EFB Dengue Literature Review

DN3 B Cell

16 min read

DN3 B Cell

Overview

DN3 B cells are a subset of IgD⁻CD27⁻ (double-negative) B cells lacking the markers that define DN2 — CD11c and T-bet — while also being CD21⁻ and/or CXCR5⁻. They were described in acute COVID-19 and active SLE and are held to be pre-plasmablasts that accumulate in tissues in autoimmune fibrosis and severe COVID-19. DN3 cells are distinct from DN2 B Cells (CD11c⁺⁺ T-bet⁺⁺) and from DN1 cells (CXCR5⁺ CD21⁺).

⚠ Two different gates are both called “DN3.” The Emory/Sanz lineage gates IgD⁻CD27⁻CD11c⁻CD21⁻ (see Woodruff2020 - EF B Cell Responses in COVID-19, Sanz2025 - Human Atypical B Cells Overview); the Pillai/Ragon lineage gates IgD⁻CD27⁻CXCR5⁻CD11c⁻, with no CD21 in the panel at all (see Allard-Chamard2023 - DN3 B Cells Infiltrate Inflamed Tissues). These exclude different cells — one drops follicle-competent CXCR5⁺ cells, the other drops CD21-normal cells — and no published study has run both axes on the same sample. The composite “CXCR5⁻CD21⁻CD11c⁻T-bet⁻” that this page previously asserted is a merge of two schemes that no single paper gates on; it is retained below only as a description of the intersection, not as any study’s operational definition.

DN3 cells should not be classified as atypical B cells or age-associated B cells (ABC) despite their CD21lo phenotype, because they lack the defining ABC markers CD11c and T-bet.

Key Points from Literature

  • Phenotype: IgD⁻, CD27⁻, CXCR5⁻, CD21⁻, CD11c⁻, T-bet⁻. The key distinguishing feature from DN2 is the absence of both CD11c and T-bet — the defining markers of ABC/DN2 identity. DN3 cells also express low CD19 (see Sanz2025 - Human Atypical B Cells Overview, review citing Woodruff et al. 2020, Allard-Chamard et al. 2023, Perugino/Pillai et al.).

  • Pre-plasmablast identity: DN3 cells convincingly represent pre-plasmablasts that accumulate in tissues in autoimmune fibrosis and severe COVID-19. They correspond to the CD27⁻ CXCR5⁻ CD19⁻low plasmablast-low population described in SLE (Szelinski et al. 2022), and likely to the early pre-plasmablasts or ASC population 1 reported in tetanus vaccination responses in healthy subjects (see Sanz2025 - Human Atypical B Cells Overview, review).

  • Expanded in acute COVID-19 and SLE: DN3 cells were first reported in the Sanz lab’s acute SARS-CoV-2 study (Woodruff et al. 2020) and subsequently confirmed in active SLE. They represent a large fraction of DN cells in some SLE patients (see Sanz2025 - Human Atypical B Cells Overview).

  • Tissue infiltration: IgD⁻CD27⁻CXCR5⁻CD11c⁻ DN3 B cells infiltrate inflamed tissues in autoimmune fibrosis and severe COVID-19, consistent with a pre-plasmablast identity and tissue-homing capacity (see Sanz2025 - Human Atypical B Cells Overview, review citing Allard-Chamard et al.). The primary is now ingested — see Allard-Chamard2023 - DN3 B Cells Infiltrate Inflamed Tissues for the actual densities, and for the composition-vs-density qualifier the reviews omit. ⚠ Lexical correction under the [2026-08-27] decision: the primary’s title word “extrafollicular” is a presumed-origin label, not an imaged location — that paper performs no follicular-vs-extrafollicular quantification and reports no Bcl6 result despite staining for it. Tissue infiltration of a salivary gland or lung is not evidence of an extrafollicular focus.

  • Durability post-vaccination: Antigen-specific DN2 and DN3 cells persist for over a year post-SARS-CoV-2 mRNA vaccination, accounting for >50% of all spike/RBD⁺ cells (see Sanz2025 - Human Atypical B Cells Overview, review citing Faliti et al. 2024).

  • Not ABC despite CD21lo phenotype: The expression of low levels of both CD21 and the defining ABC/DN2 markers (CD11c, T-bet) emphasises the importance of comprehensive phenotyping. CD21lo alone cannot be used to identify ABC/AtB (see Sanz2025 - Human Atypical B Cells Overview).

