EFB Dengue Literature Review

Double-Negative B Cell

18 min read

Double-Negative B Cell

Overview

This page is the canonical entry for the IgD⁻CD27⁻ (double-negative) population specifically. For the broader atypical / age-associated B cell umbrella — the synonymy map across the “atypical,” “ABC,” “T-bet⁺,” “CD11c⁺,” and “alternative lineage” labels, and how they route to the precise DN sub-populations — see Atypical B Cell.

Double-negative (DN) B cells are a peripheral blood memory B cell subset defined by the co-absence of surface IgD and CD27 (IgD⁻CD27⁻ CD19⁺). They are distinct from both conventional CD27⁺ memory B cells (switched and unswitched) and from naive B cells (IgD⁺CD27⁻). Despite lacking CD27 — long considered a universal memory B cell marker — DN B cells carry hallmarks of antigen-experienced memory: somatic hypermutation of VH genes, inability to extrude Rhodamine 123, and proliferative responses to TLR9 stimulation (CpG DNA) without BCR crosslinking.

In healthy peripheral blood, DN B cells are a minor population (~5% of CD19⁺ B cells). They are substantially expanded in systemic lupus erythematosus (SLE) and, by extension, serve as the foundational reference population for the “atypical B cell” or “T-bet⁺ B cell” expansions subsequently described in acute infections including malaria, SARS-CoV-2, Ebola, and — of primary interest to this wiki — dengue.

DN1/DN2/DN3 subdivision: Jenks et al. (2018) resolved the DN compartment into two functionally distinct subsets: DN1 (CXCR5⁺, CD21⁺, CD11c^lo) and DN2 (CXCR5⁻, CD21⁻, CD11c⁺, CD19^hi). DN1 cells predominate in healthy donors and transcriptionally resemble switched memory cells (only 22 DEGs by RNA-seq); DN2 cells predominate in active SLE and represent extrafollicular pre-plasmablasts. These subsets belong in separate differentiation pathways — DN1 are likely early SWM precursors that have not yet acquired CD27, while DN2 are effector cells derived from activated naive B cells via the EF pathway. A third subset, DN3 (CXCR5⁻, CD21⁻, CD11c⁻, T-bet⁻), was subsequently described in acute COVID-19 and active SLE, representing pre-plasmablasts distinct from both DN2 and ABC (see DN3 B Cell). See DN2 B Cell for the full DN2 characterisation.

A four-subset scheme (DN1–DN4): One lineage of the literature (largely the Sachinidis/Garyfallos group, building on Somers 2022 / Castleman 2022) extends the DN compartment to four subsets classified by CXCR5 / CD11c / T-bet: DN1 (CXCR5⁺CD11c⁻T-bet⁻), DN2 (CXCR5⁻CD11c⁺T-bet⁺), DN3 (CXCR5⁻CD11c⁻T-bet^low), and DN4 (CXCR5⁺CD11c⁻T-bet⁻). DN1 and DN4 are CXCR5⁺ (follicle-competent) whereas DN2 and DN3 are CXCR5⁻ and are the subsets most tied to extrafollicular responses and autoimmunity. DN4 is the least-defined subset: associated with allergic reactions, expressing Notch-signalling and protein-ubiquitination genes that distinguish it from DN1 (Allard-Chamard 2023). The wiki’s default remains the three-subset DN1/DN2/DN3 scheme (after Sanz2025); the DN4 designation is noted here as nomenclature drift rather than an independently validated fourth lineage (see Lamprinou2026 - ABCs and DN B Cells, opinion; and Atypical B Cell Contradictions).

The “atypical B cell” label is obsolete: Sanz (2025) argues that the term AtB is misleading because: (1) cells thus labelled are a normal component of immune responses, not atypical; (2) the actual nature, derivation, and function of different AtB categories depend on the immunological context; (3) inconsistent classification schemes (CD27⁻, CD21lo, CD11c⁺, T-bet⁺, FcRL5⁺ — used alone or in combinations, often without IgD) conflate fundamentally different populations. The DN nomenclature (IgD⁻CD27⁻, subdivided by CXCR5/CD21/CD11c into DN1, DN2, DN3) is recommended as the more precise classification (see Sanz2025 - Human Atypical B Cells Overview, invited review).

