EFB Dengue Literature Review

Germinal Center

27 min read

Germinal Center

Overview

Germinal centres (GCs) are specialised microanatomical structures within secondary lymphoid organs (lymph nodes, spleen, tonsils, Peyer’s patches) where antigen-activated B cells undergo rapid proliferation, somatic hypermutation (SHM) of immunoglobulin variable genes, affinity-based selection, and class switch recombination (CSR). GC reactions generate high-affinity isotype-switched memory B cells (which acquire CD27) and long-lived plasma cells. They are the canonical route for production of durable, high-quality humoral immunity.

In the context of this wiki, the germinal centre response is the primary counterpoint to extrafollicular B cell responses — understanding which populations are GC-derived vs. EF-derived is central to interpreting B cell dynamics in dengue infection.

Key Points from Literature

  • GC reactions are initiated by cognate B–T cell interactions mediated by CD40–CD154 (CD40 ligand); this interaction is required for normal GC formation and for CD27 acquisition by B cells (see Wei2007 - DN Memory B Cells in SLE, citing Maclennan 1994 and Grewal & Flavell 1998).

  • CD27 is acquired by B cells as a result of CD40–CD154-mediated signalling in GCs; absence of CD27 on DN memory B cells therefore suggests these cells bypassed or aborted GC entry (see Wei2007 - DN Memory B Cells in SLE).

  • In SLE, GC reactions appear dysregulated: germinal centre exclusion of autoreactive B cells is defective (cited as Cappione et al. 2005 in Wei2007 - DN Memory B Cells in SLE), and anti-CD154 blockade failed to reduce DN cell frequency — consistent with DN cells arising outside GC reactions.

  • Murine evidence for SHM outside GCs: William2002 - Extrafollicular Somatic Hypermutation in Autoimmune Mice demonstrated that somatic hypermutation can occur in extrafollicular reactions at GC-comparable rates (cited in Wei2007 - DN Memory B Cells in SLE). This is a critical precedent for the EF origin model of DN cells.

  • T cell-independent GC reactions produce low-level SHM compared with T cell-dependent GCs (cited as Toellner et al. 2002 in Wei2007 - DN Memory B Cells in SLE), supporting the idea that the lower mutation rate in DN cells is consistent with GC-independent or abortive-GC origin.

  • Circulating pre-GC (Bm2ʹ) cells in SLE: A subset of SLE patients have expanded circulating IgD⁺CD38^high CD20⁺CD10⁺ cells — putative GC founder cells. In a clinical cohort, 3 of 15 patients had marked pre-GC expansion (mean 9 ± 11.9% across cohort vs. 4.7 ± 2.8% in controls). These cells are phenotypically distinct from plasmablasts (they retain CD20 and CD19^bright expression) (see Anolik2004 - Rituximab and B Cell Abnormalities in SLE; see also Bm Classification).

  • Pre-GC cells may resist rituximab despite being CD20⁺: All three patients with marked pre-GC expansion had incomplete or transient B cell depletion, and pre-GC frequency increased rather than decreased after rituximab in two. This suggests either intrinsic resistance or continuous replenishment from ongoing antigen-driven GC reactions. GC-biased SLE may thus represent a mechanistically distinct disease subset (see Anolik2004 - Rituximab and B Cell Abnormalities in SLE).

  • Defective GC censoring of autoreactive B cells in SLE: VH4.34 autoreactive memory B cells are elevated 12-fold in SLE vs. healthy donors (16.2 ± 11.9% vs. 1.3 ± 0.3%; P=0.03), consistent with a GC checkpoint failure. After rituximab treatment, autoreactive memory B cell frequencies normalised (1.92 ± 0.7%), interpreted as restored GC censoring upon immune reconstitution (see Anolik2004 - Rituximab and B Cell Abnormalities in SLE).

  • acN cells can enter both EF and GC pathways: Phylogenetic clone trees in Tipton2015 show complex clones in which acN cells with 0% SHM co-exist with CD138⁻ and CD138⁺ ASCs carrying up to 21.5% VH mutation. The aggregate data are most consistent with a model of sustained and asymmetric differentiation of acN cells through both extrafollicular pathways and GC reactions — both pathways are simultaneously active rather than mutually exclusive (see Tipton2015 - ASC Diversity and Origin in SLE).

