Appanna2016 - Plasmablasts as Subset of Memory B Cell Pool

Full citation: Appanna R, Srinivasan KG, Xu MH, Toh YX, Velumani S, Carbajo D, Lee CY, Zuest R, Balakrishnan T, Xu W, Lee B, Poidinger M, Zolezzi F, Leo YS, Thein TL, Wang CI, Fink K. Plasmablasts During Acute Dengue Infection Represent a Small Subset of a Broader Virus-specific Memory B Cell Pool. EBioMedicine. 2016;12:178–188.

Raw file: [[raw/Appanna2016.pdf]]

Summary

This study investigated the developmental relationship between plasmablasts and memory B cells during acute dengue infection by comparing their BCR repertoires and antigen specificities in the same individuals longitudinally. Twelve DENV-2-infected patients (8 secondary, 4 primary) from Singapore were sampled at acute (14–71h post-fever onset), defervescent (days 3–7), and convalescent (days 15–166) timepoints. Plasmablasts (CD19⁺CD20⁻CD27^hiCD38^hi) were sorted at defervescence and DENV-specific memory B cells (CD19⁺CD20⁺CD27⁺, sorted using fluorescently-labelled live virus particles) at convalescence.

The central finding is that plasmablasts and DENV-specific memory B cells are clonally distinct — very few CDR3 sequences were shared between the two populations, and the rare shared clones were all IgM. Despite this clonal disconnect, both populations showed comparable VH mutation frequencies, indicating similar levels of affinity maturation through independent differentiation pathways. Antigen specificity diverged sharply: 85% of plasmablast-derived mAbs recognised recombinant E protein, whereas MBC-derived mAbs were predominantly complex epitope-specific (56%) and prM-specific (24%), with only 18% recognising E protein. E protein-specific antibodies were more potently neutralising than complex epitope-specific antibodies, though both were predominantly serotype cross-reactive.

The authors conclude that plasmablasts represent a small, non-representative subset of the broader memory B cell repertoire, and that the plasmablast response may not be a useful readout for long-term immune protection in dengue.

Study Design

• Type: Prospective cohort with longitudinal sampling and single-cell molecular analysis
• Sample size: n=12 dengue patients (8 secondary, 4 primary)
• Setting: Tan Tock Seng Hospital, Singapore; adult patients (>21y) presenting with acute-onset fever; endemic setting (~50% adult seroprevalence)
• Population: Adults with DENV-2 infection (11 patients) or DENV-3 (1 patient); primary vs. secondary classified by DENV-specific IgG at fever onset; no severity stratification reported (all hospitalised)
• Timepoints: Early acute (14–71h), defervescent (days 3–7), convalescent (days 15–166)

Key Findings

• Plasmablasts and MBCs are clonally distinct: Venn diagram analysis of CDR3 sequences showed very few shared CDR3s between acute-phase plasmablasts and convalescent DENV-binding MBCs. In Patient 3 (454 sequencing, 1,304 PB vs. 407 MBC-DENV unique CDR3s), only 1 CDR3 was shared. Patient 8: 10 shared of 314+2 CDR3s. Patients 11 and 12 (primary infections) showed similarly minimal overlap (7 and 8 shared CDR3s, respectively).
• Clonally related CDR3s (85% identity threshold) also showed minimal overlap: Relaxing the identity threshold to 85% to account for affinity maturation did not substantially increase shared clones. Patient 3: 7 clonally related CDR3s shared (of 97 PB clones); Patient 8: 2; Patient 11: 20; Patient 12: 6.
• Shared clones were exclusively IgM: The 8 CDR3 sequences that were ≥10× expanded in MBC-DENV relative to plasmablasts were all IgM isotype. No shared IgG CDR3 sequences were detected between plasmablasts and MBC-DENV. The authors interpret this as potential enrichment of low-affinity IgM binders during virus-based MBC sorting.
• DENV-binding MBCs are predominantly IgM⁺: A large fraction of DENV-specific MBCs expressed IgM, compared with plasmablasts which were IgG-dominated in both primary and secondary infection. Non-DENV-binding MBCs predominantly expressed IgG. Some DENV binding was observed in two non-dengue-immune individuals, suggesting low-affinity/polyreactive IgM contributes to the DENV-binding MBC pool.
• Plasmablast-derived mAbs are predominantly E protein-specific (85.3%): Of 75 DENV-specific mAbs cloned from plasmablasts across 7 patients, 85.3% recognised recombinant E protein. No prM-specific antibodies were detected among plasmablasts. 14.7% recognised complex epitopes (virus-binding but not E or prM by ELISA/western blot).
• MBC-derived mAbs have broader specificity — 55.6% complex epitope, 24.4% prM, 17.8% E: Of 45 DENV-specific mAbs from MBCs across 4 patients, the specificity profile was dramatically different from plasmablasts. This aligns with prior literature showing MBCs target prM and NS1 while plasmablasts target E.
• E-specific mAbs are more potently neutralising than complex epitope-specific mAbs: E protein-specific PB-derived mAbs neutralised DENV-1, -2, and -3 at 0.1–10 µg/ml (NT50). Complex epitope-specific mAbs typically required >10 µg/ml. Some MBC-derived complex epitope-specific mAbs showed higher serotype-specific neutralisation for the infecting serotype (DENV-2).
• Antibodies from both populations were predominantly serotype cross-reactive: Both PB- and MBC-derived mAbs bound multiple DENV serotypes. Potent serotype-specific neutralisers were rare — consistent with the cross-reactive nature of the dengue antibody response.
• VH1-69 and VH4-34 (self-antigen-binding potential) found specifically in plasmablasts: Some VH gene families with known autoantigen-binding potential (VH1-69, VH4-34) were observed among PB-derived but not MBC-derived DENV-specific mAbs.
• DENV-binding MBC-derived antibodies had significantly longer CDR3s than plasmablast-derived antibodies (p<0.0001, ANOVA). Complex epitope-specific and prM-specific antibodies accounted for the longer CDR3s. prM-specific CDR3s also contained more neutral non-polar amino acids than E-specific CDR3s.
• Comparable VH mutation frequencies between PBs and MBCs: Despite being clonally unrelated, both plasmablasts and MBC-derived antibodies showed similar levels of VH nucleotide mutations (not significantly different by one-way ANOVA across all five specificity groups). N-addition counts were also similar. This argues against a simple model where MBCs undergo more extensive GC maturation than PBs.
• VH3/VH4 usage differed between PBs and non-DENV-binding MBCs (p=0.03): VH usage between PBs and DENV-binding MBCs was not significantly different (possibly due to limited sequence numbers), but VH3 and VH4 usage differed between PBs and non-binding MBCs.
• Polyclonal plasmablast activation: Most patients showed polyclonal PB responses with little clonal expansion (except Patient 6, who had one large IgG clone specific to EDIII of DENV-1 despite DENV-2 infection). This contrasts with the oligoclonal expansion typical of influenza.
• 0.1–19% of lymphocytes were plasmablasts; 0.5–8.1% of CD19⁺CD20⁺ B cells bound DENV.

