Sanz2025 - Human Atypical B Cells Overview

Full citation: Sanz, I. (2025). Human Atypical B Cells. An Overview. Immunological Reviews, 334, e70058.

Raw file: [[raw/Sanz2025.pdf]]

Summary

This invited review from Iñaki Sanz (Emory University) argues that the term “atypical B cell” (AtB) is obsolete and should be abandoned. The central thesis is threefold: (1) cells labelled as AtB are in fact normal components of immune responses, not atypical; (2) their nature, derivation, and function depend on immunological context — naïve-derived effectors in primary responses vs. memory cells in recall settings — and cannot be inferred from phenotype alone; (3) the inconsistent use of overlapping classification schemes (AtB, ABC, CD21lo, DN2) across diseases has created a literature in which fundamentally different populations are conflated under a single label.

The review traces the historical evolution from CD27 as a universal memory marker (Klein et al. 1998), through the discovery of CD27⁻ memory cells in HIV and malaria (Moir et al. 2008; Weiss et al. 2009), to the description of murine age-associated B cells (Rubtsov et al. 2011; Hao et al. 2011), and culminates in the Sanz lab’s own DN2 definition (Jenks et al. 2018). It provides a comprehensive classification table (Table 1) and gating strategy (Figure 2) for the field’s current populations, then systematically argues that these populations are more heterogeneous than any single label captures.

A major contribution is the synthesis of evidence across diseases — SLE, SARS-CoV-2, malaria, HIV, rheumatoid arthritis, cancer — showing that the same phenotypic markers (CD11c⁺, T-bet⁺, CD21lo, CXCR5⁻) identify different cell populations with different origins and functions depending on context. In primary acute responses (SLE flares, severe COVID-19), the dominant ABC/DN2 population derives from activated naïve cells via an extrafollicular, TLR7-driven pathway. In recall settings (post-vaccination, chronic infections), antigen-specific DN2 cells may represent durable memory cells. The review also introduces DN3 cells (CXCR5⁻, CD21⁻, CD11c⁻, T-bet⁻) as pre-plasmablasts distinct from both DN2 and AtB/ABC.

Study Design

• Type: Invited narrative review
• Sample size: N/A (synthesis of published literature)
• Setting: Cross-disease review spanning SLE, SARS-CoV-2, malaria, HIV, RA, IgG4-RD, cancer; human and mouse data
• Population: N/A

Key Findings

• The AtB label conflates ≥5 distinct populations: Resting CD27⁻ memory (DN1), naïve-derived EF effectors (aNAV, DN2), pre-plasmablasts (DN3), CD11c⁺ switched memory ABC, and tissue-resident FcRL4⁺ cells are all lumped under “atypical” using subsets of CD27⁻, CD21lo, CD11c⁺, T-bet⁺, or FcRL5⁺ markers (Table 1).
• Classification inconsistency is the root problem: Different studies define AtB/ABC using different marker combinations — CD11c alone, T-bet alone, CD21lo alone, or combinations — often without IgD measurement. This makes cross-study comparison impossible and generates contradictory functional claims.
• In primary responses, ABC-phenotype cells are predominantly naïve-derived: The highest CD11c and T-bet expression is found in aNAV (IgD⁺CD27⁻CD11c⁺⁺T-bet⁺⁺) and DN2 (IgD⁻CD27⁻CD11c⁺⁺T-bet⁺⁺) cells, both of naïve origin. CD27⁺ switched memory cells express lower levels of these markers.
• DN3 cells are pre-plasmablasts, not AtB/ABC: DN3 (CXCR5⁻, CD21⁻, CD11c⁻, T-bet⁻) cells are a distinct subset expanded in acute COVID-19 and active SLE. They represent pre-plasmablasts that accumulate in tissues in autoimmune fibrosis and severe COVID-19. DN3 should not be classified as AtB or ABC despite being CD21lo.
• Memory DN2 cells exist: Antigen-specific DN2 cells persist >1 year post-SARS-CoV-2 mRNA vaccination, accounting for >50% of all spike/RBD⁺ cells, establishing durable CD27⁻ memory with a DN2 phenotype. The full properties of memory vs. effector DN2 cells remain to be elucidated.
• DN2/ABC are not exhausted or anergic: Despite overlapping CD21lo phenotype with HIV-associated exhausted B cells, DN2/ABC cells retain activation markers and robust effector functions. The autoreactivity of DN2/ABC is context-dependent — not an inherent property of the phenotype.
• ABC are powerful antigen-presenting cells: Across mouse and human studies, a consistent finding is that ABC/DN2 cells function as potent APCs, potentially driving TFH induction and sustaining GC responses — an effector function beyond antibody secretion.
• Competing EF vs. GC endotypes: SLE patients segregate into EF-dominant and memory/GC-dominant clusters. ~75% of patients fall into EF or memory endotypes, with higher severity and nephritis concentrating in the EF cluster. EF endotypes also predict reduced affinity maturation and neutralizing activity of SARS-CoV-2 vaccine responses in SLE.
• EF responses associated with worse cancer outcomes: A large single-cell analysis of human tumors found EF responses associated with worse clinical outcomes and resistance to immunotherapy, partly via immunosuppressive T cell environments.
• Self-limited EF autoreactivity is normal: In healthy subjects, acute SARS-CoV-2 infection generates naïve-derived DN2 cells producing dual-reactive (virus + self) antibodies with low SHM. These autoreactive responses subside within months in healthy individuals — self-limited EF autoreactivity. In genetically predisposed individuals, these responses may be perpetuated.
• ZEB2 is a primary driver of ABC formation: B cell-intrinsic Zeb2 is essential for the ABC transcriptional signature (including T-bet, CD11c, Zbtb32), for proinflammatory functions, and for autoimmune pathology in TLR7-driven lupus. Zeb2 represses Mef2b, a TF required for GC differentiation — providing a direct mechanistic link between EF and GC pathway antagonism. Zeb2 haploinsufficiency in humans reduces ABC.
• T-bet is important but not absolutely required for ABC: CD11c⁺ B cells can be generated and maintained in the absence of T-bet. ABC markers (CD11c, others) can be induced in vitro through diverse activation conditions without IFN-γ stimulation or T-bet expression.
• TLR7 gain-of-function causes human SLE with ABC/DN2 expansion: Monogenic TLR7 GoF mutations cause lupus in a B cell-intrinsic, GC-independent fashion with expanded ABC/DN2. Additional mutations enhancing TLR function (NOX2 loss-of-function via NCF1/NCF2; UNC93B1 instability) also expand ABC/DN2.
• Epigenetic landscape is disease-context specific: SLE DN2 and healthy donor DN2 share only 50% of differentially accessible chromatin regions (ATAC-seq), indicating a shared core epigenetic programme modified by disease-specific features. SLE DN2 chromatin is shaped by T-bet, AP-1 (ATF3), and EGR motifs.
• CXCR5 can substitute for CD21, and T-bet/FcRL5 for CD11c in gating: These alternative gating strategies identify similar populations, providing practical flexibility for panel design.
• Therapeutic targeting of ABC/DN2 is feasible: Direct targeting via FcRL5 or SLAMF7 (already approved for myeloma); indirect targeting via TLR7 inhibition (ENPATORAN phase II promising) or IFN-γ blockade. Selective ABC depletion could spare pre-existing protective memory and long-lived plasma cells.
• CD27 can be downregulated upon stimulation: CD70 on activated T cells, TLR ligands, and cytokines can all modulate CD27 expression — CD27⁻ status does not necessarily mean a cell never expressed CD27.

