Song2022 - Tfh Outside Germinal Centers Drive T-bet CD11c B Cells

Full citation: Song, W., Antao, O. Q., Condiff, E., Sanchez, G. M., Chernova, I., Zembrzuski, K., Steach, H., Rubtsova, K., Angeletti, D., Lemenze, A., Laidlaw, B. J., Craft, J., & Weinstein, J. S. (2022). Development of Tbet- and CD11c-expressing B cells in a viral infection requires T follicular helper cells outside of germinal centers. Immunity, 55(2), 290–307.e5. https://doi.org/10.1016/j.immuni.2022.01.002

Raw file: [[raw/song2022.pdf]]

Summary

This is the fate-mapping primary behind the wiki’s central claim that T-bet⁺CD11c⁺ B cells (the murine counterpart of human Age-Associated B Cells / DN2 B Cells) arise through a germinal-center-independent pathway. Until this ingest that claim reached the wiki only through a one-sentence characterization in Glaros2025 - Multilayered Identity of B Cell Memory.

Using acute LCMV-Armstrong and influenza A/PR8 infection in mice, the authors establish three things. First, T follicular helper (Tfh) cells — not Th1 cells — drive T-bet⁺CD11c⁺ B cell generation, through help delivered in spatial proximity at the follicular edge, despite the resulting B cells never entering a germinal center. Second, most of these cells develop independently of GCs, shown by four orthogonal methods: absence of GC markers, a distinct transcriptome, S1pr2-based genetic fate mapping (>80% never passed through a GC), and mixed bone marrow chimeras showing B-cell-intrinsic Bcl6-independence. Third, as infection resolves the cells relocate to and are retained at the splenic marginal zone by the integrins LFA-1 and VLA-4, forming a competitive memory subset that contributes substantially to recall antibody production.

The paper is the wiki’s first source to supply a positional mechanism for where these cells go and what holds them there, rather than only a transcriptional account of why they are excluded from follicles. It is entirely murine.

Study Design

• Type: In vivo murine infection; genetic fate mapping, adoptive transfer, mixed bone marrow chimera, imaging, RNA-seq, Ig repertoire sequencing, in vivo antibody blockade
• Sample size: 3–5 mice per group per experiment; most figures representative of 2–3 independent experiments. Repertoire analysis n=3 mice. RNA-seq on sorted populations at day 12 p.i.
• Setting: Acute LCMV-Armstrong (primary model) and influenza A/Puerto Rico/8/1934 (PR8) — the latter chosen because it preserves splenic architecture, which LCMV disrupts early
• Population: C57BL/6 WT; Tcrb⁻/⁻; Icos⁻/⁻; Sh2d1a⁻/⁻ (SAP); CD4^Cre^Bcl6^fl/fl^; CD19^Cre^Bcl6^fl/fl^; Tbet-AmCyan reporter; S1PR2-CreERT2 Rosa26^LSL-tdTomato^; MD4 (anti-HEL BCR transgenic) recipients; CD11c-DTR
• Cell definition: T-bet⁺CD11c⁺ B cells gated as B220⁺CD19⁺CD44^hi^CD11c⁺T-bet⁺

Key Findings

Kinetics

• T-bet⁺CD11c⁺ B cells emerge by day 5 p.i., rising to peak frequency ~day 8 and peak number ~day 10; Ki67^hi^ between days 5–8, with a Ki67^lo^ population from day 10 and predominant by day 14 (n=3–5/group, one-way ANOVA)

Tfh, not Th1, are the required helper subset — the decisive experiment

• Tcrb⁻/⁻ recipients given Smarta CD4 T cells showed induction of activated B cells and T-bet⁺CD11c⁺ B cells vs Tcrb⁻/⁻ alone → CD4 T cells are required
• Icos⁻/⁻ mice (defective Tfh differentiation, more Th1 than WT) had reduced T-bet⁺CD11c⁺ generation
• Sh2d1a⁻/⁻ and CD4^Cre^Bcl6^fl/fl^ — two independent Tfh-targeting models — both showed decreased percentages and numbers of T-bet⁺CD11c⁺ B cells with little reduction in Th1 cells
• Adoptive transfer of sorted Tfh (PSGL-1^lo^Ly6c^lo^PD-1^hi^CXCR5⁺) into infection-matched Tcrb⁻/⁻ mice induced T-bet⁺CD11c⁺ B cells; sorted Th1 (PSGL-1^hi^Ly6c^hi^) did not

