Cutting Edge: B Cell Specificity Contributes to the Outcome of Diabetes in Nonobese Diabetic Mice¹

Type I diabetes mellitus (TIDM)³ is an organ-specific autoimmune disease that results from the loss of insulin-producing β cells in the pancreatic islets. Many aspects of the human disease are shared with the nonobese diabetic (NOD) mouse model of the disorder, including a role for class II MHC polymorphisms (1, 2), T lymphocytes in the pancreatic infiltrates (3), and shared T cell regulatory abnormalities (4, 5). These observations and the ability of T cells to transfer the disease in NOD focus most studies on T lymphocytes. However, loss of tolerance in the B cell compartment, as manifested by the detection of autoantibodies to β cell Ags, is one of the earliest indicators of TIDM. Among these, insulin autoantibodies have predictive value in both NOD and human TIDM (3, 6, 7), and insulin-specific T cells can transfer disease in NOD (8).

Direct evidence for the importance of B lymphocytes in TIDM comes from NOD mice rendered deficient in B lymphocytes by homozygous disruption of membrane-Ig μ (μMT) (9, 10). Likewise, passive transfer of anti-IgM Ab that blocks B cell development is also effective at preventing insulitis and eliminating diabetes in NOD (11, 12). In B cell knockout mice, direct reconstitution with mature B cells was not possible because of the strong CTL response against donor B lymphocytes by μMT-NOD mice. When B cells were introduced using bone marrow chimeras, TIDM susceptibility was restored in the μMT animals while transfer of sera failed to show a role for circulating Ab in the disease (13). Although these studies clearly demonstrate the importance of B cells in the development of TIDM in NOD, there are no data on how specificity of the B cell receptor for Ag contributes to the process. The Ag receptors on B lymphocytes can capture low abundance Ag for presentation to T cells (14), and in vitro studies suggest that the Ag presenting function of B cells (15, 16) or specific Abs (17) is uniquely able to induce responses to islet Ags such as glutamic acid decarboxylase.

To examine the role of Ag-specific B lymphocytes in insulin-dependent diabetes mellitus (IDDM), we introduced V_H genes related to anti-insulin mab125 into the germline of NOD mice as IgM^a transgenes (Tg). In previous studies using site-directed mutagenesis, VH125 is shown to contain two amino acid replacements in CDRH2 that are necessary for insulin binding (18). VH281 is identical except for an unmutated CDRH2 and it has no measurable insulin binding when expressed as a soluble Ab (18). Thus, NOD mice harboring the VH125 Tg (VH125Tg) have a B cell repertoire with increased potential to bind insulin, while mice harboring the VH281 Tg (VH281Tg) have a repertoire with a decreased insulin-binding potential. Accordingly, we used these mice to examine the impact of the B cell repertoire on the development of insulitis and diabetes in NOD mice. The data show that skewing the B cell repertoire toward an islet Ag (i.e., insulin) promotes the development of diabetes, whereas a repertoire with impaired recognition of insulin limits the progression of the disease.

Mice expressing IgM^a H chain (HC) Tgs VH125 and VH281 were produced in C57BL/6 as previously described (19) and were introduced into NOD mice (Taconic Farms, Germantown, NY) by multiple backcrosses. After the sixth and eight backcrosses (N7 and N9, respectively), animals were genotyped using 40 oligonucleotide primer pairs (Research Genetics, Huntsville, AL) for 13 known diabetes susceptibility loci (9). The animals were found to be homozygous for all NOD alleles. To eliminate expression of B cells that escape allelic exclusion, NOD.MuMT mice were obtained from Dr. D. Serreze (The Jackson Laboratory, Bar Harbor, ME) and crossed with the HC-transgenic lines to produce F₂ generations that carry HC Tgs in the presence (μMT^−/+) or absence (μMT^−/−) of endogenous B cells. The deficiency of endogenous B cells was confirmed by the lack of circulating IgM^b in ELISA and by lack of IgM^b B cells in FACS. Female mice were monitored for the development of diabetes for a period of 40 wk. Mice were considered to be diabetic if two consecutive blood sugars were >200 mg/dl. All animals were housed in facilities at Vanderbilt under specific pathogen-free conditions.

Spleen cells or PBLs were analyzed using FACScan (BD Biosciences, Mountain View, CA) and mAb to B220 (6B2), IgM^a (RS-3.1), and IgM^b (AF6) (BD PharMingen, San Diego, CA). Insulin-binding B cells were identified using biotin-insulin (5 ng) and FITC-streptavidin (Sigma-Aldrich, St. Louis, MO) in FACScan (19). To exclude nonspecific interactions, insulin-specific B cells were identified by inhibition with unmodified human insulin (50 ng). Serum allotypes and specific Abs were measured in ELISA as described elsewhere (19).