  • IgM⁺ memory B cells as GC-independent first memory layer: The high frequency of IgM sequences in the IgD⁻CD27⁺ memory compartment in SLE patients (20.9–68.1%) vs. vaccinated controls (1.5–37.5%), together with the prominent naive-cell connectivity to ASCs, is consistent with IgD⁻IgM⁺ memory cells representing the first GC-independent memory layer generated from newly activated naive cells (see Tipton2015 - ASC Diversity and Origin in SLE, citing Dogan et al. 2009).

  • SM and DN2 have divergent epigenetic programmes despite similar methylation levels: DNA methylation phylogenetics place both SM and DN2 closest to ASCs, yet their chromatin accessibility landscapes are sharply distinct. SM chromatin is enriched for NF-κB, EBF, and OCT2 motifs — transcription factors associated with GC transit and canonical memory identity. DN2 chromatin is enriched for T-BET, AP-1, and EGR motifs. This epigenetic bifurcation provides the strongest evidence that SM and DN2 represent genuine alternative differentiation endpoints (GC vs. EF), not different stages of a single pathway (see Scharer2019 - Epigenetic Programming in SLE B Cells, RRBS + ATAC-seq, n=9 SLE + 12 HC).

  • DN1 cells are transcriptionally GC-derived: DN1 cells (CXCR5⁺, CD21⁺, within the IgD⁻CD27⁻ compartment) share a near-identical transcriptome with switched memory cells (only 22 DEGs by RNA-seq). They express TCF7 (the central memory TF), CXCR5 (the follicle-homing receptor), and BACH2 — all hallmarks of GC-transit. DN1 likely represent early switched memory precursors that have not yet acquired CD27 through CD40–CD154 interactions, placing them in a GC-associated differentiation pathway. CD40L stimulation does not inhibit DN1 generation in vitro (unlike DN2), consistent with their GC compatibility (see Jenks2018 - DN2 B Cells and EF Pathway in SLE, RNA-seq + in vitro differentiation).

  • CD40L actively inhibits EF pathway — antagonistic regulation between GC and EF: CD40L stimulation inhibits rNAV differentiation into aNAV and DN2 cells. IL-4 (the canonical GC-associated Th2 cytokine) also inhibits aNAV/DN2/PC generation when substituted for IFN-γ. These inhibitory effects demonstrate that GC and EF pathways are not merely parallel — they are antagonistically regulated. Conditions that promote GC entry (CD40L, IL-4) suppress EF differentiation (TLR7, IFN-γ, IL-21) and vice versa (see Jenks2018 - DN2 B Cells and EF Pathway in SLE, in vitro differentiation).

  • ZEB2 represses GC differentiation: Zeb2, the primary TF driving ABC/DN2 formation, represses Mef2b — a TF required for GC differentiation. This provides the first direct molecular mechanism linking EF commitment to GC exclusion: cells expressing high ZEB2 are actively prevented from entering GC reactions (see Sanz2025 - Human Atypical B Cells Overview, review citing Dai et al. 2024, Gao et al. 2024).

  • GC-independent autoimmunity is confirmed by monogenic evidence: TLR7 gain-of-function mutations cause human SLE with expanded ABC/DN2. In mice, the orthologous mutation causes autoimmunity that is GC-independent — ABC expansion and pathology proceed normally even when GC reactions are blocked (see Sanz2025 - Human Atypical B Cells Overview, review citing Brown et al. 2022).

  • ABC sustain GC responses in chronic infection: Paradoxically, Zeb2-driven CD11c⁺ B cells may sustain GC responses in chronic infections by functioning as APCs for GC TFH induction. Excessive ABC activity leading to abnormal TFH regulation has been proposed as a mechanism of defective antigen-specific GC responses (see Sanz2025 - Human Atypical B Cells Overview, review citing Gao et al. 2024, Zhang et al. 2019).

