Comparative analysis of the repertoire of insulin-reactive B cells in type 1 diabetes-prone and resistant mice

Abstract

Seropositivity for autoantibodies against multiple islet antigens is associated with development of autoimmune type 1 diabetes (T1D), suggesting a role for B cells in disease. The importance of B cells in T1D is indicated by the effectiveness of B cell-therapies in mouse models and patients. B cells contribute to T1D by presenting islet antigens, including insulin, to diabetogenic T cells that kill pancreatic beta cells. The role of B cell receptor (BCR) affinity in T1D development is unclear. Here, we employed single cell RNA sequencing to define the relationship between BCR affinity for insulin and B cell phenotype during disease development. We utilized immunoglobulin (Ig) heavy chain (VH125) mouse models in which high-affinity insulin-reactive B cells (IBCs) were previously shown to be anergic in diabetes-resistant VH125.C57BL/6-H2g7 and activated in VH125. NOD mice developing disease. Here, high-affinity IBCs were found in the spleen of prediabetic VH125. NOD mice and exhibited marginal zone or follicular phenotypes. Ig light chains expressed by these B cells are unmutated and biased toward Vκ4-74 and Vκ4-57 usage. Receptors expressed by anergic high-affinity IBCs of diabetes-resistant VH125.C57BL/6-H2g7 are also unmutated; however, in this genetic background light chains are polymorphic relative to those of NOD. Light chains derived from NOD and C57BL/6-H2g7 genetic backgrounds conferred divergent kinetics of binding to insulin when paired with the VH125 heavy chain. These findings suggest that relaxation of tolerance mechanisms in the NOD mouse leads to accumulation and partial activation of B cells expressing germline encoded high-affinity BCRs that support development of autoimmunity.

Keywords: B cell receptor (BCR); BCR affinity; immunoglobulin light chain; insulin; non-obese diabetes (NOD) mice; type 1 diabetes (T1D).

Figure 1

IBCs expressing Igκ accumulated in spleen of prediabetic VH125.NOD compared to diabetes-resistant VH125.C57BL/6-H2g7 mice. (A) Diagram representing magnetic bead-based methodology for enrichment of IBCs. (B, C) Frequency of splenic IBCs in the enriched fraction in prediabetic VH125.NOD and diabetes-resistant VH125.C57BL/6-H2g7: representative flow cytometric panels of IBCs gated on lymphocytes/singlets/live cells/B220⁺IgM⁺ cells are shown (B) and frequencies of the two IBC populations with insulin MFI low (insulin^lo) and high (insulin^hi) across multiple experiments are shown (VH125.NOD – n=24, VH125.C57BL/6-H2g7 – n=8) (C). (D–F) Cell surface expression of immunoglobulin kappa (Igκ⁺) (D), lambda (Igλ⁺) (E), and double positive kappa and lambda (Igκ⁺Igλ⁺) (F) on insulin^lo and insulin^hi IBCs in VH125.NOD (n=3) and VH125.C57BL/6-H2g7 (n=7). Results are pooled from at least three independent experiments and represented as the mean ± SEM. Each dot represents a mouse. One-way ANOVA followed by post hoc Tukey’s multiple comparisons tests for panels (C–F) were used to determine the significance of differences among groups and defined as *P < 0.05, **P < 0.005, ***P < 0.0005. ns, no statistical significance.

Figure 2

Marginal zone and follicular B-cell transcriptional clusters encompass most of high-affinity IBCs of VH125.NOD. Splenocytes from three prediabetic VH125.NOD were labeled with three different hashtag antibodies, cell sorted for high-affinity IBCs, pooled, and subjected to scRNA-seq on the 10X Genomics platform. (A) Flow panels representing sorted high-affinity IBCs (gated on lymphocytes/singlets/live cells/B220⁺IgM⁺ cells) from three prediabetic VH125.NOD mice. (B) Three clusters identified among high-affinity IBCs by unsupervised 10X CellRanger clustering method (blue – cluster 1, orange – cluster 2, green – cluster 3). (C) Heatmap of fifty differentially expressed genes among clusters. High-affinity IBCs from each cluster and donor mouse, overlayed on UMAP (D–F), with cell numbers for each cluster, including unassigned cells, (G), and frequency (H) (each symbol represents the same mouse as in Figure 3 : circle – mouse 1, square – mouse 2, diamond – mouse 3). One-way ANOVA, followed by post hoc Tukey’s multiple comparisons tests were used to determine the significance of differences among groups. *P < 0.05, ns, no statistical significance.

Figure 3

High-affinity IBCs of VH125.NOD expressed a biased Igκ repertoire. High-affinity IBCs from three prediabetic VH125.NOD mice from Figure 2 were additionally subjected to VDJ sequencing. (A) Heavy and light chain isotype frequency, (B) Vκ gene segment frequency, (C) Jκ gene segment frequency, (D) Vκ/Jκ gene rearrangement frequency among high-affinity IBCs (each symbol represents a mouse: circle – mouse 1, square – mouse 2, diamond – mouse 3). (E) Amino-acid composition in complementarity determining region (CDR) 1, 2, and 3 in the five predominant light chains among high-affinity IBCs. (F) Distribution of five predominant light chains on the UMAP gene expression (different colors representing five predominant light chains, other Vκ/Jκ rearrangements were grouped and designated as “other Vκ/Jκ”). One-way ANOVA, followed by post hoc Tukey’s multiple comparisons tests were used to determine the significance of differences among groups. *P < 0.05, **P < 0.005, ***P < 0.0005, ****P < 0.00005.

Figure 4

High-affinity IBCs of VH125.C57BL/6-H2g7 express unmutated and polymorphic Igκ chains. (A) Comparison of Vκ gene segments from splenic high-affinity IBCs of prediabetic VH125.NOD with degenerate primers (denoted by asterisk *) and by scRNA-seq as well as of VH125.C57BL/6-H2g7 with degenerate primers. Vκ presence (green check mark), absence (red cross) and the number of unique sequences detected are provided. (B, C) Amino acid composition of Igκ CDR3 for Vκ4-74 (B) and Vκ4-57 (C) in high-affinity IBCs of prediabetic VH125.NOD and VH125.C57BL/6-H2g7. Neutral amino acids marked in black, positively charged H (histidine) and R (arginine) in green, negatively charged D (aspartic acid) in blue.

Figure 5

Affinity of anti-insulin binding antibodies from VH125.NOD and VH125.C57BL/6-H2g7 genetic backgrounds. Igκ chains from splenic high-affinity IBCs of VH125.NOD and VH125.C57BL/6-H2g7 by scRNA-seq and sequencing with degenerate primers, respectively, were selected for recombinant antibody (recAb) expression. (A–H) ELISA with insulin coated plates for specificity measurements with recAbs containing VH125 heavy chain and light chains derived from splenic high-affinity IBCs of either VH125.NOD (A, C, E, G) or VH125.C57BL/6-H2g7 (B, D, F, H). RecAbs were grouped into Vκ4-74 (A, B), Vκ4-57 (C, D), other Vκ4 families (E, F), and non-Vκ4 (G, H). (I, J) Area under ELISA curves for different groups of recAbs containing Igκ of VH125.NOD (I) and VH125.C57BL/6-H2g7 (J). (K, L) SPR curves indicating affinity (KD) and kinetics of 0.8µM insulin binding by recAbs with Vκ4-74 (K, L) and Vκ4-57 (M). Results are pooled from at least three independent experiments and represented as the mean ± SEM. One-way ANOVA, followed by post hoc Tukey’s multiple comparisons tests. **P < 0.005, ****P < 0.00005.