  • Primary data: DN3 significantly expanded in ICU COVID-19: DN3 cells (CD11c⁻CD21⁻ within IgD⁻CD27⁻) were significantly expanded in ICU-C patients compared with HD (P ≤ 0.01). In hierarchical clustering of B cell population frequencies, DN3 consistently grouped with aN and DN2 populations — not with DN1 — supporting its association with the EF effector pathway. CD38 expression on DN3 cells from ICU patients was observable but heterogeneous (see Woodruff2020 - EF B Cell Responses in COVID-19, 24-marker spectral FCM, Extended Data Fig. 2).

  • DN3 first described in this paper: Although the Sanz2025 review cites DN3 as a known entity, Woodruff2020 is the original description. The formal gating definition — IgD⁻CD27⁻CD11c⁻CD21⁻ within the CD19⁺CD24⁻CD38⁻ DN gate — was established here. The paper notes that DN3 was “previously unreported in other conditions” at the time of publication (see Woodruff2020 - EF B Cell Responses in COVID-19).

  • DN3 cells contain SARS-CoV-2-specific cells (independent cohort confirmation): Dual-fluorophore RBD probe staining of COVID-19 peripheral blood demonstrated that DN3 cells (alongside DN2, aN, SWM, and PB) contain virus-specific B cells. DN3 was CXCR5-low (like DN2), consistent with EF pathway identity. This confirms the antigen-driven nature of the DN3 expansion originally reported by Woodruff2020 - EF B Cell Responses in COVID-19 in a separate cohort (see Kaneko2020 - GC Loss and TFH Block in COVID-19, n=68 blood cohort, 13-color FCM + RBD probes).

  • UMAP heterogeneity within DN3: UMAP projections indicated a split in the DN3 compartment driven by HLA-DR and CD19 expression — suggesting internal heterogeneity that requires further interrogation (see Woodruff2020 - EF B Cell Responses in COVID-19, Fig. 2e).

  • Distinctive transcriptomic signature — proliferation + UPR, lowest activation markers. Among DN subsets, DN3 displays a strong proliferation and unfolded-protein-response signature together with the lowest levels of CD22, CD72, CD69, and BAFFR — a profile distinguishing it from DN1, DN2, and DN4 (see Lamprinou2026 - ABCs and DN B Cells, opinion, citing Castleman 2022 / Allard-Chamard 2023).

  • Broader disease spectrum: IgG4-RD fibrosis, hypoxia, and SLE activity. Beyond severe COVID-19, DN3 has been implicated in autoimmune fibrosis in IgG4-related disease (tissue-infiltrating) and in SLE, where extrafollicular CD19^low CXCR5⁻CD11c⁻ DN3 cells correlate significantly with disease activity in female patients; DN3 has also been associated with hypoxic conditions (see Lamprinou2026 - ABCs and DN B Cells, opinion, citing Allard-Chamard 2023 / Chizzolini 2024 / Castleman 2022).

  • ★ DN3 in severe COVID-19 correlates with autoreactive antibody titres — and with systemic inflammation generally. DN3 levels in severe COVID-19 correlate positively with increased titres of autoreactive antibodies (citing Castleman 2022) and with clinical/laboratory parameters — ventilatory parameters, leukocytes, neutrophils, CRP, ferritin and D-dimers (citing Sosa-Hernández 2020). ⚠ The co-correlation with generic inflammatory markers means the autoantibody link cannot be read as DN3-specific causation from these data (see Beckers2023 - Origins and Functions of DN B Cells, review).

  • ⚠ The DN2/DN3 expansion in severe COVID-19 is TRANSIENT. DN B cell frequencies in patients who recovered from COVID-19 were similar to those who never had severe disease — the subset shift reverses (citing Reyes 2021). Sampling-window consequence for dengue: if an acute arboviral DN2/DN3 expansion decays on the same timescale, a convalescent-only or late-sampled cohort would record no expansion at all. This is a study-design constraint, not merely a caveat (see Beckers2023 - Origins and Functions of DN B Cells, review).

