(664c) Co-Evolution of Affinity and Stability of Alzheimer's Domain Antibodies Using Yeast Surface Display

An attractive method for designing lead antibodies is to mimic natural protein-protein interactions by grafting key recognition peptides into the antibody binding loops. We are using this approach to design single-domain antibodies that are specific for amyloid-forming polypeptides such as the Alzheimer’s Aβ peptide. To further improve binding properties of these lead antibody domains, we have used random mutagenesis and yeast surface display to select for mutations that enhance the binding affinity of antibodies with grafted amyloidogenic (hydrophobic) peptide segments in their binding loops. Interestingly, traditional selection methods yield domain antibodies with improved affinity and reduced stability. The highest affinity antibody domains obtained using such methods were strongly destabilized both on the surface of yeast and as autonomous domains. However, we find that stable antibody domains with improved affinity are reliably identified by replacing the display antibody that recognizes an epitope tag at the terminus of both folded and unfolded antibody domains with a conformational ligand (Protein A) that recognizes a discontinuous epitope on the framework of folded antibody domains. Importantly, selecting only for improvements in one property (e.g., stability) often resulted in reductions in the other property (e.g., affinity). Our findings highlight the importance of simultaneously optimizing both affinity and stability during the selection process to improve the rapid isolation of well-folded antibody fragments with specific binding activity.