Glaros2025 - Multilayered Identity of B Cell Memory

Full citation: Glaros, V., Francis, N., & Kreslavsky, T. (2026). The multilayered identity of B cell memory. Cellular & Molecular Immunology, 23, 150–167. https://doi.org/10.1038/s41423-025-01377-5

Naming note. Filed as Glaros2025 to match the raw filename and the DOI slug (-025-; accepted 25 Nov 2025, © 2025), but the journal citation line is 2026 (published online 6 Jan 2026; Cell Mol Immunol 2026;23:150–167). Frontmatter year: carries the journal year, the page name carries the filename year. Karolinska Institutet (Kreslavsky lab); open access, CC-BY.

Raw file: [[raw/glaros2025.pdf]]

Summary

A narrative review arguing that the memory B cell (MBC) compartment is not one population but a set of overlapping “layers” of identity, each imposed by a different axis: developmental pathway (germinal-center-derived vs GC-independent), duration and strength of antigenic stimulation, anatomical and microanatomical localization, and timing of generation in ontogeny. The authors’ thesis is that combinations of these layers — not a single marker or a single origin — determine what an MBC does when it is reactivated. The review is explicitly a synthesis: it contains no original data and no new analyses, and every claim below is the authors’ reading of someone else’s primary result.

The backbone of the review is the contrast between GC-derived MBCs (GCMBCs) and early / pre-GC MBCs (EMBCs, called eMBCs on this wiki). Activated B cells are described as tripotent “activated precursors” (APs) that can give rise to early plasma cells (ePCs), eMBCs, or GC B cells (GCBCs) — with the eMBC fate framed not as an active differentiation programme but as a passive transition to quiescence driven by the loss of activating signals (“differentiation by default”). GC-specific fate-mapping tools (tamoxifen-inducible Cre driven by S1pr2 / Gcsam) have, for the first time, allowed origin to be assigned independently of the surrogate markers (isotype, CD80/PD-L2, SHM load) that the field previously relied on — and those tools show that eMBCs outnumber GC-derived MBCs across multiple immunization scenarios (a claim the review attributes to its own prior work and to unpublished results — see the caveat in Key Findings §B), while the two subsets overlap so heavily in phenotype that no surface/isotype/SHM criterion cleanly separates them.

The section most relevant to this wiki is “Atypical MBCs” (pp. 160–161), reviewed in detail below. Its load-bearing claims for the wiki’s spine are that atypical B cells (ABCs) are antigen-experienced and T-cell-help-dependent rather than inflammatory bystanders; that T-bet is not strictly required for their formation while ZEB2 has emerged as a shared driver across mouse and human; that GC-specific fate mapping in acute viral infection assigns the majority of antigen-specific ABCs to a GC-independent pathway; and that the long-standing “ABCs differentiate poorly into plasma cells” result is now attributed substantially to an in vitro stimulation-mode artefact. The review also surveys tissue-resident MBCs (lung, skin, gut, liver, lymph node, splenic marginal zone, bone marrow) and early-life-origin MBCs.

Study Design

• Type: Narrative review / synthesis — no original data, no systematic search protocol, no meta-analysis. Not PRISMA. Reference list of 285 items; figures newly drawn (BioRender) but conceptual, not data.
• Sample size: N/A. All quantitative claims are inherited from cited primaries.
• Setting: Predominantly murine experimental immunology (fate mapping, adoptive transfer, conditional knockouts, immunization models), with human data cited for ABC phenotype, marginal-zone B cells, and chronic-infection contexts.
• Population: Not applicable. No dengue data whatsoever — DENV, dengue, and flavivirus recall are mentioned only in passing (protective antibodies against flavivirus and influenza variants arise primarily from rapid MBC→PC differentiation rather than from primary-infection LLPCs).
• Conflicts / provenance: Authors declare none. Kreslavsky-lab self-citation is present but limited (Glaros 2021 Immunity, the “limited antigen drives eMBC generation while restraining the plasmablast response” paper, is load-bearing for the “differentiation by default” model — flagged below).

