Santiago F. Elena (Instituto de Biología Molecular y Celular de Plantas [CSIC-UPV]; External Professor, Santa Fe Institute)
Abstract. Virus emergence is a complex, multilevel problem that results from a combination of ecological and genetic factors. To forecast when and how new viruses may emerge we must first identify the factors determining the distribution of genetic variants within the reservoir host as well as across all potential new ones, since this will ultimately condition the chance that different viral genotypes may have to persist in the reservoir and to successfully replicate in the new host. To this end, the following information is crucial: (i) what is the distribution of mutational fitness effects (DMFE) on the reservoir host and (ii) how it changes on different hosts (i.e., the genotype-by-host or GxE interactions), (iii) the way in which multiple mutations hitting the same genome interact in determining fitness (i.e., the genetic-by-genetic or GxG interactions, aka epistasis) in the reservoir host, and finally, (iv) how different hosts may affect the form of epistasis (i.e., the epistasis-by-host, or GxGxE interactions).
In this contribution, we will review some recent work done in our laboratory with tobacco etch potyvirus (TEV) addressing the above questions. In a first set of experiments we characterized the DMFE for genotypes carrying single nucleotide substitutions across a set of eight susceptible hosts of decreasing genetic relatedness with the primary host tobacco. We found a significant GxE interaction, which was sustained by differences in genetic variance for fitness and the pleiotropic effect of mutations among hosts. We also found that the DMFEs were markedly different between natural and non-natural hosts and, notably, the fraction of possible beneficial mutations was larger in the latter.
In a second set of experiments we generated random pairs of mutations whose separated effects were known and the combined deleterious fitness effects were determined. We found that many pairs had significant epistasis for fitness (GxG), including both positive and negative deviations from the null hypothesis of additive effects. Furthermore, we explored the contribution of sign epistasis to these non-additive effects and found that a large fraction consisted of cases of reciprocal sign epistasis, where the sign of the effect of mutations at two loci are dependent on each other.
Finally, we explored the existence of significant GxGxE interactions in determining TEV fitness by characterizing the distribution of epistasis among pairs of random mutations across four hosts that differ in their taxonomic proximity. We provide first evidence that the distribution of epistatic interactions significantly varied among hosts, and that average epistasis was stronger in the primary host but interactions became more additive as host’s genetic relatedness decreased.
The existence of significant GxE, GxG and GxGxE imply that no precise predictions on the fitness effect of an individual mutation can be made since it will depend not only on the genetic background in which it appears but also in the host wherein the virus replicates.