Insulin Autoantibodies Are Associated With Islet Inflammation But Not Always Related to Diabetes Progression in NOD Congenic Mice

NOD/MrkTacfBR (NOD) mice were purchased from Taconic Farms (Germantown, NY). The NOD congenic strains used in this study were derived as previously reported (Table 1) (9–14). The only exception is that a modified Idd3/5 strain containing a smaller introgressed B10-derived Idd5.1/Idd5.2 region to that previously published (9) was used for the current study. This newly developed Idd3/5 strain resulted from the intercross of the NOD.B10 Idd5R444(9) and NOD.B6 Idd3R450 strains (14). Like the previously described NOD.Idd3 Idd5 strain (9), this refined Idd3/5 strain displays a very low frequency of diabetes (2/100 females diabetic by 7 months of age). The NOD congenic strains are abbreviated in the text according to the control locus or loci (either B6 or B10) that is introgressed onto the NOD background (e.g., NOD.B10 Idd9 is abbreviated as Idd9). Mice were evaluated in a cross-sectional fashion at 8–10, 13–15, and 21–34 weeks of age. Only female mice were studied, and all mice were housed under specific pathogen-free conditions. All NOD congenic mice used in this study are available through the Taconic Farms Emerging Models Program.

The pancreata of mice were removed and fixed in 10% buffered formalin and processed for paraffin embedding. Tissue sections (5 μm) of each pancreas were stained with hematoxylin and eosin and examined microscopically for the presence of mononuclear cell infiltrates in the islets. Two noncontiguous longitudinal sections of each pancreas were examined. The degree of insulitis for each pancreas sample was recorded as follows: 0, no mononuclear cell infiltrate in the islets; 1, mild insulitis, <20% of islets have infiltrates; 2, moderate insulitis, 20–60% of islets have insulitis; 3, severe insulitis, affecting most islets; and 4, extensive insulitis, completely infiltrated islets and/or residual islets.

IAA analysis was performed using a competitive protein A/G based radioimmunoassay (micro-IAA [mIAA] assay) as described previously (4). In brief, human ¹²⁵I-insulin (Amersham) of 20,000 cpm was incubated with 5 μl serum (in the presence or absence of cold human insulin) in a 1:5 dilution of serum overnight at 4°C. Immunoprecipitation of the anti-insulin autoantibodies was performed by adding 50% protein A and 8% protein G-Sepharose (Pharmacia) on a MultiScreen-NOB 96-well filtration plate (Millipore), and radioactivity was counted on a TopCount 96-well plate β-counter (Packard). All of the serum samples were measured in duplicate, and a mean counts per minute was calculated. The final result, expressed as an index, was calculated based on the difference in counts per minute (Δ cpm) between the well without cold insulin and the well with cold insulin: index = (sample Δ cpm - negative control Δ cpm)/(positive control Δ cpm - negative control Δ cpm). A positive IAA measurement is set at the upper limit of normal (0.01) chosen as the 99th percentile in 106 healthy control subjects. The inter-assay coefficient of variation was 11% (n = 8).

Fisher’s exact test was used to compare the cumulative frequencies of IAA positivity among the NOD congenic mice with control NOD mice. Comparisons resulting in P < 0.05 were considered to be significant. The Wilcoxon’s rank-sum test was used to compare the positive IAA indexes and the Student’s t test was used to compare insulitis severity.

Figure 1 summarizes the incidence of IAAs in all of the mice examined in the study and also details the IAA index of the NOD and NOD congenic mice at 8–10 weeks of age. Although the incidence of IAAs in the congenic strains varied widely (Figs. 2 and 3), there was no statistically significant difference in the positive IAA indexes between NOD mice and the diabetes-resistant strains at 8–10 weeks of age (Fig. 1) or at 13–15 and 21–34 weeks of age (data not shown). These data suggest that the absolute level of IAAs is not associated with protection from diabetes.

Anti-islet autoantibodies are believed to be important markers of immune cell activation against β-cells (2). To test the hypothesis that IAA expression is associated with the prevalence of insulitis in animals with genetic susceptibility to diabetes, we measured IAAs in the sera of NOD and NOD congenic mice and examined their pancreata for the presence of infiltrating lymphocytes (Figs. 2 and 3). We previously reported that the early appearance of IAA expression in NOD mice (at 8 weeks of age) was strongly associated with the early development of diabetes (4). In agreement with this finding was a highly significant reduction in IAA prevalence (relative to NOD mice) at 8–10 weeks of age for five of six of the diabetes-protected NOD congenic strains having genes that protect from insulitis and diabetes (Fig. 2A). At this time point, only the Idd5 strain had a prevalence of IAAs that was not statistically significantly lower than that in NOD mice (48% positive [n = 21] vs. 70% positive for NOD, P = NS). Interestingly, the IAA prevalence declined with time in the Idd5 strain and was markedly reduced relative to NOD mice by 13–15 weeks (17%, P = 0.0002) and 21–34 weeks (23%, P = 0.01) (Fig. 2A). The majority of NOD mice at 8–10 weeks of age were positive for IAAs (70%, n = 20) compared with only 24% of Idd3 mice (n = 38, P = 0.002), 0% of Idd3/5 mice (n = 12, P = 0.0001), 27% of Idd10/18 mice (n = 15, P = 0.02), and 17% of Idd3/10/18 mice (n = 6, P = 0.05). With the exception of Idd3 and Idd10/18 mice, there were similar statistically significant reductions in the percent IAA positive among the NOD congenic strains at 13–15 and 22–34 weeks of age. In the Idd3 and Idd10/18 strains, there was a delayed appearance of IAAs, and although the frequency of IAA positivity appears to be reduced at 13–15 and 21–34 weeks, the decrease did not reach statistical significance.

