Investigation of the Role of B-Cells in Type 1 Diabetes in the NOD Mouse

NOD/Caj mice were used for all backcrosses. μMT^−/− mice lacking B-cells (8) were originally obtained from Klaus Rajewsky and backcrossed for 14 generations onto NOD/Caj mice and are designated NOD.μMT^−/− mice. Breeders were typed for markers at NOD susceptibility loci as described (9), and all mice observed in these studies were homozygous for these markers from the ninth backcross generation. The mice were all housed in specific pathogen-free conditions. All of the animal studies were performed under protocols approved by the Yale University Animal Care and Use Committee.

The mIg construct has been previously described (27). This consists of a rearranged variable-diversity–joining (VDJ) segment containing Vh186.2, reactive to the hapten (4-hydroxy-3-nitrophenyl) acetyl ligated to an IgM^a constant region from which the secreted exon and polyadenylation site had been excised, resulting in production of only membrane-bound but not secreted Ig. A control construct (28) was identical except that the secreted exon was intact (m+sIg). Both constructs were injected directly into NOD ova (Yale Diabetes and Endocrinology Research Center transgenic injection facility). Two founder lines each of both mIg and m+sIg were obtained, and because similar results were obtained, data were pooled. NOD mice express IgM^b, which facilitated screening by the use of fluorescein isothiocyanate–conjugated anti-IgM^a antibody (PharMingen, San Diego, CA). These mice were then crossed to the NOD.μMT^−/− mice and further intercrossed to make the μMT mutation homozygous so that the only B-cells expressed were transgenic B-cells. All mice used in the experiments were tested for the presence of the IgM^b allotype to confirm that the transgenes were expressed on a homozygous NOD.μMT^−/− background. Thus all mice used were homozygous NOD.μMT^−/− and mIg or m+sIg transgene-positive or -negative littermates.

Anti-CD4, anti-CD8, anti-CD45R (B220), anti-IgM^a, and anti-IgM^b antibodies used for flow cytometry were purchased from PharMingen.

Purified T-cells from spleen (nonimmunized mice) or popliteal lymph node (immunized mice) were resuspended in RPMI 1640 medium (Invitrogen, Carlsbad, CA) with 5% FCS. The cells were incubated in a 96-well plate at a concentration of 2.5 × 10⁵ cells/well, together with 2 × 10⁵ irradiated (3,000 rad) T-cell–depleted APCs. The antigens were used at different concentrations in a final volume of 200 μl. After 72 h, the plates were pulsed with 0.5 μCi ³H-thymidine for a further 14 h.

Splenocytes or lymphocytes from popliteal lymph nodes were incubated with anti-Fc receptor antibody (2.4G2) and anti–I-A^g7 for 30 min, washed, and then incubated on ice with BioMag anti-mouse IgG, anti-mouse IgM, and anti-rat IgG (PerSeptive Biosystems, Framingham, MA) coated beads for 45 min. The beads were then removed using a magnet and the cells washed and resuspended in culture medium. The T-cell preparations were >90% pure.

Splenocytes were incubated with anti–Thy-1 antibody, anti-CD8 antibody, and anti-CD4 antibody for 30 min. The cells were then washed and incubated on ice with BioMag anti-rat IgG (PerSeptive Biosystems) coated beads for 45 min. The beads were then removed using a magnet and the cells washed and resuspended in culture medium. The APC preparations were >90% pure.

B-cells were purified from splenocytes of NOD or mIg transgenic mice using a negative selection procedure according to the manufacturer’s protocols (Miltenyi Biotech, Cologne, Germany). B-cells obtained using this method were >97% pure.

Bovine/porcine soluble insulin (Lilly, Indianapolis, IN) was used at concentrations of 20, 5, and 1 μg/ml. Affinity-purified GAD (29) was kindly provided by Robert Sherwin and used at the same concentrations. Keyhole limpet hemocyanin (KLH) was purchased from Sigma (St. Louis, MO). Pancreatic islets were isolated from 6-week-old male NOD mice by collagenase digestion as previously described (30). Islet lysate was produced by repeated freeze/thaw cycles of islets isolated from 20 mice.

