A transcriptionally distinct subset of influenza-specific effector memory B cells predicts long-lived antibody responses to vaccination in humans

Abstract

Seasonal influenza vaccination elicits hemagglutinin (HA)-specific memory B (Bmem) cells, and although multiple Bmem cell populations have been characterized, considerable heterogeneity exists. We found that HA-specific human Bmem cells differed in the expression of surface marker FcRL5 and transcriptional factor T-bet. FcRL5⁺T-bet⁺ Bmem cells were transcriptionally similar to effector-like memory cells, while T-bet^negFcRL5^neg Bmem cells exhibited stem-like central memory properties. FcRL5⁺ Bmem cells did not express plasma-cell-commitment factors but did express transcriptional, epigenetic, metabolic, and functional programs that poised these cells for antibody production. Accordingly, HA⁺ T-bet⁺ Bmem cells at day 7 post-vaccination expressed intracellular immunoglobulin, and tonsil-derived FcRL5⁺ Bmem cells differentiated more rapidly into antibody-secreting cells (ASCs) in vitro. The T-bet⁺ Bmem cell response positively correlated with long-lived humoral immunity, and clonotypes from T-bet⁺ Bmem cells were represented in the secondary ASC response to repeat vaccination, suggesting that this effector-like population predicts influenza vaccine durability and recall potential.

Keywords: T-bet; durable immunity; influenza vaccination; memory B cells; plasmablasts.

Figure 1.. Seasonal influenza vaccination elicits distinct HA-specific Bmem cell subsets.

(A-C) FACS plots showing HA-tetramer binding versus CD19 expression (A) and FcRL5 versus T-bet expression (B) by HA-specific IgD^neg B cells from blood of a D7 post-IIV HD. Histograms (C) showing cell surface marker expression by the H1-specific IgD^neg T-bet⁺ (red) and T-bet^neg (blue) B cells. (D-E) IGH V_H BCR repertoire analysis of sort-purified circulating H3-specific FcRL5⁺ and FcRL5^neg IgD^neg B cells, IgD⁺ naïve B cells, and IgD^negCD38^hi PBs from 2 HD on D7 post-IIV. Isotype distribution (D) in each D7 subset (U = “unidentified”). Distribution of V_H sequence mutation frequencies (E) in H3-specific IgD^neg FcRL5⁺ and FcRL5^neg cells. Number of sequences in each population indicated. (F-G) Shared lineages between D120 total IgD^neg B cells (y-axis) and HA-specific FcRL5⁺ (F) or FcRL5^neg (G) IgD^neg B cells (x-axis) from 3 HDs on D7 or D14 post-IIV. Axes indicate size of shared lineages (symbols) defined as ≥2 sequences in each lineage in each population. Numbers of shared lineages (n) for each donor indicated. Larger symbols represent multiple shared lineages of the same sizes. (H-I) Frequencies of T-bet⁺ and T-bet^neg cells within H1-specific (H) or H3-specific (I) IgD^neg B cell subsets measured in each individual (n=19) on D7 post-IIV. ns = not significant. (J) V_H BCR lineages in sort-purified naïve and D7 H3-specific FcRL5⁺ or FcRL5^neg IgD^neg B cells with lineages ordered by size (y-axis) and % of total number of sequences (x-axis). Data reported as Diversity index 20 (D₂₀) values, total number of sequences per subset, and lineages per subset. (K) Lineage sharing between D7 FcRL5⁺ and FcRL5^neg subsets reported as percentage of shared non-singleton lineages. The numbers of shared and total non-singleton lineages per donor and subset also reported. Additional BCR repertoire data reported in Table S1, Fig 1F–J and Fig. S1N. See Fig. S1 for gating strategy and additional phenotypic characterization of B cell subsets.

Figure 2.. H1-specific FcRL5⁺ and FcRL5^neg Bmem cells are transcriptionally and epigenetically distinct.

