Characterization of Functional Proteins from a Combinatorial Library of de novo Designed Sequences

Abstract

Naturally occurring proteins constitute a small fraction of the possible protein sequence space. By exploring uncharted sequence space, protein design seeks to validate our knowledge of protein folding while developing novel tools for chemistry, biology,and medicine. Previously, functional proteins were selected from a combinatorial library of four-helix bundle proteins based on their ability to restore the growth of auxotrophic strains of Escherichia coli. In order to better design future combinatorial libraries, these proteins were structurally characterized by circular dichroism and nuclear magnetic resonance spectroscopy. These proteins were shown to adopt their intended a-helical conformations, but showed significant peak broadening by NMR, a characteristic of dynamic structures. Through the use of a high-throughput assay that linked protein folding to a fluorescence reporter, the properties of these functional proteins were shown to be representative of the larger combinatorial library. Two proteins that were produced through the directed evolution of Syn-IlvA-1, an auxotroph rescue protein with promiscuous activity, were also characterized, and shown to have accumulated mutations that altered their conformations and thermal stabilities. Finally, in order to better predict the structural features of molten proteins,a computational approach for modeling conformational dynamics is proposed and validated by reproducing the properties of three proteins that exhibit different dynamics. In the same way that molten globule proteins can be seen as the feedstock of evolution, combinatorial libraries directed towards these properties can serve as artificial proteomes for the selection and development of novel functional proteins.