  • GC loss confirmed histopathologically in fatal COVID-19: Post-mortem thoracic lymph nodes and spleens from COVID-19 patients showed complete absence of germinal centers, with marked reduction in Bcl-6⁺ GC B cells (LN: p<0.001; spleen: p<0.01 vs. controls) despite quantitative preservation of AID⁺ B cells. The specific block was in Bcl-6⁺ GC-type TFH differentiation — CD4⁺CXCR5⁺ pre-GC TFH were present but reduced, while CD4⁺Bcl-6⁺ GC-TFH were near-absent (LN: p<0.001; spleen: p<0.01). Aberrant TNF-α accumulation in both follicular and extra-follicular zones was implicated as the mediator — murine precedent shows TNF-α blockade rescues GC formation. FDCs were preserved, ruling out stromal destruction. This tissue-level evidence provides the anatomical basis for the peripheral EF dominance observed by Woodruff2020 - EF B Cell Responses in COVID-19 (see Kaneko2020 - GC Loss and TFH Block in COVID-19, n=11 COVID + controls, multi-color immunofluorescence).

  • EF dominance in severe COVID-19 is not due to sampling timing: CoV-A (EF-high) and CoV-B (EF-low) clusters had similar sampling times post-symptom onset, ruling out the possibility that EF dominance simply reflects early kinetics before GCs have had time to form (see Woodruff2020 - EF B Cell Responses in COVID-19).

  • Concurrent GC activity alongside dominant EF pathway in acute dengue — tentative: Despite ~75% of activated CD4⁺ T cells being CXCR5⁻PD-1⁺ Tph (extrafollicular), plasma CXCL13 is elevated in acute dengue. However, CXCL13 is not GC-specific — Tph cells themselves are a significant source of CXCL13 (Ansari2025 scRNA-seq), so elevated CXCL13 cannot be used as evidence of concurrent GC activity without independent GC markers (e.g., Bcl-6⁺ Tfh, GC B cell histology). Whether EF and GC pathways truly operate simultaneously in dengue remains unestablished, in contrast with the clear GC suppression in fatal COVID-19 (Kaneko2020 - GC Loss and TFH Block in COVID-19). The Sanz2025 endotype concept allows for mixed EF+GC responses, but this has not been demonstrated in dengue (see Ansari2025 - Peripheral T Helper Subset Drives B Cell Response in Dengue, n=170 acute dengue, plasma CXCL13 ELISA).

  • Convergent CDR3 mutation levels compatible with prior GC transit: In the first BCR repertoire analysis of acute dengue, convergent CDR3-bearing B cells carried 4.4–6.9% V gene mutation — within the range expected for GC-experienced memory B cells. The convergent CDR3s were more prevalent in secondary dengue, consistent with recall of previously GC-matured memory clones. However, the intermediate mutation level does not exclude contributions from EF maturation, as EF SHM is now documented (see Parameswaran2013 - Convergent Antibody Signatures in Dengue, 454 pyrosequencing, n=60 dengue patients — unsorted PBMCs, not cell-type resolved).

  • Delayed DENV-specific MBC peaks suggest prolonged GC reactions in 2° dengue: Peak frequency of several DENV-specific MBC subsets (IgD+/IgM+, IgD⁻ MBC, IgG+, atypical) occurred >3 months post-infection in 2° but not 1° dengue immunity (p<0.05 for 3 subsets). The authors cite Turner et al. (2021) demonstration of GC reactions persisting ~8 months post-SARS-CoV-2 vaccination as precedent. An uptick from 12–18M in 2° cases further suggests either ongoing GC output, tissue redistribution, or subclinical boosting in an endemic setting (see Singh2026 - DENV-Specific Memory B Cell Subsets, n=4/group — small sample).

  • Comparable VH mutation in dengue PBs and MBCs — consistent with both populations having GC history: Appanna2016 found that PB- and MBC-derived antibodies had comparable VH nucleotide mutation rates. The authors initially hypothesised MBCs would show higher SHM (consistent with more extensive GC maturation), but neither population showed significantly more mutation than the other. This is consistent with both populations deriving from memory B cells that had undergone similar prior GC experience, then being activated through different pathways during reinfection — one (E-specific IgG) feeding the PB wave, the other (prM/complex epitope) emerging at convalescence. The similar mutation levels do not resolve whether PBs transit through GCs during the acute response but are compatible with the Ansari2025 model of EF recall of pre-matured memory cells (see Appanna2016 - Plasmablasts as Subset of Memory B Cell Pool, n=12 dengue, IMGT analysis).