Key Points from Literature

  • DN B cells represent mean 4.6 ± 1.8% of PBL CD19⁺ B cells in healthy subjects (n=29), always below 10% (see Wei2007 - DN Memory B Cells in SLE, n=29 cross-sectional).

  • In SLE, DN cells exceed 10% of CD19⁺ PBL B cells in 50% of patients (mean 19.4% in DN-high subgroup; up to >40% in individual patients); in 25% of SLE patients they outnumber CD27⁺ memory cells (see Wei2007 - DN Memory B Cells in SLE, n=36 SLE cross-sectional).

  • DN expansion is SLE-specific: not elevated in rheumatoid arthritis (n=45) or chronic hepatitis C (n=7) (see Wei2007 - DN Memory B Cells in SLE, cross-sectional).

  • Somatic hypermutation: IgG⁺ DN cells carry ~3.2% nucleotide mutation rate in healthy donors and ~2.6% in SLE; CD27⁺ IgG⁺ memory cells carry ~5.4% (healthy) and ~5.1% (SLE). DN cells are thus antigen-experienced but less mutated than their CD27⁺ counterparts (see Wei2007 - DN Memory B Cells in SLE, VH3 family analysis).

  • Isotype composition: ~44% IgG⁺, ~15% IgM⁺ in healthy DN cells; DN IgM⁺ cells notably lack IgD co-expression, unlike the majority of CD27⁺ nonswitched memory cells (see Wei2007 - DN Memory B Cells in SLE, cross-sectional).

  • DN cells fail to extrude Rhodamine 123 (identical to CD27⁺ memory cells; naive B cells extrude it), attributable to absence of ABCB1 transporter (see Wei2007 - DN Memory B Cells in SLE, in vitro functional assay).

  • DN cells proliferate in response to CpG2006 without BCR crosslinking; naive B cells require BCR co-stimulation. Proliferating DN cells upregulate CD27 (see Wei2007 - DN Memory B Cells in SLE, in vitro functional assay).

  • FcRH4 expression: Peripheral blood DN cells are FcRH4⁻ (both healthy and SLE), distinguishing them from a tissue-resident tonsillar CD27⁻ population that expresses FcRH4 (see Wei2007 - DN Memory B Cells in SLE, cross-sectional; see also FcRH4).

  • CD38 level: DN cells express CD38 at Bm5/early-Bm5 levels — below pre-GC cells, transitional cells, and plasmablasts (see Wei2007 - DN Memory B Cells in SLE).

  • Surface phenotype (CD24, IgM, CD10, B220) is virtually identical between DN cells and conventional CD27⁺ switched memory cells; both populations lack CD10, distinguishing them from transitional and pre-GC B cells (see Wei2007 - DN Memory B Cells in SLE, 8-color flow cytometry).

  • Tonsil counterpart: the IgD⁻CD27⁻CD10⁻ fraction of Bm5 cells; contains both FcRH4⁺ and FcRH4⁻ subsets, unlike PBL DN cells which are uniformly FcRH4⁻ (see Wei2007 - DN Memory B Cells in SLE).

  • Clinical association in SLE: DN-high patients (>10%) have higher nephritis rates (p=0.025), anti-dsDNA (p=0.001), anti-RNP/Sm (p=0.009), SLAM disease activity (p=0.02), and elevated 9G4 autoreactive B cells (see Wei2007 - DN Memory B Cells in SLE, n=46 SLE patients).

  • 9G4 autoreactive B cells (encoded by VH4-34) are present at similar frequencies within the DN and CD27⁺ switched memory compartments of individual SLE patients — suggesting autoreactive specificities are not excluded from the DN compartment (see Wei2007 - DN Memory B Cells in SLE).

  • Independent cohort replication of DN expansion: In a separate clinical cohort (n=8 SLE, n=7 controls), DN cells were 14.4 ± 7.9% of CD19⁺ PBL B cells in SLE vs. 3.9 ± 1.9% in controls (P=0.01) — corroborating the Wei2007 frequency estimates with a matched control comparison (see Anolik2004 - Rituximab and B Cell Abnormalities in SLE, phase I/II trial).