At all three time points examined, NOD mice exhibited a high prevalence of IAAs. In contrast, whereas almost all NOD mice had insulitis at 8–10 weeks of age (16 of 20 females), the severity was significantly less at 8–10 weeks than at 13–15 weeks (P < 0.0001) and 21–34 weeks of age (P = 0.0001). One diabetes-protected strain, Idd3/5, had a highly significant reduction of insulitis compared with that of NOD mice at all three time points (P < 0.0001 for each). The Idd3/10/18 and Idd5 strains were protected from insulitis at 13–15 and 21–34 weeks, times by which many of the NOD mice develop severe insulitis (P values for these four comparisons ranged from <0.0001 to 0.015). The congenic strains Idd3 and Idd10/18 had significant protection from insulitis only at 13–15 weeks (P < 0.0001 and 0.006, respectively). Thus, in the strains of mice having diabetes-resistant genes that protect against insulitis most completely, Idd3/5 and Idd3/10/18, IAAs are rarely detected. When mice have fewer insulitis and diabetes protective genes, such as the Idd3, Idd5, and Idd10/18 strains, IAAs are more frequently observed, albeit with different patterns and prevalence than those in the NOD parental strain.

Idd9 mice have a delayed IAA response that is significantly lower than the IAA prevalence of NOD mice (n = 10, 20 vs. 70% in NOD mice, P = 0.05), but it becomes essentially equivalent to the prevalence of NOD at 13–15 weeks (n = 10, 80%, P = NS) and is similar at 21–34 weeks (n = 11, 45%, P = NS) (Fig. 3A). A similar pattern was observed when insulitis was assessed (Fig. 3B). A reduction in insulitis was observed at 8–10 weeks (P = 0.0031), whereas at 13–15 and 21–34 weeks a majority of Idd9 mice had severe insulitis similar to NOD mice. Despite the intense lymphocytic infiltrate, these mice are dramatically protected from diabetes (3.7% diabetes incidence at 7 months) (12).

To further evaluate the importance of IAAs as an immunologic predictor of type 1 diabetes, we have examined the correlation between IAA prevalence and diabetes frequency in NOD mice and the subset of NOD congenic strains having genes that protect from insulitis in addition to diabetes. The correlation of diabetes incidence (determined at 7 months) with IAA positivity in these mice was statistically significant at 8–10 weeks (r = 0.91, P = < 0.05) and at 13–15 weeks (r = 0.87, P = < 0.05) (data not shown). Because the genetic protection in Idd9 mice appears to occur at a stage subsequent to the development of insulitis and IAAs, these mice were excluded from the correlation analysis.

Type 1 diabetes is an immune-mediated disease in which the expression of IAAs are believed to be predictive of the risk of diabetes progression and age at onset of disease (3,4). The finding of anti-IAAs in patients who do not progress to type 1 diabetes, “nonprogressors,” may relate to the presence of genetic loci providing protection at a stage subsequent to the development of insulitis (similar to the Idd9 mice). In these patients, anti-IAAs may be poor immunologic predictors of diabetes. Idd9 mice have a high incidence and titer of IAAs in association with a high degree of insulitis but are genetically resistant to diabetes (12). The insulitis in these mice is associated with a cytokine profile that is skewed toward a Th2 “protective” phenotype (i.e., interleukin-4 dominant). We evaluated the IgG subclass of the IAAs in the Idd9 mice to determine whether it was consistent with a Th2-dominant response and found no difference compared with control NOD mice, which predominantly have IgG1 IAAs. Of note, we have found that this resistance to diabetes is not reversed with cyclophosphamide treatment (data not shown). It remains to be determined what genetic mechanism underlies this potent resistance to diabetes.

Despite significant effects on IAA prevalence, the protective loci studied here did not appear to have a significant effect on the IAA index of “IAA positive” mice. Analyzing only IAA positive sera, the range of IAA titers was similar among the NOD and NOD congenic mice, and there was no statistically significant difference between the IAA titers. It may be that the timing of IAA appearance is more important than the peak levels for predicting diabetes risk. This may relate to the extremely high risk for children developing IAAs before age 3 (4). An alternative but not mutually exclusive hypothesis is that IAA positivity in the Idd9 mouse and in the “nonprogressors” is associated with a specific yet very slow β-cell destructive process, such that diabetes may occur only after a long follow-up period. In this case, the very slow loss of β-cells may not reveal itself in the form of overt diabetes until years later and thus death from an unrelated cause may occur before the development of diabetes. The data shown here supports the concept that the prevalence of IAAs in NOD mice is related to the prevalence of insulitis in the pancreas, and analysis of congenic-inbred mice indicates that IAAs can be dissociated from diabetes (e.g., Idd9).

As it is important to advance our understanding of the immunopathogenesis of type 1 diabetes and to enhance prediction in individuals with genetic risk, it is equally important to predict which autoantibody-positive individuals will not progress to diabetes. Thus, more analyses of the genetic regulation of IAA expression and the genetic determinants of diabetes progression need to be performed, and the information gained may be useful when deciding which individuals should be targeted to preventive trials.

This study was supported by grants DK-32083 and DK-32493 and a grant from the Children’s Diabetes Foundation at Denver. L.S.W. is supported by a joint grant from the Juvenile Diabetes Research Foundation (JDRF) and the Wellcome Trust. The availability of NOD congenic mice through the Taconic Farms Emerging Models Program has been supported by grants from the Merck Genome Research Institute, the National Institute of Allergy and Infectious Diseases, and the JDRF.

We thank Dongmei Miao, MD, for expert technical assistance and Lorena Jaramillo for assistance with the data analysis.