  • Low SHM in acute dengue IgG B cells argues against dominant GC origin of the plasmablast wave: GodoyLozano2016 found paradoxically low global SHM during acute dengue, lower in DWS+ than DWS−, and lower in secondary than primary infections — the opposite of what GC-dependent memory recall would predict (increased SHM with antigen re-exposure). Influenza vaccination produced the expected GC pattern (SHM increase at day 7). The authors propose that a rapid GC-independent pathway operating concurrently with the GC pathway is responsible, consistent with the concurrent Tph + CXCL13 activity in Ansari2025 (see GodoyLozano2016 - Lower IgG SHM Rates in Acute Dengue, n=19 acute dengue + 10 TIV controls, 454 pyrosequencing).

  • LANDMARK: GCs are not required for somatic hypermutation. William2002 demonstrated that in MRL/lpr mice, RF B cells underwent active SHM at the T zone–red pulp border at rates comparable to GC mutation (~0.3 mut/gene/generation), while Id⁻ GCs in the same spleens contained no mutating RF B cells. Some mice had ongoing EF mutation with no GCs of any type. This is qualitatively different from the low-level, unidentified-site mutation in severely immunodeficient mice (CD40L⁻/⁻, LTα⁻/⁻) — AM14 RF B cells mutate at high rates at a defined anatomical site in immunocompetent mice that retain the capacity for normal GC formation. The paper proposes that SHM is induced by sufficient B cell cycling in the presence of antigen + T cell signals, regardless of anatomical location (see William2002 - Extrafollicular Somatic Hypermutation in Autoimmune Mice, in vivo murine model, 8 mice, 305 sequences).

  • IgM⁺ memory B cells re-initiate GCs while IgG⁺ memory cells yield PBs: The isotype-fate segregation model (Seifert et al. 2015, cited in Bhattacharya2016 - Memory B Cell Subset Selection in Secondary Dengue, commentary) proposes that IgM⁺ memory B cells preferentially re-initiate germinal center reactions for affinity maturation toward new pathogens, while IgG⁺ memory cells are predisposed to plasmablast/plasma cell differentiation. In dengue, this predicts that the IgM⁺ DENV-binding MBCs identified by Appanna2016 and Singh2026 may re-enter GCs upon re-exposure, contributing to the prolonged GC activity suggested by CXCL13 elevation (Ansari2025) and delayed MBC peaks (Singh2026) — while E-specific IgG⁺ memory feeds the acute PB wave.

  • All non-naive alternative lineage clusters show significant SHM — consistent with post-GC origin: BCR analysis from Smart-seq2 scRNA-seq (n=11, 163 cells) showed that all memory and atBC clusters within the alternative lineage carry somatic hypermutation. Combined with the absence of PC maintenance genes in these clusters, this suggests the alternative lineage comprises post-GC cells that do not proceed to plasmablast fate in healthy/infection contexts — they exited GC reactions and adopted the T-bet⁺/CD11c⁺ programme without committing to terminal differentiation (see Sutton2021 - Alternative Lineage B Cells in Vaccination and Infection, Smart-seq2, n=11, 163 cells — low-throughput, limiting statistical power).

  • High SHM in sorted secondary dengue PBs argues FOR GC-experienced memory origin: Priyamvada2016 found mean 18.1 VH mutations per plasmablast (range 5–39), significantly higher than IgG⁺ GC B cells (p<0.005) and comparable to influenza recall responses. CDR R:S ratios >2.9 confirmed antigenic selection. This SHM level is best explained by prior GC transit of the memory B cells that gave rise to these PBs — contrasting with GodoyLozano2016’s low-SHM finding in bulk IgG and suggesting two concurrent populations: GC-experienced memory recall (high SHM) and de novo EF differentiation (low SHM) (see Priyamvada2016 - Cross-Reactive Memory Plasmablasts in Secondary Dengue, n=4 secondary DHF, single-cell BCR sequencing).