Pancreata from transgenic NOD or nontransgenic littermates ages 6–40 wk were analyzed for the presence of islet infiltrates after Formalin fixation and H&E staining. Islets (25–50) were examined on nonsequential sections from each pancreas. Islets that had either peri-insulitis or insulitis were scored as positive for infiltration and the percentage of islets with infiltration was determined for each mouse.

IgM^a HC Tgs were introduced into NOD by multiple backcrosses and their expression was examined by flow cytometry. Representative histograms for NOD and VH125Tg NOD mice are shown in Fig. 1. B cells from nontransgenic NOD mice do not express IgM^a (Fig. 1,A) and are all IgM^b+ (Fig. 1,B). In contrast, >90% of B cells (B220⁺) in VH125Tg NOD mice express the IgM^a Tg (Fig. 1,C) and only a small number of endogenous IgM^b B cells are present (Fig. 1 D). As in B6 mice, Tg expression was stable for over 40 wk in NOD spleen (94 ± 4% IgM^a, n = 11). Similar results were obtained in VH281Tg NOD mice, although the degree of allelic exclusion is less (82 ± 5% IgM^a, n = 10).

To determine how B cell repertoires with different potentials for insulin binding effect the development of diabetes in NOD mice, cohorts of HC-transgenic female mice and their nontransgenic littermates were monitored for hyperglycemia (Fig. 2). The data show that mice harboring VH125Tg (n = 22) develop diabetes at an accelerated rate compared with the other lines. Twenty-one of 22 mice (95%) in this cohort developed diabetes by 40 wk. In contrast, mice that express VH281Tg (n = 19) are protected from developing diabetes and only 3 of 19 mice (16%) in this cohort developed diabetes. The differences are not attributable to circulating insulin Abs produced by the Tgs which are negligible (OD < 0.05) in both lines. Additional studies from different founders and at later backcrosses have identical results and support the observation that a skewed B cell repertoire can alter the outcome of diabetes in NOD mice.

Although VH281Tg NOD are protected from developing diabetes, by 40 wk all animals had extensive islet infiltrates. Since mice that carry Ig Tgs routinely have populations of endogenous B cells that escape allelic exclusion, this additional source of B cells may contribute to the development of disease in transgenic NOD. Therefore, studies on islet infiltration were conducted using mice that expressed HC Tg in the presence (μMT^−/+) or absence (μMT^−/−) of a fixed mutation in Ig-μ to eliminate endogenous B cells (Fig. 3). All NOD mice harboring HC Tgs showed evidence of islet infiltration in the late prediabetic period (14 wk) and at this time point the extent of disease (percentage of islets with infiltration) is only slightly less in VH281Tg compared with VH125Tg NOD mice. When pancreata were examined at an earlier time point (6 wk), differences in the incidence of infiltration are more apparent, VH125Tg (75–85%) vs VH281Tg (40–50%). The incidence of infiltration in each Tg line was not effected by the presence of the μMT mutation. However, the extent of involvement (percentage of islets infiltrated) was slightly greater in both VH125 and VH281Tg NOD mice when endogenous B cells are present (μMT^+/−). Thus, the additional diversity provided by endogenous B cells may enhance islet infiltration in HC-transgenic NOD mice; however, the repertoire of VH281Tg.μMT^−/− is sufficient to generate insulitis in the absence of endogenous B cells.

To extend these studies, NOD female littermates that harbor the VH125Tg in the presence (μMT^−/+, n = 15) or absence (μMT^−/−, n = 10) of endogenous B cells were monitored for the development of diabetes. The results are shown in Fig. 4 and indicate that mice whose HC repertoire is composed only of VH125 develop diabetes at a rate comparable to the original cohort. In the presence of endogenous B cells (VH125Tg.μMT^−/+), the onset of diabetes occurs slightly earlier (12 wk) but the overall incidence compared with VH125Tg.μMT^−/− NOD mice is similar by 17 wk. This outcome is consistent with the insulitis data and suggests that residual B cells escaping allelic exclusion have only a modest effect on diabetes development in VH125Tg.

