Dynamics of protein evolution within complex biophysical constraints

Abstract

Proteins evolve within complex biophysical constraints. Natural selection promotes the survival of the fittest organisms. Therefore, protein variants that decrease organismal fitness are likely to go extinct along with the strains that bare them. However selection is not destiny and the stringency of selection is not infinite. Rather, it is tuned by the number of organisms in an evolving population. Counteracting selection is the pressure of random mutations and luck; evolution is stochastic, and the fittest organisms do not always survive and reproduce. Randomly arising mutations can affect proteins in numerous ways. They can destabilize them, abolish catalytic activity, or change their intracellular abundance, and each of these can have a profound effect on fitness. Selection for stability is the most pervasive biophysical evolutionary pressure on globular proteins because these proteins are inactive and may form toxic aggregates when unfolded. In this dissertation, we examine the role of selection for folding stability in the pace of protein structure evolution. We find that the rate of structure evolution with respect to sequence divergence is slower for stable proteins than for their unstable counterparts. This observation can be explained with a simple analytical model that treats structure evolution as an activated process. We carried out simulations of model proteins to test the assumptions of the model and examine the mechanisms of protein structure evolution. Next, we lifted the assumption that evolution occurred under stability selection. Instead, protein data was visualized directly to interrogate the functional constrains under which proteins evolve. The overall strength of these constraints was manifested in the rate of protein sequence evolution, a quantity that can be calculated from genomic data. To facilitate interactive interrogation of the dataset, based on the yeast and E. coli proteomes, we developed a data visualization tool called ProteomeVis. Using this tool, users can observe correlations among protein properties and their sequence evolution rates as well as compare the sequences and structures of proteins in each proteome. We conceive of ProteomeVis as a versatile hypothesis generator that users can use as a launching point for further investigation.