Synthetic antibody engineering: i) Targeting glycans; ii) Features of D5 variants that are required for high affinity binding to HIV-1 5-Helix

Abstract

Synthetic antibody engineering (SAE) has become a valuable approach for studying antigen-antibody interactions and for the isolation of specific, high affinity binding agents. This thesis describes two areas of research that apply to SAE.;First, we developed a phage display strategy for selection of high affinity glycan-targeting antibodies with domain exchange architecture. Anti-HIV-1 antibody 2G12, which targets high molecular weight oligomannoses, contains an unusual immunoglobulin scaffold in which the two VH domains from the antigen binding fragment (Fab) arms exchange positions. This architecture provides several key features that allow for high affinity and specific glycan recognition. We functionally displayed the 2G12 Fab dimer on the surface of M13 phage and that this Fab dimer is in its domain-exchanged format. We also showed that it could be selected from a background population of non-binding clones. This approach will be valuable for generating glycan-specific antibodies for research or therapeutic applications.;Second, we studied the high affinity interaction between antibody D5 and its target (an HIV-1 gp41 mimic known as '5-Helix') using high-throughput mutagenesis. The D5-5-Helix interaction serves as a model for high affinity recognition. Furthermore, D5 originates from the VH1-69 germline segment that is prevalent in viral responses and other disease states. We created two libraries in which residues in the light chain complementarity determining regions (CDRs) and the third heavy chain CDR were allowed to vary among two randomization schemes: (i) a 'restricted diversity' scheme that allows variation among the six amino acids Tyr/Ala/Asp/Ser/Pro/His; and (ii) a 'tailored diversity' scheme that reflects the light chain amino acid occurrences in the natural antibodies that have high homology to D5. Generally, functional clones were more readily isolated from the second library than the first. Among those clones, HCDR3 and LCDR3 showed high restrictiveness. Moreover, polar and charged residues were preferred at several LCDR3 positions. Selections with other antigens using these antibody libraries are also discussed. Our results provide insight into recognition features of the VH1-69 germline segment, and the requirements for high affinity and specificity in the D5-5-Helix interaction. This information may be used in future synthetic antibody library designs.