  • ABCs are (at least partly) GC-experienced — and T-bet⁺ B cells can promote GC formation in autoimmunity. A GC-experienced origin is proposed for at least a subset of ABCs (diverse SHM⁺ repertoire, antigen-driven activation), distinguishing them from the strongly EF-tied DN2/DN3 subsets. Moreover, in autoimmunity, elevated T-bet in B cells drives not only increased antibody production and antigen presentation but also germinal-center formation — so the atypical/ABC cluster can be both a product of, and a contributor to, GC reactions, not solely an EF phenomenon (see Lamprinou2026 - ABCs and DN B Cells, opinion, citing Cancro 2020 / Rubtsov 2017). This complements the wiki’s existing Sanz2025 evidence that ABCs can sustain GC responses as APCs.

  • ★ GC cell-fate decisions are probabilistic, not deterministic — and the GC-derived memory pool is deliberately LOW-affinity. GC B cells choose between four fates (apoptosis, dark-zone reentry, gcMBC differentiation, gcPC differentiation), positioned along a single axis of increasing positive selecting-signal strength: death < gcMBC < DZ reentry < gcPC. Boundaries are “strongly blurred” by stochastic events — variable access to antigen and T cell help, and “jackpot” acquisition of antigen aggregates or FDC fragments. Consequently the gcMBC compartment is seeded across a broad affinity spectrum including very low-affinity clones, giving it greater clonal diversity than the gcPC compartment, which is affinity-enriched (see Glaros2025 - Multilayered Identity of B Cell Memory, review, no original data, citing mouse GC imaging/transfer studies). The review argues this diversity is the point: protective antibodies against flavivirus and influenza variants arise primarily from rapid MBC→PC differentiation rather than from LLPCs established in primary infection — directly relevant to dengue serotype cross-reactivity.

  • gcMBC differentiation is a quiescence programme, not an effector programme. It comprises downregulation of the GCBC transcriptional programme, transition to quiescence, and upregulation of pro-survival factors. The translating node is BACH2: lower-affinity light-zone GC B cells receive weaker T cell help → reduced mTORC1 signalling → BACH2 upregulation → quiescence, decreased MYC, and active repression of Prdm1 (BLIMP-1). The review notes this mirrors the BACH2–BLIMP1 axis in T cells — “a conserved regulatory circuit governing memory versus effector differentiation across the B and T branches” (review). Additional nodes: BCL2 (pro-survival, repressed by BCL6), HHEX (downregulates BCL6, releasing Bcl2 repression), ZBTB18 (IL-9-induced), MYC, IL-4. (These molecular regulators are covered descriptively here; the wiki deliberately does not create individual pages for them — see Glaros2025 - Multilayered Identity of B Cell Memory Relevance & Notes.)

  • The light zone appears to be the site of gcMBC commitment. MBC-like cells with quiescent G₀ status, naive/memory-typical surface markers, and reduced BCL6 are consistently enriched among GC B cells with a light-zone-associated phenotype; CXCR4-deficient GC B cells (confined to the LZ) show enhanced MBC output (see Glaros2025 - Multilayered Identity of B Cell Memory, review, mouse).

  • ★ GCs are not required to generate memory — and the wiki’s GC/EF antagonism needs qualifying. Mice lacking BCL6 in the haematopoietic compartment and patients with CD40L deficiency cannot form functional GCs yet still generate detectable MBC populations; early antigen-specific MBC-phenotype cells appear before the first GC B cells form (see Glaros2025 - Multilayered Identity of B Cell Memory, review). Combined with the fate-mapping finding that GC-independent early MBCs outnumber GC-derived MBCs, this reframes the GC as the source of high-affinity, class-switched, high-SHM memory rather than of memory as such.

  • CSR happens mostly before the GC. The long-standing use of class-switch status as indirect evidence of GC participation rests on an assumption newer work overturns: CSR takes place primarily prior to GC entry, with switched cells subsequently gaining a competitive advantage inside the GC and thereby becoming enriched over the course of the reaction (see Glaros2025 - Multilayered Identity of B Cell Memory, review). Switching is therefore a pre-GC event whose products are selected in GCs — see Class Switch Recombination.

  • Short-lived GCs can form against T-independent antigens and can support generation of gcMBCs and gcPCs, though the mechanisms and functional significance remain to be elucidated (see Glaros2025 - Multilayered Identity of B Cell Memory, review). Relevant to the assumption that GC output implies T-dependent help.

