(569b) Blueprints for Diversifying Small Protein Scaffolds in the Context of Multiple Secondary Structures

Discovering proteins with new functionality gives rise to life saving therapeutics and is a necessity for expanding beyond current diagnostic and fundamental research applications. Our ability to quickly establish novel protein function is greatly hindered by the inefficiency at which combinatorial approaches explore potentially useful proteins. The protein sequence-function landscape is a sparsely populated, barren terrain of immense size. This unfortunate truth makes it difficult to randomly stumble upon a functional protein. Moreover, even when starting with a functional protein, making small random changes or mutations will tend to destabilize the protein. Improperly choosing one too many mutations will impede proper folding of the native structure and preclude function entirely. This concept has significant implications as it relates to protein library design. Libraries composed of broadly, uniformly diversified positions experience a large fraction of unstable variants. However, these effects can be mitigated by identifying and excluding detrimental amino acids from select positions as well as biasing inter- and intra-molecularly beneficial residues at library positions on a sitewise basis. Here, we will expand upon our recent work involving sitewise diversification of a three-helix bundle protein scaffold where we find biased amino acid diversity to be 13-fold more effective for discovering novel protein function compared to the traditional broad, uniform diversification scheme. We will then explore the common conclusions that exist between multiple distinct protein scaffold topologies. High-throughput sequencing, computational prediction of stability, and structural analyses guided the design of combinatorial libraries for affibody, Fn3HP, and Gp2. Thousands of high affinity ligands, having specific binding function to several targets, were isolated and deep sequenced. Amino acid preferences throughout binding hotspots, the effects of diversification on destabilization, and the importance of predictive ability of various input datasets will be discussed.