Rationally designed mimotope library for profiling of the human IgM repertoire

A large part of the IgM repertoire is characterized by few somatic mutations, physiological autoreactivity with housekeeping function and polyspecificity. It can be a versatile source of biomarkers for immune monitoring consisting of multidimensional antibody profiles. Probing this repertoire with a set of immunodominant self proteins provides only coarse information on protein reactivity patterns. In contrast, here we describe a deep panning approach for the rational selection of a set of peptide mimotopes that represents optimally the diversity of human IgM reactivities using a relatively small library of probes. A 7mer random peptide phage display library was panned on pooled human IgM. A non-exhaustive set of 224087 mimotopes was defined using next generation sequencing of the selected phage. A Kullback-Leibler divergence based GibbsCluster algorithm grouped them in 790 sequence clusters. A set of 594 mimotopes representative of the most significant clusters was used to test the hypothesis that this approach samples uniformly the IgM repertoire. When probed with sera from patients with or without different brain tumors in an oriented peptide array, this set produced a higher dynamic signal as compared to random peptides, to random peptides purged of mimotope sequence profiles and to mimotopes from only 10 out of 790 clusters. In this respect, the representative library is considered an optimal probe of the human IgM diversity. Proof of principle predictors based on the optimized library demonstrated its capacity to produce well separable data for the randomly selected diagnoses. Thus, an optimized library of IgM mimotopes is found to address very efficiently the dynamic diversity of the human IgM repertoire, providing informationally dense, structurally interpretable, profiles of the repertoire reactivities.