Help is delivered by proximity, at the follicular edge

• Confocal imaging + histocytometry, day 8: Tfh localized around B cell follicles, Th1 dispersed in red pulp; clusters of T-bet⁺CD11c⁺ B cells sat at the edge of follicles adjacent to Tfh
• Ripley’s multitype K function: T-bet⁺CD11c⁺ B cells clustered with Tfh (K exceeded the theoretical Poisson K) but were randomly distributed relative to Th1
• Authors infer proximal delivery of IL-21 and IFN-γ plus contact-dependent CD40L — but explicitly state Tfh effector molecules were not assessed (see Limitations)

GC-independence — four orthogonal lines

• Minimal coexpression of GC markers GL-7 and EphrinB1 at days 6, 10, 15
• RNA-seq (sorted T-bet⁺CD11c⁺ vs GC vs naive follicular, day 12): PCA separates all three. GC signature genes Bcl6, S1pr2, Aicda downregulated in T-bet⁺CD11c⁺; Itgax, Tbx21, Fcrl5 lower in GC
• Histocytometry day 12: ~75% of T-bet⁺CD11c⁺ in the follicular mantle, ~0% in GC, ~20% T zone
• S1PR2-CreERT2 fate mapping (tamoxifen from day 4): >80% of T-bet⁺CD11c⁺ B cells lacked TdTomato (17.5% labelled) whereas the majority of GL-7⁺ early GC B cells were labelled (70.0%)
• Mixed BM chimera (50% CD19^Cre^Bcl6^fl/fl^ : 50% CD19^Cre^Bcl6^+/+^): GCs dominated by Bcl6-sufficient cells (~5:1), whereas T-bet⁺CD11c⁺ were populated ~equally by Bcl6-deficient cells (~1:1) → B-cell-intrinsic Bcl6 is dispensable
• The same result held in influenza PR8, which preserves splenic architecture — HA-specific T-bet⁺CD11c⁺ cells were rarely S1pr2-labelled (~15%) vs GC (~62–68%), and Bcl6 was not required → the GC-independent pathway is conserved across acute viral infections

Repertoire — class switching without affinity maturation

• Full-length Ig sequencing with UMIs; ~248,116 high-quality assembled reads per sample
• Isotype usage of T-bet⁺CD11c⁺ was not different from GC (~80% of GC sequences IgG; T-bet⁺CD11c⁺ predominantly IgG2b/IgG2c) → CSR occurs outside the GC
• Mutation: 41.6% of T-bet⁺CD11c⁺ full-length Ig sequences mutated vs 61.5% of GC; total mutation load 0.64% vs 0.99%
• CDR replacement mutations 0.66% vs 1.21% with similar silent frequencies → reduced affinity maturation
• Clones shared between T-bet⁺CD11c⁺ and GC accounted for <10% of total sequences in either population; lineage trees showed early bifurcation and infrequent interconversion

Marginal-zone relocation and integrin-dependent retention — the positional mechanism

• Day 12 clusters at the white/red pulp border; by day 15 localized to the marginal zone identified by MadCam-1⁺ sinus-lining cells
• i.v. anti-CD45 labelling (marks compartments open to circulation, incl. MZ): T-bet⁺CD11c⁺ 40.3% labelled at day 12 and 60.5% at day 15, vs naive follicular 16.8%/23.3% and GC 2.24%/3.47%
• 1,433 genes differentially expressed between days 8 and 15. S1pr3 and Cnr2 (cannabinoid receptor 2) increased as infection resolved; CXCR3 higher on T-bet⁺CD11c⁺ than naive follicular or GC
• Transwell S1P migration: FTY720 limited chemotaxis; TY52156 (selective S1PR3 inhibitor) completely abolished it
• But mixed BM chimeras (WT : S1pr3⁻/⁻) showed no competitive advantage of S1pr3⁺/⁺ at day 15 → S1PR3 loss alone is insufficient in vivo, possibly compensated by Cnr2 and CXCR3
• T-bet⁺CD11c⁺ cells showed elevated transcript and surface protein for LFA-1 (αLβ2; Itgal/Itgb2) and VLA-4 (α4β1; Itga4/Itgb1) vs naive follicular
• In vivo blockade of the LFA-1 and VLA-4 alpha subunits for 3 h at day 15 caused significant loss of T-bet⁺CD11c⁺ B cells from spleen with a concomitant increase in blood → retention at the marginal zone is LFA-1/VLA-4 dependent

