Twenty years of combinatorial antibody libraries, but how well do they mimic the immunoglobulin repertoire?

Publisher Website
Google Scholar
Abstract
Combinatorial antibody libraries celebrate their 20th anniversary this year. In 1989, the first functional antibodies were isolated from a combinatorial antibody library derived from an immunized mouse (1). It was immediately evident that the technology would also lend itself to the generation of purely human monoclonal antibodies for medical use, which was not possible to achieve with the cell-based approaches for isolation of monoclonal antibodies used at the time. In these antibody libraries, mRNAs from B lymphocytes, coding for millions or even billions of the antibody heavy and light chains, are rescued by PCR technology and expressed as functional antibodies most often on phage that allow isolation of clones with particular characteristics, e.g., binding to a target molecule, cell type, or tissue (2, 3). The libraries can be created from immune individuals, biased for clones against the immunogen, or be as diverse as possible, producing so-called naive libraries from which theoretically antibodies to any antigen may be isolated. The technology is so robust that isolation of the type of antibodies searched for most often is successful. Maybe because practical results are easy to achieve, but more importantly because the libraries have been too large to adequately analyze with available nucleotide sequencing methods, just how well these libraries mimic the immune system has not been possible to assess until now. In this issue of PNAS, Glanville et al. (4) provide the first in-depth characterization of a very large naive library of human IgM antibodies, derived from >650 individual donors and comprising >3 × 10 …