  • DN^low is very probably DN3 under another name. Szelinski 2022’s DN^low — CD19^low IgD⁻CD27⁻CXCR5⁻, CD11c⁻, plasmablast-like transcriptome, increased in SLE — matches the DN3 description already on this page from Sanz2025, which itself identifies “the CD27⁻ CXCR5⁻ CD19^low plasmablast-low population described in SLE (Szelinski et al. 2022)” as DN3. ⚠ Beckers’ body text (§3) prints DN^low as CXCR5⁺, but Table 1 of the same review prints it CXCR5⁻, and Szelinski’s own title reads “antigen-experienced CXCR5⁻CD19low B cells” — the body text is a typo, and the wiki reads DN^low as CXCR5⁻. What remains is a naming problem (two labels for one population, no agreed term), not a phenotypic conflict (see Beckers2023 - Origins and Functions of DN B Cells, review, citing Szelinski 2022).

  • ★ PRIMARY DATA — DN3 tracks plasmablasts more tightly than any other DN subset. In IgG4-related disease blood (n=38, Spearman), DN3 frequency correlated with plasmablast frequency at r = 0.6566, p < 0.0001, versus DN2 r = 0.4496 (p = 0.0067), DN4 r = 0.4037 (ns) and DN1 r = −0.1420 (ns). DN3 also reached higher frequencies than DN2 in this cohort (up to ~15% vs ~10% of B cells). This is the strongest quantitative support the wiki holds for the DN3-as-pre-plasmablast reading, and it is a correlation in blood, not a lineage relationship (see Allard-Chamard2023 - DN3 B Cells Infiltrate Inflamed Tissues, n=38 cross-sectional, 13-colour FCM).

  • ★ PRIMARY DATA — the transcriptomic case for pre-plasmablast identity. Sorted DN1–DN4 from IgG4-RD blood (n=4, SMART-Seq2, GEO GSE220582) show DN3 carrying a signature unlike any other DN subset: proliferation plus the unfolded protein response, features of plasmablasts and antibody-producing cells, and high IGHG4 message — DN3 was “the only B cells transcriptomically enriched for IgG4.” DN1 and DN2 transcriptomes broadly reproduced their lupus counterparts from Scharer2019 - Epigenetic Programming in SLE B Cells. ⚠ n=4, and transcriptomic resemblance is not lineage tracing — no pseudotime, clonal-connectivity or transfer evidence links DN3 to plasmablasts (see Allard-Chamard2023 - DN3 B Cells Infiltrate Inflamed Tissues, n=4 bulk RNA-seq).

  • ★ PRIMARY DATA — DN3 density rises in diseased end organs, but its share of the DN pool does not. This distinction is absent from every review that relays this paper, and it changes the claim:

    TissueDN3 density (disease vs control)DN3 as % of the tissue DN pool
    IgG4-RD submandibular gland (n=10 vs n=7 chronic sialadenitis)~48 vs ~3 cells/mm²not increased in disease — group means ~20% vs ~32%, but the distributions overlap almost entirely across n=7–10 and no test is reported; DN1 dominates at ~70%
    Severe COVID-19 lung (n=6 vs n=6 non-COVID autopsy)~6 vs ~1 cells/mm²~50% vs ~43% — no compositional enrichment

    Total-DN infiltration is significantly increased (SMG: ~35% vs ~16% of B cells, p<0.05; ~255 vs ~40/mm², p<0.01. Lung: ~29% vs ~11%, p<0.05; ~16 vs ~2/mm², p<0.05). But no significance markers are printed on any subset-level tissue panel (Figs. 5F, 5G, 6D, 6E), n=6–10 per arm, and the paper states p-values were not corrected for multiple comparisons. What is disease-associated is how many DN3 cells are present, not what fraction of the DN compartment they represent (see Allard-Chamard2023 - DN3 B Cells Infiltrate Inflamed Tissues, multi-colour IF).

  • ★ PRIMARY DATA — DN3 is the dominant DN subset in COVID-19 draining lymph nodes, and DN2 is nearly absent there. In late-phase (15–36 d) rapid-autopsy thoracic lymph nodes (n=6), DN cells were 18–66% of all CD19⁺ B cells, and by absolute density DN3 (~400/mm²) and DN1 (~230/mm²) dominated while DN2 sat at ~7/mm² — roughly fifty-fold below DN3 (DN4 ~17/mm²). No control lymph nodes are shown in this panel. ⚠ This is a late/resolving-phase window; nothing here speaks to the acute febrile period that matters for dengue (see Allard-Chamard2023 - DN3 B Cells Infiltrate Inflamed Tissues, n=6 autopsy).

