A variable number of tandem repeats (VNTR) polymorphism upstream of the insulin promoter is strongly associated with type 1 diabetes. The short class I alleles are predisposing and the long class III alleles are protective. As a possible mechanism for this effect, we previously reported a two- to threefold higher insulin transcription from class III than from class I chromosomes in thymus where insulin is expressed at low levels, presumably for the purpose of self-tolerance. In this article, we confirm this finding with independent methodology and report studies testing the hypothesis that class III alleles are associated with T-cell tolerance to (pro)insulin. Cytokine release in vitro after stimulation with 21 overlapping preproinsulin epitopes was assessed in blood mononuclear cells as well as naive and memory CD4+ T-cell subsets from 33 individuals with the high-risk DRB1*04, DQ8 haplotype (12 type 1 diabetic patients, 11 healthy control subjects, and 10 autoantibody-positive subjects). No significant differences between genotypes (24 I/I subjects versus 10 I/III or III/III subjects) were observed for γ-interferon, tumor necrosis factor-α, or interleukin (IL)-4. By contrast, the I/III + III/III group showed a significant threefold higher IL-10 release in memory T-cells for whole proinsulin and the immunodominant region. Given that IL-10 is a marker of regulatory function, our data are consistent with the hypothesis that higher insulin levels in the thymus promote the formation of regulatory T-cells, a proposed explanation for the protective effect of the class III alleles.

The importance of understanding the antigenic targets in the autoimmune process that destroys the β-cells in type 1 diabetes is generally accepted and the matter is the focus of intense research efforts. At least three of the autoantigens initially discovered on the basis of humoral autoimmunity in human patients (insulin, GAD, and IA-2) seem to play a causative role in rodent models, as evidenced by the successful prevention of the disease by tolerization to each specific antigen (13). These results, which have yet to be reproduced in humans, also indicate the possibility of disease prevention by targeting antigens that are not necessarily “primary.” Of the three, insulin may be the most important: its tissue specificity mirrors most closely the exquisite β-cell specificity of the T-cell–mediated autoimmunity in type 1 diabetes, it is the major protein product of these cells, birth-cohort studies show that antibodies to it are the most likely to be the first to appear (4), and it is the only autoantigen whose gene maps to a genetic susceptibility locus (5,6). Thus, understanding how self-tolerance to insulin is achieved and how it fails in type 1 diabetes is likely to be crucial in devising interventions to prevent the autoimmune destruction of β-cells (or its reoccurrence after potential future β-cell regeneration/stem cell therapies).

Until recently, it was not clear how T-cells acquire tolerance to proteins for which expression is highly tissue specific, with the assumption being that this happens mostly in the periphery. Mounting evidence now suggests that a substantial part of tolerance to these antigens is acquired in the thymus, through the ability of specialized thymic cells to synthesize small amounts of a wide variety of tissue-specific antigens (7,8). The case of insulin is particularly interesting. Long before the phenomenon was generalized to other antigens, it was known that small amounts of insulin are expressed in the thymus (9). More interestingly, tolerance to allo-antigens (10) or xeno-antigens (11) transgenically expressed under the insulin promoter can be transferred by thymus transplant to nontransgenic recipients, indicating that such expression has a functional role in tolerance. Multiple mouse and human studies have localized thymic insulin expression to a small subset (not much more than 1%) of medullary stroma cells (7,1113).