Key Findings

A. Atypical MBCs — the section this wiki cares about (pp. 160–161)

• Phenotype as the review states it. Human ABCs “typically lack classical MBC markers such as CD27 and CD21” and often express T-bet together with CD11c, CXCR3, FCRL4, and FCRL5. The murine counterpart is CD11c⁺CD21^lo “age-associated” B cells (see Age-Associated B Cell). Note this marker list is CD27/CD21-negative-based, not IgD/CD27-based — it does not map one-to-one onto this wiki’s IgD⁻CD27⁻ DN gate.
• ABCs expand in acute infection and after vaccination, then contract over weeks to months. The review explicitly extends ABCs beyond the chronic-infection/autoimmunity settings where they were first described, citing acute expansion in mice and humans and post-vaccination expansion in humans, followed by contraction (review, citing multiple mouse + human primaries). This is the kinetic frame the dengue pilot needs — an acute, self-resolving expansion, not a chronic accumulation.
• A conserved transcriptional programme across contexts. ABCs from individuals with malaria, HIV, and autoimmune disease show similar transcriptional profiles, suggesting a shared differentiation trajectory rather than disease-specific look-alikes (review, citing Holla et al. 2021 Sci Adv, cross-disease human transcriptomics).
• ★ T-bet is NOT strictly required. Although T-bet was long “believed to be essential” for ABC development, CD11c⁺ ABCs still arise in both infection and autoimmune models in the absence of T-bet (review, citing Du 2019 Eur J Immunol and Levack 2020 J Immunol, both mouse conditional-KO). This directly qualifies the wiki’s standing description of T-bet as the defining ABC transcription factor.
• ★ ZEB2 has emerged as a key ABC transcription factor in both mice and humans, and is proposed as the imposer of the shared cross-context programme (review, citing Dai 2024 Science and Gao 2024 Sci Immunol). Functionally dual-edged: B-cell-specific Zeb2 deletion improved disease in a lupus mouse model, but decreased GCBC numbers during persistent Plasmodium infection — i.e. ABCs are pathogenic in autoimmunity and supportive of humoral immunity in persistent infection.
• Mechanism of formation. ABCs arise from BCR stimulation in an inflammatory cytokine milieu, requiring additional signals from endosomal TLRs (see TLR7), IFN-γ, and IL-21 (review, citing a large mouse + in vitro literature).
• ★ ABCs are antigen-experienced and at least partly T-help-dependent — not inflammatory bystanders. Supporting lines assembled by the review: ABCs are found within the antigen-specific pool after immunization in mice and humans; some express classical MBC markers; some carry somatically hypermutated BCRs; they respond to antigen restimulation; ABC frequencies fall significantly in patients with CD40/CD40L mutations; and ABCs do not develop in mice with a fixed BCR specificity under steady state. In autoimmunity, ABCs are proposed to derive from B cells that have engaged self-antigen — anergic B cells can be converted into ABCs and are more prone than naive B cells to acquire the phenotype in vitro (review, citing Imabayashi 2025 Sci Adv, mouse + in vitro).
• ★★ Origin: GC-specific fate mapping assigns the majority of antigen-specific ABCs to a GC-INDEPENDENT pathway. Following acute viral infection in mice, most antigen-specific ABCs are generated without transiting a germinal center (review, citing Song 2022 Immunity, mouse GC-specific fate mapping). The review carries the caveat plainly: “whether this holds true across other immune contexts that give rise to ABCs remains to be elucidated.” The relative GC-dependent vs GC-independent contribution to the ABC pool is listed as an open question, as is whether the compartment is maintained by self-renewal vs replenishment from newly activated B cells.
• ★★ The “ABCs can’t become plasma cells” result is substantially a stimulation-mode artefact. Early in vitro work reported limited PC-differentiation capacity for ABCs (review, citing Portugal 2015 eLife and Sullivan 2015 PLoS Pathog, both malaria-associated human atypical MBCs). Later work attributes this to how the BCR was engaged: soluble anti-Ig fails, whereas membrane-bound anti-Ig effectively drives PC differentiation, because inhibitory receptors — FcγRIIB, and per the review also FCRL5 — dampen BCR signalling unless they are physically excluded from the immune synapse, which only membrane-associated antigen achieves (review, citing Ambegaonkar 2020 Sci Adv). Primary source abstract verified during ingest (2026-08-16): direction of effect confirmed — atypical MBCs “are unable to respond to soluble antigens” but “robustly respond to antigens that associate with cell surfaces, such as antigens in immune complexes.”
• Positive evidence that ABCs are primed for PC fate. ABCs express PC-associated genes (Song 2022) and show increased propensity to differentiate into PCs in T cell coculture (review, citing Wang 2018 Nat Commun, SLE CD11c^hi T-bet⁺ B cells + IL-21; and Louis 2021 JCI Insight, T-bet⁺CD27⁺CD21⁻ B cells in kidney-transplant antibody-mediated rejection).
• But ABCs are functionally plastic, not obligate pre-PCs. Adoptive transfer of ABCs between mice of the same autoimmune strain shows they can additionally differentiate into GCBCs and self-renew (review, citing Nickerson 2023 J Exp Med, mouse adoptive transfer). Whether this plasticity tracks the degree of prior antigenic stimulation is unresolved.
• Therapeutic read. Depletion of CD11c⁺ or T-bet⁺ B cells decreased autoantibody levels and ameliorated disease in autoimmune models — with the review’s own caveat that such depletion also hits other activated B cell subsets.
• Unresolved effector route. The review closes the section by stating it “remains to be investigated” whether ABCs act predominantly through differentiation into antibody-producing PCs, through antigen presentation to T cells, or both.

