⁹⁷) appears to have arisen later in Herpestidae (figure 1). Empirical work has shown that electrostatic repulsion resistance is less effective than either glycosylation or proline resistance [9,12,20]. Our data suggest that electrostatic repulsion resistance arose first at the base of Eupleridae + Herpestidae, and subsequent selection for increased resistance in Herpestids likely led to glycosylation and proline resistance. Interestingly, the Malagasy carnivores (Euplerids) are not sympatric with any αNTX producing snake besides sea snakes (for which we do not have any evidence of predation). As the Euplerids are restricted to Madagascar, and the R¹⁸⁷ appears to have preceded the split between Euplerids and Herpestids (figure 1), the presence of this mutation in these species is most likely a remnant of selection pressure imposed prior to the isolation of this group on Madagascar, though further work should explore if these ancient substitutions have facilitated (or are maintained by) a contemporary ecological interaction such as exploitation of poisonous or venomous prey (e.g. frogs, scorpions and centipedes).

Across the clades Herpestidae, Eupleridae and Viverridae, mutations known to confer resistance have arisen at least four times (five, if we include K¹⁸⁷ in civets). These results strongly suggest that species among all three clades have substantial αNTX-resistant nAChRs and warrant further investigation into their ecological interactions with venomous snakes, as well as their physiological ability to cope with venom.

(b) . Positive selection identified for substitutions

To test for adaptive evolution on nAChR in the clades we suspected of being venom resistant, we used codon model-based (CODEML) branch-site tests of positive selection in PAML v4.6 and explored additional methods and results in the electronic supplementary material [21]. Convergent foreground branches were specified as Eupleridae + Herpestidae, the base of Viverridae, and singular branches leading to honey badger, hedgehog and pig [3]. Because the latter three species and the mongoose have previously been shown to be under positive selection and may inflate the overall signal of selection, we performed branch-site tests with and without these species (figure 1). In both cases, we recovered a strong signal of selection in foreground lineages (2ΔL = 30, d.f. = 1, p < 0.0001; without honey badger, hedgehog and pig 2ΔL = 20, d.f. = 1, p < 0.0001; table 1). ‘Drop-out’ site tests showed no selection for a tree pruned of all lineages suspected of being under selection, (2ΔL = 2.4, d.f. = 1, p = 0.121), indicating that the signal of positive selection can be attributed to the foreground lineages [36].

Bayes empirical Bayes tests identified four sites to be evolving under positive selection in foreground lineages (table 2). Of these, three (187, 189 and 197) modulate αNTX binding in empirical studies [12–16,19,20]. Site 182 was identified as a site under positive selection, though no functional studies exist for this site (figure 1 denoted in grey).

Table 2.

Sites identified to be evolving under positive selection via different maximum-likelihood analyses. Italics indicate sites for which the probability that omega is greater than 1 is 95% or more.

site	foreground specified PAML	foreground specified MEME	foreground unspecified PAML	foreground unspecified MEME	mutations	function
182	0.96	—	—	—	W, Q	unknown
187a	1.00	—	0.965	—	Rb, Nc, K	Shb/Erc
189a	1.00	0.9911	0.776	—	L, I, E, Tc	Shc
195	—	—	0.990	0.955	L
197a	0.95	—	—	—	H	Prd

^aSite directly associated with αNTX binding [18].

^bReduced binding via steric hindrance [13,16,18,20].

^cReduced binding via electrostatic repulsion [18,19].

^dReduced binding via proline binding disruption [11,18].

