Widespread convergence in toxin resistance by predictable molecular evolution

Significance

Convergence has strong bearing on the fundamental debate about whether evolution is stochastic and unpredictable or subject to constraints. Here we show that, in certain circumstances, evolution can be highly predictable. We demonstrate that several lineages of insects, amphibians, reptiles, and mammals have utilized the same molecular solution, via the process of convergence, to evolve resistance to toxic cardiac glycosides produced defensively by plants and bufonid toads. The repeatability of this process across the animal kingdom demonstrates that evolution can be constrained to proceed along highly predictable pathways at molecular and functional levels. Our study has important implications for conservation biology by providing a predictive framework for assessing the vulnerability of native fauna to the introduction of invasive toxic toads.

Abstract

The question about whether evolution is unpredictable and stochastic or intermittently constrained along predictable pathways is the subject of a fundamental debate in biology, in which understanding convergent evolution plays a central role. At the molecular level, documented examples of convergence are rare and limited to occurring within specific taxonomic groups. Here we provide evidence of constrained convergent molecular evolution across the metazoan tree of life. We show that resistance to toxic cardiac glycosides produced by plants and bufonid toads is mediated by similar molecular changes to the sodium-potassium-pump (Na⁺/K⁺-ATPase) in insects, amphibians, reptiles, and mammals. In toad-feeding reptiles, resistance is conferred by two point mutations that have evolved convergently on four occasions, whereas evidence of a molecular reversal back to the susceptible state in varanid lizards migrating to toad-free areas suggests that toxin resistance is maladaptive in the absence of selection. Importantly, resistance in all taxa is mediated by replacements of 2 of the 12 amino acids comprising the Na⁺/K⁺-ATPase H1–H2 extracellular domain that constitutes a core part of the cardiac glycoside binding site. We provide mechanistic insight into the basis of resistance by showing that these alterations perturb the interaction between the cardiac glycoside bufalin and the Na⁺/K⁺-ATPase. Thus, similar selection pressures have resulted in convergent evolution of the same molecular solution across the breadth of the animal kingdom, demonstrating how a scarcity of possible solutions to a selective challenge can lead to highly predictable evolutionary responses.

Convergent evolution is the process by which phenotypic similarities evolve independently among disparate species (1–3). Although convergence is sometimes distinguished from parallel evolution, both are part of a continuum, which often makes attempting to distinguish between them problematic and potentially misleading (4); here, we simply use the term “convergence” throughout. Convergence has strong bearing on a fundamental and heated debate on the predictability of evolution. Epitomized by the writings of Gould (5) and Conway Morris (6), this debate centers on whether evolution is stochastic and unpredictable (5) or subject to constraints that limit the available options for evolution, resulting in frequent convergence and a degree of predictability (6). However, evidence of convergence at the genetic level, where similar molecular changes confer the same change in phenotype, is currently limited to only a few taxonomically restricted examples (7–11).

Cardiac glycosides offer an ideal model system to investigate the extent to which evolution can be constrained to predictable changes throughout the animal kingdom. These organic compounds are highly toxic molecules that inhibit the sodium-potassium-pump (Na⁺/K⁺-ATPase), which disrupts ion transport and thereby perturbs membrane potentials, often resulting in lethal cardiotoxicity (12, 13). Cardiac glycosides are produced independently by a number of plants and bufonid toads as secondary metabolites and are used for defense against natural enemies. For example, milkweed, foxglove, and oleander plants produce “cardenolides” (e.g., ouabain) that protect against herbivorous insects (13), whereas bufonid toads secrete structurally and functionally similar defensive compounds, “bufotoxins” (e.g., bufalin), from their parotoid glands (14).

