Experimental Evolution of Venom Resistance

The therapeutic properties offered by venom-derived toxins have led to the development of many life-saving medications. As this medicinal library continues to grow, so does the need to fully understand these toxins and their potential targets, many of which remain unknown. Venom target identification hypotheses have previously been derived from trial-and-error methods based on known toxin families and symptoms exhibited by affected animals, and have generally been limited to single toxin-target pairs. Although honing in on toxin-target pairing is ultimately necessary in the venom-to-drug pipeline, current approaches have neglected the evolutionarily favored complex venom phenotype, leaving many potential targets undiscovered. The advantages of Drosophila genetics enable the identification of genetic (allele frequency) changes resulting from evolved resistance in comparison to a control population, narrowing down potential targets to a set of candidate genes that have responded to the strong selective pressure of venom. Partnered with the complete venom characterization of the giant Florida blue centipede, Scolopendra viridis, our experimental evolution approach to venom target identification brings molecular evolution and therapeutic discovery together to identify candidate toxin targets which can then be used in downstream analyses, leading to both drug development and a greater understanding of the coevolutionary dynamics of predator and prey. In Chapter 1, we performed a complete venom characterization of two individual S. viridis from North Florida using high-throughput transcriptomic and proteomic approaches. Because venom is known to be variable within and among species, and centipede speciation is largely based on morphology, we then compared our results to a previously published venom characterization of S. viridis from Mexico. In this comparison we found evidence of local adaptation, or possibly speciation, as the venoms from the two regions were distinct to each population. Because selection, and therefore evolution, requires trait variation to act, in Chapter 2, we first established variation in the trait of venom resistance by scoring 89 inbred lines of Drosophila melanogaster for this trait. We then experimentally evolved venom resistance to S. viridis venom in multiple replicates of a genetically mixed D. melanogaster population. After 18 months of selection, we sequenced all experimentally evolved (venom-injected) and control (both PBS-injected and non-injected) replicates, and compared allele frequency differences between the two groups at two time points (middle and end) throughout selection. Using this data, we were able to identify genes of interest by their significance in two independent tests for allele frequency differences, partnered with their presence among genes with significantly enriched protein domains. However, results of experimental evolution studies are often noisy, and it can be difficult to distinguish true positives (i.e. alleles that have changed in frequency as a result of selection) from false positives (i.e. alleles that have changed in frequency as a result of drift or other environmental selection pressures), despite imposing stringent criteria. To assist in validation of true positives, in Chapter 3, we performed a full genome wide association study (GWAS) by scoring 192 inbred fly lines for the trait of venom resistance and used quantitative trait loci (QTL) mapping to identify genomic regions associated with venom resistance. Genomic regions identified in both experimental evolution and GWAS data sets will have a higher probability of being true positives and therefore genes located within these regions are more likely to be venom targets. Additionally, in both Chapter 2 and Chapter 3, we identified human disease-associated orthologs of our candidate genes to illustrate their potential as therapeutic targets of venom-derived drugs. Among the candidate genes identified using our experimental evolution and GWAS approaches, we found several that would be expected based on the known toxins within the venom. We also found several unexpected targets implicated in a range of human diseases including neurological disorders such as autism and Alzheimer's disease, multiple cancer types, and diseases of sensory systems including vision and auditory. Our results provide strong evidence that, as predicted, our current toxin-target identification methods have limited our ability in exploiting the full therapeutic potential of venoms and more robust methods using whole-venom, whole-animal systems should be considered.