Energy Landscape of a Bridge between Protein Folds

Abstract

The entire structure space of the protein universe, so called fold-space, is estimated to contain thousands of different folds. Most proteins displaying high sequence identity are homologous and thus share the same fold, however, remarkable exceptions continue to be discovered. Some naturally occurring protein sequences can switch between different folds triggered by environmental changes. In addition, recent protein design experiments on two domains of different structure (all-alpha GA and alpha/beta GB from streptococcal protein G) have delineated a sequence of mutations linking two folds, suggesting how a new fold could arise from an existing one. This raises many interesting questions, including: How difficult is it to evolve from one fold to another? How many possible paths are there between folds and what are their common features? We have addressed such questions using a theoretical model of protein evolution, which builds on recent developments in statistical energy functions capturing co-evolutionary variation within a set of related proteins. We have applied the model to an experimentally well-characterized set of mutant sequences which bridge the fold landscape between GA and GB. It can capture both the trends in stability of the folded state within each fold, and also the propensity of the individual sequences to fold to a given structure. By using computer simulation to mimic the evolutionary dynamics, the mutational bridge between GA and GB in sequence space is characterized. Without a constraint to remain folded, the favored path between the two folds go via unfolded sequences. This suggests that the switch may occur via intrinsically disordered intermediate sequences which only fold on encountering their cognate ligands. However, there are still many accessible bridge sequences after a requirement of folded state stability is imposed. This model provides a new strategy of designing fold-switching proteins.