B. GC-derived vs early (GC-independent) MBCs — the review’s backbone

• Tripotent activated precursors (Fig. 1). Naive B cells activated at the T–B border become APs coexpressing naive/memory markers (CD38, CCR6) and activated/GCBC markers (CD95, GL7), with intermediate IRF4 and absent or low BCL6. Limiting-dilution transfer shows a single naive B cell can generate all three lineages — GCBCs, ePCs, and eMBCs. The three-way split is evident by day 4 after immunization.
• eMBC formation is “differentiation by default.” eMBC generation is marked by minimal transcriptional change — in sharp contrast to the extensive divergence in cells committing to GCBC or ePC fates — and is driven by early cell-cycle exit following a rapid decline in antigen abundance outside GCs in the first days of the response. Supplying additional antigen prevents the quiescent transition and pushes early activated B cells toward the ePC lineage instead (review, citing Glaros 2021 Immunity — author self-citation, load-bearing for this model). A similar early quiescent-eMBC dominance was seen in nonhuman primates, suggesting evolutionary conservation.
• Affinity biases the branch point. Higher-affinity BCRs and stronger T-derived signals favour ePC generation; eMBCs were generated more frequently in response to lower-affinity antigen. GCBC fate, unlike PC fate, cannot be induced by simply lowering stimulus strength — it requires a more complex regulatory framework.
• GC-independent MBCs are real, not an artefact. Mice lacking BCL6 in the haematopoietic compartment and patients with CD40L deficiency — neither able to form functional GCs — still generate detectable MBC populations. Early antigen-specific, largely unswitched cells with an MBC phenotype appear before the first GCBCs.
• ★ eMBCs OUTNUMBER GC-derived MBCs — but see the attribution caveat. GC-specific fate mapping unexpectedly showed that eMBCs — “often considered a minor memory subset” — outnumber their GC-derived counterparts across multiple immunization scenarios. ⚠ Attribution caveat: the review attributes this jointly to Glaros 2021 Immunity and to the authors’ own unpublished results — the numerical-dominance claim is not independently published, and Glaros 2021 is a self-citation. Treat as the authors’ position, not as an established finding; do not let it carry weight alongside the published fate-mapping origin result (Song 2022).
• Surrogate markers cannot separate the two. Class-switch status (long used as a proxy for GC transit) fails because CSR occurs primarily prior to GC entry, with switched cells then gaining a competitive advantage inside GCs; fate mapping shows eMBCs include a significant fraction of class-switched cells while gcMBCs can remain IgM⁺ or even IgM⁺IgD⁺. SHM load overlaps — gcMBCs average higher mutation loads, but the range within each population is highly variable and some gcMBCs sit at eMBC levels and vice versa. Recent transcriptomics finds the two “closely related at the transcriptional level, with only relatively subtle differences” — again attributed to two published primaries plus the authors’ own unpublished observations.
• ⚠ TERMINOLOGY COLLISION — “DN MBC” here means CD80⁻PD-L2⁻, NOT IgD⁻CD27⁻. The Shlomchik-lab classification splits MBCs on CD80 and PD-L2 into DP (CD80⁺PD-L2⁺), SP (CD80⁻PD-L2⁺), and DN (CD80⁻PD-L2⁻). GC fate mapping shows DP is enriched for gcMBCs and DN consists almost exclusively of eMBCs — though not absolutely, as a considerable proportion of DP cells originate from the non-GC pathway. This “DN” is a different axis entirely from this wiki’s Double-Negative B Cell (IgD⁻CD27⁻). See the synonymy map on Atypical B Cell.
• Function on reactivation. IgG1⁺ MBCs (gcMBC-enriched) and DP MBCs are biased toward PC differentiation; IgM⁺ MBCs (eMBC-enriched) and DN MBCs more efficiently reenter secondary GCs. But GC-specific fate mapping complicates the simple story: upon booster immunization, fate-mapped gcMBCs were predominantly biased toward PC differentiation and contributed only minimally to secondary GC formation, while the non-fate-mapped compartment (naive + eMBCs) dominated the secondary GCBC pool — and a subsequent study found those secondary GC cells were derived primarily from naive B cells, not eMBCs. Net: gcMBCs have greater intrinsic capacity to enter secondary GCs, but their overall contribution remains smaller than that of naive B cells.
• ★ Epigenetic “recording” of cumulative stimulation history (Fig. 4). Accumulated stimulation progressively increases chromatin accessibility at PC-associated gene loci including Prdm1, thereby increasing baseline BLIMP-1 expression and biasing MBCs toward PC differentiation over GC reentry — mechanistically routed through epigenetic regulation of BACH2 and BLIMP1. The proposed model: MBC fate on reactivation may be programmed not by a binary GC vs non-GC origin but by the cumulative history of stimulation recorded epigenetically. DN and DP MBCs carry distinct epigenetic profiles that may predetermine their responsiveness to secondary antigen encounter. The review flags this as “intriguing… remaining to be tested.”