  • ★ PRIMARY DATA — tissue T-B conjugates are DN-dominated and CD4-restricted, but the DN3 assignment is inferred, not measured. Cell-cell contacts detected by cytoplasmic overlap (StrataQuest) formed primarily between DN B cells and T cells and between plasmablasts and T cells, rarely with switched CD27⁺CD20⁺ memory cells; almost all involved CD4⁺ rather than CD8⁺ T cells (n=4). The authors attribute most of these conjugates to DN3 by elimination — “As DN2 B cells are relatively infrequent in tissues, most of these T-B interactions are likely with DN3 B cells” — having stained only CD19/IgD/CD27/SLAMF7, which does not separate DN2 from DN3. Fig. S4C shows conjugates with both SLAMF7⁺ and SLAMF7⁻ DN cells (see Allard-Chamard2023 - DN3 B Cells Infiltrate Inflamed Tissues, n=4 tissue IF).

  • ★ PRIMARY DATA — DN3 in blood independently discriminates IgG4-RD, with a statistically fragile effect size. Logistic regression gave DN3 OR 113.5 (95% CI 4.509–2857.1), p = 0.004, against plasmablasts OR 2.839 (1.214–6.643) and DN2 OR 5.381 (1.58–18.32). ⚠ The paper reads this as DN3 discriminating “with higher odds ratios than plasmablasts,” but an OR scales with its predictor’s units and these subsets span very different frequency ranges, so cross-predictor magnitudes are not comparable; the DN3 interval spans a 634-fold range and is uninterpretable as a point estimate. Direction and significance survive; the magnitude ranking does not (see Allard-Chamard2023 - DN3 B Cells Infiltrate Inflamed Tissues, n=38).

  • ⚠ The “DN3 expresses low CD19” claim is not verifiable in the primary. Sanz2025 - Human Atypical B Cells Overview attributes CD19^low to Woodruff 2020, Allard-Chamard 2023 and Perugino/Pillai jointly (this page’s Phenotype bullet). The Allard-Chamard primary never describes DN3 as CD19-low — CD19 is used only as a lineage gate. The claim may rest entirely on the other two citations; it is retained above but should not be attributed to Allard-Chamard 2023.

  • ⚠ SLAMF7 is expressed by DN3 but does not distinguish it from DN2. Protein-level MFI (n=4): DN2 2123 > DN3 1536 > DN4 956 > DN1 895. DN3 also shares a cytotoxic-gene module with DN2 (GZMA, GZMH, GZMB, GNLY, NKG7, KLRB1, KLRD1, KLRF1, FCGR3A), though perforin was generally low in DN3. Practical consequence: SLAMF7 cannot be used as a DN3-specific tissue marker (see Allard-Chamard2023 - DN3 B Cells Infiltrate Inflamed Tissues, n=4). See SLAMF7.

  • ⚠ Sanz2019 lists a third DN population that may or may not be this one — and the marker that would settle it has never been measured here. Table 1 carries an IgD⁻CD27⁻CD38⁻CD24⁻CD21⁻FcRL4⁺ row annotated “atypical/tissue-based memory”, listed separately from DN2 and identical to it on every core marker. It is CD21⁻, as this page’s DN3 is. But the DN3 literature (Lamprinou2026 - ABCs and DN B Cells, Allard-Chamard2023 - DN3 B Cells Infiltrate Inflamed Tissues) has never measured FcRL4, and Allard-Chamard’s DN3 is defined CXCR5⁻CD11c⁻ with no CD21 in the panel. Whether Sanz2019’s FcRL4⁺ row, Emory’s CD21⁻CD11c⁻ DN3 and Pillai’s CXCR5⁻CD11c⁻ DN3 are one cell, two or three is open (see Sanz2019 - Consistent Classification of Human B Cell Populations, review — no original data).

  • ★ [2026-08-29] The Emory DN3 gate is now held first-party, not relayed. A Sanz-lab primary states it in a figure legend: within IgD⁻CD27⁻ cells, DN1 = CD21⁺CD11c⁻, DN2 = CD21⁻CD11c⁺, DN3 = CD21⁻CD11c⁻. Until now the wiki carried the Emory definition through Sanz2025 - Human Atypical B Cells Overview and Woodruff2020 - EF B Cell Responses in COVID-19. DN3 rose together with DN2 to account for DN expansion in all three lupus groups (see Jenks2021 - B Cell Subset Composition in Cutaneous Lupus, n=207 + 46 HCD, cross-sectional, 8-marker conventional flow). ⚠ This does not reconcile the two DN3 axes. CXCR5 appears nowhere in this paper’s own data, so the supportable claim is as of 2021 a Sanz-lab primary gates DN3 on CD21 × CD11c with no CXCR5 — not that the Emory axis is confirmed correct. The reconciliation still requires both axes in one tube (B Cell Panel Variant 1).