Dependence of insulin self-tolerance on thymic expression is also supported by an independent line of evidence based on genetics. Genetic susceptibility accounts for about half of type 1 diabetes risk and behaves as a complex polygenic trait. HLA (especially the DR-DQ region) explains about half of the genetic risk (14); the other half is attributed to an unknown number of minor loci of smaller individual effect. Of the three such loci that have been multiply confirmed, the one conferring the highest relative risk maps to a variable number of tandem repeats (VNTR) polymorphism situated 0.5 kb upstream of the insulin promoter on chromosome 11p15.5. It consists of slight variants of the consensus sequence unit ACAGGGGTGTGGGG repeated 28–44 times (class I alleles) or 138–159 times (class III alleles) (15). Intermediate class II alleles are very rare. Class I alleles are found in ∼80% of Caucasian chromosomes and, in the homozygous state, confer a more than twofold relative risk for type 1 diabetes (15). Conversely (and more meaningfully) stated, the less common (∼20%) class III alleles are dominantly protective. Because the VNTR does not alter peptide sequence, it must affect biology by modulating expression levels of INS, or the adjacent downstream IGF2 gene. We have excluded any substantial effect on IGF2 in tissues relevant to type 1 diabetes (1618), and, in heterozygous human fetal pancreas, the class III allele was associated with a modestly (15–20%) lower mRNA levels than the class I allele (18). Bennett et al. (19) found similar results for insulin expression in adult pancreas. Although lower expression of an autoantigen might protect from autoimmunity, in our opinion, this marginal loss of function would be a poor explanation for a dominant effect.

A better explanation was found in the thymus where, in heterozygous human fetal samples, class III alleles enhance insulin transcription two- to threefold (20,21) (Fig. 1). This led to the obvious hypothesis that the INS-VNTR exerts its effect on type 1 diabetes risk through more secure T-cell tolerance induction to insulin epitopes achieved by higher levels in the thymus. This model was subsequently supported by observations in mice with targeted disruption of the insulin genes.

Murine insulin is encoded by two homologous but unlinked genes: Ins1 (not present in humans) and Ins2 (syntenic to human INS). Whereas in the pancreas, the two genes are expressed at about equal levels, Ins1 is minimally expressed in the thymus (22). Ins2(−/−) mice have normal pancreatic insulin and glucose levels (23) but drastically reduced insulin expression in the thymus (22). They display spontaneous T-cell autoreactivity to human insulin even against a diabetes-resistant background (22,24). When crossed with the NOD mouse, diabetes occurs more frequently and at an earlier age, an enhancement of the autoimmune process that is specific to insulin (25). Conversely, NOD-Ins1(−/−) mice are remarkably resistant to diabetes (26), compatible with the notion that in that model, the autoimmunity is primarily against the insulin peptide not expressed in thymus.

The constraints of human research, essentially limiting samples to peripheral blood, make it difficult to test predictions of this model in the human. INS-VNTR status had no effect on the frequency of anti-insulin autoantibodies in one small study (27), whereas another much larger study found a significantly higher proportion of class III alleles in subjects who had anti-GAD but not anti-insulin antibodies (28), suggesting a protective effect of class III against insulin autoimmunity, even though β-cell autoimmunity directed against other antigens could occur. In any event, the relevance of humoral immunity to what is essentially a cell-mediated process is not clear.

Direct evaluation of cellular reactivity to insulin is much more challenging. Recent publications, including our own work, support the concept that type 1 diabetes autoimmunity reflects a defect in an integrated system of immunoregulation mediated by distinct T-cell subsets (2935). In recent studies using peripheral blood lymphocytes and two CD4+ T-cell subsets (CD45RA memory and CD45RA+ naive and recently primed) of subjects with the high-risk DRB1*04,DQ8 haplotype, we have shown that the cytokine production patterns of preproinsulin reactive T-cells differed significantly between autoimmune subjects (autoantibody-positive [Ab+] and type 1 diabetic patients) and healthy HLA-matched control subjects (31). In insulin-treated type 1 diabetic patients, naive and recently primed CD4+ cells were characterized by an increase in T-helper (Th)-1 cytokines (interferon [IFN]-γ and tumor necrosis factor [TNF]-α). In contrast, in CD4+ T-cell subsets of Ab+ subjects, T regulatory (IL-10) and Th2 (IL-4) phenotypes dominated, whereas Th0 responses were present in the control subjects. The CD45RA+IL-10hi T-cell phenotype has been associated with regulatory function in subjects with type 1 diabetes and in experimental models of autoimmunity (29,34,36,37).

The purpose of the current study was to see how the in vitro response of peripheral blood mononuclear cells (PBMCs) and CD4+ T-cell subsets to insulin peptide epitopes, in terms of induction of the most important T-cell cytokines, correlates with the genotype at the INS-VNTR locus. The results show substantial differences in IL-10 production, consistent with promotion of regulatory cells in subjects with the protective genotype. In parallel, we also confirm our initial findings of the transcriptional allelic effect of the VNTR in human thymus in a new batch of thymus samples, using an independent and more reliable technique.

