William2002 - Extrafollicular Somatic Hypermutation in Autoimmune Mice

Full citation: William J, Euler C, Christensen S, Shlomchik MJ. Evolution of autoantibody responses via somatic hypermutation outside of germinal centers. Science. 2002;297(5589):2066-2070. doi:10.1126/science.1073924

Raw file: [[raw/william2002.pdf]]

Summary

This paper provides the first direct demonstration that somatic hypermutation occurs at extrafollicular sites in vivo. Using MRL.Fas^lpr (MRL/lpr) lupus-prone mice carrying a transgenic immunoglobulin heavy chain (AM14) encoding a rheumatoid factor (RF) autoantibody specific for IgG2a^a, the authors traced RF-producing B cells using an anti-idiotype (4-44) monoclonal antibody. Unexpectedly, these RF B cells were found proliferating at the splenic T zone–red pulp border rather than in germinal centres.

Microdissection of these extrafollicular RF B cell clusters followed by PCR amplification and sequencing of Vκ8 light chain genes revealed extensive somatic mutations organised into genealogical trees — the hallmark of ongoing SHM. The mutation rate was estimated at ~0.3 mutations per gene per generation, comparable to rates estimated for GC hypermutation. The pattern of V gene mutations showed clear signs of antigen-driven selection. Critically, Id⁻ GCs in the same spleens contained few or no RF B cells and yielded no mutated Vκ8 sequences, confirming that the mutation was occurring at the extrafollicular site, not in GCs.

The authors propose that SHM is induced whenever B cells undergo substantial proliferation in the presence of antigen and T cell-derived signals — a principle that explains why mutation occurs in GCs, in vitro, and at extrafollicular sites. They further suggest that TLR co-stimulation (via TLR9-mediated recognition of chromatin-containing immune complexes) may be a unifying feature of dominant autoantigens, and that EF mutation may allow B cells to escape the tolerance censoring mechanisms that normally operate within GCs.

Study Design

• Type: In vivo murine study (immunohistochemistry + microdissection + BCR sequencing)
• Sample size: 8 mice analysed by microdissection/sequencing (853 cells picked across 45 libraries, 305 total sequences, 125 unique sequences); additional mice for immunohistochemistry
• Setting: AM14 H chain transgenic MRL.Fas^lpr mice (lupus-prone); IgH^b congenic controls (lacking autoantigen); non-autoimmune-prone controls
• Population: Aged mice with established autoimmune disease; RF autoantibody response tracked by anti-idiotype (4-44)

Key Findings

• RF B cells localise to the T zone–red pulp border, not GCs. Immunohistochemistry with anti-Id (red) and CD3 (blue) showed RF B cells in clusters at the splenic T zone–red pulp border, interdigitated with T cells. PNA⁺ GCs were present but did not colocalise with Id⁺ areas. CR1 staining confirmed absence of FDC networks at the Id⁺ sites.
• High proliferative rate at EF sites. 15% of splenic RF Id⁺ B cells incorporated BrdU within 2 hours of a single injection, indicating rapid cell cycling at the T zone–red pulp border.
• CD11c⁺ DCs interact with RF B cells at EF sites. CD11c⁺ dendritic cells were abundant within the RF B cell clusters and showed close physical interaction with Id⁺ cells, in contrast to GCs where CD11c⁺ DCs are rare.
• Somatic hypermutation occurs at EF sites at GC-comparable rates. Vκ8 light chain sequences from microdissected EF clusters carried substantial somatic mutations. Mathematical modelling estimated the mutation rate at ~0.3 mutations per gene per generation — comparable to GC hypermutation rates estimated by the same method.
• Genealogical trees confirm ongoing diversification. Independent sequences from the same microdissected area shared some mutations but differed by others, generating hierarchical genealogical trees with shared trunk mutations and unique branch mutations — the definitive signature of ongoing in situ SHM.
• Mutations show antigen-driven selection. The pattern of replacement:silent mutations across framework and CDR regions was consistent with antigen-driven positive selection (Table S2).
• Clonal relatedness is local, not migratory. Sequences from the same or nearby picks shared common mutations and VJ junctions, while distant sites had different mutation sets — ruling out mutation elsewhere followed by migration to the EF site.
• GCs lack mutating RF B cells. Of 7 GCs examined in one spleen, all contained few or no visible Id⁺ cells. Vκ8 PCR yielded product from only 1/7 GCs (vs. 10/10 control Id⁺ areas), and the lone positive GC’s sequences were unmutated with a different Jκ segment — further evidence against GC mutation followed by EF migration.
• Some spleens with active EF mutation have no GCs at all. In certain mice, ongoing EF somatic mutation was detected despite complete absence of GCs of any type, eliminating even residual GCs as a possible mutation site.
• Process is autoantigen-dependent. RF B cell accumulation and EF proliferation were never observed in IgH^b congenic mice lacking the IgG2a^a autoantigen, nor in non-autoimmune-prone controls.
• Weighted average: 4.3 mutations per unique sequence across all mice, with 3.2 trunk mutations (shared by all clones in a tree) and 0.7 unique mutations per sequence.
• TLR co-stimulation proposed as a general mechanism. The authors invoke Leadbetter et al. (2002, Nature) showing that chromatin-containing immune complexes co-stimulate AM14 RF B cells via TLR9, and propose that TLR co-signalling may be a unifying feature of dominant autoantigens that enables EF mutation.
• EF mutation may escape GC tolerance. Because GCs are thought to harbour mechanisms that censor autoreactive mutants, mutation outside GCs may allow autoreactive B cells to escape tolerance checkpoints — a mechanism for autoantibody diversification in autoimmune disease.

