Double-negative-2 B cells are the major synovial plasma cell precursor in rheumatoid arthritis

Abstract

B cells are key pathogenic drivers of chronic inflammation in rheumatoid arthritis (RA). There is limited understanding of the relationship between synovial B cell subsets and pathogenic antibody secreting cells (ASCs). This knowledge is crucial for the development of more targeted B-cell depleting therapies. While CD11c+ double-negative 2 (DN2) B cells have been suggested as an ASC precursor in lupus, to date there is no proven link between the two subsets in RA. We have used both single-cell gene expression and BCR sequencing to study synovial B cells from patients with established RA, in addition to flow cytometry of circulating B cells. To better understand the differentiation patterns within the diseased tissue, a combination of RNA-based trajectory inference and clonal lineage analysis of BCR relationships were used. Both forms of analysis indicated that DN2 B cells serve as a major precursors to synovial ASCs. This study advances our understanding of B cells in RA and reveals the origin of pathogenic ASCs in the RA synovium. Given the significant role of DN2 B cells as a progenitor to pathogenic B cells in RA, it is important to conduct additional research to investigate the origins of DN2 B cells in RA and explore their potential as therapeutic targets in place of the less specific pan-B cells depletion therapies currently in use.

Keywords: antibody secreting cells; double-negative-2 (DN2) B cells; rheumatoid arthritis; single-cell sequencing; synovium.

Figure 1

DN2 B cells are significantly enriched in the blood and synovium of RA patients. (A) Peripheral blood mononuclear cells (PBMCs) were stained for flow cytometry, live CD19+ve B cells were gated before selecting either the double negative (DN; CD27^-ve IgD^-ve) or the naïve (CD27^-ve IgD^+ve) cells. The expression of CD21, CD24, and CD38 were used to gate to double negative 2 (DN2) and activated naïve B cells (aNAV). Representative plots show the expansion of DN2 and activated naïve B cells (aNAV) in RA. (B) Percentage of DN2 B cells within the CD19^+ve B cell population. n=34 RA patients and 15 healthy controls. P<0.0001 by Mann-Whitney test. (C) Percentage of aNAV cells within the CD19^+ve B cell population. n=34 RA patients and 15 healthy controls. P=0.0329 by Mann-Whitney test. (D) Representative histogram plots show the expression of relevant surface markers on DN2 (filled grey), resting naïve (red line), and switched memory B cells (dotted line). (E) Flow cytometry of paired peripheral blood and synovium. DN2 B cells are significantly increased in the synovium of RA patients compared to blood (n=5). P=0.0257 by Paired t test. (F) The ratio of DN2:DN1 B cells is significantly increased as a result of DN2 cell expansion. P<0.0001 by Mann-Whitney test. * P ≤ 0.05 and **** P ≤ 0.0001. (G) Scatterplot of the percentage of DN2 cells and age of RA donors. Spearman’s r coefficient. * P ≤ 0.05 and **** P ≤ 0.0001.

Figure 2

Single-cell sequencing of synovial B cells unveils the heterogeneity of B cells in the rheumatoid joint. (A) Overview of sample preparation: CD19+ve B cells were isolated from the synovial tissue of RA patients (n=3) before single cell RNA-sequencing with paired BCR sequencing. Created with BioRender.com. (B) Uniform manifold approximation and projection (UMAP) projection of all 27,053 synovial B cells from RA patients (n=3). Unsupervised clustering identified 12 clusters, the B cell subsets were annotated using the expression of known subset markers and gene signatures. DN2, double negative 2; HSP, heat shock protein; ASC, antibody secreting cell. (C) Stacked bar plot showing the distribution of the 12 clusters over the 3 samples. (D) Dotplot demonstrating the expression of key gene markers for each cluster. (E) Violin plots demonstrating the expression of MS4A1 (CD20), JCHAIN, ITGAX (CD11c), and CD86 across the 12 clusters.

Figure 3

DN2 B cells are a heterogenous group of cells, primed to present antigen and become ASC. (A) Volcano plot showing the transcriptome analysis DEGs (red, log2FC is >|1| and P value > 10e-6; blue, log2FC is <|1| and P value > 10e-6; grey, nonsignificant genes) of DN2 B cells compared to all other non-ASC cells. (B) UMAP projection of scRNA-seq data from 27,053 synovial B cells from RA patients (n=3) with the module scores for (i) human SLE DN2 B cells and (ii) CD21^lo CD23^lo B cells from mice, the contour overlay represents the position of the top 10% of module scores. (C) The top 8 pathways enriched in the DN2 B cells as identified by gene set enrichment analysis using gene sets derived from the GO Biological Process ontology.

Figure 4

Trajectory inference shows DN2 B cells are ASC precursors in the RA synovium. (A) RNA velocity stream plot, projected onto the UMAP of scRNA-seq data from 27,053 synovial B cells from RA patients (n=3). (B) Minimum spanning tree demonstrating the full lineage structure identified by Slingshot. Lineage 1 starts in the Naïve 1 cluster, Lineage 2 in the Cell-cycling cluster, and Lineage 3 in the Early Activation cluster. (C) Lineage 1 identified by Slingshot with pseudotime colouring. (D) Slingshot trajectory identified after clustering 1,891 DN2 B cells and 5,786 ASCs from RA synovial tissue (n=3). (i) The trajectory starts in the DN2 cluster before transitioning into ASCs. (ii) The lineage identified by Slingshot with pseudotime colouring. (E) Expression of MS4A1 (CD20), ITGAX (CD11c), and JCHAIN in DN2 B cells and ASCs.

Figure 5

BCR sequences from synovial B cells show hallmarks of autoimmunity. (A) CDR3 length distribution for the (i) heavy chain and the (ii) light chains for Sample 1 (red column), Sample 2 (green column), and Sample 3 (blue column). (B) (i) UMAP plot of the 7,825 B cells with a productive heavy and light chain coloured by the BCR isotype. (ii) Stacked bar plot showing the distribution of the isotypes across the 12 clusters. (C) (i) UMAP plot of the 7,825 B cells with a productive heavy and light chain coloured by the BCR mutation count. (ii) Stacked bar plot showing the distribution of mutation counts across the 12 clusters. (D) Distribution of mutation frequencies for IGHM, IGHA, and IGHG for the three RA samples.

Figure 6

BCR sequences from DN2 B cells share identical CDR3 sequences with ASCs. (A) Stacked barplot showing the clonal family sizes across the three samples. (B) Network graphs of a representative repertoire from each RA patient. Each node represents a cell, coloured according to the isotype, with up to three edges representing the most closely related heavy CDR3 amino acid sequences according to Levenshtein distance. (C) Four clonal lineage trees where a DN2 cell shares an identical heavy CDR3 amino acid sequence with an ASC. Orange circles indicate a cell from the DN2 cluster, dark blue IgD^-ve memory, dark brown ASC 1, bright blue ASC 2, pale brown ASC 3, white inferred sequence and black is germline. (D) Characteristics of the 4 clonal families where a DN2 cell shares an identical junction sequence with and ASC.