T-bet+ CXCR3+ B cells drive hyperreactive B-T cell interactions in multiple sclerosis

Abstract

Multiple sclerosis (MS) is an autoimmune disease of the central nervous system (CNS). Self-peptide-dependent autoproliferation (AP) of B and T cells is a key mechanism in MS. Here, we show that pro-inflammatory B-T cell-enriched cell clusters (BTECs) form during AP and mirror features of a germinal center reaction. T-bet+CXCR3+ B cells are the main cell subset amplifying and sustaining their counterpart Th1 cells via interferon (IFN)-γ and are present in highly inflamed meningeal tissue. The underlying B cell activation signature is reflected by epigenetic modifications and receptor-ligand interactions with self-reactive T cells. AP+ CXCR3+ B cells show marked clonal evolution from memory to somatically hypermutated plasmablasts and upregulation of IFN-γ-related genes. Our data underscore a key role of T-bet+CXCR3+ B cells in the pathogenesis of MS in both the peripheral immune system and the CNS compartment, and thus they appear to be involved in both early relapsing-remitting disease and the chronic stage.

Keywords: B-T cell-enriched clusters; BTEC; CXCR3; IFN-gamma; T cells; T-bet+ B cells; autoproliferation; autoreactivity; hyperreactive B-T cell interaction; meninges; multiple sclerosis; self-peptides.

Graphical abstract

Figure 1

AP results in BTEC formation, which is BTK dependent (A and B) Staining of B cells (CD19 and CD20) and T helper cells (CD3 and CD4) after 7 days of (A) anti-IgM stimulation or (B) AP in PBMCs (natalizumab-treated RRMS = NAT) by flow cytometry (top panel) and by 4i (bottom panel). Nuclei were labeled with DAPI (blue). Images show increasing magnification from left to right, 49-(well), 4-, and 1-site view. CFSE stands for carboxy fluorescein succinimidyl ester. Scale bar: 200 μm. (C) AP response upon vehicle (top) or BTK inhibitor (BTKi) treatment (bottom) assessed by flow cytometry (left) and 4i (right) including staining for pERK (cyan), pAKT (green), and PCNA (red). Representative image from one donor (NAT, n = 3). Scale bar: 50 μm. (D) 2D projection of local morphology centroids, colored by the mean fluorescence intensity (MFI) values of PCNA, without (AP) or with BTKi treatment to visualize the spatial intensity distribution of cell proliferation. (E) Computation of nearest neighbor statistics using G(r) function on local morphology centroids, for AP upon vehicle or BTKi treatment to measure deviations from complete spatial randomness in favor of clustering or regular patterns. G(r) values above or below the simulation envelope imply clustering or dispersion, respectively. (F) Representative 4i image of BTECs following AP (n = 3 NAT). Images on the right show magnified 1-site view. Scale bar: 50 μm. (G) Top: 2D projection of cell morphology centroids, colored by B cells (green), T cells (red), and unclassified (gray). Bottom: G(r) function computed on cell morphology centroids based on (F). See also Figure S1 and Table S1.

Figure 2

AP+ B cells are increased in MS and bear characteristic features of germinal center-like formation (A and B) (A) Frequency of AP+ B cells in HD (n = 32), RRMS (n = 68, REM/REL nihil and NAT), and PMS patients (n = 13, nihil) (mean ± SEM; Kruskal-Wallis test). (B) Frequency of AP+ B cells in RRMS group, untreated RRMS in relapse (REL, n = 18) or remission (REM, n = 35), and natalizumab-treated RRMS (NAT, n = 15) (mean ± SEM; Kruskal-Wallis test). Dotted line indicates mean value of AP+ B cells and below the number of donors with >2.6% mean value. (C) RNA-seq of CFSE^hi and CFSE^dim (AP+) B cells (n = 6 REM) for a predefined gene set characteristic for germinal center (GC) formation. The differential expression is expressed by the Z score based on the reads per kilobase of transcript per million reads mapped (RPKM) values. The gene set was tested as customized KEGG pathway for significance (Fisher’s exact test). (D) Intracellular AID expression in CFSE^hi and CFSE^dim (AP+) memory B cells. (Left) Representative histogram plot; (right) MFI of AID expression corrected by isotype control (n = 6 REM; mean ± SEM; Mann-Whitney U test). (E) Gating strategy of switched memory B cell (G1), plasmablasts (G2), and plasma cells (G3) in CFSE^hi and CFSE^dim B cells upon AP. (F) Frequency of plasmablasts/-cells following AP (n = 6 REM; mean ± SEM; Mann-Whitney U test). (G and H) Expression of IgG and IgM upon AP. (G) Representative histogram plot from one donor and (H) distribution across B cell subsets and donors (n = 5 REM). (I) scVDJ-sequencing-derived BCR isotype usage in AP+ and resting B cells across donors (n = 16 donors; HD: n = 5, REM: n = 6, NAT: n = 5) and the proportion of clonally expanded B cells. (J) SHM counts in AP+ and resting B cells stratified by their BCR isotype (n = 16) (t test with false discovery rate [FDR] < 0.01 and a location shift of 1 base). See also Figures S2 and S6 and Table S1.

