Detection of Occurrent Somatic Hypermutation and Associated Gene Expression Profile in Single B Cells

Introduction
Somatic Hypermutation (SHM) is an endogenous mutator mechanism in B lymphocytes that preferentially targets the immunoglobulin V(D)J coding region and facilitates antibody affinity maturation. The exact frequency and impact of SHM in individual B cells remains elusive, as previous quantifications only considered cumulative outcomes at population level. We therefore employed single-cell RNA sequencing on primary follicular lymphoma (FL) cells, a B-cell malignancy retaining the germinal center (GC) B-cell phenotype, to identify and quantify individual SHM events. This approach allowed us to detect neovariants, defined as single nucleotide differences that originate from recent SHM events, and to explore the gene expression profiles of the B lymphoma cells with neovariants, particularly focusing on genes and pathways effectuating SHM.

Methods
We collected and processed samples from 12 follicular lymphoma (FL) and five chronic lymphocytic leukemia (CLL) patients as non-germinal center controls. Single cell whole cDNA libraries were prepared by 10X Genomics. Immunoglobulin gene single cell libraries were prepared by enrichment with semi-nested amplification using 3′ constant domain primers followed by 10X Genomics prep. Both single cell libraries were sequenced in paired-end mode on Illumina HiSeq 4000. We developed a customized method to preprocess scBCR data designed specifically for highly mutated FL BCR sequences. To ensure data quality and validity of putative neovariants, we applied a comprehensive filtering system, including stringent thresholds in terms of numbers of individual immunoglobulin (IG) mRNA molecules sequenced as defined by their UMI, and reads per position obtained from the median of the estimator's distribution, base quality scores, exclusion of cell doublets, position within the V(D)J region, and proportion of cell-level variants. Based on a reported IG mRNA half-life of 3 h and assuming a first-order decay, the filtering criteria implied that detected neovariants were created between 1 and 7 hours prior to the time of sampling.

Results
A median of 1112 cells (range: 215-16432) per case were analyzed. 1239 IG neovariants were detected in a total of 334 FL cells in 7 of the 12 FL cases. Neovariants were undetectable in control CLL cells. A remarkably high proportion (95,4%) of neovariants were found within known AID motifs (WRCY, WA, and RCG), suggesting active AID involvement in the hypermutation process. The observed fractions of FL cells carrying recent SHM events indicated initiation of FL pathophysiology at least 11,2 months prior to sampling.

To further evaluate the role of AID, we compared gene expression profiles between FL cells with and without neovariants. The results demonstrated not only higher expression of AID but as well of pathways involved in SHM. Mismatch repair (hsa03430, 9 enriched genes with p.adjust < 0.001) and base excision repair (hsa03410, 17 enriched genes with p.adjust = 0.01), together with DNA replication (hsa03030, 14 enriched genes with p.adjust < 0.001), were the three most prominent pathways with upregulated genes. In cells with detectable neovariants, the MLH1/MSH2 genes, associated with the recognition of the deamination process, as well as POLH gene, which is involved in error-prone mutations in MMR, were highly expressed. Overall, we observed a more prominent expression of genes related to MMR than BER in cells with neovariants supported by a higher normalized enrichment score (NES) in MMR (MMR:1.93 and BER:1.6). This suggests not only an active SHM process but also potential interplay between AID and MMR in driving SHM in FL.

Conclusions

Here we report for the first time quantitative detection of ongoing somatic hypermutation at single cell level in FL. Based on the lifespan of IG mRNA, simultaneous detection of two IG mRNA transcripts differing by one (or a few) nucleotides only within a single cell permits to estimate the timing when this SHM event occurred. The detection of this phenomenon in 58% of FL samples and its association with AID expression suggests that AID-induced mutagenesis might be acting at a much higher rate than expected. Quantification of SHM at the single-cell level should facilitate the understanding of the cellular heterogeneity and dynamics in physiological GC reactions but also may offer novel insights into the AID-dependent shaping of the genomic landscape of germinal center B-cell neoplasms.