Despite the toxicity of these molecules, natural resistance exists in many different herbivores and predators and is predominately mediated by molecular alterations to the H1–H2 extracellular domain of the Na⁺/K⁺-ATPase, resulting in target-site insensitivity to cardiac glycosides (9, 15, 16). For example, we recently showed that two amino acid replacements in the Na⁺/K⁺-ATPase α3 subunit (previously denoted α1) are responsible for a 3,000-fold increased resistance to bufalin in toad-feeding African and Asian varanid lizards, compared with toad toxin-susceptible Australian varanids (17). However, until now, comparisons of the molecular basis of resistance and mechanisms of action across the diverse range of resistant taxa reported (9, 15, 17, 18) have been lacking. Here we present an analysis of convergent molecular changes to the Na⁺/K⁺-ATPase H1–H2 domain in cardiac glycoside-resistant invertebrates and vertebrates. Our results support the view that molecular evolution in the animal kingdom can be heavily constrained, resulting in convergent processes canalizing evolution along highly predictable pathways.

Results and Discussion

To examine the molecular mechanism of cardiac glycoside resistance, we sequenced the nucleotides coding the resistance-conferring 12 amino acids of the H1–H2 extracellular domain of the α3 subunit of the Na⁺/K⁺-ATPase in 47 squamate reptiles. These squamates represent 22 species that are known to be resistant to the toxic effects of the cardiac glycosides produced by bufonid toads, 18 that are susceptible and 7 that have not been recorded feeding on toads (SI Appendix, Table S1). Four amino acid replacements were observed in the 47 taxa. Remarkably, all toad toxin-resistant squamates have the same amino acid replacements at two codons, leucine (L) at position 111 and arginine (R) at position 120, as those previously demonstrated to confer resistance to toad toxins in African and Asian varanid lizards (17). Furthermore, we find that squamates that are susceptible to toad toxins (or that do not feed on toads) share the same amino acids at these codons as susceptible Australian varanid lizards [glutamine (Q) at 111 and glycine (G) at 120]. The remaining two amino acid positions (113 and 119) vary in composition across both susceptible and resistant squamates, strongly suggesting that these replacements do not affect toad toxin resistance.

We next applied an evolutionary approach to test whether the resistant genotypes resulted from homology or homoplasy by reconstructing the evolutionary history of sequence change in the Na⁺/K⁺-ATPase and estimating the ancestral character state across the squamate tree. Our analyses demonstrate that ancestral squamates were susceptible to toad toxins and that resistance has evolved convergently on at least four occasions, each governed by a single, identical point mutation at two codons (111 and 120) (Fig. 1). In addition to Afro-Asian varanids (17), we find convergent resistance-conferring amino acid changes in the phylogenetically distinct viperid, elapid, and natricine snake lineages (including members of the toad toxin sequestering genus Rhabdophis) (19), representing convergent molecular evolution over ∼165 million years of separation (20). This constitutes compelling evidence of adaptive molecular convergence underlying phenotypic convergence. Although 85% of the H1–H2 domain of the Na⁺/K⁺-ATPase remains highly conserved under the influence of strong negative selection pressures [nonsynonymous to synonymous rate ratio (ω) of 0.07], one of the codons that governs resistance (position 111) was found to escape these evolutionary constraints in toxin-consuming species and experience episodic bursts of diversifying selection (SI Appendix, Fig. S1). Furthermore, the two toxin resistance governing codons (111 and 120) were found to be coevolving [posterior probability (pp) = 0.83] and directionally evolving toward the amino acid targets that confer resistance to cardiac glycosides (L111 and R120).

Fig. 1.

Convergent molecular evolution of resistance to toad toxins in squamate reptiles and reversal to susceptibility in Australian varanid lizards. The timing of changes to resistant amino acids in the H1–H2 extracellular domain of the Na⁺/K⁺-ATPase gene correlates with taxa that feed on toads. Pictures of toads indicate clades of taxa that are known to feed on toads without ill effects. The picture of a toad circled in red highlights that Australian varanid lizards have reverted back to being susceptible to toad toxins. Colored branches indicate the amino acid composition at key positions (susceptible, Q111 and G120; resistant, L111 and R120), and changes in color represent the reconstructed timings of amino acid replacements. Sites 111 and 120 were found to be coevolving (pp = 0.83). The character state (resistant or susceptible) at all key nodes in the tree, including those relevant for timings of character change, are strongly supported (pp ≥ 0.95); nodes with asterisks represent those falling beneath this threshold. Species tree was generated from refs. 57, 58.