(A-J) RNA-seq and ATAC-seq analyses performed on D7 post-IIV B lineage subsets from 5–6 HDs. (A-B) PCA of RNA-seq (A) and ATAC-seq (B) data sets from D7 B lineage subsets. (C) PCA of cell cycle genes expressed by D7 B cell subsets. (D-F) PRDM1 (D), XBP1 (E) and IRF4 (F) mRNA expression by D7 B cell subsets. (G-H) Accessibility and expression of published IRF4-regulated ASC genes by D7 B lineage subsets. Chromatin accessibility surrounding IRF4 ASC regulon genes (G) is reported as the mean peak accessibility for all peaks mapping to IRF4 ASC regulon genes. GSEA (H) comparing the RNA-seq ranked gene list from D7 H1-specific IgD^neg FcRL5⁺ vs FcRL5^neg B cells to the IRF4 ASC regulon gene set. (I-J) PR identified TFs predicted to regulate gene networks in D7 PBs and IgD^neg H1-specific Bmem cell subsets. Data reported as logFC in PR score and logFC in mRNA expression of each TF with D7 FcRL5⁺ Bmem cells over PB (I) or D7 FcRL5^neg Bmem cells over PBs (J). Selected TFs indicated. (K-P) Single cell RNA-seq, BCR-seq and CITE-seq analyses of sort-purified HA-specific IgD^neg B cells from 2 HD between D7-D21 post-IIV. (K) UMAP showing transcriptional clusters (C0-C6) from single cell RNA-seq data. (L) BCR-seq analysis of single cells in Clusters C0-C4. Data reported as V_H gene mutation frequency. Unswitched (IgM⁺, red dots) and switched (black dots) cells are shown. (M) TBX21 expression reported as % of cells in each cluster with at least one TBX21 transcript. (N) CITE-seq analysis of FcRL5 protein in HA-specific IgD^neg in Clusters C0-C4. Expression in single cells normalized using scTransform and shown as violin plots. (O-P) GSEA comparing the ranked gene list from D7 H1-specific IgD^neg FcRL5⁺ vs FcRL5^neg B cells to the upregulated or downregulated DEGs in each single cell cluster. Data shown include single cell DEG sets enriched in D7 H1-specific IgD^neg FcRL5⁺ (O) or FcRL5^neg (P) B cells. p_nom values indicated. Data sets for RNA-seq, ATAC-seq, PR analysis, GSEA gene sets, and sc-seq datasets are provided in Tables S2–S4. Fig. S2 has additional RNA expression analysis from D7 (S2A-B) and D14 (S2C-I) B lineage subsets. Statistical analyses performed using one-way ANOVA (D-F, N). *p< 0.05, **, p<0.01, *** p<0.001, **** p <0.0001 ns= non-significant.

Figure 3.. Division of H1-specific B cells into FcRL5⁺ effector-like and FcRL5^neg stem-like Bmem cells.

(A) Volcano plot showing DEG upregulated in D7 H1-specific IgD^neg FcRL5⁺ (red) and FcRL5^neg (blue) Bmem cells. (B) GSEA comparing the ranked gene list from D7 H1-specific IgD^neg FcRL5⁺ and FcRL5^neg B cells to DEG that are either upregulated (KAECH) or down-regulated (GOLDRATH) in effector vs memory CD8⁺ T cells. (C) TFs identified by PR as regulators of the D7 H1-specific IgD^neg FcRL5⁺ and D7 H1-specific IgD^neg FcRL5^neg gene networks. (D-E) Shared BCR V_H lineages between H1-specific IgD^neg Bmem cells isolated from the same donor at sequential timepoints. Data reported as percentage of shared lineages between different Bmem cell populations over time. The numbers of non-singleton lineages identified in each subset at each timepoint and the numbers of shared lineages for each pairwise comparison indicated. (F) TFs identified by PR as regulators of D14 H1-specific IgD^neg FcRL5⁺ and D14 H1-specific IgD^neg FcRL5^neg gene networks. Data sets for BCR repertoire, RNA-seq, PR, and GSEA genesets provided in Tables S1–S3.

Figure 4.. The transcriptome and epigenome of FcRL5⁺ Bmem cells display T-bet associated changes.

(A-B) Box plots showing chromatin accessibility surrounding (within 50 bp) defined T-bet motifs (A) and previously identified T-bet ChIP-seq peaks (B) in memory FcRL5⁺, FcRL5^neg and PB populations. (C) Expression of transcriptional regulators (n=17) that were identified as DEG between FcRL5⁺ and FcRL5^neg Bmem cells and were also assigned to at least one DAR that contained a T-bet binding motif (green dots) and/or ChIP-seq T-bet binding site (purple dot). Data shown as a heat map displaying per-regulator gene z-scores of log expression in the indicated subsets. PR scores for each regulator shown as absolute value log₂ FC of FcRL5⁺ over FcRL5^neg cells. (D-G) RNA-seq expression and genome ATAC-seq plots for TCF7 (D), BACH2 (E), TOX2 (F) and ARID3A (G). Genome ATAC-seq plots aligned with ChIP-seq T-bet binding sites and with ATAC-seq from resting naïve B cells. Boxes indicate loci with significant DAR. Vertical black bars indicate HOMER-predicted consensus T-bet binding motifs. Super-enhancers in BACH2 locus identified in CD19⁺ and CD20⁺ human B cells shown as horizontal black bars. Shaded gray triangles indicate location of DAR-containing regions within each locus. Statistical analyses were performed using one-way ANOVA (D-G). *p< 0.05, **, p<0.01, *** p<0.001, **** p <0.0001 ns= non-significant. Data sets for RNA-seq, ATAC-seq, transcriptional regulators and PR analyses are provided in Tables S2–S3.