  • Quantitative correlation with autoantibody titers: DN expansion correlates with VH4.34 IgG autoreactive antibody levels (R²=0.8, P<0.05) — a stronger correlation than that seen for naive lymphopenia (R²=0.6 for VH4.34). This positions DN expansion as the closer correlate of autoreactive B cell biology (see Anolik2004 - Rituximab and B Cell Abnormalities in SLE).

  • Reversibility after B cell depletion: After effective rituximab-mediated B cell depletion and immune reconstitution (≥1 year), DN expansion resolved significantly (P=0.05 vs. baseline) in patients with effective depletion. Patients with incomplete depletion did not show resolution. This demonstrates that DN accumulation is driven by ongoing B cell dysregulation, not irreversible programming (see Anolik2004 - Rituximab and B Cell Abnormalities in SLE).

  • DN compartment composition in severe COVID-19 mirrors active SLE: In critically ill COVID-19 patients (CoV-A cluster), DN composition was skewed heavily toward DN2 (80.3% of DN cells in a representative ICU patient vs. 9.5% in HD). DN1 contracted correspondingly (9.8% vs. 68.0% in HD). DN3 cells were also expanded (8.5% in ICU vs. 15.1% in HD). The overall DN compartment was greatly expanded (19.3% of CD19⁺ B cells in ICU vs. 3.0% in HD). Direct comparison showed DN profiles in CoV-A were highly similar to active SLE — both showed strong DN2 skewing with concordant DN1 contraction (see Woodruff2020 - EF B Cell Responses in COVID-19, 24-marker spectral FCM, n=10 ICU-C, n=7 OUT-C, n=17 HD, n=7 SLE).

  • DN2:DN1 ratio elevated in severe COVID-19 to SLE-comparable levels: log₂(DN2:DN1) ratios were significantly higher in both CoV-A and active SLE than in HD and CoV-B (P ≤ 0.0001 and P ≤ 0.001 respectively). DN2:DN1 ratio was not significantly different between CoV-A and SLE, confirming that the EF pathway achieves equivalent DN skewing in acute viral infection as in chronic autoimmunity (see Woodruff2020 - EF B Cell Responses in COVID-19).

  • Unswitched memory contraction accompanies DN2 expansion: usM cells were significantly reduced in CoV-A, a feature consistently observed in SLE and other autoimmune diseases. This suggests that EF pathway activation may come at the cost of the unswitched memory compartment (see Woodruff2020 - EF B Cell Responses in COVID-19).

  • FIRST DENGUE DATA — DENV-specific “atypical” (CD27⁻CD21⁻) MBCs accumulate in 2° dengue: DENV-specific atypical MBCs (CD20+/IgD⁻/CD27⁻/CD21⁻) — corresponding to DN B cells — were significantly higher in 2° than 1° dengue at early convalescence (p<0.01) and at 18 months (p<0.05), with a significant main effect of infection history across all timepoints (Two-way RM-ANOVA p<0.001). Resting MBCs (CD27+/CD21+) did not differ by infection history, indicating the DN/atypical compartment selectively expands with repeat DENV exposure. Temporal correlation analysis suggests DENV-specific atypical MBCs at acute/3M correlate with later class-switched and activated MBC levels in 2° immunity, implying functional responsiveness rather than exhaustion (see Singh2026 - DENV-Specific Memory B Cell Subsets, n=58 samples, 18 pediatric patients, conventional 12-color FCM).

  • Limitation — DN1/DN2/DN3 resolution not possible: The Singh2026 panel lacks CXCR5 and CD11c, so the DENV-specific “atypical” (CD27⁻/CD21⁻) population cannot be subdivided into DN1, DN2, or DN3. Per the Sanz2025 criterion, the expanded population could be EF-derived DN2 effectors, GC-derived DN1 memory, or a heterogeneous mix. The panel does include IgD, passing the Sanz2025 IgD audit (see Singh2026 - DENV-Specific Memory B Cell Subsets).

  • FIRST EVIDENCE OF CD21⁻CD11c⁺ EF B CELLS IN DENGUE: Within the IgD⁻CD27⁻ (DN) gate, CD21⁻CD11c⁺ B cells — phenotypically consistent with DN2 — are significantly expanded during acute dengue infection compared to healthy donors and convalescence. These cells are CD19^hi and CXCR5^lo (by inference from parallel T cell data). This is the first direct demonstration that the EF B cell phenotype characterised in SLE (Jenks2018) and COVID-19 (Woodruff2020) is also present in dengue (see Ansari2025 - Peripheral T Helper Subset Drives B Cell Response in Dengue, multi-color FCM, n=170 acute dengue).

