NOD B-cells Are Insufficient to Incite T-Cell-Mediated Anti-islet Autoimmunity Free

NOD/LtJ, NOD.NONThy1.1, C57BL/6, and C57BL/6 μMT^−/− mice were purchased from The Jackson Laboratories (Bar Harbor, ME). All other strains (F1.g7, F1μMT^−/−, NOD μMT^−/−, and B6.g7) were derived from stocks maintained in our facility. All mice were housed under specific pathogen-free barrier conditions and maintained according to the guidelines for use and care of laboratory animals as set forth by the University of Pennsylvania. All NOD mice and chimera were monitored weekly for the development of diabetes by blood glucose measurement with Accu-Check Advantage test strips (Boehringer Mannheim, Indianapolis, IN). Two consecutive daily glucose measurements >250 mg/dl constituted a diagnosis of diabetes. Control groups were compared with experimental at the end point of each diabetes incidence experiment with the χ² test.

Chimeric mice were produced as described previously by Matthieu et al. (19). To generate mixed chimera, F1 recipients received sublethal irradiation (1,000 rad) in a divided dose (500 × 2) on consecutive days. For full NOD chimera, lethal irradiation (1,200 rad) was employed in a divided dose (600 × 2) on consecutive days. Immediately after the second irradiation treatment, mice received 20 million nondepleted bone marrow cells from the appropriate donor. Mixed inocula were prepared as a 1:1 ratio with 10 million of each donor type. All recipient mice were female to maximize the penetrance of diabetes within these groups.

One million splenocytes were suspended in biotin-free RPMI containing 0.1% azide and 3% FCS and surface stained in 96-well plates with the appropriate monoclonal antibody: RA3–6B2 biotin (anti-CD45R/B220), RM4–5 allophycocyanin (anti-CD4), 10–3.6 PE (anti-I-Ag⁷), OX-7 PE (anti-CD90.1/Thy-1.1), 30-H12 biotin (anti-CD90.2/Thy-1.2) (BD PharMingen). Biotin-conjugated monoclonal antibodies were subsequently stained with streptavidin-RED670 (Life Technologies); cells were washed no fewer than two times before the addition of the secondary reagent. All samples were analyzed on a FACSCalibur flow cytometer (Becton Dickinson, Mountain View, CA) using CellQuest software.

Splenocytes were labeled with 5-(and −6)-carboxyfluorescein diacetate succinimidyl ester (CFSE) (Molecular Probes, Eugene, OR) as previously described (20). Briefly, splenocytes were resuspended at a concentration of 10 × 10⁶ cells/ml in serum-free IMDM (Gibco/BRL, Gaithersburg, MD) at 37°C. An equal volume of a 1:350 dilution of the CFSE stock (5 mmol/l in DMSO) in 37°C serum-free IMDM was then added to the cell preparation, which was subsequently incubated for 5 min at 37°C. CFSE labeling was quenched by adding an equal volume of heat-inactivated FCS, whereupon cells were washed twice and resuspended in IMDM containing 10% heat-inactivated FCS.

CFSE-labeled cells as described in the previous section were plated in 24-well plates at a density of 1 × 10⁶ total cells in 1 ml of media containing 10% heat-inactivated FCS with the designated amount of anti-CD3 (145–2C11) and 4 μg/ml anti-CD28 (37.51) antibodies. All cells were incubated for 65–70 h at 37°C in 5% CO₂. After incubation, the cultured cells were harvested and stained with allophycocyanin-conjugated anti-CD4 (RM4–5) to allow the identification of CFSE-labeled CD4+ T-cells using FACSCalibur (Becton Dickinson, Mountain View, CA). In all, 10,000 CD4+ events were collected and division was tracked utilizing a live cell gate that included the blasting cells, as determined by forward and side scatter. Division history of CD4+ T-cells was analyzed as previously described, based on the property of CFSE-labeled cells to lose half of their fluorescence intensity with each round of division (21).

The data generated by CFSE-labeled cultures was analyzed to determine the potential for activation inherent in the precursor pool. To perform this calculation, we determined the number of activated precursors required to produce all the daughter cells in a given division peak (P), a factor we term “precursor equivalents,” by using the following formula:

where P is the peak number.

The probability of generating a daughter cell in peak P is given by dividing the number of precursor equivalents for that division by the total number of equivalents for all divided peaks (P > 1).

Previous data from our laboratory has indicated that the B-cell is the only efficient coactivator of T-cells within the NOD system (13,14). Therefore, characteristics of B-cell APCs determine the nature of T-cell activation. Given the crucial role of this cell class in diabetogenesis, we undertook a careful analysis of the cell surface expression of molecules involved in antigen presentation on the NOD B-cell.