  • ★ The GC-origin case for ABCs is inference, not demonstration — Cancro’s own words. The evidence usually cited for ABCs being GC-experienced — somatic mutation in sorted ABC Ig genes, plus a requirement for MHC-II, CD40, and CD154 signalling (neither MHC-II-deficient nor CD40-deficient FO B cells yield ABCs; CD154-deficient mice fail to develop natural ABCs with age) — is, in the author’s own assessment, inference: “both the requisite for cognate T cell interactions and the assertion that ABCs have entered GCs are based on inference, but have not yet been directly demonstrated.” Anatomic localisation of ABCs within GCs has never been shown, even though T-bet⁺ B cells express the GC-associated markers PNA and CD95 during the first weeks of a response (see Cancro2020 - Age-Associated B Cells, review — no original data; mouse). This directly qualifies this page’s existing “ABCs are (at least partly) GC-experienced” bullet from Lamprinou2026 - ABCs and DN B Cells — both draw on the same somatic-mutation evidence but reach different confidence levels about what it establishes.

  • The ABC phenotype is achievable without CD40 ligation — bystander IFN-γ may suffice. Even though CD40/CD154 signalling is required for the natural, age-driven accumulation of ABCs, the ABC phenotype itself can be induced without CD40 ligation if sufficient bystander cytokine — particularly IFN-γ — is present (citing Naradikian 2016; Zumaquero et al. 2019 eLife, human T-bet^hi B cells; plain text, no wiki page) (see Cancro2020 - Age-Associated B Cells, review — no original data; mouse + human, in vitro). This is the mechanistic basis for treating the CD40 requirement as necessary for the natural pathway but not sufficient evidence that ABCs are GC-transited.

  • A genuinely GC-associated ABC route exists and does not undercut the inference caveat above: B6.Sle1 ABC-phenotype cells emerge in IFN-γ-dependent spontaneous GCs. In the B6.Sle1 lupus-prone model, ABC-phenotype cells emerge within IFN-γ-dependent spontaneous GCs — a route that is genuinely GC-associated rather than inferred. The two observations (GC entry as inference in the general case; a demonstrated GC-associated route in one autoimmune model) are treated as compatible rather than contradictory: ABCs likely arise via a combination of formative routes whose relative contribution varies with context (see Cancro2020 - Age-Associated B Cells, review — no original data; mouse).

  • ★ “Premature exit from the GC reaction” is a live hypothesis for the origin of DN1 cells — and the evidence is a mutation-load sandwich. DN1 cells express CXCR5 (follicle-homing competent); the number of cell divisions and Ig mutations in IgG⁺CD27⁻ DN cells resembles GC B cells, while the mutation load sits below switched memory cells. Class-switched (IgA⁺/IgG⁺) CD27⁻ B cells carry significantly more VH mutation than naive and unswitched memory cells but less than switched memory. The reading: these cells entered the GC and left before accumulating a full mutation load — i.e. they are GC alumni, not GC avoiders (see Beckers2023 - Origins and Functions of DN B Cells, review, citing Berkowska 2011 for the division/mutation comparison with GC B cells, and Wu 2011 for the isotype-specific mutation levels).

  • CDR3 physicochemistry places DN between naive and switched memory. DN CDR3 regions are smaller, more hydrophilic and more basic than those of naive cells, but larger, more hydrophobic and more acidic than those of unswitched and switched memory cells — an independent repertoire-level signal consistent with partial GC passage (see Beckers2023 - Origins and Functions of DN B Cells, review, citing Wu 2011; see BCR Sequencing).

  • ⚠ The counter-evidence: one repertoire study found NO GC-mechanism difference between DN and switched memory. An AIRR analysis reported no significant differences in VH gene family usage or in the types and locations of Ig mutations between class-switched DN and switched memory cells, concluding that most class-switched DN cells are memory cells that acquired their mutations by the same GC-derived mechanism as SM cells. On that reading DN cells are not premature GC emigrants at all, but ordinary GC output that subsequently lost CD27 (see Beckers2023 - Origins and Functions of DN B Cells, review, citing Wu 2011).