Figure 5.. Metabolic gene expression differs between FcRL5⁺ effector and FcRL5^neg stem-like Bmem cells.

(A-C) Identification of signaling and metabolic gene modules in FcRL5⁺ Bmem cells. Clustering (A) of leading-edge genes from 185 GO gene lists (mSigDB v.7) that were enriched (FDR q < 0.01) for expression in D7 FcRL5⁺ relative to D7 FcRL5^neg Bmem cells. Nine clusters (A-I) with overlapping leading edge gene sets were identified. Eleven prototypic GO gene lists (lists i-xi) representing the 9 GO gene set clusters were selected and the leading-edge genes from gene lists were functionally annotated and used for PCA (B) comparing D7 and D14 H1-specific IgD^neg FcRL5⁺ and FcRL5^neg Bmem cells. The leading-edge genes were grouped into 14 signaling and metabolic modules (C) with representative genes from each module shown as the difference in log₂ FC in expression by D7 H1-specific IgD^neg FcRL5⁺ and FcRL5^neg Bmem cells. Red dots indicate individual samples with FDR q<0.05. (D) PCA using the 14 module gene sets comparing naïve B, D7 PB and D7 or D14 H1-specific IgD^neg FcRL5⁺ and H1⁺ IgD^neg FcRL5^neg Bmem cells. (E-I) GSEA comparing the RNA-seq ranked gene list from D7 H1-specific IgD^neg FcRL5⁺ and FcRL5^neg B cells to curated gene lists for mTORC1 signaling (E), fatty acid metabolism (F), mitochondrial metabolism (G), oxidative stress (H) and proteasome complex (I). GSEA p_nom-values indicated. Table S2 and Fig. S3 include clustering and module data, leading-edge genes from prototypic GO-gene lists, and gene lists for the signaling and metabolism modules and GSEA genesets.

Figure 6.. FcRL5⁺ Bmem cells are metabolically poised to secrete Ab.