Methods Used

• Conventional Flow Cytometry — 5-colour panel (CD20, CD27, CD19, CD38, CD138) for PB identification; fluorescently-labelled live virus (Alexa Fluor 594-DENV-1, 488-DENV-2, 647-DENV-3) for antigen-specific MBC identification
• FACS Sorting — single-cell sorting into 96-well plates (FACSAria) for Sanger sequencing and mAb cloning; pool sorting for 454 sequencing
• BCR Sequencing — Sanger sequencing of single-cell VH/VL from PBs and MBCs; 454 pyrosequencing (GS-FLX Titanium, 250,000 reads/sample) of pooled sorted populations using iRepertoire primers

Entities Mentioned

Plasmablast, CD19, CD20, CD27, CD38, CD138, IgG, IgM, IgA

Concepts Addressed

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

Relevance & Notes

This paper provides the first direct clonal comparison of plasmablasts and memory B cells within the same dengue patients — filling a critical gap between the magnitude data (Wrammert2012, GarciaBates2013) and the BCR repertoire data (Parameswaran2013). The clonal disconnect has several implications for the wiki:

1. Supports distinct developmental pathways: The finding that PBs and MBCs are clonally unrelated supports a model where only a specific subset of pre-existing memory B cells is activated as plasmablasts during reinfection, while the broader MBC pool follows a separate trajectory. This is consistent with the Ansari2025 finding that Tph preferentially drive memory B cell→PB differentiation — if Tph selectively activate E-specific IgG⁺ memory cells, the non-E-specific MBCs (prM, complex epitope) would follow separate activation pathways and would not appear in the PB compartment.

2. VH4-34 in plasmablasts connects to autoreactivity literature: The presence of VH4-34 (self-antigen-binding) and VH1-69 specifically in PB-derived mAbs echoes the Woodruff2020 finding of autoreactive VH4-34 clones in COVID-19 EF-derived ASCs and Tipton2015/Anolik2004 data on VH4-34 in SLE. This suggests the acute dengue PB wave may include autoreactive/polyreactive antibodies — relevant to the ADE and autoimmunity questions in dengue.

3. Comparable SHM challenges the GC-divergence model: If MBCs had undergone additional GC maturation relative to PBs, higher mutation rates would be expected. The comparable mutation levels suggest either (a) both populations derive from similarly matured precursors with different specificities, or (b) PBs undergo some EF-associated SHM that brings them to a similar level. This is consistent with the Parameswaran2013 intermediate SHM (4.4–6.9%) and the Ansari2025 model where memory cells (already mutated) are the PB precursors.

4. Plasmablast response as poor correlate of long-term protection: The E-protein dominance of PBs (85%) vs. the broader specificity of MBCs (prM, complex epitope, E) means measuring the PB wave during acute infection captures only one arm of the B cell response. For vaccine evaluation, both PBs and MBCs should be assessed.

Limitations: Small sample size (n=12, with different analyses on different patient subsets); the DENV-binding MBC sort may enrich for low-affinity IgM binders (evidenced by IgM dominance and DENV binding in naive individuals); no severity stratification; CD27⁺ gate for MBC sorting may miss DN memory B cells; convalescent timepoints varied widely (15–166 days).

Questions Raised

• If plasmablasts are derived from a specific MBC subset (E-specific IgG⁺ memory cells per the Ansari2025 model), what activates prM- and NS1-specific MBCs? Do they differentiate into tissue-resident plasma cells without circulating as plasmablasts?
• The VH4-34/VH1-69 enrichment in PBs — is this transient autoreactivity analogous to the self-limited EF autoreactivity described in COVID-19 (Woodruff2020/Sanz2025)? Does it resolve post-dengue or contribute to dengue-associated autoimmune phenomena?
• The IgM⁺ clones shared between PBs and MBCs — are these genuine cross-compartment clones or sorting artefacts from low-affinity polyreactive IgM? If genuine, they would represent the only evidence of direct PB→MBC (or MBC→PB) clonal lineage in dengue.
• Would single-cell paired VH/VL sequencing with modern methods (10x Chromium) reveal more clonal overlap than the 454/Sanger approach used here, given the deeper sampling and paired-chain resolution?