Methods Used

Conventional Flow Cytometry (classification schemes reviewed; Figure 2 gating strategy)

Entities Mentioned

Double-Negative B Cell, DN2 B Cell, DN3 B Cell, Activated Naive B Cell, Plasmablast, CD27, CD21, CD11c, T-bet, CXCR5, FCRL5, FcRH4, SLAMF7, ZEB2, IRF4, BLIMP-1, BACH2, TRAF5, TLR7, CD19, CD38, IgD, IgG

Concepts Addressed

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

Relevance & Notes

This is the authoritative synthesis from the lab that defined DN2 cells (Jenks2018), the aNAV→DN2→PB pathway (Jenks2018), and the naive-derived SLE ASC programme (Tipton2015). It consolidates all prior Sanz lab findings and extends them with cross-disease evidence.

For this wiki, the review has three major impacts:

1. Classification authority: The wiki’s existing decision to use “Double-Negative B Cell” and “DN2 B Cell” as entity names — rather than “atypical B cell” or “ABC” — is directly validated by this review’s argument that the AtB/ABC labels should be abandoned. The DN nomenclature (IgD⁻CD27⁻, subdivided by CXCR5/CD21/CD11c) is Sanz’s recommended classification.

2. Context-dependent interpretation: The review establishes that the same phenotype (CD11c⁺ T-bet⁺ CD21lo) means different things depending on disease context and whether the response is primary or recall. For dengue — where acute responses likely involve primary naïve activation — this predicts that ABC-phenotype cells would be predominantly naïve-derived EF effectors (aNAV → DN2 → PB), not exhausted memory.

3. DN3 as a new entity: The review defines DN3 cells as pre-plasmablasts (CXCR5⁻ CD21⁻ CD11c⁻ T-bet⁻), distinct from both DN2 and AtB/ABC. This population should be tracked in dengue studies, where it may represent part of the acute plasmablast wave.

The review is notably frank about the limitations of the field’s current understanding: the full properties of memory vs. effector DN2 cells are unresolved; the origins and fates of ABC in most autoimmune diseases beyond SLE are poorly understood; and the relationship between ABC and long-term protective immunity vs. pathogenic autoimmunity remains context-dependent.

Questions Raised

• Do dengue B cell studies that report “atypical B cell” expansions capture the same population as DN2, or are they measuring a heterogeneous mix (DN2 + DN3 + CD21lo activated memory) under a single label?
• Does the self-limited EF autoreactivity model (healthy subjects generate then resolve naïve-derived autoreactive DN2 responses) apply to dengue — i.e., does acute dengue transiently generate autoreactive B cells that resolve in convalescence?
• Can the competing EF/GC endotype concept be applied to dengue patient stratification — do some patients have dominant EF plasmablast responses while others have GC-biased responses, and does this correlate with clinical severity or antibody quality?
• Is the DN3 pre-plasmablast population (CD11c⁻ T-bet⁻ within IgD⁻CD27⁻CXCR5⁻) detectable in acute dengue PBMCs, and if so, does it expand alongside or independently of DN2?
• Given that ZEB2 is the primary driver of ABC formation and represses the GC-required TF Mef2b, is ZEB2 expression elevated in acute dengue B cells — and does this predict EF dominance?