(A-D) Metabolic flow assays using tonsil-derived matched (n=6–8 donors) FcRL5⁺ and FcRL5^neg IgD^negCD27⁺ Bmem cells. ROS activity (A) measured with 2, 7–dichlorohydrofluorescein diacetate (H₂DCF). Mitochondrial mass (B) measured with Mitotracker Green. Membrane lipid composition (C) measured with BODIPY510. mTORC1 activation (D) measured with anti-phospho-S6 (pS6) kinase Ab. Data reported as geometric mean fluorescence (gMFI) and shown as paired analysis between FcRL5⁺ and FcRL5^neg cells purified from the same donor. (E-I) Proliferation and differentiation of purified donor-matched naïve (brown), FcRL5⁺ (red) or FcrL5^neg (blue), IgD^negCD27⁺CD38^lo/med tonsil Bmem cells. CTV labeled cells were stimulated with R848, IL-21, IL-2 and IFN-γ and assessed on D2. Cell survival (E) and frequency of cells per cell division (F) on D2 were determined by flow cytometry and reported as the mean ± SD (n = 5 donors). ASCs in D2 cultures enumerated by flow cytometry (G) with frequencies of total CD27^hiCD38^hi ASCs (H) and percentage of ASCs in each cell division (I) reported. (J-K) GSEA comparing the ranked gene list from D7 H1-specific IgD^neg FcRL5⁺ and FcRL5^neg Bmem cells to genes induced by the mTORC1-dependent B cell activation UPR (J) or the mTORC1-dependent PC inductive UPR (K). (L) Flow cytometry showing H1 tetramer-binding by D7 T-bet⁺ and T-bet^neg IgD^neg gated B cells. (M) Extracellular expression of IgG and IgA by D7 IgD^neg HA-specific FcRL5^neg and FcRL5⁺ Bmem cells after sequential staining with Abs to FcRL5, IgD, CD19 and HA tetramers (H1 and H3) followed by anti-IgG and anti-IgA Abs. (N) RNA-seq analysis of IgH transcripts (IgHG1, IgHG2, IgHG3, IgHG4, IgHA1, IgHA2) from sort-purified D7 IgD^neg HA-specific FcRL5^neg and FcRL5⁺ Bmem cells. (O-R) Image Stream analysis of sort-purified D7 post-IIV CD19⁺IgD^neg cells measuring expression of CD38, plasma membrane Ig (detected with H1-APC tetramer), intracellular T-bet, and intracellular Ig (detected with H1-PE tetramer). Representative images (O) of individual CD38^hi PBs (n=211 cells), H1 tetramer-binding T-bet⁺ (n=195 cells) and T-bet^neg (n=62 cells) cells. Intensity of intracellular (IC) H1 tetramer staining (P) and extracellular (EC) H1 tetramer staining (Q) and the ratio of EC/IC H1 tetramer staining (R) in each T-bet^neg, T-bet⁺ and CD38^hi cell was determined. Ratio in (R) shown as median for the individual cells within each subset and error bars indicating the 95% confidence interval. Fig. S4 shows additional analyses of tonsil B cell subsets, gated or purified as in (Fig. S4A–C) and examined in metabolism assays (Fig. S4D–G) or after in vitro stimulation for 2–3 days (Fig. S4H–N). Table S2 and Fig. S5A provide mTORC1-controlled UPR gene set data. Table S5 and Fig. S5C–D report BCR affinity measurements. Fig. S5B shows HA-binding expression over 8 weeks post-IIV and Fig. S5E shows expression of transmembrane and secretory Ig exon transcripts. Statistical analyses were performed using Wilcoxon matched-pairs signed rank tests (A-D, N), one-way ANOVA with Kruskal-Wallis Multiple Comparisons Testing (P,Q) and one-way ANOVA with Tukey’s multiple comparison testing (E, H, R). *p< 0.05, **, p<0.01, *** p<0.001, **** p <0.0001 ns= non-significant.

Figure 7.. HA-specific IgD^neg T-bet⁺ eBmem cells contribute to secondary ASC responses and predict enduring humoral immunity.

(A) Composition of H1 and H3 Ags in 2015–2017 IIV. (B-D) Enumeration of D7 H1- or H3-specific IgD^neg T-bet⁺ and T-bet^neg B cells in HD sequentially immunized with IIV between 2015 and 2016 (n=6) or between 2016 and 2017 (n=5). FACS analysis showing H1-CA09⁺ Bmem cells over 2 years (2015–2016) from a HD immunized with the same CA09-H1 Ag each year (B) or showing HK14-H3 Bmem cells over 2 years (2016–2017) from a HD vaccinated with the same HK14-H3 Ag each year (C). Frequencies (D), represented as mean with standard error, of the H1-specific or H3-specific T-bet⁺ or T-bet^neg B cells for 11 HD vaccinated over 2 years with IIV containing matched H1 or H3 Ags. ** p = 0.002 Wilcoxon matched-pairs signed rank tests. (E-G) V_H BCR repertoire analysis performed on H3-specific IgD^neg FcRL5⁺ and FcRL5^neg Bmem cells isolated from HDs (n=3) on D7 (E-F) or D14 (G) post-IIV and PBs isolated from the same donors one year later on D7 following re-vaccination. Data reported as percentage of shared non-singleton lineages between year 1 Bmem cells and year 2 PBs. The number of shared lineages between the PBs and Bmem cell subsets and the number of non-singleton lineages identified by population and timepoint are indicated. (H-K) Correlation analysis of D7 IgD^neg HA-specific Bmem cell responses (frequencies in blood, x axis) and vaccine-specific Ab responses (FC in titers between D0 and D120, Y axis) after IIV in HD (n=19). Comparisons include FC in H1-specific IgG titers vs % D7 H1-specific T-bet^neg Bmem cells (H); FC in H3-specific IgG titers and % D7 H3-specific T-bet^neg Bmem cells (I); FC in H1-specific IgG titers and % D7 H1-specific T-bet⁺ Bmem cells (J); and FC in H3-specific IgG titers and % D7 H3-specific T-bet⁺ Bmem cells (K). Spearman correlation (r) and significance (p) are provided. ns = no significant correlation. See Table S1 for additional correlation analyses.