Identification of an anergic BND cell-derived activated B cell population (BND2) in young-onset type 1 diabetes patients

Abstract

Recent evidence suggests a role for B cells in the pathogenesis of young-onset type 1 diabetes (T1D), wherein rapid progression occurs. However, little is known regarding the specificity, phenotype, and function of B cells in young-onset T1D. We performed a cross-sectional analysis comparing insulin-reactive to tetanus-reactive B cells in the blood of T1D and controls using mass cytometry. Unsupervised clustering revealed the existence of a highly activated B cell subset we term BND2 that falls within the previously defined anergic BND subset. We found a specific increase in the frequency of insulin-reactive BND2 cells in the blood of young-onset T1D donors, which was further enriched in the pancreatic lymph nodes of T1D donors. The frequency of insulin-binding BND2 cells correlated with anti-insulin autoantibody levels. We demonstrate BND2 cells are pre-plasma cells and can likely act as APCs to T cells. These findings identify an antigen-specific B cell subset that may play a role in the rapid progression of young-onset T1D.

Graphical abstract

Figure 1.

Unsupervised clustering reveals novel B cell subpopulations among T1D and HC donors. (A) Overall schematic of the study and unsupervised clustering analysis. Insulin-binding and tetanus-binding B cells were enriched from the peripheral blood of T1D (n = 38) and HC (n = 40) donors, stained with a panel of B cell–focused antibodies, and processed for mass cytometry. (B) UMAP plots of B cell populations identified through clustering by surface marker expression for T1D and HC donors using mass cytometry. 12 unique B cell subsets were identified for the T1D and HC groups. IgM memory B cells were found only in HC, while IgD memory B cells were found only in T1D subjects (cluster 6). Autoreactive anergic B_ND cells (CD27⁻ IgM^lo IgD⁺) were divided into two subpopulations, B_ND1 and B_ND2, based on differences in activation markers. B_ND2 cells exhibit increased surface marker expression of CD69, CD80, CD95, and CD11c compared with B_ND1 cells. (C) Violin plots of insulin-binding levels for each B cell population identified for T1D and HC donors. The red dotted line indicates the cutoff for true insulin-binding versus non-binding B cells. The cutoff was determined using T cell binding and unenriched B cells as negative controls. B_ND1, B_ND2, and IgD memory B cells were enriched in insulin-reactivity in T1D subjects, while B_ND2, CD11c⁺ MN, and DN2 cells were enriched in insulin-reactivity in HC. (D) Violin plots of tetanus-binding levels for each B cell population for comparison to insulin-binding levels.

Figure S1.

Heatmap of surface marker expression levels in the B cell subsets identified by unsupervised clustering for T1D and HC donors.

Figure S2.

UMAP projections for insulin-binding-only, tetanus-binding-only, and non-binding-only B cells from T1D and HC donors.

Figure 2.

Surface marker expression of B_ND2 cells demonstrates these cells are recently activated and enriched in insulin reactivity. (A) Manual gating scheme for identification of MN, B_ND1, B_ND2, and DN2 populations. (B) Comparison of surface marker expression levels of various activation and inhibitory molecules for MN, B_ND1, B_ND2, and DN2. Antibody staining intensity is represented as median metal intensity (MdMI). B_ND2 cells exhibit significantly increased expression of markers of activation compared to B_ND1 cells, which is similar to DN2 cells. B_ND2 cells maintain surface expression of the inhibitory receptors CD72 and CD22 compared with DN2 cells. (C) Comparison of B_ND2 cells with surface markers associated with PB/plasma cells. B_ND2 cells have significantly reduced levels of markers associated with PB/plasma cells but are increased in expression of CD71 and CD38 compared with DN2 cells. PB were gated as CD19⁺ CD38hi CD27hi. (D) Comparison of the frequency of insulin- and tetanus-positive B cells among various B cell subsets. B_ND1 and B_ND2 cells are enriched in insulin reactivity, but not tetanus reactivity, compared with DN2 cells. (E) Comparison of the frequency of IgG⁺ B cells among the various B cell subsets. B_ND2 cells have increased expression of surface IgG compared to B_ND1 cells but is decreased compared with DN2 cells. Data generated from T1D (n = 38) and HC (n = 40) donors. Statistical significance determined by mixed-effects model. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Figure 3.