A total of 33 subjects with the high-risk HLA-DRB1*04, DQ8 haplotype were analyzed: 11 healthy control subjects without family history of type 1 diabetes (median age 23 years, range 2–43), 10 Ab+ schoolchildren without family history of type 1 diabetes from the Karlsburg type 1 diabetes risk study (38) (median age 14.5 years, range 8–24), and 12 recently diagnosed insulin-treated patients with type 1 diabetes (median age 24 years, range 2–56; median duration of insulin treatment 3 months, range 1–12). Of the 10 Ab+ individuals, 8 were classified as “high-risk” subjects, since they were positive for more than one antibody specificity, i.e., insulin autoantibodies, antibodies against GAD, or islet thyrosine phosphatase and/or had a high titer of cytoplasmic islet cell antibodies (>20 Juvenile Diabetes Foundation units) (39,40). Subject characteristics are described in Table 1.

Autoantibody assays.

Subject sera were tested for autoantibodies in the Institute of Pathophysiology, Karlsburg, Germany, as described (38,41). HLA typing of DRB1* and DQB1* alleles was performed using allele-specific PCR amplification procedure as described elsewhere (42). Informed consent was obtained from all individuals before analysis, and studies were carried out in accordance with the Declaration of Helsinki.

Cell isolation.

Peripheral blood mononuclear cells were isolated from heparinized blood by Ficoll-Paque (Pharmacia, Freiburg, Germany) density centrifugation, aliquoted, and cryopreserved in liquid nitrogen until use. Enrichment of CD45RA memory and CD45RA+ naive/recently primed Th cells of all investigated individuals was performed in two steps using a CD4+ T-cell isolation kit (Miltenyi Biotec, Auburn, CA), and stimulation assay was modified for enriched Th cell subsets as previously described (30).

Autoantigens and peptides.

Human proinsulin (86 amino acids; a gift from Eli Lilly, Indianapolis, IN) and insulin (51 amino acids; a gift from Aventis, Frankfurt, Germany) were tested simultaneously with 21 overlapping preproinsulin peptides. Preproinsulin peptides were synthesized according to the primary preproinsulin structure (GenBank accession number P01308). They were 16–17 amino acids long and overlapped by 12 amino acids. In each T-cell assay, control peptide CASSSDRLGNQPQHF (T-cell receptor peptide, 43), tetanus toxoid (Connaught Laboratories, Toronto, Canada), and phytohemagglutinin M (Difco, Detroit, MI) were included as described previously (30).

Cytokine assays.

To measure the level of secreted TNF-α, IFN-γ, IL-2, IL-4, IL-5, and IL-10, after stimulation with proinsulin, insulin, and 21 overlapping preproinsulin peptides, 175 μl supernatant was taken from each well of a microtiter plate on day 5 of the culture as previously described (30). Supernatants of replicate cultures were pooled and stored at −20°C until assay. They were analyzed according to the manufacturer’s instructions using an antigen-capture enzyme-linked immunosorbent assay from PharMingen (San Diego, CA). Detection limits for cytokines were as follows: 19.5 pg/ml for TNF-α, 34.2 pg/ml for IFN-γ, 9.8 pg/ml for IL-2, 19.5 pg/ml for IL-4, 48.8 pg/ml for IL-5, and 4.9 pg/ml for IL-10. Quantification of spontaneous cytokine release was performed by incubating the cells of each individual under the same conditions as preproinsulin peptides in the absence of antigen. Positive cytokine responses were defined as two adjacent peptide (antigen)-stimulated wells, which gave responses greater than the mean + 2 SD of unstimulated wells. The intra-assay coefficient variation for IL-10, calculated on five replicate tests, was 27.8%. The amount of secreted cytokines was expressed in picograms per milliliter.

VNTR typing.