C. Molecular regulation of gcMBC differentiation (Fig. 3)

• gcMBC differentiation = downregulation of the GCBC programme + transition to quiescence + upregulation of pro-survival factors, rather than a proliferative effector programme.
• BACH2 is the “translation” node for weak selection signals: LZ GCBCs with lower-affinity BCRs receive weaker T cell help → less mTORC1 signalling → BACH2 upregulation → quiescence, reduced MYC, and active repression of Prdm1 (BLIMP1), i.e. antagonism of the PC programme. The review notes this mirrors the established BACH2–BLIMP1 axis in T cells, “a conserved regulatory circuit governing memory versus effector differentiation across the B and T branches.”
• Additional nodes covered descriptively: BCL2 (pro-survival; repressed by BCL6; forced expression lets GCBCs with nonfunctional BCRs become quiescent and acquire MBC markers), HHEX (downregulates BCL6, releasing BCL6-mediated repression of Bcl2), ZBTB18 (IL-9-induced; represses cell-cycle and some GC programme components, promotes gcMBC precursor survival), MYC, IL-9, and IL-4 (IL-4-mediated BCL6 downregulation may facilitate GC exit; overall effect on gcMBCs contested — reported to both promote and suppress depending on timing).
• gcMBCs are enriched for relatively LOW-affinity clones and show greater clonal diversity than the gcPC compartment; some bind the immunizing antigen only when valency is increased. This diversity is argued to be the substrate for cross-variant recall — protective antibodies against flavivirus and influenza variants arise primarily from rapid MBC→PC differentiation rather than from LLPCs laid down in primary infection.
• GC selection is stochastic, not strictly deterministic (Fig. 2): affinity shifts fate probabilities, with boundaries “strongly blurred” by noise, jackpot antigen-acquisition events, and variable access to T cell help.

