(190aw) Identifying an Individual’s Comprehensive Epitope Repertoire

Antibodies are essential immune system proteins. They act as markers and guide an immune response by binding to foreign molecules with high specificity and affinity. The precise and strong nature of their binding interactions have led them to be extensively used both experimentally and in therapeutic purposes. Many diagnostic tests work by determining the presence of antibodies that bind to specific antigens. In recent years, it has also become possible to identify the amino acid sequences of the most common antibodies in an individual through genetic sequencing.

Although the immune system and its functions are understood to a great extent, the identification of all the epitopes bound by an individual’s antibodies has not been made to date. We are developing and testing a diagnostic screen using a designed peptide library and complementary computational methods to enable this identification. Recently, we completed a computational analysis of a non-redundant database of 492 antibody – protein complexes. A key finding was that the energetic contributions to binding of antigen amino acids follow an exponential decay (R² = 0.96). In other words, only a few antigen amino acids contribute most of the binding energy. This is in agreement with previous experimental reports that five amino acids is sufficient to define an epitope for most antigens.

Based on this finding, we have developed a designed peptide library of all patterns of 7 amino acids in peptides of 12 total residues, where a pattern is specific amino acids possibly interspersed with undefined positions. Alanine is widely considered to be a generic amino acid, hence its use in alanine scanning mutagenesis, and is used for the undefined positions in the peptide library. The challenge of this library with an individual’s antibodies coupled with next-generation sequencing techniques enables the identification of millions of peptides recognized by those antibodies. Because the peptides are non-random, the important amino acids in each bound peptide are directly known. We have developed novel similarity matrices and clustering techniques to group the patterns into the epitope motifs. Once matured, this technology should simplify the diagnosis of many diseases and enable the discovery of novel biomarkers. We will present on the rationale for developing this screen, the statistics of the designed peptide library, the novel computational methods required for its analysis, and preliminary results from its use.