The VNTR genotype was deduced from the genotype at the −23 A/T single nucleotide polymorphism (SNP) at the INS promoter, whose A allele is in very tight linkage disequilibrium with the class I and the T with class III (r2 = 1). The SNP was typed by single-nucleotide primer extension on PCR-amplified DNA according to the PE AcycloPrime-fluorescence polarization protocol in a GeneAmp PCR system 9700 (MJ Research, Waltham, MA). Final detection of the SNP by fluorescence polarization used the Criterion System with Analyst HT (Molecular Devices, Sunnyvale, CA). In a study of 2,500 samples, the genotyping call rate for this assay was 97%, Mendelian error <0.001, and no divergence from the Hardy-Weinberg equilibrium was detected.

Tissues.

Twelve human fetal thymus and pancreas samples were obtained at the time of pregnancy interruption for reasons other than health problems in mother or fetus. The sampling was approved by the Research Ethics Board of the McGill University Health Center, and written consent was obtained from all mothers. Samples were snap-frozen and kept at −70°C. After pulverizing frozen tissue in a liquid N2–filled mortar and pestle, Trizol (Invitrogen) was added for extracting DNA and RNA.

Allelic imbalance assay.

INS transcripts corresponding to each VNTR allele in heterozygous samples were distinguished using the +1,127 C/T SNP at the INS 3′ UTR.

Levels of each allele in PCR or RT-PCR product (primers and cycling as described by Vafiadis et al. [20]) were quantitatively evaluated by single-nucleotide primer extension, using ddNTPs labeled with different fluorochromes (SnapShot kit, Applied Biosystems International) followed by capillary electrophoresis.

Statistical analysis.

The Kalmogorov-Smirnov test indicated that all cytokine results significantly diverged from a normal distribution. Therefore, all comparisons between means were performed by permutation, using resampling without replacement (Resampling Stats software, http://www.resample.com). Results are summarized as means ± SE. SEs were calculated by resampling with replacement.

Confirmation of the INS-VNTR transcriptional effect on thymus.

Our initial report of the allelic INS-VNTR effect on thymic insulin transcripts was based on the relative intensity of radioactive bands of the two alleles by RT-PCR restriction fragment-length polymorphism (20).

PCR versus RT-PCR products could differentially affect the efficiency of digestion by the PstI endonuclease used in that study (for example, through differential formation of heteroduplexes that the enzyme cannot digest). We rigorously controlled for this by spiking the PCR product with RNA and digesting the RT-PCR product with RNAase (20). However, the investment in effort to pursue this original finding justifies rigorous confirmation, which we undertook in a new set of 12 heterozygous human fetal samples with a completely independent method that does not involve restriction digestion. Figure 1 shows the results, which are in concordance with our previous observation, showing a statistically highly significant average 2.6-fold higher level in the allele derived from the class III chromosome compared with the class I chromosome.

In vitro lymphocyte response to proinsulin and immunodominant epitopes.

As we have previously reported, measurable cytokine responses to preproinsulin peptides are observed in a number of subjects, whether they are type 1 diabetic patients, normal control subjects, or healthy but Ab+ individuals. We compared in vitro cytokine release to preproinsulin peptides by PBMCs as well as the naive (CD45RA+) or memory (CD45RO+) CD4+ T-cell subsets from high-risk DRB1*04, DQ8 individuals. I/I homozygotes were compared with those with at least one class III allele (I/III and III/III) without distinction as to disease or autoantibody status.