(c) . Mutations recovered in other mammals

Within our 199-mammal dataset, we found five new instances of independently evolved substitutions known to confer electrostatic repulsion resistance (R¹⁸⁷) outside the three clades initially examined. In two of these instances, the Eurasian otter (Lutra lutra) and the oryx antelope (Oryx dammah), R¹⁸⁷ is not secondarily accompanied by the substitution I¹⁸⁹ or L¹⁸⁹ (figure 1). Empirical work has shown that nAChR loses affinity for αNTX with the R¹⁸⁷ mutation alone, and that addition of the I¹⁸⁹, or I¹⁸⁹ by itself does not confer resistance despite its propensity to be paired with R¹⁸⁷[19]. However, the presence of the single mutant R¹⁸⁷ in these data suggests that the accompanying mutation I¹⁸⁹ is likely not a necessary epistatic mutation and raises the possibility that it may have some function in resistance that is apparent in vivo that is not recovered in vitro [19]. Both the Eurasian otter and the oryx antelope are sympatric with αNTX producing snakes, though direct ecological links were not found in the literature for either species. While snake venom is by far the best-known source of nAChR agonists, we should resist the temptation to consider it the sole source of selection pressure for this receptor. Notably, the Eurasian otter is known to eat venomous catfish (Ameiurus nebulosus), and while catfish venom is not well studied, the acetylcholine receptor agonist atropine partially diminished the effects of bullhead venom in mice, suggesting the nAChR may be the targeted receptor for this venom [38–41]. Khan et al. [10] also found two species of freshwater fish that exhibit substitutions that confer αNTX resistance and suggest that this may be an adaptation to exposure to αNTXs secreted by freshwater cyanobacteria [15]. These hypotheses are both speculative and demand further examination on the source of nAChR-targeting toxins which may be driving selection for these substitutions.

Further investigations of species closely related to Oryx dammah revealed that all available species in the families Alcelaphinae and Hippotraginae also had an R¹⁸⁷ mutation (electronic supplementary material, figure S1). Notably, Kazandian et al. [42] suggest that venom spitting did not evolve as a response to trampling by ungulates, but instead was an adaptation that arose convergently as a response to predation by great apes [42]. These data may suggest an alternate source of selection pressure and a potentially coevolutionary relationship driving the evolution of spitting.

We found that the warthog, Phacochoerus africanus, shares the R¹⁸⁷ and L¹⁸⁹ mutation with its pig sister taxa (Sus scrofa), and that this mutation arose in an ancestor of these two species. Additional African species with charge-resistant mutations are the Cape golden mole (Chrysochloris asiatica, R¹⁸⁷), a subterranean blind animal which may be especially susceptible to predation by burrowing snakes and is known prey of several species of Elapids, and the thicket rat (Grammomys surdaster; K¹⁸⁷) [43–45].

The only non-African species with R¹⁸⁷ was the Australian brushtail possum (Trichosurus vulpecula). While this species is sympatric with αNTX producing elapids, mention of ecological interactions with venomous snakes in the literature is lacking. However, some species of elapids (Hoplocephalus stephensii) have been known to prey upon closely related pygmy opossums [46]. Mammalian resistance in Australian mammals is generally understudied and should be further explored. Additionally, several other species (wombats, bush babies and aardvarks) showed convergent mutations (K¹⁸⁷, Q¹⁸⁷ and C¹⁸⁷) of unknown function, and further biochemical assessment is needed.

(d) . Conclusions

The evolution of resistance to αNTX in mammals has previously been categorized as a relatively rare adaptation only known in six mammals [3]. This work revealed 27 new species with amino acid changes known to cause resistance (experimentally validated across diverse taxa), and an additional 17 species with substitutions that are suspected to result in αNTX resistance [10–20]. In total, our work shows that convergent αNTX resistance has evolved at least 11 times within mammals. Further investigation of fauna that interact with αNTX producing snakes is needed to determine whether αNTX resistance translates to whole venom resistance, and whether these substitutions are the result of present or past ecological interactions with venomous species. Our analyses, along with other recent work, suggest that snake and other animal venoms are a source of strong selection pressure likely facilitated via complex coevolutionary interactions that may be the rule rather than the exception, particularly for animals which share habitat with many venomous snakes [5,19,47].

Data accessibility

All data (trees, alignments, control files, ancestral sequences, and gene trees) used in this manuscript can be found on the Dryad Digital Repository: https://doi.org/10.5061/dryad.jh9w0vtf8 [48]; the accompanying READme.txt file provides a full description of all files included.

The data are provided in the electronic supplementary material [49].

Authors' contributions

D.H.D.: conceptualization, data curation, formal analysis, funding acquisition, investigation, methodology, project administration, resources, supervision, validation, visualization, writing—original draft and writing—review and editing; J.H.: data curation, formal analysis, funding acquisition, investigation, resources and writing—original draft; S.E.M.: funding acquisition, project administration, resources, supervision and writing—review and editing.

All authors gave final approval for publication and agreed to be held accountable for the work performed therein.

Conflict of interest declaration

We declare we have no competing interests.

Funding

This work was generously funded by the NIH TREM fellowship (grant no. K12GM119955), the UMN UROP and the Minnesota Herpetological Society.