Convergent evolution of pain-inducing defensive venom components in spitting cobras

Abstract

Convergent evolution provides insights into the selective drivers underlying evolutionary change. Snake venoms, with a direct genetic basis and clearly defined functional phenotype, provide a model system for exploring the repeated evolution of adaptations. While snakes use venom primarily for predation, and venom composition often reflects diet specificity, three lineages of cobras have independently evolved the ability to spit venom at adversaries. Using gene, protein, and functional analyses, we show that the three spitting lineages possess venoms characterized by an up-regulation of phospholipase A2 (PLA2) toxins, which potentiate the action of preexisting venom cytotoxins to activate mammalian sensory neurons and cause enhanced pain. These repeated independent changes provide a fascinating example of convergent evolution across multiple phenotypic levels driven by selection for defense.

Conflict of interest statement

Competing interests: Authors declare no competing interests.

Copyright © 2021, American Association for the Advancement of Science.

Figures

Fig. 1. Reconstruction of the evolutionary origin of venom spitting and comparative analysis of venom composition.

(A) Multilocus-derived multispecies coalescent species tree, pruned to display the taxa whose venoms were analyzed in this study. Node support is indicated by colored circles, representing Bayesian posterior probabilities: black = 1.00, grey >0.90, white >0.70. Purple node labels indicate estimated divergence times (see Fig. S14 for credibility intervals). Spitting species are highlighted by red tip labels, and the three independent origins of venom spitting are indicated by the red-boxed spitting images. Pie charts adjacent to tip labels represent proteomic toxin composition of each species as a percentage of total toxins. (B) Principal Coordinate Analysis (PCoA) of cobra (Naja spp.) and rinkhals (H. haemachatus) venom toxins reveal major distinctions between spitting and non-spitting lineages. Asterisk highlights the Asian spitting species N. philippinensis, which exhibits greater similarity to non-spitting species than to its nearest relatives. Note each species is represented by two, typically overlapping, data points, which represent technical proteomic duplicates. (C) PCoA of cobra (Naja spp.) and rinkhals (H. haemachatus) cytotoxic three-finger toxins (CTXs) derived from top-down venom proteomics reveals that the most abundant CTXs detected in venom exhibit little sequence diversity among spitting and non-spitting lineages. Circle sizes reflect relative abundances of CTXs detected.

Fig. 2. Spitting cobra venoms cause significantly greater activation of sensory neurons than non-spitting cobras, mediated via potentiation by PLA2 toxins.

(A) Ancestral state estimation of half maximal effective concentrations (EC50) of venom-induced activation of neuronal cells shows a significant association between increased potency and venom spitting (PGLS, t = -4.48, p = 0.0004). EC50 values are expressed as the mean of triplicate measurements and colored branches are scaled accordingly (red, low EC50 and thus high venom potency; blue, high EC50 and thus low venom potency). Filled or empty circles at nodes/tips represent estimated ancestral states of non-spitting or spitting, respectively, and colored tip labels correspond to the different lineages. (B) PLA2 toxins in spitting cobra venoms potentiate the activating effect of CTXs on sensory neurons. A CTX fraction from each venom was added to dissociated mouse DRG neurons in the presence or absence of a corresponding PLA2 fraction (added 1 min prior), neuronal activation (i.e. a rapid increase in [Ca2+]i) monitored, and data presented as mean ± SEM of the resulting proportion of viable cells from 2-3 independent experiments. Statistical comparisons were performed using unpaired parametric t-tests; *, p < 0.05; **, p < 0.01. (C) The PLA2 inhibitor varespladib reduces neuronal activation stimulated by spitting cobra venoms. Calcium influx in F11 cells was measured on a FLIPR instrument incubated in the presence of venom from spitting species (2.4 μg or 4.8 μg [in the case of H. haemachatus and N. philippinensis] venom) and in the presence or absence of varespladib (13 μM). The data displayed represents the percentage of venom only cell activation stimulated by treatment with venom and varespladib. Error bars represent SEM and box mid-lines median values.

Fig. 3. The abundance, enzymatic activity and diversity of PLA2 toxins are associated with convergent evolution of venom spitting.

(A) Ancestral state estimation of PLA2 proteomic abundance, expressed as percentage of all toxins in venom proteomes, revealed a significant association with venom spitting (PGLS, t = 4.24, p = 0.0007). Colored branches are scaled according to PLA2 abundance (blue, low abundance; red, high abundance), filled or empty circles at nodes/tips represent estimated ancestral states of non-spitting or spitting, respectively, and colored tip labels correspond to the different lineages. (B) Ancestral state estimation of enzymatic PLA2 activity, expressed as area under the curve of concentration curves (AUCC) from kinetic in vitro colorimetric assay, revealed a significant association with venom spitting (PGLS, t = 2.24, p = 0.04). Colored branches are scaled according to PLA2 activity (blue, low activity; red, high activity). Labels as in (A) and see Fig. S12 for PLA2 activity concentration curves. (C) Principal Coordinate Analysis (PCoA) of cobra (Naja spp.) and rinkhals (H. haemachatus) PLA2 toxins derived from top-down venom proteomics reveals major variation between African spitting and non-spitting lineages, but little variation between Asian cobras. Note the convergent placement of Hemachatus PLA2 toxins with those of African spitting cobras. Circle sizes reflect relative abundances of PLA2s detected in the venom proteomes.