Figure 3

AP+ B cells reveal abundant expression of IFN-γ-driven markers and dependency on JAK-STAT pathway (A) Workflow of isolated, pre-stained, and sample barcoded CFSE-labeled B cells following AP (n = 4 REM) for surface marker screening using LEGENDscreen. (B) Gating strategy for sample deconvolution based on CD45 barcoding, subgating on CD27⁻CFSE^hi naive resting B cells (gray), CD27+CFSE^hi memory resting B cells (blue), and CD27+CFSE^dim memory AP+ B cells (red). Histogram plots of CD80 expression and respective isotype control. (C) Heatmap summarizing surface marker expression levels in B cell subsets of all 4 donors (D1–D4). Categorization in low, intermediate, and high expressing proteins for better visualization. The target protein median fluorescence intensity (MFI) was corrected with its respective isotype control. Only surface markers with MFI > 250 and MFI ratio changes of >1.5 or <0.5 are shown. The red box highlights characteristic protein expression of age-associated B cells. (D) Intracellular T-bet expression in CXCR3-CFSE^hi (G1), CXCR3+CFSE^hi (G2), and CXCR3+CFSE^dim (AP+) (G3) B cells and T helper cells. (Upper) Representative plots from one patient and (bottom) MFI of each donor (n = 6 REM; mean ± SEM; Kruskal-Wallis test). (E) Intracellular phosphorylation of STAT1, STAT4, and STAT5 in CFSE^hi and CFSE^dim B cells upon AP. (Left) Representative histogram plot and (right) MFI in all donors (n = 4 REM; mean ± SEM; Mann-Whitney U test). (F) AP of B and T cells upon treatment with vehicle control or JAK-STAT inhibitor baricitinib (100 nM). (Upper) Representative dot plots from one patient and (bottom) summary of all donors (n = 6 REM; mean ± SEM; Kruskal-Wallis test). See also Figure S3 and Table S1.

Figure 4

AP+ CXCR3+ B cells are enriched in RRMS and PMS and promote Th1 responses (A) Frequency of CXCR3− or CXCR3+ B cells in HD (n = 13), REM (nihil; n = 16), and PMS (nihil; n = 5) upon AP (left) or IgM stimulation (right) (mean ± SEM; Kruskal-Wallis test). Representative gating strategy is shown from one donor. (B and C) Correlation of CFSE^dim B and T cells after (B) AP and (C) anti-IgM in HD (n = 13), REM (n = 16), and PMS (n = 5) (Pearson’s correlation test). (D) Activation of CD4⁺ T cells (HLA-DR+) upon AP co-culture with autologous CD27⁻ naive (nB), CXCR3-CD27⁺, or CXCR3+CD27⁺ memory (mB) B cells (NAT n = 3; Kruskal-Wallis test). (E) Secretion of Th1, Th2, and Th17 cytokines after AP of B-T cell co-cultures (NAT n = 3). Levels of cytokine secretion are shown as heatmap. (F) Frequency of Th1 and Tfh cells upon AP. (Left) Representative dot plots from one patient and (right) distribution across donors in CFSE^hi (non-proliferating = NP) and CFSE^dim (AP) CD4⁺ T cell phenotypes (n = 7 NAT; mean ± SEM; Mann-Whitney U test). (G) Intracellular expression of CXCL9 in CFSE^hi (NP) and CFSE^dim (AP+) B cells. (Top) Representative dot plot from one donor and (bottom) summary of all donors (REM n = 6, mean ± SEM; Mann-Whitney U test). (H) Secretion of CXCL9 upon AP in HD (n = 30), RRMS (n = 47: REM n = 32; NAT n = 15), and rituximab-treated RRMS (RTX n = 14) (mean ± SEM; Kruskal-Wallis test). See also Figure S4 and Table S1.

