Using single-cell sequencing and isolation of specific antibodies, we found that the memory B cell repertoire was dominated by large IgM, IgA and IgG2 clonal families, whereas IgG1 families specific for recall antigens were small. Analysis of multiyear samples demonstrated stability of memory B cell families and revealed that a large fraction of recently generated plasmablasts derives from long-term memory B cell families and is found recurrently in the absence of antigenic stimulation. We also found that, within large clonal families, somatic mutations generate individual clones with distinct specificities for microbial pathogens or commensals. Surprisingly, the predicted unmutated ancestor did not bind to the same antigen suggesting that somatic mutations diversify the naive B cell repertoire before rather than after antigen encounter. Collectively, these studies provide a systematic description of the structure, stability and dynamics of the human memory B cell pool and support a new model for the maintenance of plasma cells and antibody levels (Figure 1).
Using two complementary approaches we further investigated receptor-based antibodies. On one hand, we searched for insertion of inhibitory receptors other that LAIR1 and found antibodies generated by insertion of LILRB1 domains that were mapped to distinct families of RIFINS. On the other hand, we used an unbiased sequencing approach to identify different classes of non-VDJ inserts in 80% of individuals at frequencies of 1 in 104 to 105 B cells.
To dissect the antibody response to SARS-CoV2 we used a new screening strategy and isolated from immune donors several monoclonal antibodies that bind to all human-infecting coronavirus spike proteins. All these antibodies recognize the fusion peptide of the S protein and acquire affinity and breadth through somatic mutations and two of them neutralize Omicron BA.1 and BA.2 viruses and reduce viral burden and pathology in vivo. Interestingly, the analysis of the unmutated ancestor indicates that affinity and breadth were acquired through somatic mutations. Structural and functional analyses showed that these antibodies bind to a cryptic epitope hidden in prefusion stabilized spike, which becomes exposed upon ACE2 binding. Using the structural information gained in this study we plan to engineer the S protein to better expose the fusion peptide region.
