Negative Selection with a de Novo Library of Amphiphilic Peptides from Unevolved Sequence Space

Abstract

The sequence space of possible protein sequences is staggeringly large; both evolution by natural selection and design by re-engineering natural proteins are effective search strategies for finding useful sequences near explored points in sequence space, but they have left some regions largely unexplored. One underexplored sequence type is that containing stretches of alternating polar and non-polar amino acids, perhaps selected against because such sequences tend to be amyloidogenic and toxic. Here is reported the design and characterization of the RA3S library, a library of peptide sequences containing three strands of alternating polar and non-polar amino acids, designed to take the amyloid fold. Sequences within the library did not show soluble, well-folded expression with a folding reporter GFP tag, and were toxic to the cells expressing them with both the FRGFP tag and a H6-SUMO expression tag. The toxicity and poor folding of proteins in this library complicates the hypothesis that amyloid folds were the first protein folds, sheds light on the mechanisms by which natural selection has avoided this region of sequence space, and suggests application of these sequences as negative selection sequences on vectors for expression or allelic exchange.