  • Class switching and somatic hypermutation occur outside follicles — so neither is proof of GC passage. The review states this directly as the enabling premise for the GC-independent origin hypothesis, which is the same caution the wiki already carries from Glaros2025 - Multilayered Identity of B Cell Memory on the CSR→GC proxy (see Beckers2023 - Origins and Functions of DN B Cells, review; see Class Switch Recombination, Somatic Hypermutation).

  • “Follicular response” is not a synonym for “GC response” — the follicle is much larger than the GC within it. The consensus Perspective flags this as an active source of confusion, since a secondary follicle contains a GC plus a surrounding mantle zone of IgD⁺ naive B cells, and a primary follicle has no GC at all. A B cell inside a follicle is therefore not necessarily in a germinal centre (see Eisenbarth2025 - A Roadmap for Defining Extrafollicular B Cell Responses, consensus Perspective, 12 authors, no primary data).

  • ⚠ GC-derived plasma cells complete their maturation outside the follicle, in the splenic bridging channels and red pulp or the LN medulla, and GC-derived plasmablasts can still undergo T cell-based selection in the LN medulla after GC exit. These are the same anatomical sites that define classical extrafollicular responses. Locating an ASC at an “EF site” therefore does not establish an EF origin (see Eisenbarth2025 - A Roadmap for Defining Extrafollicular B Cell Responses, consensus Perspective, 12 authors, no primary data).

  • T-bet is expressed by GC B cells and GC-derived memory B cells, so T-bet positivity does not exclude GC transit; likewise, CD95 is present on both GC and memory B cells, and even the well-vetted GC shorthand (GL7, CD95, BCL6) requires marker combinations rather than any single marker (see Eisenbarth2025 - A Roadmap for Defining Extrafollicular B Cell Responses, consensus Perspective, 12 authors, no primary data).

  • Mutational load indicates a probability of GC participation, not a fact. SHM occurs outside GCs, AID is also required for class switching (which typically precedes GC entry), and mutational content is the product of SHM rate and time rather than of GC transit as such. In prolonged GC-independent responses the load approaches GC levels (see Eisenbarth2025 - A Roadmap for Defining Extrafollicular B Cell Responses, consensus Perspective, 12 authors, no primary data). See Somatic Hypermutation and GC-Independent Response.

  • ⚠ [2026-08-27] A tissue study of “extrafollicular” B cells that stains for Bcl6 and never reports the result. Allard-Chamard2023 - DN3 B Cells Infiltrate Inflamed Tissues lists anti-Bcl6 (clone LN22) in its immunofluorescence antibody table but reports no Bcl6 data anywhere in the paper, and performs no follicular-versus-extrafollicular quantification in the COVID-19 lymph nodes it images. Its “extrafollicular” designation is therefore a presumed origin, not an imaged location — the exact usage Eisenbarth2025 - A Roadmap for Defining Extrafollicular B Cell Responses warns against. The GC status of those specimens is known only because a companion study, Kaneko2020 - GC Loss and TFH Block in COVID-19, documented loss of germinal centres and Bcl-6⁺ Tfh cells in the same rapid-autopsy cohort — which is what makes a GC-independence argument by exclusion available at all (see Allard-Chamard2023 - DN3 B Cells Infiltrate Inflamed Tissues, n=6 autopsy). See GC-Independent Response.

  • GC cells are the one human B cell population the review says blood cannot show you. “With the possible exception of GC cells, all human B cell populations found in lymphoid tissues can also be demonstrated in the peripheral blood lymphocytes” — a structural reason why blood panels cannot adjudicate GC-versus-EF questions directly. Two related points: the Bm1–Bm5 scheme fails to separate transitional (IgD⁺CD38ʰⁱ) from pre-GC (Bm2′) cells in blood; and Table 1 assigns activated naive cells the function “precursor of short-lived PB and GC reactions”, i.e. the same phenotype feeds both pathways (see Sanz2019 - Consistent Classification of Human B Cell Populations, review — no original data).