D. MBCs in tissues, and other layers

• ★ Liver. In murine Ehrlichia muris infection — a model in which GCs are absent in the spleen — somatically hypermutated IgM⁺ T-bet⁺ MBC-like cells are found in both liver and spleen, generated through a GC-independent pathway, persisting after infection clearance, and residing in the tissue (labelling-resistant to intravenous antibody). Repertoire analysis suggested ~half of liver clones were liver-exclusive; liver but not splenic MBCs expressed Cd69. Separately, atypical MBCs have been reported in the livers of chronic hepatitis B patients. See Tissue-Resident Memory B Cell.
• Lung. The best-characterized BRM compartment: CD69↑ / S1PR1↓, CCR6 and CXCR3 expression, IgA⁺-predominant after respiratory infection with a smaller IgG⁺ fraction; IgG⁺ BRMs mount faster and more localized PC responses than circulating MBCs. Some BRMs are recruited to infection sites irrespective of antigen specificity (“bystander” screening) and can differentiate into PCs in response to innate signals alone.
• Splenic marginal zone. A large fraction of human MZ B cells are MBCs — CD27⁺ and carrying SHM — unlike rodent MZ, thought to be mostly naive innate-like cells. Long-lived anti-smallpox MZ-phenotype MBCs dominated the switched memory compartment decades after vaccination.
• Lymph node. MBCs are strongly biased to the outer follicular area near the subcapsular sinus, a niche enriched for antigen and memory TFH; draining-LN MBCs reenter secondary GCs more efficiently than non-draining-LN MBCs — implying the anatomical site of a booster shapes response quality.
• Bone marrow. BM MBCs carry a transcriptional signature and BCR repertoire distinct from splenic MBCs, adhere via α4β1/α6β1 to VCAM-1⁺ stroma, and appear to be genuine residents rather than a mixing pool.
• Early-life-origin (ELO) MBCs. Neonatal B cells have a distinct repertoire (minimal TdT → germline-proximal V-gene usage). B cell clones generated in the first two weeks of life contribute substantially to adult antigen-experienced compartments across organs, especially mucosal; ELO cells are enriched among splenic MZ B cells.
• Methodological caveat the review raises for fate mapping. Many tissue-BRM studies used Aicda-based fate-mapping to infer GC origin — but AID-driven CSR and SHM also occur outside the GC, so these systems do not definitively establish GC provenance. This is the same logic the wiki already carries from William2002 - Extrafollicular Somatic Hypermutation in Autoimmune Mice and Kaneko2020 - GC Loss and TFH Block in COVID-19.

Methods Used

The review reports no experiments of its own. The methods it relies on for its strongest claims, and which are relevant to this wiki:

• FACS Sorting — MBC subsetting by CD80/PD-L2, IgM/IgG, tetramer-based antigen-specific sorting.
• BCR Sequencing — SHM load comparisons between eMBC and gcMBC; liver vs spleen clonal overlap.
• Single-Cell RNA Sequencing — transcriptional comparison of eMBCs and gcMBCs; ABC cross-disease profiling.
• ATAC-seq — chromatin accessibility at Prdm1 as a function of cumulative stimulation (the Fig. 4 model).
• In Vitro B Cell Stimulation — the soluble vs membrane-bound anti-Ig comparison that reframes ABC PC-differentiation capacity; TLR/IFN-γ/IL-21 ABC differentiation.
• T-B Coculture Assay — ABC→PC differentiation propensity.
• Conventional Flow Cytometry — all ABC and MBC subset phenotyping cited.
• GC-specific genetic fate mapping (tamoxifen-inducible Cre under S1pr2 / Gcsam) — the enabling technology behind the review’s central claims. No wiki method page yet; see Questions Raised.

Entities Mentioned

Atypical B Cell, Age-Associated B Cell, Double-Negative B Cell, DN2 B Cell, Early Memory B Cell, Tissue-Resident Memory B Cell, Switched Memory B Cell, Plasmablast, T-bet, ZEB2, BACH2, BLIMP-1, Bcl-6, IRF4, AID, CD11c, CD21, CD27, CXCR3, FCRL5, FcRH4, IgM, IgG, IgA, IgD, IL-21, TLR7

Concepts Addressed

Memory B Cell, Extrafollicular Response, Germinal Center, Somatic Hypermutation, Class Switch Recombination, Original Antigenic Sin

Relevance & Notes

What this adds to the wiki. Glaros2025 is the first source in the wiki to give the GC-independent memory pathway a formal, mouse-genetics-grade identity rather than treating it as the shadow of the GC. That matters because this wiki’s spine — the atypical/DN cluster and the EF pathway that generates it — has until now rested on human phenotypic correlations (Jenks2018, Woodruff2020, Ansari2025) plus one murine SHM proof (William2002 - Extrafollicular Somatic Hypermutation in Autoimmune Mice). The review supplies the missing developmental scaffold: a tripotent branch point (AP → ePC / eMBC / GCBC), a fate-mapping technology that assigns origin without surrogate markers, and the finding that GC-independent memory is the larger, not the smaller, compartment.