Plates (96 well) were coated with insulin or affinity-purified GAD at a concentration of 50 μg/ml or islet lysate overnight at 4°C. After washing with PBS + 0.1% Tween-20, the plates were blocked for at least 1 h with PBS + 1% BSA. Serum (50 μl) was incubated at concentrations of 1 in 50. The plates were further washed with PBS + 0.1% Tween-20 and then incubated with 50 μl anti-IgM conjugated to alkaline phosphatase and streptavidin, followed by development with disodium para-nitrophenyl phosphate substrate (Sigma) and read at OD₄₀₅ (optical density at 405-nm wavelength).

mIg transgenic and transgene-negative NOD.μMT^−/− littermates or NOD mice were immunized with 50 μg antigen (GAD, insulin, or KLH) with complete Freund’s adjuvant in the hind footpads. Three weeks later, they were boosted with the same antigen emulsified with incomplete Freund’s adjuvant. One week later, the mice were killed, and the popliteal lymph nodes and spleen were removed for assay. Peripheral blood was taken and serum separated for analysis.

Pancreatic tissue was fixed in formalin, paraffin embedded, and stained with hematoxylin and eosin. The sections were examined microscopically, and insulitis of individual islets was assessed by two independent observers according to the scale of 0–4 as follows: 0, no insulitis; 1, peri-insulitis; 2, peri-insulitis with less than one-half of the islet infiltrated; 3, more than one-half of the islet infiltrated; and 4, complete islet destruction. Additionally, a proportion of pancreata was fixed in periodate-lysine-paraformaldehyde, sucrose infused, and then frozen in Tissue-Tek OCT (Bayer, Elkhart, IN). Sections (7-μm thick) were stained with biotinylated YT4.3 antibody recognizing CD4, TIB 105 antibody recognizing CD8, and B220 antibody, which stains the majority of B-cells. The color was developed using diaminobenzidine tetrahydrochloride and nickel ammonium sulfate. The sections were then counterstained with hematoxylin.

Animals were tested weekly for glycosuria using Diastix (Bayer), and if present, diabetes was confirmed by a blood glucose measurement of >250 mg/dl (13.9 mmol/l) using OneTouch test strips (LifeScan, Milpitas, CA) (30).

The transgene was identified using antibodies against IgM^a. It can be seen from Fig. 1 that the transgenes restored B220⁺ (CD45R) B-cells in both types of transgenic mice. However, although the levels of B-cells in the mIg transgenic mice was normal, the number of B-cells in both lines of m+sIg transgenic mice was reduced. The ratio of CD4 to CD8 in all of these mice was normal (data not shown).

Our previous experience had shown that NOD.μMT^−/− mice develop mild insulitis (10) in accordance with an earlier report (11) and have a low incidence of diabetes. Insulitis in these NOD.μMT^−/− mice can be seen from the age of 8 weeks onwards, and an example of the insulitis at 12 weeks of age is shown here; although, in general, NOD.μMT^−/− mice have milder insu-litis compared with NOD mice. A fraction of the transgenic mice showed an increased amount of insulitis compared with NOD.μMT^−/− mice (Fig. 2B). Expression of the transgenes significantly altered the severity of insulitis overall (P < 0.001), especially in the male transgenic animals (Fig. 2C).

A number of mice expressing the mIg transgene developed diabetes (Table 1). This was significant (P < 0.05) in female mice. Diabetes was not, however, restored to the level seen in NOD mice, and no diabetes was seen in the m+sIg control transgenic mice.

In keeping with the fact that the NOD.μMT^−/− mice develop insulitis, they can, in the absence of B-cells, also generate spontaneous T-cell proliferative responses to islets, as shown in Fig. 3A. NOD.μMT^−/− mice have low spontaneous T-cell responses to GAD and insulin, and the introduction of the mIg transgene did not lead to increased responses to these antigens (data not shown). In order to check that the B-cells from the mIg transgenic mice were able to function normally, T-depleted spleen cells (predominantly B-cells) from NOD and mIg mice were stimulated with lipopolysaccharide (LPS) as a nonspecific stimulus. Results shown in Fig. 3B indicate that the T-depleted spleen cells respond normally to this stimulus.

T-depleted spleen cells from the mIg transgenic mice were used as APCs to stimulate islet responses from NOD T-cells. The APCs were able to present exogenous antigens derived from islets to NOD T-cells (Fig. 3C). When this experiment was performed with purified B-cells, it can be seen that there is no difference in the ability of the mIg transgenic B-cells to present islet antigen to NOD T-cells compared with wild-type NOD B-cells (Fig. 3D).

mIg transgenic mice were immunized with KLH, insulin, or affinity-purified GAD in complete Freund’s adjuvant. The T-cell proliferative responses to the KLH after immunization were similar in the transgenic, compared with the nontransgenic, B-cell–deficient and NOD mice shown in Fig. 4. However, immunized T-cell responses to GAD did not increase in the presence of the transgene compared with transgene-negative mice, and responses to insulin were uniformly low.