  • T-bet⁺ B cells are not GC-excluded — mouse primary data. A Tbx21-ZsGreen reporter after influenza infection was expressed by NP-specific germinal centre B cells as well as ASCs and memory B cells (see Stone2019 - T-bet Promotes ASC Differentiation by Limiting IFN-gamma Inflammation, mouse, reporter flow cytometry, n=3–6/group). This supplies primary-data support for the point Eisenbarth2025 - A Roadmap for Defining Extrafollicular B Cell Responses makes on consensus grounds: T-bet positivity is not evidence of a GC-independent origin. Any inference from B cell T-bet expression to extrafollicular derivation is therefore unsafe on its own — see GC-Independent Response, T-bet.

  • ★ Four orthogonal demonstrations that a T-bet⁺CD11c⁺ B cell response develops without transiting a germinal centre. (i) Minimal coexpression of the GC markers GL-7 and EphrinB1 at days 6, 10 and 15; (ii) sorted-population RNA-seq separating the cells from GC B cells by PCA, with Bcl6, S1pr2 and Aicda downregulated; (iii) histocytometry placing ~75% in the follicular mantle and ~0% inside GCs at day 12; (iv) S1pr2-CreERT2 fate mapping with tamoxifen from day 4 — >80% unlabelled (17.5% TdTomato⁺) against 70.0% of GL-7⁺ early GC B cells. A mixed bone marrow chimera added that B-cell-intrinsic Bcl6 is dispensable (~1:1 vs ~5:1 in GCs), and Ig repertoires overlapped by <10% of sequences with early lineage-tree bifurcation. Replicated in influenza PR8, which preserves splenic architecture (see Song2022 - Tfh Outside Germinal Centers Drive T-bet CD11c B Cells, mouse, n=3–5 mice/group; Ig-seq n=3).

Contradictions & Debates

  • The DN1/DN2 subdivision resolves much of the original debate: DN2 cells are EF-derived (TLR7-dependent, CD40L-inhibited), while DN1 cells are GC-associated (SWM-like transcriptome, CD40L-tolerant). The remaining question is whether DN1 cells complete full GC reactions or represent early GC emigrants.
  • ABC can be both GC-sustaining and GC-excluded: Sanz2025 reveals a paradox — Zeb2-driven ABC/DN2 cells are molecularly excluded from entering GCs (Zeb2 represses Mef2b), yet ABC can sustain GC responses as APCs. These functions need not be contradictory if ABC provide TFH support from extrafollicular positions without themselves entering GCs. [2026-08-16] Now with a functional genetic result on the GC-sustaining side: B-cell-specific Zeb2 deletion decreases GC B cell numbers during persistent Plasmodium infection, indicating ZEB2-driven atypical B cells are required to sustain that GC response (see Glaros2025 - Multilayered Identity of B Cell Memory, review, citing Gao 2024 Sci Immunol, mouse conditional-KO). The cell-intrinsic-exclusion / cell-extrinsic-support reconciliation above remains the best available, but it is not experimentally established. Tracked in full on ZEB2.
  • ★ Does the wiki over-read “low SHM” as evidence of GC bypass? Several wiki claims infer GC-independence from low mutation load — most consequentially the dengue SHM paradox (GodoyLozano2016 - Lower IgG SHM Rates in Acute Dengue). Glaros2025 - Multilayered Identity of B Cell Memory weakens that inference twice over: (a) gcMBCs are specifically enriched for low-affinity, and by extension often low-SHM, clones — the GC deliberately exports a diverse, unselected memory tail alongside its affinity-matured PCs; and (b) fate mapping shows SHM loads overlap substantially between GC-derived and GC-independent memory, with “some gcMBCs displaying SHM levels similar to those of eMBCs and vice versa.” Low SHM is therefore consistent with but not diagnostic of a GC-independent origin. This does not overturn the dengue finding — the Monte Carlo ASC-dominance argument in GodoyLozano2016 is independent of this inference — but any wiki bullet resting on mutation load alone should be read as suggestive rather than demonstrative.

Extrafollicular Response, Somatic Hypermutation, Class Switch Recombination, Memory B Cell, Double-Negative B Cell, Age-Associated B Cell, DN2 B Cell, CD27, T-bet, ZEB2, BACH2, BLIMP-1, Bcl-6, Plasmablast, Early Memory B Cell, GC-Independent Response

Sources

Graph View

Backlinks