Evolutionary Relationships Beyond Fold Boundaries

Abstract

The comparative study of protein sequences and structures is traditionally used to better understand protein fold evolution. The insights we gain from our evolutionary analyses are applied in protein design projects. At the same time, we engineer proteins to test evolutionary assumptions; thus, we establish a feedback loop between both aspects of protein science. First, we compared Profile Hidden Markov Models, state of the art tools for homology detection, that represent all structures that adopt the (βα)8-barrel and the flavodoxin-like fold to discover an evolutionary relationship between these basic structural forms. Moreover, we located the region of the sequence space where both folds are most closely related. Having found this interface, we performed remote homologous searches and protein clustering to find sequences with intermediate features between the (βα)8-barrel and the flavodoxin-like fold.

We determined the x-ray crystal structure of one of these sequences to learn possible scenarios of fold change during the evolution of these ancestral structures. The intermediate sequence, named NTM0182, displayed features towards both folds. Moreover, and by structurally superimposing the three structures, we found classical evidences of homology among the three folds: high sequence identity over long aligned fragments. Our approach then starts by using very sensitive novel tools for homology detection (probability scores), to find intermediary links that provide classical evidences of common ancestry (high sequence identity). Next, we extended the Profile Hidden Markov Model comparisons to include all folds classified as α/β in the Structural Classification of Proteins (SCOP). Our comparisons showed that high scoring pairwise alignments are correlated with high local structural similarities between different folds. This observation inspired us to look for interchangeable sub-domain size protein fragments, related by sequence and structure, to build chimeric proteins and mimic protein fold evolution. Our global comparisons revealed that both, the flavodoxin-like and the (βα)8-barrel folds were related to Periplasmic binding protein-like I proteins.

We then employed sequence comparisons, structural superpositions, homology modeling and computational assessments to engineer a novel chimeric protein by fusing a flavodoxin-like fold protein into a PBP-like scaffold. The chimera turned out to be a well-folded protein with native-like properties. Thus, our research allowed us to gain evolutionary insights that are applied to protein engineering. In this respect, we establish a feedback loop between fold evolution and protein engineering. We finally contribute to an emergent vision where proteins from different folds are evolutionary related.