While basal Afro-Asian varanids possess the resistant genotype (L111 and R120), the Australian varanids are characterized by the susceptible genotype (Q111 and G120), indicating that the latter may have reverted back to the ancestral state of “toad toxin susceptibility” (Fig. 1). Although the alternative hypothesis, that Asian and African varanids have independently evolved resistance to toad toxins, appears similarly parsimonious (two gains versus one gain and one loss), our ancestral sequence reconstruction analyses support the reversal hypothesis. Indeed, a combination of Bayesian and maximum likelihood reconstruction methods provided robust support (all key nodes pp > 0.95) for an early origin of resistance in basal varanids followed by a reversal to the susceptible genotype in varanids that migrated to the toad-free Australian continent (Fig. 1). Examples of such reversals are seemingly rare, as they usually relate to complex morphological characters (21, 22) rather than molecular changes to specific genes. This apparent reversal suggests a level of constraint acting on the Na⁺/K⁺-ATPase, most likely in the form of a trade-off between reduced Na⁺/K⁺-pump efficiency in the resistant state (23) and the ability to feed on toxic toads. In this context, these constraints lend support to Conway Morris’ concept of few “engineering optima” for biological functions, which may result in predictable evolution (6).

Our analysis of the Na⁺/K⁺-pump in squamate reptiles demonstrates the repeatability of molecular changes that underpin the gain and secondary loss of resistance to cardiac glycosides. We next assessed whether convergent changes to the H1–H2 extracellular domain occur in the diversity of other phyla reported to have Na⁺/K⁺-pump–mediated resistance to cardiac glycosides, specifically (i) insects that feed on and sequester toxins from cardenolide-producing plants (9), (ii) muroid rodents that feed on those plants and insects (and in some cases bufonid toads) (24, 25), (iii) the frog Leptodactylus latrans (formally referred to as Leptodactylus ocellatus) that preys on bufonids (18), and (iv) bufonid toads themselves (18) to prevent autotoxicity.

First, we reconstructed the evolutionary history of the multilocus Na⁺/K⁺-ATPase gene family and found that, in contrast to squamate reptiles, resistance in the above-described taxa is governed by changes to the H1–H2 extracellular domain of the α1 subunit (SI Appendix, Fig. S2). Reconstructions of the evolutionary history of α1 sequences revealed that, in all lineages, changes at two amino acid positions (mediated by one or two point mutations per codon) of the H1–H2 domain confer toxin resistance (SI Appendix, Fig. S3–S6). Notably, one of these codons (111) is found consistently across all taxa, whereas the second varies between phyla (119, 120, or 122) (Fig. 2). However, we find that previously described changes in the rat (Q111R and N122D) that synergistically confer 1,000-fold resistance to the cardenolide ouabain (15) are present in additional rodent taxa (SI Appendix, Fig. S4) and are identical to those observed in the bufonid-eating frog L. latrans, thus demonstrating molecular convergence between amphibians and mammals (Fig. 2) after ∼330 million years of separation (26).

Fig. 2.

Cross-phyla molecular convergence in the H1–H2 extracellular domain of the α Na⁺/K⁺-ATPase resulting in resistance to cardiac glycosides. Key amino acid residues found in the extracellular region are labeled by number. Resistance conferring residues are found grouped at position 111 at the N-terminal end and at positions 119, 120, and 122 at the C-terminal end. All individual amino acid changes have been demonstrated to contribute to resistance to cardiac glycosides in prior functional studies (9, 15–17), with the exception of 111E proposed by Dobler et al. (9) and 119D proposed and validated here (Fig. 4). Convergent changes observed within listed taxa are indicated by boxes, with numbers reflecting the number of independent changes within that lineage (e.g., convergent changes from Q to L at 111 has occurred three times in snakes). Amino acid changes are highlighted by letters (see also Fig. 1 and SI Appendix, Figs. S4–S6 and S8) and changes in charge by color (green, neutral charge; blue, positive charge; red, negative charge). Schematic of the Na⁺/K⁺-ATPase was modified from ref. 15.