  • EF B cell expansion driven by Tph-IL-21 axis: The CD21⁻CD11c⁺ B cell expansion occurs in the context of massive Tph (CXCR5⁻PD-1⁺) T cell activation providing IL-21. Blocking IL-21 reduces plasmablast output by ~60%. The Tph→IL-21→memory B cell→plasmablast axis represents the T cell help arm of EF activation in dengue (see Ansari2025 - Peripheral T Helper Subset Drives B Cell Response in Dengue).

  • TISSUE-LEVEL DN B CELLS IN COVID-19 — present at both follicular and extrafollicular sites: Post-mortem immunofluorescence of COVID-19 lymph nodes and spleens identified IgD⁻CD27⁻ DN B cells within and outside follicles, with T-B conjugates at both sites. Peripheral blood analysis using dual-fluorophore RBD probes confirmed that DN populations (DN2 and DN3 as CXCR5-low; DN1 and DN4 as CXCR5-high) contain SARS-CoV-2-specific cells, establishing that the DN expansion is antigen-driven. Switched memory cells were the largest RBD⁺ subset (53.6% convalescent, 39.1% severe) (see Kaneko2020 - GC Loss and TFH Block in COVID-19, tissue: n=11 COVID + controls, multi-color immunofluorescence; blood: n=68 patients, 13-color FCM).

  • CD21⁻CD27⁻ gating captures only 44.7% of transcriptomically-defined atypical B cells: CITE-seq (combined transcriptome + surface protein) on >12,000 B cells revealed that the conventional CD21⁻CD27⁻ flow cytometry gate captures fewer than half of the cells that cluster transcriptomically as atBC1 (the most prominent alternative lineage cluster). CD11c protein expression was a superior single marker for identifying the alternative lineage. This has major implications for all prior flow cytometry studies using the DN gate — including dengue studies — which likely underestimate the true size of this population (see Sutton2021 - Alternative Lineage B Cells in Vaccination and Infection, n=4, CITE-seq).

  • Alternative lineage present at ~20% of B cells in healthy non-exposed donors: Transcriptomic clustering identified alternative lineage cells (atBC1, atBC2, atBC3, MBC1) in non-malaria-exposed Australian donors at ~20% of total B cells — far above the ~5% typically reported for CD21⁻CD27⁻ DN cells by flow cytometry. The discrepancy is explained by the gating limitation above (see Sutton2021 - Alternative Lineage B Cells in Vaccination and Infection, n=4, 10x Chromium; validated by flow cytometry n=18).

  • MBC1 cluster — transcriptomic evidence for quiescent alternative lineage memory: The MBC1 cluster sits at the base of the alternative lineage pseudotime branch, expresses memory markers but lacks activation markers, and represents a quiescent memory state. This provides transcriptomic support for the “memory DN2” population predicted by Sanz2025 and Faliti2024 (see Sutton2021 - Alternative Lineage B Cells in Vaccination and Infection, n=4, 10x Chromium).

  • IgG3 overrepresented in alternative lineage in malaria-exposed donors: Alternative lineage B cells were enriched for IgG3 isotype compared to classical lineage cells in malaria-exposed Kenyan donors, consistent with the IgG3 enrichment on DN2 cells reported in SLE by Jenks2018 (see Sutton2021 - Alternative Lineage B Cells in Vaccination and Infection, n=4, 10x Chromium).

  • Only DN2 corresponds to the ABC/T-bet⁺ population — and only one ABC subset corresponds to DN. Among DN subsets, only DN2 highly expresses T-bet and CD11c and efficiently differentiates into plasma cells; DN1, DN3, and DN4 lack CD11c/T-bet and so fall outside the ABC definition. Conversely, ABC is a heterogeneous superset (CD27⁺ + IgD⁺ + predominantly IgD⁻CD27⁻ cells), so only its IgD⁻CD27⁻ fraction maps onto DN — CD27⁺ ABCs are excluded because DN is CD27⁻ by definition (see Lamprinou2026 - ABCs and DN B Cells, opinion, citing Jenks 2018 / Tangye 2023 / Rubtsov 2011). ABC and DN frequencies nonetheless correlate in both health and lupus (citing Sachinidis 2025 / Chizzolini 2024).