Figure 5

AP+ B cells display transcriptional and epigenetic changes in key inflammatory markers of CXCR3+ B cells (A) Gating strategy and purity from sorting of resting (CFSE^hi) and AP+ (CFSE^dim) memory B cells (representative, n = 6 NAT). (B) Global transcriptome (n = 5) and methylome changes (n = 6) in AP+ vs. resting memory B cells. (C) Pathway analysis of DEGs in AP+ B cells. ∗: FDR < 0.01, ∗∗: FDR < 0.001, ∗∗∗: FDR < 0.0001. (D) Gene set enrichment analysis of genes characteristic for age-associated B cells (ABCs). The enrichment score is illustrated in green with differences in expression ranked from high (red) to low (blue). (E) Heatmap illustrating expression of ABC-related genes in AP+ vs. resting memory B cells. The Z score is based on DESeq2 normalized gene expression values. (F) Heatmap illustrating methylation levels of CpGs in key ABC genes in AP+ vs. resting memory B cells. All CpGs located within promoter, promoter-flanking, and enhancer regions of genes with p < 0.01 are shown; differential methylation positions (DMPs) with p value < 0.001 and difference ≥ 10% are highlighted in bold. See also Figures S2 and S4 and Tables S1 and S2.

Figure 6

Single-cell RNA-seq of AP+ CXCR3+ B cells reveals abundant enrichment of IFN-γ-response genes in MS (A) Uniform Manifold Approximation and Projection (UMAP) of 99.678 single cells from equal sample size collected from resting and AP+ B cells of HD (n = 5), REM (n = 6), and NAT (n = 5) samples, showing the annotation of 8 clusters: naive B cells, intermediate memory B cells (iMBCs), memory B cells (MBC_1 - MBC_4), and plasmablasts (PB_1 - PB_2). Each dot corresponds to a single cell, colored according to cell cluster. (B) Key phenotypic marker gene expression and abundance across B cell subsets. The color code depicts the normalized mean expression of the gene, and the numbers denote the frequency of cells for which each gene is detected. (C) Clonotypes that span various CXCR3+ B cell subsets. The numbers indicate the number of clonotypes spanning across two respective B cell subsets. (D) IFN-γ response gene set enrichment analysis for CXCR3+ B cells in HD,REM and NAT. (E) DEGs (FDR < 0.01, mean expression < 5% quantile) of the IFN-γ response pathway (Gene ID: 3458) between AP+ and resting CXCR3+ B cells across the different B cell subsets. Color coding depicts fold change of AP vs. resting. (F) Differential gene expression analysis between AP+ and resting within CXCR3− (red) and CXCR3+ (blue) B cells, respectively. Only upregulated genes in AP with logFC > 0.5 are shown with significant DEGs highlighted in blue (Wilcoxon test). (G) DEGs that are exclusively upregulated in AP+ CXCR3+ B cell subsets of MS and not in AP+ CXCR3− B cells. Boxplots describe the pseudobulked expression level and donor-to-donor variability in HD (green), REM (yellow), and NAT (purple). The heatmap shows condition-wise mean expression of the DEGs with logFC (AP vs. resting) values indicated inside the boxes. ∗: FDR < 0.05. See also Figures S5 and S6 and Tables S1, S3, and S4.

Figure 7

Single-nuclei RNA-seq reveals age-associated B cell phenotypes and IFN-γ⁺ T cells in inflamed meningeal MS tissue (A) UMAP of 24,939 single nuclei from highly inflamed meningeal and adjacent gray matter tissue sections (n = 3 MS, 2 biological replicate samples per donor and brain tissue block) and the 16 cell clusters identified by Harmony. Each dot corresponds to a single cell. (B) Annotation of the 16 clusters with respective labels for cell populations based on key marker genes. (C) Key phenotypic marker gene expression for all 16 clusters and their respective cell lineage annotations. The color code depicts the scaled mean log expression of the gene in each cluster. The mean expression of the respective gene in the cluster is denoted by the circle size. (D) UMAP of T cells with expression of CD3E, CD4, and CD8. Each dot corresponds to a single cell, colored by a gradient for expression level, which is scaled so that 0 equals zero expression and 1 equals 95^th percentile of log expression. (E) Relative and normalized expression levels of key characteristic T cell markers. (F) UMAP plot of B cells with expression of IGHG1, IGHD, and IGHM. (G) Relative and normalized expression levels of characteristic B cell, plasma cell, and age-associated B cell markers. See also Figure S1 and Tables S1 and S5.