To understand the functional significance of these mutations, we analyzed the specific amino acid replacements underpinning resistance to cardiac glycosides across all phyla. We find that significant increases in the isoelectric point of the Na⁺/K⁺-ATPase H1–H2 domain occur in resistant taxa compared with their susceptible counterparts (t test, P = 1.42 × 10⁻¹⁷; Fig. 3 and SI Appendix, Table S2). In all vertebrates and many insects, resistance to cardiac glycosides is mediated by the addition of charged amino acids to the ends of the H1–H2 extracellular domain (Fig. 2). Indeed, compared with all possible amino acid replacements, we find a significant excess of those that result in changing the charge of the H1–H2 domain in species that are resistant to cardiac glycosides (binomial test, P = 6.9 × 10⁻⁷). This provides very strong evidence that resistance is a consequence of shifts from neutral amino acids to charged amino acids in this particular domain throughout the extremely divergent taxa investigated herein.

Fig. 3.

Convergent evolution of animal resistance to cardiac glycosides is mediated by changes in the isoelectric point of the H1–H2 extracellular domain of the α Na⁺/K⁺-ATPase. A schematic tree of animal life (central) displays divergence times (in Myr) of major animal lineages based on paleontological constraints (26). Arrows on the tree represent the four major phyla analyzed here (insects, anuran amphibians, squamate reptiles, and mammals). For each phylum, a 3D phylogenetic tree displays reconstructed changes in the isoelectric point of the H1–H2 extracellular domain of the α Na⁺/K⁺-ATPase. Increases in the z axis reflect increases in isoelectric point, which are largely mediated by the replacement of amino acids with charged residues (Fig. 1 and SI Appendix, Table S2 and Figs. S4–S6 and S8). Animal pictures represent taxa found in each phylum that are resistant to cardiac glycosides. Note that the substantial decrease in isoelectric point observed in the squamate tree (red arrow) represents Australian varanid lizards, which have reverted back to the susceptible state.

To elucidate how resistance is mediated by changes in charge, we modeled the binding interaction between bufalin and the Na⁺/K⁺-pump of various animal taxa, using the pig Na⁺/K⁺-ATPase/bufalin crystal structure as our template (27). We compared the protein–ligand interactions of resistant, wild-type Na⁺/K⁺-ATPase from the rat, Leptodactylus frog, and bufonid toad with mutants designed to reflect the ancestral, susceptible genotype. Due to the paucity of full-length sequences for squamates, we used the susceptible, wild-type Na⁺/K⁺-ATPase isolated from the genome of the Burmese python (28) as our model and compared it with mutants designed to confer the resistant genotype. In all cases, bufalin was found to bind in a similar orientation to that observed in the crystal structure of the pig Na⁺/K⁺-ATPase/bufalin complex (27); that is, bufalin wedges into an extracellular-facing cavity formed between helices αM1–M2 (encoded by the H1–H2 extracellular domain) and αM4–M6, with the lactone moiety deeply buried in a hydrophobic funnel formed by αM4–M6 (Fig. 4 and SI Appendix, Fig. S7). The β-surface of the steroid core in bufalin faces polar side chains in αM1–M2 (Fig. 4B and SI Appendix, Table S3). However, there are critical differences in intermolecular contacts between susceptible and resistant genotypes of the Na⁺/K⁺-ATPase (SI Appendix, Table S3). Most strikingly, hydrogen bonds between the OH14β group of bufalin (a substituent that is conserved in all cardiac glycosides) and one or both of the carbonyl oxygen atoms of D121 are present in all susceptible models but absent from all of the resistant models, regardless of the nature of the proximal resistance mutation at positions 119, 120, or 122. We predict that the net effect of these subtle alterations in intermolecular contacts will be decreased affinity of bufalin binding, leading to significantly reduced inhibition of Na⁺/K⁺-ATPase activity. Our data therefore provide mechanistic insight into the convergent molecular evolution of cardiac glycoside resistance and validate mutagenesis studies describing specific amino acid changes that confer resistance in rodents and squamates (15–17).