But note a distinction the wiki must not blur. The review’s eMBC is a quiescent, low-SHM, largely unswitched memory cell formed by default when antigen runs out. The EF response this wiki tracks in SLE/COVID/dengue is a proliferative, ASC-generating effector arm. Both descend from the same pre-GC branch point, but they are different outputs — eMBC ≈ the memory limb, DN2→plasmablast ≈ the effector limb. Glaros himself keeps these separate (the ePC and eMBC fates are alternative destinations of the same precursor). Treating “GC-independent” as a synonym for “extrafollicular effector” would be a category error, and the tripotency framing is the correction.

Direct qualification of an existing wiki claim. T-bet is described across this wiki as the defining ABC transcription factor. Glaros states plainly that T-bet is not strictly required — CD11c⁺ ABCs arise without it in both infection and autoimmune models — and elevates ZEB2 to the shared cross-context driver. This has an immediate consequence for B Cell Panel Variant 1 Panel 4, which adds intracellular T-bet specifically to confirm DN2/ABC identity: a T-bet-gated definition will undercount a T-bet-independent CD11c⁺ fraction. CD11c should stay the primary axis (consistent with Sutton2021 - Alternative Lineage B Cells in Vaccination and Infection, where CD11c was the best single surface marker), with T-bet used as a confirmatory, not defining, parameter.

Third position on a contradiction the wiki actively tracks. The wiki records a live tension: Sutton2021 found no PC-maintenance gene upregulation in atBC clusters, challenging the EF pre-plasmablast model; SLE data (Jenks2018) and dengue data (Ansari2025) support DN2→plasmablast. Glaros adds a methodological third position: the foundational “ABCs differentiate poorly into PCs” results (Portugal 2015, Sullivan 2015 — both malaria) used soluble anti-Ig, and ABCs’ high inhibitory-receptor load (FcγRIIB, FCRL5) blocks signalling from soluble antigen specifically. With membrane-associated antigen the block is relieved and PC differentiation proceeds. This does not resolve the contradiction — Sutton’s is a transcriptomic observation, not a stimulation assay — but it means the field’s functional prior was built on an assay that systematically disadvantaged the cell type being tested.

A constraint for the cross-wiki bridge thesis — worth taking seriously. The verified Ambegaonkar 2020 abstract states that atypical MBCs “robustly respond to antigens that associate with cell surfaces, such as antigens in immune complexes, but are unable to respond to fully soluble antigens, such as self-antigens.” The bridge-wiki/ thesis proposes atypical B cells / plasmablasts as the cellular source of dengue autoantibodies. If atypical MBCs are constitutionally poor responders to soluble self-antigen, the straightforward “atypical cell meets soluble self-antigen → autoantibody” route is weakened, and the mechanism would need to run through membrane-associated or immune-complexed self-antigen instead. Dengue secondary infection is an immune-complex-rich setting, so this is not fatal — but it changes the required mechanism and should be stated as such rather than assumed away.

Relation to the severity axis. The liver findings intersect the wiki’s severity work: hepatic involvement (AST/ALT >1000) is one of only two severe-dengue signs with a consensus operational definition (see Morra2018 - Defining Warning Signs and Severe Dengue). Glaros documents GC-independent, somatically hypermutated IgM⁺T-bet⁺ MBCs resident in liver in a GC-absent infection model, plus atypical MBCs in chronic-HBV liver. No dengue liver B cell data exist — this is a precedent and a hypothesis, not evidence.

Editorial decisions taken during this ingest.