A measurable response of different cytokines to different preproinsulin epitopes was seen in a significant proportion of samples. The most striking and consistent difference between genotypes was observed for IL-10. Responses were higher in PBMC samples from subjects with the I/III or III/III genotype than in I/I homozygotes, most of whom had a weak or no response to whole proinsulin (Fig. 2). Mean responses were 4.2 ± 1.5 vs. 14.1 ± 6.6, respectively (P = 0.02). Whole insulin responses were not different between genotype groups (Fig. 3). The genetic effect on proinsulin response could be accounted for entirely by the response of the CD4+CD45RA set of memory T-cells, as the response of the CD4+CD45RA+ set was no different between genotype groups (Fig. 2). Response of memory cells to all four peptides covering the DR4-restricted immunodominant proinsulin 64–95 region (30,33,44) elicited a three- to fourfold higher response in the class III–containing genotypes than in I/I. Averaged response to the four peptides was 17.5 vs. 4.4 pg/ml for class I/III + III/III vs. class I/I (P = 0.007). Response to the 79–95 peptide was 16.3 vs. 1.8 pg/ml (P = 0.0002), a P value that remains significant even after Bonferroni correction for three cell types and 21 epitopes (0.0002 × 3 × 21 = 0.0126; however, such correction would have been much too stringent because peptide responses highly correlated with each other). Increased IL-10 secretion of the subjects with the protective class III genotypes, preferentially to the peptides covering the enzymatic cleavage site between C-peptide and insulin A-chain (present only in the intact proinsulin molecule), demonstrate that the reactivity is directed primarily to the insulin autoantigen and not to the exogenous insulin. Somewhat less striking differences between genotypes were also seen in several of the epitopes outside the 64–95 region, such as the two epitopes on the insulin B-chain (30–46 and 35–51) overlapping the previously reported B9–23 region (45).

The number of epitopes for which each subject’s cells were positive for IL-10 was again significantly higher in PBMCs of I/III+III/III subjects (12.5 ± 2.5 epitopes vs. 5.2 ± 1.9 for I/I subjects, P = 0.037). Comparison of the CD4+CD45RA memory IL-10 responses to preproinsulin epitopes of class III positive autoimmune subjects with control subjects (despite relatively low numbers) revealed higher numbers of recognized epitopes in the control subjects (18.5 ± 2.8 peptides vs. 11 ± 10.1 in autoimmune subjects). In contrast, in class I positive subjects there was no difference in the number of recognized epitopes between autoimmune subjects and control subjects (11.3 ± 6.8 vs. 10 ± 7.7).

The few differences between genotypes observed in the other cytokines were small and of borderline significance (summarized in Table 2), except for IL-2 and IL-5 responses, which were higher in I/III individuals; IL-2 in memory cells for proinsulin (42.5 vs. 9 pg/ml, P = 0.002); and IL-5 in PBMCs for the immunodominant proinsulin 64–95 region (40 vs. 4 pg/ml, P = 0.009).

It is generally accepted that elucidating the complex genetics of type 1 diabetes susceptibility is an important part of the effort to devise safe and effective preventive interventions, through better understanding of the factors responsible for the autoimmune process. In addition, as has been argued recently (46,47), an additional potential advantage of genetic prediction could be the distinction between different susceptibility genotypes, each encoding a different immune dysregulation phenotype that may respond differently to available alternative interventions.

Toward this end, detecting disease association with DNA sequence variants is only the first step. The ultimate goal is to understand the mechanism by which the variant alters biology. These mechanisms are often far from obvious. For example, decades after the strong effect of HLA haplotypes on type 1 diabetes risk was first observed, the biological mechanism of this effect is largely unknown. In the case of insulin, based on the evidence currently available, the most plausible mechanism would involve allele-dependent levels of insulin expression in the thymus. Central T-cell tolerance to β-cell autoantigens, determined in the thymus, is of obvious importance in type 1 diabetes, a T-cell–mediated disease.

The classical model of central tolerance would involve more efficient negative selection of insulin-autoreactive T-cells made possible by more abundant autoantigen. Indeed, insulin and other tissue-specific autoantigens are expressed only in thymic medulla (7,8,11), the site of negative selection. This model would predict a different effector cytokine phenotype of T-cell response to insulin epitopes by individuals with a class III allele. In the small sample examined, with admittedly limited statistical power, we did not observe such an effect. Our most striking finding involved IL-10, a cytokine now recognized as an important marker of regulatory function by T-cells.