Fig. 4.

Interactions between susceptible and resistant Na⁺/K⁺-ATPase and bufalin. (A) Chemical structure of bufalin. (B) Bufalin binding pocket in the crystal structure of bufalin bound to pig Na⁺/K⁺-ATPase-α1 (Sus scrofa; susceptible genotype; PDB ID code 4RES) (27). Bufalin wedges into a cavity formed by helices αM1–M2 (orange) (encoded by the H1–H2 extracellular domain), αM3–M4 (red), and αM4–M6 (gray). (C) The best structure from docking bufalin into a model of native (cardiac glycoside resistant) bufonid toad Na⁺/K⁺-ATPase. The β-surface of bufalin interacts with residues Q111, E117, E327, and T797 but makes no interactions <4 Å with D121. (D) The best structure from docking bufalin into a model of bufonid toad Na⁺/K⁺-ATPase with substitutions R111Q and D119N, thereby forming the susceptible genotype. The β-surface of bufalin interacts with residues E117, E327, and T797, and the OH14β group forms two hydrogen bonds (<3 Å) with the side-chain carboxyl of D121. (E) The best structure from docking bufalin into a model of native (cardiac glycoside resistant) rat Na⁺/K⁺-ATPase. The β-surface of bufalin forms a hydrogen-bond with residues E327 and T797 but not D121. (F) The best structure from docking bufalin into a model of rat Na⁺/K⁺-ATPase with substitutions R111Q and D122N, thereby forming the susceptible genotype. The β-surface of bufalin forms hydrogen bonds with both T797 and D121 (H bonds < 3 Å). Predicted binding modes and ligand–protein interactions in resistant and susceptible genotypes of the Leptodactylus frog, hedgehog, and python can be found in the SI Appendix, Fig. S7 and Table S3.

Bufalin is only a moderately potent inhibitor of Na⁺/K⁺-ATPase (K_d ∼43 nM for human Na⁺/K⁺-ATPase) (29), and therefore, even a 10-fold reduction in binding affinity is likely to significantly impair its ability to induce a pathological level of Na⁺/K⁺-ATPase inhibition. This situation is comparable to kdr and superkdr mutations in arthropod voltage-gated sodium (Na_V) channels that confer resistance to pyrethroid insecticides and DDT. Single point mutations in domain II of arthropod Na_V channels (mostly commonly a semiconservative L1014F mutation in the pore-forming domain II S6 helix) reduce deltamethrin affinity by only 17-fold, but this is sufficient to confer a high level of resistance to this pyrethroid (30). Analogous to bufalin resistance, a higher level of insecticide resistance is conferred by the combination of just two Na_V channel mutations in superkdr mutants (31).

Because the evolution of resistance to cardiac glycosides appears to be highly predictable, we next interrogated Na⁺/K⁺-ATPase sequence data from a previously unstudied taxon, the European hedgehog (Erinaceus europaeus), to establish whether these predictions hold in general. European hedgehogs frequently feed on bufonid toads and even anoint themselves with toad toxins for antipredator defense (32, 33). Our analyses show that the European hedgehog has the same resistance-conferring amino acid replacements in the H1–H2 extracellular domain of the α1 subunit (R and D) as those observed in rodents (SI Appendix, Fig. S8). Remarkably, these changes add identical charged residues at the same amino acid positions (111 and 119) as those found in bufonid toads (Figs. 2 and 3) and hence disrupt cardiac glycoside binding interactions in a highly similar manner (SI Appendix, Fig. S7). These results further support strong molecular convergence between amphibians and mammals and again highlight the predictability of evolution in this system.