• No entity pages created for BCL2, HHEX, ZBTB18, MYC, IL-9, IL-4, CD80, or PD-L2. These are GC-molecular-regulator or MBC-subsetting nodes with no current dengue, atypical-cell, or flow-panel application in this wiki; creating eight 1-source pages would add orphan/lint debt against a lean-infrastructure constraint. They are covered descriptively here and on Germinal Center / Memory B Cell. The CD80/PD-L2 axis specifically is documented as a synonymy-map row on Atypical B Cell because its “DN” label collides with this wiki’s DN gate.
• Two new entity pages created: Early Memory B Cell (the review’s central population; will accumulate sources) and Tissue-Resident Memory B Cell (earns its place on the liver hook; splenic-MZ material folded in rather than given a separate page).

Limitations to carry.

• Zero original data. Every number and result is inherited. Weight bullets by the primary’s design, not by the review’s authority.
• Predominantly murine. The two claims this wiki will lean on hardest — GC-independent ABC origin and eMBC dominance — are both mouse fate-mapping, and the ABC-origin result is from one acute viral infection model. The review says so explicitly; the wiki should not drop that caveat.
• Author self-citation is load-bearing for the “eMBC differentiation by default” model (Glaros 2021 Immunity). It is corroborated by the nonhuman-primate observation and by the CD40L-deficiency human data, but it is the authors’ own result carrying their own framework.
• ~10 named primaries are threaded through this page (Song 2022, Holla 2021, Dai 2024, Gao 2024, Du 2019, Levack 2020, Ambegaonkar 2020, Nickerson 2023, Wang 2018, Louis 2021, Portugal 2015, Sullivan 2015, Imabayashi 2025). Only Ambegaonkar 2020 was independently verified (abstract) during this ingest. The rest are the review’s characterizations. Added to External Citation Audit.

Questions Raised

1. Does the GC-independent ABC origin result hold outside acute viral infection in mice? The review flags this itself. If it holds in human acute flavivirus infection, it converts the wiki’s central EF hypothesis from analogy to mechanism; if it does not, the dengue DN2 expansion could be substantially GC-derived, which would collide with the low-SHM data (GodoyLozano2016).
2. ★ Does the epigenetic-recording model explain the secondary-dengue triad? (Wiki-generated hypothesis — Glaros makes no dengue claim.) Secondary dengue = MBCs that have been repeatedly stimulated by prior DENV exposure. Glaros’s Fig. 4 model predicts such cells carry increased accessibility at Prdm1, biasing them toward PC differentiation over GC reentry. That would predict exactly the observed triad: a massive plasmablast burst (GarciaBates2013 - Plasmablast Response and Dengue Severity, 87% peak), lower SHM in secondary than primary (GodoyLozano2016 - Lower IgG SHM Rates in Acute Dengue), and OAS-biased output (Priyamvada2016 - Cross-Reactive Memory Plasmablasts in Secondary Dengue) — because a PC-biased recall bypasses the GC entirely and therefore adds no new mutations. Testable with ATAC-seq on sorted DENV-specific MBCs from primary vs secondary donors. This is a wiki-generated synthesis, not a finding in Glaros2025.
3. Is T-bet⁻CD11c⁺ ABC a real population in human acute infection, and how large? If so, every T-bet-gated ABC count in the human literature — and the wiki’s own Panel 4 — is a systematic undercount of unknown magnitude.
4. Are the CD21⁻CD11c⁺ cells in dengue (Ansari2025) ZEB2⁺? ZEB2 is now the strongest candidate for a cross-context ABC identity marker and no dengue study has stained it. It is intranuclear, so it fits the lab’s newly available intracellular-staining capability.
5. Does a soluble-vs-membrane antigen distinction change dengue B cell assay design? If atypical MBCs cannot respond to soluble antigen, in vitro DENV-antigen restimulation assays using soluble E/NS1 may systematically under-report the functional capacity of exactly the population this wiki cares about.
6. Should the wiki carry a method page for GC-specific genetic fate mapping? It is the enabling technology behind this review’s strongest claims but is murine-only and cannot be applied to the curator’s human cohort — arguably background context rather than a wiki method.
7. Is there a liver-resident atypical B cell compartment in dengue? Hepatic involvement is a consensus severe-dengue criterion and the Ehrlichia precedent shows GC-independent T-bet⁺ MBCs residing in liver. Untestable in the current pilot (no tissue), but relevant to interpreting hepatic severity signals.