Convergent evolution of venom gland transcriptomes across Metazoa

Abstract

Animals have repeatedly evolved specialized organs and anatomical structures to produce and deliver a mixture of potent bioactive molecules to subdue prey or predators-venom. This makes it one of the most widespread, convergent functions in the animal kingdom. Whether animals have adopted the same genetic toolkit to evolved venom systems is a fascinating question that still eludes us. Here, we performed a comparative analysis of venom gland transcriptomes from 20 venomous species spanning the main Metazoan lineages to test whether different animals have independently adopted similar molecular mechanisms to perform the same function. We found a strong convergence in gene expression profiles, with venom glands being more similar to each other than to any other tissue from the same species, and their differences closely mirroring the species phylogeny. Although venom glands secrete some of the fastest evolving molecules (toxins), their gene expression does not evolve faster than evolutionarily older tissues. We found 15 venom gland-specific gene modules enriched in endoplasmic reticulum stress and unfolded protein response pathways, indicating that animals have independently adopted stress response mechanisms to cope with mass production of toxins. This, in turn, activates regulatory networks for epithelial development, cell turnover, and maintenance, which seem composed of both convergent and lineage-specific factors, possibly reflecting the different developmental origins of venom glands. This study represents a first step toward an understanding of the molecular mechanisms underlying the repeated evolution of one of the most successful adaptive traits in the animal kingdom.

Keywords: convergent evolution; evolutionary novelties; gene expression; stress response; venom.

Fig. 1.

Global patterns of gene expression differences between multiple tissues and lineages. PCA based on the normalized expression levels (TPM) of 2,528 shared orthogroups among all taxa. The proportions of variance explained by the components are in parenthesis. “Body tissues” include the following: abdomen, body tissue, cephalothorax, and viscera; “Other glands” include the following: accessory venom gland, hypopharyngeal gland, rictal gland, salivary gland, and silk gland; “Muscle tissues” include the following: muscle, proboscis, heart, and leg; and “Other organs” include the following: liver, kidney, and pancreas.

Fig. 2.

Transcriptome similarity between venom glands and other tissues. For each species, interspecific similarity (Spearman rank correlation coefficient ρ) between venom glands is compared to intraspecific coefficients between the venom gland and all other tissues for that species. Significant comparisons (Wilcoxon signed-rank test, P < 0.05 before correction) are indicated with an asterisk, although they were not significant after Benjamini–Hochberg correction. The low data points correspond to the pairwise correlations with echidna.

Fig. 3.

Comparison of venom transcriptomes and species phylogeny. (A) Phylogenetic species tree (Left), with circles marking the independent origins of venom in relation to venom gland expression tree (Right) with 100% bootstrap support throughout (not shown). (B) Sequence-based phylogenetic distances versus venom gland expression distances (1-Spearman coefficient). Pair distances between echidna and the other species are marked with triangles; all the others are circles. The dotted line indicates the positive correlation between expression and phylogenetic distances, excluding the echidna data points; the corresponding correlation test values are in parentheses.

Fig. 4.

Tissue-specific modules and GO enrichment results. (A) Heatmap based on Spearman correlation coefficients between the 62 modules; most modules are tissue-specific and cluster together. Definitions of tissue groups can be foundin Fig. 1. (B) Enrichment of the top three biological process GO terms of the tissue-specific modules. Color bar representing the tissues indicated in A. (C) Visualization of biological process GO terms enriched in the venom gland core gene set (modules 13, 24, and 40) produced using GO-Figure (59). Each bubble represents a cluster of similar GO terms summarized by a representative term reported in the legend and sorted by the average P values of the representative GO term across the three modules. Bubble size indicates the amount of GO terms in each cluster, and the color is the average P value of the representative GO term across the gene modules. Similar clusters plot closer to each other.

Fig. 5.

Semantic similarity scatterplot of GO biological process terms enriched in venom glands. Functional enrichment of up-regulated genes was performed separately for each species, and all significant GO terms (P value < 0.01) were summarized using GO-Figure (59). Each bubble represents a cluster of similar GO terms summarized by a representative term reported in the legend and sorted by the number of species with at least one term in the cluster. Bubble size indicates the number of terms in the cluster, and the color corresponds to the number of species in the cluster. Similar clusters plot closer to each other.