As a result of the repeatability of mechanisms of resistance described herein, the molecular composition of the Na⁺/K⁺-ATPase could be used as a predictive tool to anticipate the ecological impact of invasive species harboring cardiac glycoside defenses on naïve and potentially vulnerable animal populations. This is important, because the introduction of the cane toad (Bufo marinus) to the toad-free Australian continent in 1935 has resulted in severe declines of numerous naïve predators due to the toad’s high toxicity (34–36). Currently, toads have invaded, or are close to invading, other toad-free biodiversity hotspots, such as Madagascar and parts of Indonesia (37, 38), where they are also likely to threaten native fauna susceptible to cardiac glycosides, such as the world’s largest and most iconic lizard, the Komodo dragon (Varanus komodoensis) (39).

In summary, although evolutionary processes are likely to be predominately stochastic, in the present study we have outlined a fascinating example of predictable convergent molecular evolution. Our data further emphasize how the application of similar, strong, selective pressures appear to underpin the convergent evolution of resistance to both naturally occurring (e.g., cardiac glycosides, tetrodotoxin, venom toxins) (7–9, 40) and anthropogenically applied (e.g., insecticides, pesticides) (10, 30, 31, 41) toxic molecules. Interestingly, many of these toxic molecules appear to interfere with essential components involved in the transport of ions across cell membranes, typically by impinging upon ion channel or ion pump function (7–9, 30, 31, 40, 42). In many cases, convergent evolution of resistance to these toxins is mediated by a small number of similar molecular alterations to the target molecule across unrelated taxa (Fig. 2) (7–9, 30, 40). These observations suggest that certain protein types appear to be highly constrained in terms of the evolutionary solutions available to overcome similar pressures, even when from highly divergent phylogenetic backgrounds. This concept is further reinforced by multiple reports of ion transporters evolving via the process of convergence in other systems, including acid nociception in mammals (43) and electric communication in fish (44). It therefore appears that ion transport proteins are heavily constrained and liable to canalization. Although it has been demonstrated that constraints can apply simultaneously to many proteins in complex systems (45), in the case of ion transporters, it seems likely that their strict ion selectivity and necessity for neurotransmission may result in most mutations being deleterious, thereby funneling molecular changes down the very few, similar, physiologically feasible pathways available in a repeatable manner. Our findings with the animal Na⁺/K⁺-ATPase encapsulate this process and prompt a general reconsideration of the pervasiveness of constraint limiting the options available for evolution, thereby promoting predictable molecular convergence.

Methods

SI Appendix, SI Materials and Methods has additional information relating to the methodologies described below.

Sequence Data.

Sequence data for the H1–H2 domain of the α3 subunit of the Na⁺/K⁺-ATPase of 43 squamate reptiles were generated as previously described (17). We supplemented these data with sequences generated from genome sequencing projects for squamates and nonsquamate outgroups. A list of the sequenced taxa and their propensity to feed on bufonid toads is displayed in SI Appendix, Table S1, and the DNA sequence data generated in this study have been submitted to GenBank with the accession numbers KP238131–KP238176. Sequence data for the H1–H2 domain of the α subunit of the Na⁺/K⁺-ATPase from insects and anurans were obtained from previous studies (9, 18) and the mammalian dataset generated by similarity searching various National Center for Biotechnology Information databases (www.ncbi.nlm.nih.gov) with the previously isolated sequence from the rat (15, 16) to identify homologous genes in related taxa.

Ancestral Sequence Reconstruction and Evolution.

