Distinct repertoire expansion patterns during affinity maturation of antigen responding camelid-derived single-domain VHH antibody

Abstract

Antibody maturation is the central function of the adaptive immune response. This process is driven by the repetitive selection of mutations that increase the affinity toward antigens. We hypothesized that precise observation of this process by high-throughput sequencing along time course of immunization will enable us to predict the antibodies reacting to the immunized antigen without any additional in vitro screening. An alpaca was immunized by IgG fragments using multiple antigen injections and antibody repertoire development was traced via high-throughput sequencing periodically for months. The sequences were processed into clusters and the antibodies in the 16 most abundant clusters were generated to determine whether the clusters included antigen-binding antibodies. The sequences of most antigen-responsive clusters resembled those of germline cells in the early stages. These sequences were observed to accumulate significant mutations and also showed continuous sequence turnover throughout the experimental period. The foregoing characteristics gave us 75% successful prediction of clusters composed of antigen-responding VHHs against IgG fragment. In addition, the prediction method was applied to the data from other alpaca immunized by epidermal growth factor receptor and 77% successful prediction confirmed the generality of the prediction method. Superior to previous studies, we identified the immune responsive but very rare clusters or sequences from the immunized alpaca without any empirical screening data. Significance StatementWe have developed a method for selecting VHH antibody sequences that react to antigens without in vitro screening by performing a large-scale sequence analysis of VHH along the immunization time course, clustering the sequences, and tracking the evolution of the sequences in the clusters. This method enables the identification of antibodies with low frequencies of occurrence, which has been difficult with existing in silico antibody identification methods.