Methods Used

Immunohistochemistry, BCR Sequencing

Entities Mentioned

CD11c

Concepts Addressed

Extrafollicular Response, Somatic Hypermutation, Germinal Center

Relevance & Notes

This is the foundational paper establishing that somatic hypermutation is not restricted to germinal centres. It is the most-cited direct evidence for EF SHM across the entire wiki’s literature base, referenced in Wei2007, Sanz2025, and multiple other sources as the key murine precedent for EF origin of somatically mutated B cells. The study has 585 citations (Semantic Scholar) and is published in Science.

For the dengue EF model: William2002 provides the mechanistic precedent for two key features of the dengue plasmablast response: (1) the intermediate SHM levels (4.4–6.9%) found in acute dengue BCR data (Parameswaran2013, GodoyLozano2016) could reflect genuine EF-derived mutation rather than exclusively GC origin; (2) the TLR9 co-stimulation mechanism maps directly onto the TLR7 pathway proposed for dengue (GodoyLozano2016, Jenks2018) — dengue ssRNA is a physiological TLR7 ligand, analogous to the chromatin immune complex TLR9 ligand in the RF model.

Limitations: (1) Murine model with a transgenic BCR — the artificially elevated precursor frequency may facilitate EF responses that would not occur at natural frequencies. (2) Fas deficiency (lpr) is not representative of wild-type immunity; it prevents apoptosis that normally resolves EF proliferative foci. (3) Light chain (Vκ8) sequencing only — heavy chain SHM at EF sites was not assessed. (4) The RF autoantibody system may not generalise to all autoantibody or antiviral B cell responses. (5) The paper does not identify what molecular signals sustain SHM at EF sites (AID expression, T cell help quality, etc.).

Relationship to later work: Jenks2018 confirmed that DN2 cells (the human EF pre-plasmablast) carry SHM at levels comparable to plasmablasts but lower than switched memory cells, consistent with the William2002 model of EF-derived mutation at reduced levels. Tipton2015 established that ~30–33% of human SLE ASCs have <3% VH mutation, providing the quantitative human benchmark that the William2002 murine model predicted. Sanz2025 explicitly cites William2002 alongside MacLennan et al. 2003 and Roco et al. 2019 as evidence that SHM and CSR are not restricted to GCs.

Questions Raised

• Does the EF SHM rate observed in this murine RF model (~0.3 mut/gene/generation) translate to human extrafollicular B cell responses? The human DN2 SHM data (Jenks2018) is consistent but not directly comparable (different quantification method).
• Is AID expression directly detectable at the T zone–red pulp border, or is it inferred from mutation outcomes? Direct AID localisation would strengthen the conclusion.
• Does the TLR9 co-stimulation mechanism extend to TLR7 (ssRNA sensing) for viral autoantigen responses? This is directly relevant to dengue, where TLR7 ligands are physiologically abundant during viraemia.
• Can the EF tolerance escape mechanism explain the autoreactive VH4-34/VH1-69 enrichment seen in dengue plasmablasts (Appanna2016)?