High-throughput sequencing error and bias correction increases the quantitative resolution of human naïve and memory B-cell receptor repertoires

Accurate high-throughput sequencing of immunoglobulin (Ig) chains (Ig-Seq) is often problematic due to primer bias and sequencing errors. Human Ig sequencing is further complicated by factors such as greater population-level germline allelic diversity, longer CDR3 regions relative to murine sequences, and a more complex antigenic history combined with higher frequency of somatic hypermutation (SHM), particularly in affinity-matured memory B-cell subsets. As a result, Ig heavy chain repertoire analysis tends to underestimate combinatorial diversity while simultaneously overestimating SHM. To overcome these issues, we developed a workflow for highly accurate human antibody heavy chain sequencing. First, we designed a set of 85 synthetic (in vitro transcribed RNA) Ig heavy chain standards representing all known IGHV and IGHJ alleles, unique CDR3s, and incorporating point mutations to mimic SHM. These standards are used in both isotype-dependent and -independent manners at predetermined ratios as spike-ins with biological samples to control for sequencing accuracy. Next, we prepared antibody libraries from purified circulating human B cells and spike-in RNA using a protocol known as molecular amplification fingerprinting (MAF), which incorporates unique molecular identifiers before and during multiplexed PCR amplification. We then performed MAF-based error and bias correction, and cellular replicate sampling to generate a robust, reliable, and highly accurate analysis of human antibody repertoires. We applied the workflow to estimate clonal diversity, gene segment usage, and SHM in naïve (IgM+ CD27-) and memory (IgG+ CD27+) B-cell subsets isolated from three different donors. Based on the sampling size, we are able to estimate the clonal diversity of the human naïve B-cell repertoire and that of the IgG memory B-cell repertoire combined with the level of SHM.