Design and evolution of Alzheimer's and Parkinson's antibodies

Abstract

In vitro antibody display methods such as phage and yeast surface display hold great promise for isolating antibodies with high affinity for a wide range of target molecules (antigens). These methods are particularly important because lead antibodies identified using in vivo methods (immunization) typically do not have high enough affinity for therapeutic applications and need to be further affinity matured in vitro. Other attractive properties of in vitro display methods include the ability to precisely control antigen presentation, conformation and concentration as well as the ability to perform negative selections against various types of non-antigens to eliminate non-specific variants. However, antibody display technologies generally produce antibodies with reduced stability and specificity relative to natural antibodies. Thus, we have sought to develop novel antibody design and in vitro display methods that overcome these challenges to generate antibodies in vitro with properties that rival those of natural antibodies generated in vivo.

We have also developed methods for efficiently enhancing the affinity of antibody variable domains using synthetic libraries based on natural diversity mutagenesis. Our approach involves first performing alanine-scanning mutagenesis to identify CDR positions permissive to mutagenesis. Next, we vary these permissive sites using degenerate codons that encode the wild-type residue and a few (1-5) residues that most frequently occur at each CDR site in natural antibodies. Finally, we use yeast surface display and stringent selection conditions to isolate antibodies with improved affinity and specificity. Using a variable domain specific for the Parkinson’s protein (α-synuclein) as a model system, this strategy led to a high success rate of isolating antibodies with significantly improved affinity while maintaining (or even improving) specificity. Computational modeling revealed that the enhanced affinity was not due to direct contacts between the CDR mutations and target molecule. Instead, the mutations appear to enhance affinity due to subtle structural rearrangements that enhance existing interactions or establish new interactions between wild-type CDR residues and the target molecule. These findings highlight new approaches to overcome challenges associated with in vitro antibody evolution and enable robust generation of antibodies with high specificity and stability in addition to affinity.

We first developed a platform to discover and evolve antibodies specific for amyloid-forming proteins. Our approach is to first design lead antibodies (single-chain antibody fragments, scFvs) specific for target proteins such as the Alzheimer’s peptide (Aβ) by grafting self-recognition peptides into the binding loops (complementarity-determining regions, CDRs) of a highly stable antibody scaffold. Next, we introduce mutations into the CDRs to generate focused antibody libraries, display these libraries on the surface of yeast, and sort for antibody variants with high affinity for Aβ. Interestingly, we find that the solution environment used during antibody selection strongly impacts the specificity of the resulting antibodies. Selection of antibodies in a non-stringent environment leads to the identification of antibody variants enriched in arginine mutations in the CDRs. The affinity of these antibodies is strongly dependent on the arginine mutations, which leads to low specificity. In contrast, selection of antibodies in a stringent environment also leads to antibody variants enriched in arginine CDR mutations, but the affinity of such antibodies is much less dependent on their arginine mutations and thus the specificity is significantly higher. Moreover, the evolved antibodies retained high stability, which was similar to the original antibody scaffold. This demonstrates that the use of stable antibody scaffolds and targeted mutagenesis in the CDRs can maintain high antibody stability during in vitro evolution.