Different B cell subpopulations show distinct patterns in their IgH repertoire metrics

Abstract

Several human B-cell subpopulations are recognized in the peripheral blood, which play distinct roles in the humoral immune response. These cells undergo developmental and maturational changes involving VDJ recombination, somatic hypermutation and class switch recombination, altogether shaping their immunoglobulin heavy chain (IgH) repertoire. Here, we sequenced the IgH repertoire of naïve, marginal zone, switched and plasma cells from 10 healthy adults along with matched unsorted and in silico separated CD19+ bulk B cells. Using advanced bioinformatic analysis and machine learning, we show that sorted B cell subpopulations are characterised by distinct repertoire characteristics on both the individual sequence and the repertoire level. Sorted subpopulations shared similar repertoire characteristics with their corresponding in silico separated subsets. Furthermore, certain IgH repertoire characteristics correlated with the position of the constant region on the IgH locus. Overall, this study provides unprecedented insight over mechanisms of B cell repertoire control in peripherally circulating B cell subpopulations.

Memory persistence and differentiation into antibody-secreting cells accompanied by positive selection in longitudinal BCR repertoires

“…Most IGHV-3 family members were observed more frequently in antigen-experienced B cells compared to naive subsets in all donors and time-points, while most of the other V genes that are well-represented in the naive subset decreased in frequency. These differences in V usage frequencies between naive and antigen-experienced B cell subsets have also been reported in several previous studies, even though different FACS gating strategies were used (Mitsunaga and Snyder 2020;Ghraichy et al 2021). Our findings further support the idea that initial recruitment of B cells to the immune response is in many cases determined by the germline-encoded parts of the BCR, presumably CDR1 and CDR2.…”

Deciphering the language of antibodies using self-supervised learning

“…Following pre-training, AntiBERTa outputs a distributed vector representation, or embedding, per residue for each BCR sequence (see experimental procedures ). To visualize the AntiBERTa embeddings, 1,000 BCR heavy chains were randomly selected from a well-characterized public dataset 40 and then averaged over the length dimension before projection by uniform manifold approximation and projection (UMAP). 26 , 29 , 41 Despite only being given BCR sequences and no other information, we find that the BCR embeddings naturally separate according to mutational load and the underlying BCR V gene segments used ( Figure 1 ).…”

Deciphering the language of antibodies using self-supervised learning

“…Machine learning methods have previously been used to classify B cells based on the BCR sequence and its features, such as the CDR3 region's physicochemical properties (44). While we have not explicitly predicted B cell subsets using AntiBERTa, its embeddings can already separate naïve and memory B cells.…”

“…Following pre-training, AntiBERTa outputs a distributed vector representation, or embedding, per BCR sequence (see Methods). To visualise the AntiBERTa embeddings, 1000 BCR heavy chains were randomly selected to embed from a well-characterised public dataset (44), then averaged over the length dimension (26,29). Despite only being given BCR sequences and no other information, we find that the BCR embeddings naturally separate according to mutational load and the underlying BCR V gene segments used (Figure 1).…”

“…For testing the representation capabilities of AntiBERTa, two datasets were used: a BCR repertoire dataset containing information on which B cell type each BCR sequence was derived from (44), and 191 non-redundant therapeutic antibodies from TheraSAbDab (45) with anti-drug antibody (ADA) scores. Of these, 65 antibodies also have known source information (46).…”