Aged B cells as a candidate cell of origin for diffuse large B cell lymphoma: A multi-modal analysis of somatic mutation, clonality, and transcriptional programs

Abstract

Abstract Background: The human B cell repertoire is shaped by V(D)J recombination and somatic hypermutation (SHM) for diverse adaptive immunity, but prone to off-target mutagenesis1,2. Activation-induced cytidine deaminase (AID), central to SHM, is classically confined to germinal centers (GCs) but emerging data suggest activity beyond GCs, contributing to off-target mutations3,4. Memory B cells (mB Cells), compared to their naïve counterparts, accumulate more somatic mutations, with patterns reflecting off target SHM—patterns that mirror those seen in B cell lymphomas5. Aged B Cells (ABCs; CD21low/CD11c+/CD18+/T-BET+/ZEB2+), are a memory B cell subset enriched in aging and autoimmunity (AI) shown to express AID outside GCs6-8. Diffuse large B cell lymphoma (DLBCL) is an aggressive malignancy, prevalent in older adults and those with AI or chronic infection9,10. Recent large-scale genomic studies have proposed a complementary classification of DLBCL based on recurrent somatic mutations and indel profiles—with characteristic activation-induced cytidine deaminase (AID) mutational footprints—rather than relying solely on cell-of-origin categories. Defining DLBCL subtypes by mutational patterns links the contribution of mutagenesis to DLBCL pathogenesis and suggests opportunities for mutation-based MRD monitoring. We hypothesized that ABCs may serve as a cell of origin for DLBCL. Methods: We analyzed single-cell RNA sequencing (scRNA-seq) data from peripheral B cells (ABCs, mB cells and Naïve B cells) of 148 healthy donors (aged 23–91 years) to characterize transcriptional states and assess AID and associated machinery in each B cell population. To determine whether ABCs exhibit features of a premalignant precursor, we screened ABCs for mutations in lymphoma-related genes in 36 healthy older adults, 27 patients with AI, and 31 with DLBCL, and compared them to matched mB cells and germline controls (CD3⁺/CD14⁺ cells from the same individuals). We integrated deep immunoglobulin CDR3 sequencing to characterize clonal architecture. To establish a biological connection between ABCs and DLBCL, we analyzed paired blood and tumor samples to identify shared mutations and CDR3 clones. Longitudinal immune monitoring in autoimmune patients treated with biologic agents (adalimumab, belimumab, etanercept) was used to assess treatment-related changes in ABC dynamics. Results: ABCs exhibited significantly higher expression of AICDA and its associated DNA repair and error-prone polymerase genes (log₂ fold change >1.5), suggesting active AID-driven mutagenesis occurring outside germinal centers, independent of T cell regulation. Mutations screening revealed that ABCs harbored pathogenic somatic mutations in canonical lymphoma driver genes, with notable enrichment for KMT2D, SPEN, PRDM1 (BLIMP1), and ARID1A, in contrast to matched mB cells. Variant allele frequency (VAF) analysis showed that ABCs carried larger clones compared to memory B cells and germline controls (mean VAF 4%, 1.5%, and 0.9%, respectively; p = 1.92×10⁻³ and 1.03×10⁻²). Mutational context analysis identified a higher burden of mutations occurring within the AID hotspot motif (WRCY) in ABCs compared to memory B cells and germline controls, highlighting an AID-associated mutational footprint. CDR3 sequencing revealed increased oligoclonality and frequent clonal dominance in ABCs relative to memory B cells (Shannon entropy 3.6 vs. 5.4; p = 3.79×10⁻⁸). In 5 of 18 DLBCL patients with paired blood and tumor samples, expanded ABC-derived clones—but not memory B cell clones—were identified within the tumor tissue. In longitudinal samples from autoimmune patients treated with BAFF- or TNF-targeted biologics (adalimumab, belimumab, etanercept), ABC clonality was dynamically altered. Clone size declined globally, yet diversity was reduced, indicating therapeutic pressure and selective expansion of resistant ABC clones. Conclusions: Our findings implicate ABCs as a likely reservoir of lymphoma-initiating clones and a candidate cell of origin for DLBCL. The enrichment of AID-driven mutations, clonal dominance, and overlapping between ABC and tumor clones supports a model in which aberrant persistence and ongoing mutagenesis in ABCs promote lymphomagenesis. This work establishes a framework for mutation-informed MRD surveillance and highlights the clinical relevance of ABC biology across both AI and B cell malignancy.