Genetics of Adaptation of ssRNA Viruses

The process of adaptive evolution is complex, determined in part by the set of beneficial mutations available to adapting populations, the distribution of their effects, and the extents to which pleiotropy and epistasis impact the relationships between genotype, phenotype, and fitness. Assumptions about epistasis, pleiotropy, and the availability of beneficial mutations are made by theories and models of adaptation and determine their predictions. Through the experimental evolution of bacteriophages, we tested model assumptions and predictions about epistasis, hybrid viability, the causes of parallel evolution, the distribution of beneficial fitness effects, the genotype-fitness landscape, antagonistic pleiotropy, and the cost of complexity. We did this in an effort to answer standing questions in the study of evolution regarding the process of adaptation and the forces that govern it. To characterize the effects of epistasis on determining the viability and persistence of novel hybrids, we generated experimental hybrids of highly divergent and highly related pairs of phage strains. We found that incompatibilities resulting from disruption of favorable epistatic interactions within the genome manifested as fitness costs, even in very closely related genotypes, but that ancestral fitness levels could be recovered through only a handful of compensatory mutations. Remarkably, we found that by permitting movement over wide expanses of sequence space in a single step—coupled with subsequent adaptation—hybridization allowed one phage genotype to explore a substantially higher and previously inaccessible peak in its fitness landscape. Thus, if conditions permitted hybrid genotypes to persist long enough to allow for compensatory evolution, even very low fitness hybrids could be viable in populations with their ancestral genotypes, potentially facilitating the establishment of stable hybrid zones and the creation of new species. Competing selectionist and mutationist viewpoints argue that parallel evolution is the result of either selection repeatedly fixing the same, optimal adaptive solution to a recurring evolutionary problem or a product of mutational constraint presenting selection with a biased subset of potentially suboptimal adaptive strategies. To characterize the rate of parallel evolution and the underlying forces driving parallelism, we performed 20 replicate first steps of adaptation for four phage genotypes. We found parallel and convergent adaptation to be common at every level of biological organization both within and among genotypes, from the level of mutational effects to the levels of protein, gene, amino acid, and nucleotide substitution. We also found that contrary to expectations, the mutation of greatest benefit in each genotype was never the most frequently fixed, and only fixed once in each of two genotypes. Patterns of parallel evolution are often offered as strong evidence of adaptation. However, we found that parallelism was driven at least as much by mutational biases and constraints as by selection, indicating that frequent observations of par- allel adaptive evolution in natural and experimental populations may not be driven primarily by selection, but rather by selection acting on a biased subset of all available mutations, possibly, as in the case with our results, yielding suboptimal adaptive outcomes. We also fit a distribution to the set of fitness effects for each set of 20 replicates. The form of the distribution of beneficial fitness effects shapes the predictions of models of adaptation and affects the properties of adaptive walks, particularly the average size of mutational steps and the likelihood of parallel evolution. We found that all four genotypes were best described by a distribution of beneficial fitness effects with a finite upper bound, conflicting with the assumption of some models that this distribution is exponential. In another experiment, we constructed double-mutant genotypes containing all possible pairings of three sets of five beneficial mutations generated under three different selective regimes: selection on growth rate only, selection on growth rate and thermal stability, and selection on growth rate and pH stability. The purpose of this experiment was to characterize patterns of pairwise epistasis for beneficial mutations, and in particular to measure interactions for each of the phenotypes underlying fitness. We found that mutations interacted antagonistically with regard to growth rate and fitness, in line with other recent work which has generally found a trend of antagonistic interactions between beneficial mutations. However, we found that mutations behaved on average more additively with regard to capsid stability. We concluded that mutations interact additively with regard to phenotype when considered at a basic, biophysical level, and that epistasis for fitness emerges from an intermediate phenotypic optimum and pleiotropy between its underlying phenotypes, manifested in adapting populations as a pattern of diminishing returns and antagonism between beneficial mutations. Finally, we performed adaptive walks for two pairs of wild-type and growth-optimized phage genotypes under selection on growth rate and capsid stability to test the hypotheses that adaptation of these two traits is constrained by antagonistic pleiotropy and that organisms evolving under complex environmental or selective pressures incur a cost of complexity manifested as a lower rate of adaptation. Despite expectations that antagonism would prevent optimization of growth rate and stability, both traits were significantly improved over the course of replicate adaptive walks. Additionally, we found no evidence for a cost of complexity, as the rate of adaptation under complex selection was actually higher than under simple selective conditions and did not require any additional mutational steps relative to one-trait selection. Our results indicated that increased organismal complexity, or an increase in the number of traits under selection, may not decrease an organism's rate of adaptation, even when mutations affect all of the traits under selection simultaneously, contrary to model predictions.