High affinity, developability and functional size: the holy grail of combinatorial antibody library generation

Abstract

Since the initial description of phage display technology for the generation of human antibodies, a variety of selection methods has been developed. The most critical parameter for all in vitro-based approaches is the quality of the antibody library. Concurrent evolution of the libraries has allowed display and selection technologies to reveal their full potential. They come in different flavors, from naïve to fully synthetic and differ in terms of size, quality, method of preparation, framework and CDR composition. Early on, the focus has mainly been on affinities and thus on library size and diversity. Subsequently, the increased awareness of developability and cost of goods as important success factors has spurred efforts to generate libraries with improved biophysical properties and favorable production characteristics. More recently a major focus on reduction of unwanted side effects through reduced immunogenicity and improved overall biophysical behavior has led to a re-evaluation of library design.

Figure 1

Display methods as tools for antibody selection. Antibody (e.g., scFv, Fab, IgG) display methods include ribosome and phage display as well as cell surface display, such as yeast display. A: Ribosome display; in vitro evolution and selection method for scFv fragments. B: Phage display; protein selection method, where antibody derived fragments are genetically fused to pIII phage-coat protein and presented on the surface of filamentous phage. C: Yeast display; cell surface selection method, capable of presenting full length IgGs. The a-agglutinin yeast adhesion receptor, composed of two proteins Aga1 and Aga2, is used to display recombinant proteins on the surface of S. cerevisiae.

Figure 2

Four types of antibody libraries can be distinguished by source and design. A: Naïve libraries are amplified from a natural source, such as primary B-cells of non immunized donors. One library can be used for a wide variety of antigens. B: Immune libraries are generated from B-cell derived antibody repertoire of immunized or immune donors. Libraries are predisposed for a limited panel of antigens.C: Synthetic libraries are based on computational in silico design and gene synthesis. CDR design and composition is precisely defined and controlled.D: Semi-synthetic libraries comprise both CDRs from natural sources as well as in silico design of defined parts.

Figure 3

Representative examples of semi-synthetic and fully synthetic library design. Semi-synthetic libraries combine synthetic design and natural sources to different degrees whereas fully synthetic libraries are based on in silico design and de novo synthesis. Three examples are schematically shown.Library A comprises natural diversity within all CDR regions integrated into one selected single scaffold. CDR sequences are gained via PCR amplification from all naturally occurring frameworks (adapted from semi-synthetic N-CoDeR^® library).Library B comprises synthetic diversity in CDR-H1 and –H2 combined by natural diversity within CDR-H3. Whereas one single heavy chain framework was selected, the whole light chain repertoire has been used (adapted from semi-synthetic Dyax library). C: The fully synthetic HuCAL^® library concept is based on consensus sequences representing major germline families yielding 7 VH and 7 VL mastergenes. CDRs are designed to match amino acid composition of naturally occurring antibodies.