NOD B-cell function was assessed with respect to antigen presentation by comparing surface levels of antigen presenting and costimulatory molecules to the levels observed on the control B6.g7 strain (Fig. 1), which is congenic to NOD at the major histocompatibility complex (MHC) locus. Class II, CD40, CD19, and CD21/35 were upregulated on splenic NOD B-cells; we have previously reported on the role of the complement receptor complex in the development of diabetes (22). Levels of CD80, CD86, CD54, and CD1 were indistinguishable between these strains (data not shown). Together, these data paint a picture of a B-cell that has extended its ability to acquire antigen (complement receptor), to present those antigens (Class II), to respond to stimulation (CD19), and to activate those T-cells which it might engage (CD40). Analysis of this increased expression in splenocytes from T-cell deficient (Cα^−/−) NOD mice revealed a similar increase in surface levels, indicating that these alterations did not result from interaction with autoreactive T-cells (data not shown).

The upregulation of several key mediators of antigen presentation invited the possibility that NOD B-cells may inappropriately activate NOD T-cells. Moreover, as B-cells are the only competent APCs within the lymph node microenvironment, alteration in B-cell functional status may have profound consequences for the maintenance of peripheral tolerance (13). We hypothesized that the activated nature of NOD B-cells may permit them to actively disrupt the tolerant state. To test this hypothesis, we constructed chimera in which all B-cells were NOD derived. Bone marrow with this reconstituting profile was obtained from F1 μMT^−/− mice which were produced from crosses between NOD μMT^−/− and B6 μMT^−/− B-cell deficient parents. Mice reconstituted with mixtures of F1 μMT^−/− and NOD bone marrow demonstrated an absence of F1 B-cells in the periphery despite the presence of NOD- derived B-cells (Fig. 2). In addition, fluorescence-activated cell sorter analysis to track the MHC class I disparities indicated that the T-cell compartment was evenly divided among the donor strains in the mixed chimera by 15 weeks after reconstitution (Fig. 2). No differences were detected in the percentages of CD4 cells, CD8 cells, B-cells, or macrophage/dendritic cells between the reconstituted groups; analysis of absolute splenocyte numbers at the termination of the experiment produced expected numbers of cells (>100 × 10⁶ per spleen) and no significant difference between the groups (data not shown). Although the B-cell activity of these mice was derived wholly from the autoimmune-prone NOD strain, these mice were protected from spontaneous diabetes for the >70-week duration of the study (Fig. 3). Further analysis of splenocytes from these chimera revealed the same alterations in class II, CD21/35, CD19, and CD40 present in the NOD strain (data not shown). These data indicate that NOD-derived B-cells do not incite autoimmunity and suggest that another APC subset likely participated in the establishment of a tolerant state.

The data from F1μMT^−/− chimera indicate that diabetes resistance can be maintained even in the absence of F1 B-cells; this finding suggests that the function of bone-marrow-derived F1 professional APCs may provide the protective quality we have observed. However, the mechanism of action of these cells remained unknown. It was also plausible that the presence of MHC molecules distinct from I-A^g7 leads to the selection of a peripheral lymphocyte repertoire devoid of islet reactivity. This explanation is supported by studies in which the presence of alternative MHC molecules in the NOD mouse prevents diabetes (23–26). Nonetheless, there is also ample evidence of pathogenic specificities that are not deleted in these hybrids (27,28). Alternatively, the presence of normal F1-derived professional APCs could lead to T-cell tolerance following contact with peripheral autoantigens by dint of other tolerizing functions. We sought to exclude the first possibility as the primary cause of protection seen from F1 bone marrow. To analyze this contention, we created NODxB6.g7 progeny, a substrain we have designated F1.g7. This strain expresses only NOD-derived MHC molecules and hence could be more permissive to diabetes induction than F1s derived from the B6 strain in which the MHC will be heterozygous (g7 and b).

Chimera were generated with the irradiation protocols described above. NOD bone marrow was collected from NOD.Thy1.1 donors to permit tracking of T-cell reconstitution as these F1.g7 mice are MHC congenic to the NOD. Mixed chimeras reconstituted with an average of 31% NOD T-cells; T-cells in full chimera were more than 95% NOD-derived (Fig. 4); no differences were detected in overall proportion of B-cells or CD8 cells (data not shown). As shown in Fig. 5, full NOD chimera became diabetic around 25 weeks after reconstitution. Mixed NOD/F1.g7 mice were completely protected from diabetes (P < 0.05) despite the presence of significant numbers of NOD-derived T-lymphocytes. These data indicate that the presence of non-NOD MHC alleles is not solely responsible for the protection from diabetes seen previously but rather that some other function possibly within the professional APC subclass serves to establish or maintain peripheral tolerance.

Previous data from our laboratory and others has demonstrated a hypoproliferative T-cell phenotype in the NOD strain, which develops around 4–6 weeks of age and precedes the onset of diabetes (13,14,29,30). As these chimeric mice were protected from diabetes, we sought to analyze their proliferative response to determine whether NOD proliferation characteristics tracked with the development of diabetes. Both the proliferative distribution of splenocytes at maximal stimulation and the dose response physiology were determined. The parameters studied in these assays did not serve to distinguish these groups in any way. As shown in Fig. 6, splenocytes from both protected and susceptible mice at maximal stimulation demonstrated a level of proliferation intermediate between that seen for NOD and B6 CD4 T-cells. These cultures generated some daughter progeny in the latter cell divisions, a feature that demonstrated an improvement over NOD stimulation levels but did not reach that seen in B6 cultures. In addition, the dose-response curves of both groups were not distinguishable from each other and were identical to that seen in the NOD strain (Fig. 7). This analysis indicates that dose response and proliferative phenotypes do not predict progression to disease. The possibility remains that these protected chimera may retain susceptibility to some of the autoimmune syndromes that can be induced in NOD mice (31–37). This aspect will require further investigation.