To characterize the functional potential of the B cell repertoire of nontransgenic and HC-transgenic NOD mice, spleens were examined for insulin binding (Fig. 5,). In nontransgenic NOD mice, some insulin-binding B cells are observed (Fig. 5,A); however, binding is dim (mean fluorescence intensity, <200) and principally in B220-low or -negative regions. This binding is not Ag specific as indicated by the inability of insulin to inhibit binding (Fig. 5,B). Low levels of noninhibitable insulin binding are also seen in spleens from VH281Tg mice (Fig. 5, C and D). The insulin-binding profile of B cells was different in VH125Tg mice (Fig. 5, E and F). In these mice, a distinct population of B220⁺ cells is shown to bind insulin (mean fluorescence intensity, >200; Fig, 5E), and inhibition with soluble insulin (Fig. 5 F) indicates binding is specific. This population accounts for 3.1 ± 1.3% of B220⁺ cells in VH125Tg NOD mice (n = 9). Insulin-specific B cells were uniformly IgM^a+ and similar results were obtained in mice with the μMT mutation (data not shown). Insulin-specific cells were very rare in VH281Tg NOD (1 of 9 at 0.5%) and in nontransgenic NOD (1 of 21 at 0.6%). Since NOD mice are known to produce insulin Abs (7), we interpret these findings to indicate that the frequency of insulin-specific B cell is increased at least 10–20 times in VH125Tg compared with nontransgenic NOD mice. This finding in NOD mice differs from our observations in B6 mice harboring VH125Tg where insulin-specific B cells were not observed (19).

In this study, NOD mice harboring Ig HC Tgs are used to examine the role of Ag-specific B cells in TIDM. In these mice, B cell receptor usage is limited to a single HC VDJ while maintaining the potential for diversity at the L chain locus. The data show the importance of Ag-specific B cells in the development of diabetes in NOD mice and reveal that divergent outcomes for diabetes may be determined by differences in the B cell repertoire. A repertoire skewed toward insulin binding in VH125Tg mice favors the selection of insulin-specific B cells into the peripheral repertoire of NOD mice. These events may be obscure in nontransgenic NOD but are detectable in VH125Tg NOD because of the increased frequency of trackable B cells. Furthermore, B6 mice that express the same VH125 HC Tg do not demonstrate an increase in insulin-binding B cells over background (19); thus, the entry of insulin-specific B cells into the repertoire represents a breach of tolerance in the B cell compartment of NOD that is unmasked by the availability of the VH125 HC Tg. This observation is analogous to the presence of high-affinity insulin autoantibodies in human pre-IDDM that is linked to progressive β cell destruction (7, 20). Thus, the selection of insulin-specific B cells into the repertoire of VH125Tg NOD and the differentiation of B cells to produce high-affinity IgG anti-insulin Ab in human pre-IDDM are similar processes that denote a critical threshold for β cell destruction has been reached. Since our fixed HC Tgs do not produce measurable insulin Abs, the role of anti-insulin B cells may be as APCs to expand insulin-specific T cells which are known to contribute to TIDM (8).

In contrast, a B cell repertoire that is impaired for insulin binding by a HC Tg that differs in two amino acids protects VH281Tg NOD mice from diabetes. These mice, however, still develop islet infiltration (both peri-insulitis and insulitis). Since Ig Tgs do not completely exclude the expression of endogenous B cells (IgM^b) that may include other specificities, we intercrossed our HC-transgenic NOD with B cell-deficient NOD.μMT to examine the role of endogenous B cells. The data show that additional diversity provided by residual endogenous B cells contributes modestly to insulitis in VH281Tg NOD; however, insulitis develops when VH281 is the only HC available. Likewise in VH125Tg mice, endogenous B cells may have a small effect on insulitis and diabetes but the overall outcome is determined principally by the function of the Tg. Recent studies examining anti-hen egg lysozyme Tgs in NOD also show the protection from diabetes in the presence of insulitis (D. Serreze, unpublished data). This observation is consistent with the data in VH281Tg NOD and indicate that a B cell repertoire with impaired recognition of islet Ags has limited the progression of diabetes. Ag-specific B cells may not be required for early islet infiltration but their presence may be associated with enhanced β cell destruction. The current findings suggest that diverting even a small portion of the B cell repertoire may be beneficial in preserving β cell function. Since insulin is only one of several autoantigens in TIDM, the approach may be readily extended to other islet Ags in the future.

We are grateful to Dr. David Serreze for making NOD.MuMT mice available for this study and for sharing prepublication data. We thank Drs. Tom Aune, Nancy Olsen, and Peggy Kendall for their critical reviews of this manuscript.