Effector and memory output

• Genes highly expressed in T-bet⁺CD11c⁺: Prdm1 (Blimp-1), Xbp1, Zeb2, Bhlhe41, Zbtb32, Tnfrsf17 (BCMA — covered on BAFF), Sdc1 (CD138), Sub1, Tle3, Mndal
• Surface phenotype resembled memory B cells: increased CD38 but not CD138 at days 10, 12, 15; IgD^lo^CD38⁺GL-7⁻ memory phenotype maintained to at least day 35
• Gene-set enrichment at day 12 did not match conventional memory B cell signatures in existing databases
• ELISpot: T-bet⁺CD11c⁺ contained antibody-secreting cells producing anti-LCMV IgG prior to TLR7 (R848) stimulation, increasing after — unlike IgD^lo^CD38⁺GL-7⁻ memory B cells, which were reliant on TLR stimulation, and unlike PCs, which were independent of it
• Autoreactivity: on autoantigen microarray, T-bet⁺CD11c⁺ and IgD^lo^CD38⁺GL-7⁻ cells secreted similar repertoires of autoantigen-binding IgG compared with PCs. However, on anti-dsDNA and anti-chromatin ELISpot, T-bet⁺CD11c⁺ rarely secreted autoantibodies in the absence of R848 despite containing a higher frequency of autoreactive producers → further differentiation toward ASC is necessary before they contribute to serum autoreactivity
• BrdU (day 4, 10 days), assessed day 30: T-bet⁺CD11c⁺ were ~5% of BrdU⁺ IgD^lo^CD38⁺GL-7⁻ memory B cells → a small fraction of the memory compartment
• i.v. anti-CD45 labelled the majority of T-bet⁺CD11c⁺ memory cells but only ~30% of total memory B cells → MZ localization is maintained into memory
• Recall: sorted transfer into MD4 recipients — T-bet⁺CD11c⁺ generated a rapid, robust anti-LCMV IgG response at days 5 and 10, higher than naive and comparable to memory B cells lacking the phenotype; naive and memory recipients made more IgM. All transferred subsets gave rise to GC B cells at day 10 → these memory cells can enter GCs on antigen re-encounter
• CD11c-DTR depletion at days 28/30 with rechallenge: serum anti-LCMV IgG trended higher in PBS vs DT (p=0.0535 at day 5 post-rechallenge)

A methodological warning the authors document

• The CD11c-DTR system is unsuitable for depleting CD11c-expressing B cells immediately following viral challenge: activated B cells increased DTR expression independently of Itgax in vitro and CD11c protein in vivo, so DT caused global ablation of proliferating B cells

Methods Used

Conventional Flow Cytometry, FACS Sorting, RNA Sequencing, BCR Sequencing, ELISpot, Multi-color Immunofluorescence, In Vitro B Cell Stimulation

Also used but not separately updated: Immunohistochemistry — trimmed 2026-09-05 during the C1 repair pass: no immunohistochemistry was performed. The words “immunohistochemistry”, “paraffin”, “HRP” and “DAB” appear nowhere in the paper. All imaging is multi-colour immunofluorescence on PLP-fixed frozen sections stained with directly conjugated antibodies and read by confocal microscopy and histocytometry; that content was propagated to Multi-color Immunofluorescence instead. Also used and not listed in this triad: an anti-LCMV IgG ELISA on mouse sera, and a 95-autoantigen microarray (UTSW core facility) for which the wiki holds no method page.

Entities Mentioned

T-bet, CD11c, CD38, CD138, BLIMP-1, ZEB2, FCRL5, CXCR3, Bcl-6, AID, CD40L, ICOS, IL-21, LFA-1, VLA-4, S1PR3, XBP1, BAFF, IgD, IgG, IgM, B220, Age-Associated B Cell, Atypical B Cell, DN2 B Cell, Plasmablast, Switched Memory B Cell, Peripheral Helper T Cell

Concepts Addressed

Germinal Center, Extrafollicular Response, Class Switch Recombination, Somatic Hypermutation, Memory B Cell, Follicular Exclusion, Extrafollicular T Cell Help, Atypical B Cell Effector Output

Relevance & Notes

What it settles. The GC-independent origin of T-bet⁺CD11c⁺ B cells was, before this ingest, held in the wiki only through Glaros2025 - Multilayered Identity of B Cell Memory’s one-sentence characterization, with Cancro2020 - Age-Associated B Cells independently supporting the same direction. Song2022 is the direct evidence: four orthogonal methods, two different acute viral infections.