  • DN4 — a poorly defined, allergy-associated CXCR5⁺ subset. In the four-subset scheme, DN4 (CXCR5⁺CD11c⁻T-bet⁻) is allergy-associated and expresses Notch-signalling and protein-ubiquitination genes distinguishing it from DN1; being CXCR5⁺ and T-bet⁻, it is not an EF effector and not an ABC. Its evidence base is largely secondary/self-cited (see Lamprinou2026 - ABCs and DN B Cells, opinion, citing Somers 2022 / Castleman 2022 / Allard-Chamard 2023).

  • A cytoplasmic-FOXO1⁺ DN population in SLE awaits subset assignment. A DN B cell population marked by cytoplasmic FOXO1 (a TF central to B-cell development) has been described in SLE, but whether it corresponds to DN2, DN3, or a distinct subset is unknown (see Lamprinou2026 - ABCs and DN B Cells, opinion, citing Hritzo Ahye & Golding 2018).

Proposed Origin and Relationship to Extrafollicular Response

Wei et al. propose that DN cells represent B cells that failed to complete a productive germinal centre reaction and instead differentiated via extrafollicular pathways. The reasoning is:

  1. CD27 is normally acquired via CD40–CD154-mediated B–T cognate interactions within the GC.
  2. DN cells show lower SHM rates than CD27⁺ cells, consistent with less extensive GC passage or GC-independent hypermutation.
  3. Murine studies demonstrate that SHM can occur outside GCs at GC-comparable rates (see William2002 - Extrafollicular Somatic Hypermutation in Autoimmune Mice).
  4. CD11c⁺ dendritic cells can activate extrafollicular B cells and induce CD40-independent class switching via BLyS-BAFF-R interactions.

This EF origin model directly links DN B cells to the concept of Extrafollicular Response and makes them the prototypical human EF memory B cell population. See Wei2007 - DN Memory B Cells in SLE for the full mechanistic argument.

Resolution by Jenks2018 — DN1 and DN2 are separate lineages: The DN1/DN2 subdivision resolves the origin debate. DN1 cells (CXCR5⁺, TCF7⁺) share a near-identical transcriptome with SWM (22 DEGs) and likely represent early switched memory precursors that have not yet acquired CD27 — a GC-derived population. DN2 cells (CXCR5⁻, T-bet⁺, ZEB2⁺, TCF7⁻) are extrafollicular effector cells with a distinct transcriptional programme. CD40L stimulation inhibits rNAV differentiation into aNAV and DN2 but does not affect DN1 generation, supporting separate pathway origin. The original Wei2007 “EF origin” hypothesis applies to DN2 cells specifically, not to the undivided DN compartment (see Jenks2018 - DN2 B Cells and EF Pathway in SLE).

Relationship to atypical/T-bet⁺ B cells: The later literature (post-2010) increasingly characterises a CD21⁻CD27⁻ or IgD⁻CD27⁻ population in acute infections as “atypical B cells” or “age-associated B cells (ABCs)” with T-bet expression, FcRL5⁺, and CD11c⁺ features. These populations substantially overlap with DN2 B cells as defined by Jenks2018. The DN2 phenotype (CXCR5⁻, CD11c⁺, T-bet⁺, FCRL4⁻, FCRL5⁺) provides the most precise current definition and resolves several prior ambiguities in the field. DN2 cells lack FCRL4 (distinguishing them from HIV exhausted memory cells) but retain intact BCR signalling (distinguishing them from malaria atypical memory cells). See DN2 B Cell for full characterisation.

Contradictions & Debates

  • Wirths & Lanzavecchia (2005, Eur J Immunol — cited but not yet ingested) identified a CD27⁻ PBL memory population using R123 extrusion, but their population was almost exclusively IgG⁺ (no IgM memory cells) and represented only ~1% of PBL B cells. Wei et al.’s DN cells include IgM and IgA subsets and represent ~5% — the discrepancy may reflect differences in isolation method, R123 threshold, or healthy donor selection. This potential definitional inconsistency has not been resolved.

