Toll-like receptor mediated inflammation directs B cells towards protective antiviral extrafollicular responses

Abstract

Extrafollicular plasmablast responses (EFRs) are considered to generate antibodies of low affinity that offer little protection from infections. Paradoxically, high avidity antigen-B cell receptor engagement is thought to be the main driver of B cell differentiation, whether in EFRs or slower-developing germinal centers (GCs). Here we show that influenza infection rapidly induces EFRs, generating protective antibodies via Toll-like receptor (TLR)-mediated mechanisms that are both B cell intrinsic and extrinsic. B cell-intrinsic TLR signals support antigen-stimulated B cell survival, clonal expansion, and the differentiation of B cells via induction of IRF4, the master regulator of B cell differentiation, through activation of NF-kB c-Rel. Provision of sustained TLR4 stimulation after immunization shifts the fate of virus-specific B cells towards EFRs instead of GCs, prompting rapid antibody production and improving their protective capacity over antigen/alum administration alone. Thus, inflammatory signals act as B cell fate-determinants for the rapid generation of protective antiviral extrafollicular responses.

Fig. 1. Primary influenza infection induces strong early EFRs prior to GC formation.

Shown are flow cytometric analyses of mediastinal lymph nodes (medLN) from C57BL/6 mice infected with influenza A/PR8 intra-nasally (i.n.) at seven days post-infection (dpi). a Identification of extrafollicular plasmablasts (EF PBs) by gating first on CD19lo/CD45Rlo then CD24+/CD38- and pre-GC/GC B cells on CD19+/CD45R+ then CD24hi/CD38lo. Presence of CD19lo/CD45Rlo and CD24+/CD38- populations at 5 dpi (left) and 7 dpi (right). b IRF8 and GL7 expression shown to confirm GC identity and IRF4 and CD138 expression shown to confirm EF PB identity (left), along with IgM and IgD expression of B cell subsets (right). c–e C57BL/6 mice (n = 3–4) were infected and medLN were collected on the days specified, measuring B cell frequencies of total cells (c), pre-GC/GC frequency of B cells (d), and EF frequency of B cells (e). Data in (c–e) represent mean ± 95% confidence interval (CI) of two independent experiments. Statistical significance determined by one-way ANOVA. ****p < 0.0001. Source data are provided as a Source Data file.

Fig. 2. EFRs generate influenza-specific antibody-secreting cells.

a Influenza-specific ELISPOTS of sorted EF PBs and pooled non-EF cells from influenza-infected C57BL/6 mice (n = 6) for total Ig (left) and IgG2c (right). b Flow plots of HA-specific B cells identified using double HA-tetramer staining with subsequent phenotyping by additional markers as outlined in Fig. 1a. c–e Time course of HA-specific B cell subsets during influenza infection in C57BL/6 mice (n = 3–4), measuring frequency of HA-specific clones (c), HA-specific pre-GC/GC clones (d), and HA-specific EF PBs (e). Data in (a, c–e) represent mean ± 95% CI of two independent experiments. Statistical significance determined by two-tailed Student’s t-test with Welch’s correction or one-way ANOVA. *p < 0.05, **p < 0.01, ****p < 0.0001. Source data are provided as a Source Data file.

Fig. 3. Subcutaneous immunization with influenza and alum does not elicit EFRs.

a–e C57BL/6 mice (n = 4) were immunized s.c. with 1 × 10⁷ PFU influenza A/PR8 in alum and inguinal LNs were analyzed on days indicated. a Comparison of EF PB and GC B cell frequencies at multiple timepoints post-immunization. b Mice (n = 4) were immunized s.c. with indicated HAU sucrose-gradient purified influenza A/PR8 virion emulsified in Complete Freund’s adjuvant and analyzed at 7 dpi. GC B cell (left) and EF PB (right) frequency were compared at each antigen concentration. c Representative flow plots gating for HA-specific B cells at specified timepoints after immunization in (a). d Quantification of HA-specific B cells at each timepoint post-immunization. e Frequencies of HA-specific B cells that were either GC or EF compared by timepoint. Data in (a, b, d, e) represent mean ± 95% CI of two independent experiments. Statistical significance determined by one-way ANOVA and two-tailed Student’s t-test with Welch’s correction. *p < 0.05. **p < 0.01. Source data are provided as a Source Data file.

Fig. 4. Optimal EFR kinetics and protective antibodies require MyD88 and TRIF.

Knockout and WT mice (n = 4–5) were infected with 10 PFU A/PR8 and medLNs were collected at 7 days post-infection (dpi). a Fold-difference of B cell subsets in TLR-deficient versus WT mice at 7 dpi. b–d Serum was transferred to C57BL/6 mice prior to infection with a lethal dose (100 PFU) of influenza A/PR8 the next day from naïve WT, DKO, and TKO mice (n = 4) (b) and influenza-infected, age/sex matched- WT and MyD88/TRIF-deficient (DKO) mice (n = 10) (c) or WT and TLR2/4/unc93b-deficient (TKO) mice (n = 10) (d) from 10 dpi. Shown is percent change in weight over the course of infection. Data in (a–d) represent mean ± 95% CI of two independent experiments. Statistical significance determined by two-way ANOVA and two-tailed Student’s t-test with Welch’s correction. **p < 0.01, ***p < 0.001, ****p < 0.0001, *****p < 0.00001 or indicated in subfigures. Source data are provided as a Source Data file.