Insulin-binding B_ND2 cells are increased in young-onset T1D subjects and correlate with insulin autoantibody levels. (A) Gating strategy to determine insulin and tetanus reactivity among antigen enriched fractions. PBMCs were enriched for insulin-binding and tetanus-binding B cells simultaneously using magnetic beads. The enriched fractions were labeled with anti-CD45-113Cd, while the depleted fractions were stained with anti-CD45-89Y. Enriched and depleted fractions were combined and processed through the mass cytometer as one sample. To define/gate on insulin-binding and tetanus-binding B cells, antigen-binding levels in the enriched fraction were compared to the depleted fraction, as well as to T cells (not shown). (B) Insulin-binding B_ND2 cells are significantly increased in young-onset T1D (≤10 yr old; n = 12) compared with 11–17 yr old (n = 13) T1D and ≥18 yr old (n = 13) T1D groups. Insulin-binding B_ND2 cells were not increased compared to young ≤10 yr old HC (n = 13), 11–17 yr old HC (n = 12), or ≥18 yr old HC (n = 15). Tetanus-binding B_ND2 cells are not increased in young-onset T1D, demonstrating specificity of an increase in insulin-reactive B cells. Statistical significance determined by one-way ANOVA with Tukey multiple comparisons post-test. **P < 0.01. (C) Absolute number of insulin- and tetanus-binding B_ND2 cells in T1D and HC. *P < 0.05 (D) The frequency of insulin- and tetanus-binding DN2 cells were not significantly increased in the young-onset T1D group. (E) Absolute number of insulin- and tetanus-binding DN2 cells are not significantly different. (F) The frequency of insulin-binding B_ND2 cells, but not DN2 cells, in T1D subjects, irrespective of their age of onset, correlates with anti-insulin antibody titers (mIAA Levels). The black dotted line indicates cutoff value for insulin autoantibody positivity. Statistical significance determined by Spearman’s Correlation, ***P < 0.001. (G) Increased frequency of insulin-binding B_ND2 cells is associated with carriers of the high-risk HLA-DR4-DQ8 allele (red), whereas decreased frequency of insulin-binding B_ND2 cells is associated with the T1D protective DQ*0602 allele (green). Black solid bar indicates mean for each group.

Figure 4.

Insulin-binding B_ND2 cells are increased in the pLN of T1D donors and express increased CD86 expression compared with DN2 cells. (A) The percentage of insulin-reactive B cells among each B cell subset. Insulin-binding B cells were enriched from the pLN of T1D (n = 6) and ND (n = 3) donors using magnetic beads and analyzed using flow cytometry. B_ND2 cells are enriched in insulin-reactive B cells in T1D donors compared with MN, B_ND1, DN2, and memory (gated as CD27⁺) B cells. ND control donors have decreased levels of insulin-reactive B cells. Differences were determined using a mixed-effects model. (B) The percent of insulin-binding (IBC) B_ND2 cells is increased in T1D donors compared with ND donors. The percentage of non-IBC B_ND2 cells is not increased, demonstrating islet-antigen specificity of the finding. Statistical significance determined by unpaired Student’s t test. (C) Insulin-binding DN2 cells are also increased in T1D donors, but not non-IBC DN2 cells. Statistical significance determined by unpaired Student’s t tests. (D) IBC B_ND2 cells exhibit increased expression of the activation molecule CD86 on their cell surface compared to non-IBC B_ND2 cells in T1D donors. CD86 expression among IBC DN2 cells is decreased in 4/6 T1D donors compared to non-IBC DN2 cells. (E) CD86 expression is significantly increased in insulin-binding B_ND2 cells compared to DN2 cells in T1D donors. (F) The percent of IBC B_ND2 cells that are CXCR3⁺ is significantly increased compared to non-IBC B_ND2 cells. Similarly, the percentage of IBC DN2 cells that are CXCR3⁺ is significantly increased compared to non-IBC DN2 cells. (G) The percentage of CXCR3⁺ insulin-binding B_ND2 cells is increased, though not significantly, compared to insulin-binding DN2 cells in T1D donors. Statistical significance determined by paired Student’s t tests for D–G. *P < 0.05, **P < 0.01, ****P < 0.0001.

Figure 5.

