(280h) Optimizing Combinatorial Diversity with High Throughput Selections and Computation

Discovering new protein functionality via combinatorial library screening currently lacks efficiency, particularly in small ligand scaffolds in which sufficient mutations are needed to introduce novel function while maintaining intramolecular interaction integrity. This study hypothesizes that combinatorial protein libraries would benefit from sitewise optimization of amino acid distributions – rather than the traditional binary design with a heavily diversified paratope and fully conserved framework – and posits efficient methods to identify the optimal diversity. We investigated the evolvability of small proteins in the context of binding affinity using the beta-sandwich fibronectin and the three-helix bundle affibody domains. 

We took two complementary approaches: (a) use yeast surface display for magnetic and fluorescent selection and sequencing of thousands of binding ligands towards numerous targets; and (b) use in-house code, Rosetta, and FoldX for high throughput computation of structural and stability metrics of diversified protein mutants to assess side-chain accessibility and stability tolerance. The resulting analyses enlighten the relative importance of inter- and intra-molecular interactions at each site as well as the degree and type of diversification bias that optimizes evolutionary potential. Within the 25 and 15 sites within the fibronectin and affibody paratopes we found a gradient of the optimal extent of diversification.  Select positions benefited from extensive diversity whereas, for example, fibronectin T76 – a semi-buried position in the paratope periphery – gained from mild diversity, and restricted diversity was observed at structurally critical positions both within (V29 and G79) and distal to (D23 and S85) the paratope. In addition to diversity biases that reduce destabilization within the library, we will discuss the stabilization of the parental protein via mutation of its conserved framework. In particular, we compared stability engineering via random mutation, structurally-guided mutation, and consensus – both natural and synthetic – design.

All throughout, we will balance discussion of technical advances in the particular scaffolds and generalizable lessons including the extent to which each of the aforementioned efforts aids evolutionary efficiency.