Abstract

The nature of the genotype to phenotype connection, the fitness landscape, remains poorly understood. In my talk I will first present a review of why this question is particularly relevant for molecular evolution. The rest of the talk I will focus on an experimental assay of the fitness landscape of a specific protein, the GFP , We assayed the GFP native function, fluorescence, of tens of thousands of genotypes allowing us to characterize its fitness landscape including genotypes with multiple mutations. We find its fitness landscape is locally narrow, with half of genotypes two mutations from the wildtype showing reduced fluorescence and fluorescence completely eliminated in half of genotypes with five mutations. The narrowness is enhanced by epistasis, which was detected in 30% of genotypes. Complex interactions were present in 1/6 genotypes, blocking 20% of the paths between fluorescent genotypes, demonstrating substantial ruggedness of the fitness landscape. A model of orthologous sequence divergence spanning hundreds of millions of years predicted the extent of epistasis in our data, indicating congruence between the local and global evolutionary scales of the fitness landscape. Time permitting, I will present data on an ongoing project involving an assay of the fitness landscape of two interacting proteins.