Fig. 5. BCR-mediated survival and proliferation are defective in the absence of TLR signaling.

a Mixed bone-marrow chimeras (BMC) (n = 12) established with irradiated CD45.1 C57BL/6 host mice reconstituted with µMT donor BM and BM from either DKO or TKO, then infected with 10 PFU A/PR8 6 weeks later. b Quantification of DKO and TKO BMC compared to WT BMC controls of B cell subsets at 7 dpi. c Negatively enriched (>98% purity), pooled splenic and LN B cells from WT, DKO, or TKO mice (n = 2–3) were pulsed with graded levels of anti-IgM for 3 h, then stimulated with CD40L and BAFF for 48 h. d Quantification of cell viability (top) and cell proliferation (bottom). e Ki67+ non-EF/GC B cells in chimeras (n = 5–7) from 5 dpi. Data in (b, d, e) represent mean ± 95% CI of two (d, e) or three (b) independent experiments. Data in (d) contain n = 6 total replicates per group. Statistical significance determined by one-way ANOVA and two-tailed Student’s t-test with Welch’s correction. *p < 0.05, ***p < 0.001, ****p < 0.0001 or indicated in subfigures. Source data are provided as a Source Data file.

Fig. 6. Lack of functional TLR signaling leads to altered BCR complex dynamics and failure to upregulate IRF4.

a Representative flow plots showing IRF4 and IRF8 expression in infected mice, highlighting clustering of EF PBs (left). Fold-difference compared to WT controls in IRF4 and IRF8 of non-EF/GC B cells from chimeras (n = 5–7) at 5 dpi (right). b Pre-enrichment baseline of IRF4 and IRF8 in B cells of each strain (left) and representative IRF4 versus IRF8 flow plots from cells stimulated with indicated anti-IgM concentrations (right). Colored numbers in plots correspond to each like-colored axis. c–d Fold-difference compared to non-stimulated WT B cells of IRF4 (c) and IRF8 expression (d) after treatment outlined in Fig. 5c. e Fold-difference of knockout versus non-stimulated WT B cells in cytoplasmic c-Rel measured by flow cytometry after 30-minute anti-IgM or LPS treatment (n = 3). f Fold-difference compared to non-stimulated WT B cells of total c-Rel expression after a 3 h anti-IgM pulse and 48 h culture in complete media only (n = 3). Data in (a, c–f) represent mean ± 95% CI of two independent experiments. Data in (c, d) contain n = 6 total replicates per group. Statistical significance determined by one-way ANOVA and two-tailed Student’s t-test with Welch’s correction. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Stars in (g, h) are Student’s t-test comparison to respective WT control. Source data are provided as a Source Data file.

Fig. 7. Sustained TLR-mediated inflammation generates strong EFRs in the draining LN after immunization.

a Mice (n = 7–12) were immunized s.c. with or without influenza in alum and with or without LPS, then boosted with either LPS or PBS on days specified, followed by analysis of draining LN. b Counts of major B cell subsets. c Quantification of HA-specific B cell subsets as in (b). d Flow plots of HA-specific B cells from each regimen in terms of proliferation and plasma cell differentiation (left) and IRF4 vs IRF8 signature (right, HA-sp. highlighted in red). e Quantification of HA-specific EF PBs, proliferation, and relative expression of IRF4. Data in (a, c–e) represent mean ± 95% CI of two (e[right]) or three (b, c, e [left, middle]) independent experiments. Statistical significance determined by one-way ANOVA and two-tailed Student’s t-test with Welch’s correction. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Source data are provided as a Source Data file.

Fig. 8. Repeated antigen exposure alone biases antigen-specific B cells towards a GC fate, requires sustained LPS exposure to polarize towards an EF fate.

a C57BL/6 mice (n = 7) were immunized s.c. with influenza and LPS in alum, then boosted with antigen alone or antigen with LPS and LPS alone on days specified, followed by analysis of draining LN. b–e Quantification of total HA B cells (b), Ki67+ HA B cells (c), HA GC B cells (d) and HA EF PBs (e). f Concentration of influenza-specific serum IgG at 10 days post-prime (n = 10–12) expressed in relative units (left) and fold difference relative to Antigen Only group (right). g, h Serum from primed/boosted mice at 10 days post-prime (n = 10) and naïve mice (n = 6) was transferred to naïve C57BL/6 mice prior to infection with a lethal dose (100 PFU) of influenza A/PR8 the next day. Shown is survival probability (g) and percent change in weight (h) by average (top) and individually (bottom) over the course of infection. Data in (b–f, h) represent mean ± 95% CI of two (b–e, g, h) or three (d) independent experiments. Statistical significance determined by one-way ANOVA and two-tailed Student’s t-test with Welch’s correction. *p < 0.05, **p < 0.01 ***p < 0.001, ****p < 0.0001. Source data are provided as a Source Data file.