B_ND2 cells do not exhibit an anergic B cell phenotype, produce modest amounts of pro-inflammatory cytokines, and display features consistent with pre-ASCs. (A) B_ND2 and DN2 cells have higher basal levels of phosphorylated Syk, PLCy2, AKT compared to B_ND1 and MN cells, likely reflecting recent in vivo activation, both in T1D (n = 5) and HC (n = 5) donors. B_ND2 cells are still able to mount a signaling response similar to MN B cells after stimulation with anti-IgG F(ab′)2 H&L, suggesting they are not anergic. B_ND2 and DN2 cells have elevated levels of the inhibitory signaling molecule PTEN, which may also reflect recent in vivo activation. (B) B_ND2 cells produce increased levels of IL-10, IL-6, TNF-α, and IFN-γ compared with B_ND1 cells after 6 h in the presence of a protein transport inhibitor. Cytokine production by B_ND2 cells is similar to DN2 and memory (CD27⁺) B cells, with the exception that B_ND2 cells produce some IFN-γ, but not DN2 cells. (C) B_ND2 and DN2 cells have increased expression of the transcription factor XBP-1, which is associated with plasma cell differentiation. DN2 cells, but not B_ND2 cells, have increased expression of IRF4, another transcription factor for plasma cell differentiation. B_ND2 cells have similar levels of the transcription factor T-bet to DN2 cells, which is significantly elevated compared to B_ND1 cells. (D) B_ND2 cells display increased differentiation into ASCs compared to MN and B_ND1 cells, but not to the level of DN2 or memory B cells. B cells from T1D donors differentiate into more ASCs compared to HC donors. (E) The level of proliferation after in vitro culture for ELISpot assay was determined. B_ND1 cells undergo cell death, while B_ND2 cells were maintained or slightly proliferated. Dashed red line represents proliferation (above) or loss of cells (cell death). Data presented are from T1D (n = 5) and HC (n = 5) donors. Statistical significance determined by mixed-effects model. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Figure 6.

sc-RNA-seq reveals B_ND2 and DN2 cells share a similar transcriptome, consistent with pre-ASCs. (A) UMAP of B cell clusters identified using scRNA-seq. Naming of B cell clusters was based on varying levels of gene expression of markers associated with activation, naive B cells, memory B cells, and/or isotype. (B) Heatmap of top 10 DEGs for each B cell cluster identified in A. (C) Heatmap of handpicked genes of interest that are known to be associated with plasma cell differentiation, activation or inhibitory receptors, the DN2 B cell subset, and immunoglobulin isotype. Transcriptionally, B_ND2 cells are similar to the DN2 subset. (D) Volcano plot of DEGs that are significantly different between DN2 and B_ND2 cells. Blue dots indicate genes that are increased in DN2 cells, while red dots indicated genes that are increased in B_ND2 cells. Labels for a few hand-picked genes are included. (E) Volcano plot of DEGs that are significantly different between B_ND1 and B_ND2 cells. Red dots are genes increased in B_ND2 cells, while dark yellow dots are genes that are increased in B_ND1 cells. Data are from an HC donor.

Figure 7.

Differentiation of B_ND1 cells into B_ND2 cells is driven by a combination of CD40L, IFN-γ, IL-21, and R848. (A) Example cytograms of the gating strategy used to identify the percent of retained B_ND cells and the percent that differentiated into B_ND2 cells after stimulation with varying reagents for 4 d. B_ND1 cells were sorted from HC PBMC samples (n = 4) and stimulated with a combination of different reagents. (B) The frequency of B_ND1 cells that retained a B_ND phenotype (CD27⁻, IgM^lo, IgD⁺) after stimulation. (C) The frequency of B_ND1 cells that differentiated into B_ND2 cells after stimulation. B_ND2 cells were defined as CD19⁺ CD27⁻ IgM^lo, IgD⁺, CD21⁻, CD11c⁺. Overall, using a combination of CD40L, IFN, IL-21, and R848 maintained the B_ND phenotype the best, and induced the most differentiation into B_ND2 cells after stimulation. Data presented are from HC (n = 3) donors, and samples were treated in triplicates. Statistical significance determined by mixed-effects model. *P < 0.05.

Figure S3.

Cytograms of B cell phenotype of B_ND1 cells after stimulation with the indicated reagents after 5 d. The majority of B_ND1 (CD19⁺ CD27⁻ IgM^lo IgD⁺ CD21⁺ CD11c⁻) cells revert to an MN B cell phenotype (CD19⁺ CD27⁻ IgM⁺ IgD⁺) after stimulation. A combination of CD40L, IFN-γ, IL-21, and R848 provide the best signals to maintain a B_ND phenotype and drive differentiation of B_ND1 cells into B_ND2 (CD19⁺ CD27⁻ IgM^lo IgD⁺ CD21⁻ CD11c⁺) cells.