  • Whether DN cells represent a developmental dead-end (failed GC entrants) or a bona fide lineage with distinct progenitors remains unresolved. The CpG-driven CD27 upregulation by proliferating DN cells suggests plasticity rather than a fixed fate.

  • CD27 absence in DN cells may not always indicate failure to enter GCs — CD27 can be downregulated upon stimulation with CD70 (on activated T cells), TLR ligands, and cytokines. Some CD27⁻ memory cells may have originally expressed CD27 (see Sanz2025 - Human Atypical B Cells Overview, review).

  • Whether DN2 cells in different disease contexts (SLE, dengue, COVID-19, malaria) represent identical or merely phenotypically similar populations remains unresolved. The Sanz2025 review emphasises that context determines function: naïve-derived DN2 in primary responses vs. potential memory DN2 in recall settings.

  • Alternative lineage reframing challenges the EF pre-plasmablast model outside SLE: Sutton2021 scRNA-seq data show that no atBC cluster upregulates PC maintenance genes (XBP1, IRF4, PRDM1), and PCs are detached from the pseudotime manifold — arguing against DN/atBC cells being obligate pre-plasmablasts in healthy or infection contexts. However, Sutton’s own Discussion reconciles this as context-dependent: in SLE, chronic TLR7 stimulation can drive atBCs toward PC fate, while in vaccination/infection they remain within an alternative memory lineage. This reframes the DN2→PB pathway as a pathological possibility rather than a default property (see Sutton2021 - Alternative Lineage B Cells in Vaccination and Infection, n=4, 10x Chromium + pseudotime).

  • COVID-19 validates the SLE EF pathway in infection — but pathogenesis mechanism unclear: Woodruff2020 shows that CoV-A patients have DN profiles (DN2:DN1 ratio, DN composition) statistically indistinguishable from active SLE. This validates the SLE-derived EF model in acute viral infection, but whether EF activation drives pathogenesis or simply correlates with severity remains unresolved (see Woodruff2020 - EF B Cell Responses in COVID-19).

  • Distinction from acN cells: During SLE flares, the circulating pool contains both DN memory B cells (IgD⁻CD27⁻) and a newly characterised activated naive (acN) population (IgD⁺CD27⁻, CD19^hi, MTG⁺, CD24⁻). These are distinct: acN cells retain IgD surface expression (not yet class-switched), while DN cells are IgD⁻. The Tipton2015 data show that acN cells — not pre-existing DN memory cells — are the primary precursor of circulating ASCs during SLE flares (see Tipton2015 - ASC Diversity and Origin in SLE). Jenks2018 subsequently showed that aNAV cells share a near-identical transcriptome and phenotype with DN2 cells and demonstrated the developmental link: aNAV → DN2 → plasmablast (see Jenks2018 - DN2 B Cells and EF Pathway in SLE).

  • Are DN B cells and ABCs “two sides of the same coin”? The opinion that anchors this debate answers partially, and context-dependently: the IgD⁻CD27⁻ ABC subset ≈ DN2, but ABC is a broader population (also CD27⁺/IgD⁺) and DN is broader on its non-DN2 subsets — so neither label contains the other. Even within their T-bet⁺CD11c⁺ overlap, ABCs are transcriptomically distinct from DN2 (Maul 2021). The wiki therefore keeps DN and ABC as related-but-non-identical (see Lamprinou2026 - ABCs and DN B Cells, opinion; Age-Associated B Cell; Atypical B Cell).

  • Three-subset vs four-subset DN taxonomy. Sanz2025 (and most wiki pages) use DN1/DN2/DN3; the Sachinidis/Garyfallos lineage adds DN4 (allergy-associated, CXCR5⁺). This is granularity/nomenclature drift, not a factual disagreement, but consumers of dengue data should be aware that CXCR5⁺ DN subsets (DN1, DN4) are systematically discarded by CXCR5⁻-focused EF gating.

Atypical B Cell, Age-Associated B Cell, DN2 B Cell, DN3 B Cell, CD27, IgD, FcRH4, FCRL5, CD38, CD11c, CXCR5, T-bet, Memory B Cell, Extrafollicular Response, Germinal Center, Somatic Hypermutation, B220

Sources

Graph View

Backlinks