Overall, these data indicate that the presence of normal non-B-cell APCs may prevent the development of autoimmunity. We have further demonstrated that the protective function of these “normal” bone marrow derivatives as characterized by previous investigators does not rely on the expression of normal MHC molecules but rather can still be exerted even when these cells also expressed the diabetes associated MHC antigen, I-A^g7. Although our prior data had indicated that NOD B-cells were primarily responsible for the activation of diabetogenic T-cell specificities, these autoimmune strain B-cells were not able to promote diabetes in the presence of other normal APCs (9,13,14,38). Rather, only in the case of the NOD mouse (where B-cells provide the majority of coactivating stimuli in the presence of an effete professional APC compartment) does diabetes ensue.

The mechanism by which these professional APC compartments promote stable tolerance remains to be elucidated. We and others have previously suggested that tolerance induction in the NOD mouse may be impeded by the hypoproliferative phenotype illustrated in the CD4 T-cell compartment (4,14,39,40). However, the data here suggest that the proliferative phenotype of the peripheral T-cell compartment is not predictive of disease development. That both diabetic and resistant F1 chimera were indistinguishable by all estimates of their proliferative capacity suggests that at least part of the tolerance-inducing mechanism operates independently of proliferative achievement. Whether these chimera are stably tolerant or can be induced to develop other aspects of autoimmunity will be an important future assessment of their immunologic status (32–37,41,42). The presence of normal APCs may hinder the targeting of islet tissue while leaving open the possibility that latent tendencies to autoimmunity are unrealized in the absence of other environmental triggers.

Prior studies have amply demonstrated the requirement of B-cell APC function for diabetogenesis and T-cell activation in NOD mice but have not indicated the mechanism by which B-cells participate in the development of autoimmunity. We have considered whether NOD-derived B-cells may actively disrupt the tolerant state and our study has negated the hypothesis that autoimmune diabetes results directly from interaction with B-cells from the NOD background (8–12,38). Our initial analysis of the B-cell compartment indicated a number of anomalies in cell surface molecule expression and painted a picture of heightened costimulatory activity which might promote autoreactivity. Nonetheless, when F1 chimera were constructed in which all B-cells were NOD derived, diabetes did not result despite the maintenance of these features within the B-cell compartment. These data confirm the critical role of other APCs in establishing or maintaining peripheral tolerance. In addition, B-cell-deficient NOD mice may be protected from the development of diabetes by the ineffectual properties of the remaining professional APC compartments rather than by virtue of being tolerant to autoantigens (13). The data presented here may also have important implications for understanding clinical disease entities and for designing preventative therapies. There has been a report of the development of diabetes within a B-cell-deficient kindred (43). Although prior data in murine systems suggests that autoreactive T-cells would not be activated in the absence of B-cells, the data here suggest that the B-cell-deficient NOD may not be tolerant to autoantigens (8,11–13,38,42,44). Moreover, the clinical data may suggest that the T-cell-activating and tolerizing capacity of professional APCs are separable traits. Therefore, it may be possible to restore the T-cell-activating capacity of professional APCs within the NOD system and thereby permit the development of diabetes in the absence of B-cells. Secondarily, these data also suggest that although B-cell depletion may prevent the progression of diabetes, it may not induce a stable state of tolerance without additional manipulations aimed at restoring the function of the professional APC compartment. In this regard, clinical trials that consider B-cell depleting therapies (such as rituximab for the prevention of diabetes) should examine manifestations of tolerance during prolonged follow-up.

Intriguingly, deficiencies in APC activation and function are emerging as critical factors in other models of autoimmunity. The recent characterization of the CARD15/NOD2 gene in inflammatory bowel disease suggests that mutations in the promoter of this toll-like gene product limit macrophage function in those human populations most susceptible to Crohn’s disease (45–49). The inability to produce optimal inflammatory signaling may hinder T-cell targeting and trafficking and result in an increase in the likelihood of inappropriate T-cell responses.

That the presence of a small population of surrogate APCs derived from a nonautoimmune strain can halt the development of diabetes is an exciting observation that may be extended to yet other models of autoimmunity. These data suggest that the deficiencies within the professional antigen-presenting compartments prevent the establishment or maintenance of tolerance and suggest that tolerance induction requires contact with these professional APC lineages. Future studies should determine the activity of these cells and whether their mechanism of action impinges directly on the T-cell compartment or possibly modifies the function of the B-cell pool.

This research was supported by Grants DK54215 and DK49814 from the National Institutes of Health. D.J.M. is supported by the American Diabetes Association Physician Scientist Training Grant.