For each group of taxa (squamates, insects, anurans, and mammals), separate sequence alignments were generated using the Multiple Comparison by Log-Expectation (MUSCLE) algorithm (46). Species trees were constructed from previously published studies (see Fig. 1 and SI Appendix, Figs. S4–S6 and S8 for details) and ancestral sequences reconstructed at various nodes of the Na⁺/K⁺-ATPase phylogenies using the marginal sequence reconstruction method (47) with the Ancestral Sequence Reconstruction (ASR) algorithm on the Datamonkey web server (48). The rate of evolution of the Na⁺/K⁺-ATPase gene was estimated using the maximum-likelihood model (M8) of PAML (phylogenetic analysis by maximum likelihood) (49) and the influence of episodic adaptive selection assessed using the mixed effects model of evolution in HyPhy (50). Coevolving amino acid sites were detected using the spidermonkey algorithm (51) in HyPhy; the Fast, Unconstrained Bayesian AppRoximation method (52) was used to detect sites in each dataset evolving under the pervasive influence of selection; and the Directional Evolution in Protein Sequences algorithm (53) was used for identifying sites that are the subject of directional evolution.

Changes in Isoelectric Point and Charged Residues.

The isoelectric point and changes in charge of the H1–H2 extracellular domain of the αNa⁺/K⁺-ATPase were calculated for all susceptible and resistant taxa in each dataset (SI Appendix, Table S2) using the ProtParam tool (web.expasy.org/protparam/). Statistical comparisons of changes in isoelectric point between resistant and susceptible taxa were performed using an unequal variance two-tailed t test. We next investigated whether resistance was associated with a shift from neutral to charged amino acids using binomial tests in R v.3.1.0 (54). Binomial tests compare an observed proportion, in this case the proportion of resistance mutations that involved a neutral to charged amino acid shift, to an expected proportion. In effect, this test asks whether shifts to charged amino acids (as we observe) occur more frequently than shifts to other amino acids (that we do not observe). In our test, the expected proportion was based on a null model of equal nucleotide-base substitution, such that transitions were weighted based on the number of individual base changes required to shift from one amino acid to another, thereby accounting explicitly for silent mutations in the evolutionary process.

Molecular Modeling.

All docking simulations were performed with AutoDock vina 2.0 (55) using an approach similar to that described by Zhen et al. (8). To establish a docking protocol we first redocked the cardiac glycoside bufalin into the crystal structure of bufalin bound to the pig Na⁺/K⁺-ATPase [Protein Data Bank (PDB) ID code 4RES] (27). The bufalin ligand was modeled with explicit polar hydrogens and torsional flexibility. The side chains of the ATPase residues Q111, E115, E116, E117, P118, D121, N122, L125, V322, A323, E327, E779, T797, I800, and D804 were treated as flexible, whereas the remaining residues were held rigid. The “best” structure from the redocking experiment was defined as the structure from the top 10 highest affinity solutions that is closest to the coordinates of the bufalin ligand in the cocrystal structure [measured as heavy-atom root mean square deviation (RMSD) in Ångstrom]. This best redocking structure was used as a reference structure for docking runs using the Na⁺/K⁺-ATPase isolated from other taxa. We next modeled the binding of bufalin with wild-type and H1–H2 extracellular domain substitution mutants of the α1 Na⁺/K⁺-ATPase in the rat (mutants R111Q, D122N), hedgehog (R111Q D119Q), Leptodactylus frog (R111Q D122N), and bufonid toad (R111Q D119N) and the α3 Na⁺/K⁺-ATPase in the python (Q111L G120R). All homology models were prepared with Modeler v9.10 (56) using the bufalin/pig Na⁺/K⁺-ATPase crystal structure (27) as a template and model quality checked using Swiss-PdbViewer (spdbv.vital-it.ch). Bufalin was next docked to each of the wild-type Na⁺/K⁺-ATPases and their substitution mutants. For each docking run, we calculated the RMSD between the best structure, defined as the top 10 high-affinity solutions closest to the coordinates of bufalin in the cocrystal structure, and the best structure from the redocking of bufalin to the pig Na⁺/K⁺-ATPase/bufalin crystal structure. RMSD only provides information about the position of the ligand but not about the interactions between the ligand and the protein. In each model, we thus determined the contacts between the ligand and the Na⁺/K⁺-ATPase and compared them to those in the cocrystal structure.