Islet lysate was coated onto enzyme-linked immunosorbent assay plates to represent multiple islet antigens. Nonimmunized NOD mice have a greater spontaneous antibody response to these islet antigens than m+sIg transgenic mice (P < 0.05). As expected, mIg transgenic mice as well as transgene-negative mice (B-cell deficient) do not secrete any detectable antibody (Fig. 5). Serum was also taken from the mice that were immunized with KLH and insulin and tested against these antigens coated onto enzyme-linked immunosorbent assay plates. Antibody responses to both of these antigens were low in all mice compared with NOD mice (data not shown).

In order to separate the role of the B-cells in antigen presentation from antibody production in the development of diabetes, we generated NOD transgenic mice expressing B-cells (mIg) that were only able to present antigen but not secrete antibody due to a mutation in the secreted exon. Although the transgenic construct had a fixed heavy chain, the potential for selection of a diversity of light chains was maintained. Therefore, in theory, the cells bearing the transgenic construct should, at albeit reduced frequency, be able to present the autoantigens important in diabetes to the autoreactive T-cells. The expression of the mIg transgene significantly increased insulitis in the male animals. There was also a sevenfold and statistically significant increase in the number of female mIg mice developing diabetes (5 of 47 [10.6%]) compared with the female nontransgenic NOD.μMT^−/− littermate (1 of 64 [1.5%]) mice (P < 0.05). As the mIg transgenic mice do not secrete antibody, this increase in diabetes is clearly antibody independent and not related to the antibody-secreting function of B-cells. Thus we establish, using a genetic approach, that B-cells do, in fact, have an antibody-independent role in promoting diabetes.

Although the nonsecreting Ig transgenic B-cells increase the incidence of diabetes, it is not restored to the level of wild-type NOD mice. This could be due to lack of a diverse repertoire, lack of antibody, or both. While repertoire almost certainly plays a role (see below), it is less clear whether antibody also has a role. The m+sIg transgenic mice did not develop diabetes, although they had insulitis comparable with the mIg mice (data not shown). This does not necessarily imply that the ability of these mice to produce antibody has an adverse effect on the development of diabetes. Rather, as the restoration of B-cells in both lines of m+sIg mice did not occur to the same extent as in the mIg transgenic mice, it is quite possible that the low total number of B-cells present in the m+sIg transgenic mice contributed to the lack of development of diabetes. For this reason, no major comparisons have been made with the m+sIg transgenic mice; however, the m+sIg mice were useful in assessing antibody responses to islet and other antigens.

In keeping with the fact that the NOD.μMT^−/− mice developed insulitis in the absence of B-cells, NOD.μMT^−/− mice had some spontaneous proliferative responses to islets, although responses specifically to both insulin and GAD were low. This indicated that the ability to generate anti-islet T-cell responses was not entirely dependent on B-cells to prime these responses. This agrees with a previous investigation (14) showing the development of diabetes in NOD.scid mice transplanted with NOD.mMT^−/− fetal thymus. The presence of the mIg transgene did not increase the response to the specific islet autoantigens tested in transgenic mice. Although the B-cells from the transgenic mice were normally responsive to the nonspecific stimulus LPS and presented antigen normally to NOD T-cells, they were not able to allow the diversification to increase GAD and insulin responses. We further investigated the ability of T-cells from transgenic mice to respond after immunization with a nonislet-specific, foreign immunogen KLH as well as with diabetes autoantigens, insulin, and GAD. The nontransgenic B-cell–deficient mice were able to respond to KLH immunization as equally well as the transgenic NOD.μMT^−/− mice. However, none of the mice responded well to immunization with either GAD or insulin, whereas nontransgenic NOD mice can respond to GAD but poorly to insulin after immunization. Thus, it is likely that the fixed heavy chain did not facilitate sufficient diversity for the recognition of these self-antigens.