What it adds that nothing else in the wiki had. A positional mechanism. Before this ingest, follicular exclusion was explained only transcriptionally (CXCR5-low, ZEB2→Mef2b per Sanz2025 - Human Atypical B Cells Overview); nothing accounted for where the cells actually go or what holds them there. Song2022 supplies marginal-zone relocation with LFA-1/VLA-4-dependent retention demonstrated by in vivo blockade — the wiki’s first adhesion-receptor mechanism, and the first use of ICAM-1/VCAM-1 ligand biology anywhere in the corpus. See Follicular Exclusion.

Where it agrees with the human data. The reduced-SHM finding (0.64% vs 0.99%) parallels GodoyLozano2016 - Lower IgG SHM Rates in Acute Dengue and the <3% EF benchmark from Tipton2015 - ASC Diversity and Origin in SLE, and the CSR-without-affinity-maturation pattern matches William2002 - Extrafollicular Somatic Hypermutation in Autoimmune Mice. Antibody-secreting capacity prior to TLR stimulation is consistent with the pre-ASC framing in Jenks2018 - DN2 B Cells and EF Pathway in SLE.

Where it complicates the wiki. Two places.

1. Xbp1 and Prdm1 are highly expressed in murine T-bet⁺CD11c⁺ cells here, whereas Sutton2021 - Alternative Lineage B Cells in Vaccination and Infection reported XBP1/IRF4/PRDM1 as a negative result in human atypical B cells. Flagged under Contradictions on BLIMP-1 and XBP1.
2. The Tfh dependence sits awkwardly beside the wiki’s Tph framing from Ansari2025 - Peripheral T Helper Subset Drives B Cell Response in Dengue. Song2022’s helper is CXCR5⁺PD-1^hi^ (canonical Tfh) acting outside the GC; Ansari2025’s is CXCR5⁻PD-1⁺ (Tph). These are different cells reaching a similar outcome — the wiki should not merge them. See Extrafollicular T Cell Help.

Limitations — as stated by the authors.

• Tfh effector molecules driving generation were not assessed; IL-21/IFN-γ/CD40L are inferred from proximity, not tested here
• A small fraction was labelled by S1pr2 fate tracking, so GC development is possible under inflammatory conditions other than the acute viral infections studied
• Whether the unlabelled fraction localized within white pulp is uncertain
• The absent in vivo S1PR3 phenotype may reflect compensation by Cnr2 and CXCR3; this was not further investigated

Limitations — curator/wiki view. Entirely murine, with LCMV and influenza; no dengue, no human validation. The i.v. anti-CD45 labelling and histocytometry require intact splenic architecture, which is why LCMV alone would have been insufficient. The recall experiments use MD4 recipients whose endogenous B cells carry a fixed irrelevant BCR, which is a clean but artificial competitive setting.

Questions Raised

• Which Tfh effector molecules actually drive T-bet⁺CD11c⁺ generation? The authors name this as their first limitation, and it is the direct question a blockade experiment would answer.
• Does the LFA-1/VLA-4 retention mechanism operate in humans, and would it be detectable as an absence from blood? If circulating DN2 frequency partly reflects failure of marginal-zone retention, blood frequency is a mobilisation readout — the same concern Cancro2020 - Age-Associated B Cells raises from splenic/blood disequilibrium.
• If T-bet⁺CD11c⁺ memory cells can enter secondary GCs on re-challenge, is the EF/GC distinction a property of the response rather than a fixed lineage boundary?
• Does the requirement for further ASC differentiation before autoantibody secretion mean circulating DN2 frequency is a poor proxy for autoantibody burden?
• The acute-versus-chronic caveat the authors raise: do these developmental rules hold in a short febrile illness like dengue, which is neither a resolving 10-day murine model nor a chronic autoimmune state?