Animal venoms are outstanding evolutionary innovations used to subdue and digest prey, with other important functions, such as self-defense or intraspecific conflicts [1
]. Venoms are secretions consisting of organic and inorganic components that alter cellular processes in the target organism. Several arthropod groups are well known for their venoms, such as bees and wasps [6
], centipedes [8
], remipedes [11
], and arachnids (e.g., see [12
]). Some of the most efficient venoms occur in scorpions, which possess venom cocktails rich in peptides that affect ion channels [14
]. Toxins are thought to have arisen from diversification of paralogs of housekeeping genes, followed by neofunctionalization driven by positive selection in coding regions [15
Our knowledge of scorpion toxins has benefitted from high-throughput sequencing technologies, though taxonomically, sampling favors the diverse scorpion family Buthidae. In recent years, however, transcriptomic analyses have expanded beyond the Buthidae to the other 19 scorpion families (e.g., see [21
]). These efforts have demonstrated remarkable phylogenetic inertia in scorpion venom components at higher-level categories, but little is known about how venom composition and toxin evolution vary among species below the family level.
The North American scorpion family Hadruridae is represented by nine species in two genera, including several of the largest species worldwide [27
]. Of these, the diversity and effects of venom components from Hoffmannihadrurus gertschi
have been extensively studied [28
]. More recently, a transcriptome analysis, conducted on the venom gland of Hadrurus spadix
, was published [31
]. Though both species showed similar venom composition, a comparative analysis can shed light on the evolution of toxin components among these and other closely related species.
In this study, we focused on five hadrurid species: Hadrurus arizonensis, Hadrurus spadix, Hadrurus concolorous, Hoffmannihadrurus aztecus, and Hoffmannihadrurus gertschi. We used RNA-Seq to sequence de novo the venom gland transcriptome of Hoffmannihadrurus aztecus and Hadrurus concolorous.
We assessed the commonality of venom composition by comparing the number of putative transcripts found in the five species. Specifically, we focused on the evolutionary history of scorpine-like peptides (SLP, sensu [22
]). Originally, SLPs were called ‘orphan peptides’ due to the presence of two domains with dual functionality. The N-terminal region has cytolytic activity, whereas the C-terminus blocks potassium channel activity [32
]. SLPs were first isolated from the venom of Pandinus imperator
]. They have now been described in one or two gene copies, from 21 scorpion species, including nine buthids and 12 species from 10 other, non-buthid families [22
]. Intriguingly, buthid SLPs (with the exception of one sequence) and non-buthid SLPs form two mutually monophyletic clusters that are phylogenetically restricted to the parvorders Buthida and Iurida, respectively [22
], suggesting that the origin of SLPs could predate the diversification of scorpions. To assess SLP evolution in the context of comprehensive hadrurid phylogeny, we inferred internal hadrurid relationships and branch lengths using phylogenomic datasets, and surveyed venom gland transcriptomes to discover and map the distribution of non-buthid SLP homologs. We then inferred the direction of selection acting on the codon sequences of non-buthid SLP genes and evaluated the evolutionary dynamics of paralogs of SLPs within the Hadruridae.
The study of the early evolutionary origins of scorpion toxins has been complicated by issues of homology inference and taxonomic sampling. Recently, the use of transcriptomics to resolve the scorpion tree of life has yielded sequence data that can be used to assess venom composition concomitantly with reconstruction of phylogenetic relationships. Our analyses, based on nearly 2000 loci, provide the first resolution of relationships within the scorpion family Hadruridae. Most notably, we provide evidence of the conservation of venom composition up to the family level. Furthermore, our results detected different evolutionary selective pressures on the two domains of scorpine-like peptides. These results strongly contrast with previous hypotheses that scorpion cystine-stabilized alpha/beta fold toxins (CSαβ) peptides are evolving under positive selection to increase potency. Instead, more sites are evolving under negative selection, suggesting an evolutionary conservation in function for these peptides since the estimated Permian diversification of scorpions [37
], continuing through the estimated Neogene period of diversification of the Hadruridae. This study provides parameters to test the significance of toxic evolutionary dynamics in the extraordinary diversification of this culturally iconic group of arachnids.
Maximum likelihood and species tree analyses supported the monophyly of the family Hadruridae and its two constituent genera. Our results are congruent with those presented in a more sparsely sampled phylogenomic study [27
]. The molecular phylogeny presented here demonstrates the monophyly of Hoffmannihadrurus
as a lineage distinct from Hadrurus
, in agreement with previous hypotheses based on morphology [38
]. Our results suggest that the tradeoff between missing data and number of genes does not adversely impact reconstruction of the major scorpion clades (compare results in [27
]). Molecular dating supported the divergence of the Hadruridae 7–11 Mya, in partial agreement with formation of the Trans-Mexican Volcanic Belt, which could have been a vicariance event that isolated Hoffmannihadrurus
in the south from Hadrurus
in the north [41
Of marked interest to us in the transcriptomic analyses of the Hadruridae were the evolutionary dynamics of, and selective pressures on, scorpine-like peptides, which exhibit paralogous copies in a subset of hadrurid species. Buthid scorpine-like peptides differ from non-buthid SLPs with respect to the size of the mature peptide (<75 sites in buthids vs. >80 sites in non-buthids). Gene tree topologies, however, do not reflect a phylogenetic signal due to additional duplications in two hadrurid species (H. concolorous and H. aztecus) or other scorpions (e.g., in vaejovids). Only SLP2 was recovered from buthid scorpions.
Comparative transcriptomic analyses of hadrurid species highlights the evolutionary process that contributes to the diversity of CSαβ scorpion toxins. Results suggest that the SLP gene underwent duplication, which is concordant with the traditional model of venom toxin gene evolution [17
]. However, evolutionary analyses of scorpine-like peptides showed that few non-cysteine amino acid substitutions occurred between the cysteines of the C-terminus, including insertion/deletions. Consequently, our results do not support a previous hypothesis [15
] that CSαβ toxins have evolved under the influence of positive selection. Fewer than five sites in each SLP copy have experienced diversifying selection. Moreover, the majority of these sites were found at the N-terminus (a putative alpha helix), suggesting that positive selection might play an important role in evolution of this domain, but not at the C-terminus (Figure 4
). Thus, this highly conserved conformation of both SLP copies, along with the incidence of several sites evolving under negative selection to reduce alternate states, is consistent with the tendency to preserve function (e.g., venom potency or target specificity [19
Parametric analyses of the predicted peptides also suggest that both SLP copies possess similar biochemical properties. For example, SLP1 homologs have shorter sequences (<84 amino acids) with lower molecular weights and volumes (with two exceptions) and more negative net charges (with three exceptions, p
< 0.01). In contrast, SLP2 homologs have longer sequences (>84 amino acids) with correspondingly greater molecular weights and volumes (with three exceptions) and more positive net charges (with three exceptions; p
< 0.01). To test the evolutionary dynamics of these protein traits, we assessed the role of selection during hadrurid evolution by modeling shifts in SLP biochemical traits under an Ornstein–Uhlenbeck (OU) process. Several studies have used this process to model evolution of continuous traits by detecting shifts to different regimes with different adaptive optima, along a time calibrated phylogeny (e.g., in [45
], with a recent application to 3D structures of toxins [32
]). The incidence of shifts in hadrurid SLP2s may be biased by sequence lengths from H. aztecus
and H. concolorous
(smaller than other SLP2) and the divergence time of these species (between 7 and 11 Myr). In contrast, the incidence of shifts in Cercophonius
suggests a unique evolutionary regime in the family Bothriuridae.