The ability of the transgenes to recognize and diversify the immune response was further tested by investigating the antibody responses to islets. To test responses to antigens other than GAD and insulin, we used islet lysate. For this experiment, we investigated the antibody production from the m+sIg control transgenic mice because this could not be examined in the mIg transgenic mice. NOD mice generated antibody responses to islet lysate that were greater than those from the m+sIg transgenic mice. This was further evidence that antibody repertoire was reduced, and the failure to develop diabetes may, in part, be due to the fact that the mice could not sufficiently diversify the response to stimulate the T-cell repertoire necessary for diabetes to occur.

In the natural history of autoimmune diabetes, at the time of development of overt disease, there is a T-cell response to a diverse range of autoantigens that include GAD, insulin, HSP60 (heat shock protein 60) (31) and likely other autoantigens that have yet to be fully defined. In addition, autoantibodies are also produced, although there is not the same diversity as that seen for T-cells (32). Insulin autoantibodies are readily detected (33), but anti-GAD antibodies are only found in some NOD mice. One of the important aspects of the function of B-cells as APCs is the ability to concentrate antigen manyfold by virtue of the antigen specificity of the B-cell receptor (16,34–36). This would allow antigen-specific T-cells to be expanded if appropriately stimulated by relevant B-cell populations. Therefore the ability to diversify the repertoire of B-cells that recognize different autoantigens might be important. Our study suggests that the reduced Ig repertoire resulting from the fixed heavy chain transgenes used here was not sufficient to produce islet-specific antigen receptors to allow appropriate diversity. This complements the other studies showing that B-cells expressing insulin-specific antibody receptors develop accelerated diabetes (37) and, the converse, that expression of a fixed antibody receptor to hen egg lysozyme, which does not allow any B-cell diversity to develop, retards the onset of diabetes (38).

Despite a number of contributory studies, the exact role of B-cells in diabetes development has still not been completely elucidated. The question of whether antigen presentation or production of autoantibodies is important in this process has been much debated but, as yet, has not been conclusively answered. Transfer of total serum from NOD mice does not precipitate disease (22). However, this experiment has not been done with antibodies tested to be specific for diabetes autoantigens. Certainly, autoantibodies are good predictive markers for the development of disease both in humans (7) and in NOD mice (33). In addition, the presence of specific autoantibodies may aid in the presentation of autoantigen (39). Recently, evidence has been provided that maternally transmitted antibodies are important in development of diabetes in NOD mice (24) as shown by experiments where NOD.μMT^−/− females bred with NOD males had offspring that developed a lower incidence of diabetes than offspring from NOD females mated with NOD.μMT^−/− males. Further, when NOD embryos were transferred to superovulated DBA/2 females or control NOD females, a lower incidence of diabetes was found in the mice of the DBA/2 surrogate mothers. While these experiments are compelling evidence that maternally transmitted antibodies from NOD mice can influence the development of disease, it does not directly show that it is the anti-islet antibodies that are specifically involved in the disease process. Analysis of the natural autoantibody repertoire of NOD mice has indicated that the repertoire has broad specificity for multiple antigens and is low avidity (40). In addition, experiments analyzing the autoantibody repertoire in NOD mice indicated that IgM may be important early in the process and that the role of IgG antibodies may increase after 14 weeks as pancreatic β-cells are destroyed (40).

Although there has been speculation over an antigen-presenting role for B-cells in diabetes, this is the first report that directly and experimentally separates the antigen-presenting role of B-cells from the ability to secrete antibody in the development of autoimmune diabetes. The data clearly show that antibody secretion is not essential for some of the diabetes-promoting role of B-cells. Our experiments also indicate, not surprisingly, that the ability of B-cells to recognize a diversity of islet autoantigens is likely to be important in the development of diabetes. Taken together, these results strongly implicate an antigen-specific antigen-presenting function for B-cells. On the other hand, these experiments do not address the role that antibody may play in the development of diabetes. This issue will require separate approaches. Our results are highly relevant to potential human studies to treat or prevent early diabetes via depletion of B-cells with anti-CD20 treatment; such therapy has a much greater effect on B-cells and presumably antigen presentation by B-cells than it does on antibody levels. Thus, our mechanistic data provide some experimental basis for the consideration of such a trial.

This work was supported by National Institutes of Health Grants PO1-DK53015 (to C.A.J.) and AI43603 and by the Yale DERC pilot grant (to M.J.S.). F.S.W. and L.W. held Career Development Awards from the Juvenile Diabetes Research Foundation. F.S.W. is currently a Wellcome Trust Senior Fellow in Clinical Science. C.A.J. was an investigator of the Howard Hughes Medical Institute.