  • [2026-08-29] The CD21⁺CD11c⁺ quadrant is left unnamed — which is not the same as DN4 being absent. This paper defines three DN subsets and does not label the fourth quadrant of its own 2×2. It therefore does not arbitrate the DN4 marker conflict between Allard-Chamard2023 - DN3 B Cells Infiltrate Inflamed Tissues (DN4 = CXCR5⁺CD11c⁺) and Lamprinou2026 - ABCs and DN B Cells (CXCR5⁺CD11c⁻), which remains blocked (see Jenks2021 - B Cell Subset Composition in Cutaneous Lupus).

Contradictions & Debates

  • The relationship between DN3 cells and the plasmablast maturation continuum (CD138⁻ → CD138⁺ ASC) described in Tipton2015 - ASC Diversity and Origin in SLE is not fully resolved. DN3 cells may represent the earliest circulating pre-plasmablast stage — before CD27 and CD38 upregulation — but this requires direct comparison with sorted populations.
  • Whether DN3 cells are an obligate intermediate in the EF pathway (aNAV → DN2 → DN3 → PB) or arise independently of DN2 is unclear. [2026-08-27] The tissue data argue against an obligate DN2→DN3 step: in COVID-19 draining lymph nodes DN3 outnumbers DN2 ~fifty-fold (~400 vs ~7 cells/mm²), and in IgG4-RD salivary gland DN3 is ~48/mm² against DN2’s ~6/mm². A precursor that scarce is hard to reconcile with a linear DN2→DN3 pathway unless the transit is very fast or DN2 is retained elsewhere (see Allard-Chamard2023 - DN3 B Cells Infiltrate Inflamed Tissues).
  • ★ [2026-08-27] Two incompatible DN3 gates, unreconciled. Emory/Sanz: IgD⁻CD27⁻CD11c⁻CD21⁻. Pillai/Ragon: IgD⁻CD27⁻CXCR5⁻CD11c⁻. Neither panel contains the other’s discriminating marker — Allard-Chamard2023 - DN3 B Cells Infiltrate Inflamed Tissues has no CD21 and no T-bet; Woodruff2020 - EF B Cell Responses in COVID-19 resolves CD21 but the two axes have never been run together. Every cross-study DN3 comparison in this wiki therefore rests on an untested assumption that CXCR5⁻ and CD21⁻ select the same cells within the CD11c⁻ DN pool. Directly testable with B Cell Panel Variant 1, which carries CXCR5, CD21, CD11c and T-bet in one tube.
  • ⚠ [2026-08-27] Is HLA class II high or low on DN3? The primary contradicts itself. The Results text states plasmablasts expressed lower HLA class II than DN3 and that DN3 was comparable to switched memory; the Fig. S3 legend of the same paper states DN3 cells “down regulate the expression of class II HLA.” The direction cannot be resolved from the published article, and it is load-bearing — the paper’s proposed mechanism has DN3 re-activating CD4⁺ T cells in tissue by antigen presentation, which requires HLA-II to be high. Quarantined to the source page under the [2026-08-23] inconsistent-reporting rule; no synthesised HLA-II claim is carried here (see Allard-Chamard2023 - DN3 B Cells Infiltrate Inflamed Tissues, n=13).
  • [2026-08-27] Is the tissue DN3 signal disease-specific at all? Absolute DN3 density is clearly higher in diseased organs, but DN3’s share of the tissue DN pool is not increased in disease in either arm — group means run the other way in IgG4-RD salivary gland vs non-fibrotic chronic sialadenitis (~20% vs ~32%, though the distributions overlap almost entirely across n=7–10 and no test is reported) and are essentially the same in COVID-19 vs inflamed control lung (~50% vs ~43%). One reading: DN3 accumulation is a generic feature of inflamed tissue and what differs in disease is total B cell infiltration. The paper does not address this (see Allard-Chamard2023 - DN3 B Cells Infiltrate Inflamed Tissues).

Double-Negative B Cell, DN2 B Cell, Atypical B Cell, Age-Associated B Cell, Plasmablast, Extrafollicular Response, GC-Independent Response, CXCR5, CD11c, T-bet, SLAMF7, B Cell Panel Variant 1

Sources

Graph View

Backlinks