In recent years, it has become increasingly obvious that elimination of autoreactive T-cells, either centrally in the thymus or through peripheral processes, only partly explains T-cell tolerance. Self is actively protected by T-cells with regulatory function. The best-examined paradigms involve the generation of the TH2 subset of CD4+ cells and, more recently recognized, CD4+ cells constitutively expressing CD25. One important difference between the two, relevant to the mechanism of action of the insulin VNTR, is that TH2 cells are generated on the basis of signals received in the periphery, whereas the CD4+CD25+ subset is already present and active in the thymus (48). Even more importantly, the generation of these cells in the thymus appears to require high-avidity interaction of their T-cell receptor with self-peptide (49). Because avidity depends on both affinity and abundance of the antigen, the more than twofold higher expression from the class III VNTR could explain the more efficient selection of these cells in the thymus of individuals carrying this allele.

The cells producing the excess IL-10 we observed in subjects with a class III allele could have also been generated in the periphery. In that case, there is no obvious explanation for the genetic effect. There are reports of immunoreactive insulin expression by dendritic cells in peripheral lymphoid tissues (12) but, in our hands, rigorously controlled RT-PCR has failed to detect such expression. If it occurs, the effect of the INS-VNTR genotype on this peripheral expression has not been studied and it is not known whether it is enhanced by the class III allele as in thymus, diminished as in pancreas, or behaving yet in a different way.

Peripherally generated regulatory T-cells act through secretion of regulatory cytokines, notably IL-4 and IL-10. Thymus-generated CD4+CD25+ cells, on the other hand, regulate via a mechanism that involves cell contact but also express IL-10 (rev. in 50 and 51). Thus, although this preliminary study presents evidence suggesting that insulin-specific regulatory function is involved in the genetic effect of the INS-VNTR, it is not clear that this regulation is necessarily determined in the thymus. A thymic explanation is certainly the most attractive because a transcriptional mechanism compatible with the observation has been observed and confirmed.

In future studies, after we confirm the data presented here in a larger set of samples, we propose to dissect the question of thymic versus peripheral effects by repeating the studies on the CD4+CD25+ subset, as well as on lymphocytes obtained from fetal and surgical samples of human thymus.

FIG. 1.

Confirmation of the allelic transcriptional effect of the INS-VNTR on thymus insulin expression in 12 heterozygous samples. Capillary electrophoresis peaks corresponding to two alleles of an SNP, labeled by single-nucleotide primer extension with ddNTPs labeled with different fluorochromes. Because of perfect linkage disequilibrium with the VNTR, the A allele (black curve) is transcribed from the class I chromosome and the T allele (gray curve) from the class III chromosome. Heterozygous DNA is used to define 1:1 allele stoichiometry. A: Individual values for 10 samples of DNA, pancreas RNA, and thymus RNA as ratios of class III expression level over class I. Values are normalized taking the average DNA value as 1:1, to compensate for different ddNTP-fluorochrome efficiencies. B: A representative capillary electrophoresis profile.

FIG. 1.

Confirmation of the allelic transcriptional effect of the INS-VNTR on thymus insulin expression in 12 heterozygous samples. Capillary electrophoresis peaks corresponding to two alleles of an SNP, labeled by single-nucleotide primer extension with ddNTPs labeled with different fluorochromes. Because of perfect linkage disequilibrium with the VNTR, the A allele (black curve) is transcribed from the class I chromosome and the T allele (gray curve) from the class III chromosome. Heterozygous DNA is used to define 1:1 allele stoichiometry. A: Individual values for 10 samples of DNA, pancreas RNA, and thymus RNA as ratios of class III expression level over class I. Values are normalized taking the average DNA value as 1:1, to compensate for different ddNTP-fluorochrome efficiencies. B: A representative capillary electrophoresis profile.

Close modal
FIG. 2.

Genotype dependence of IL-10 expression by blood lymphocytes stimulated with proinsulin (♦ ⋄, control subjects; ▪ □, Ab+ subjects; • ○, type 1 diabetic patients). In PBMCs, most I/I subjects (marked as I) show little or no IL-10 production, whereas subjects with at least one class III allele (marked as III) show significantly higher responses. This is observed in the memory but not the naive subset.

FIG. 2.

Genotype dependence of IL-10 expression by blood lymphocytes stimulated with proinsulin (♦ ⋄, control subjects; ▪ □, Ab+ subjects; • ○, type 1 diabetic patients). In PBMCs, most I/I subjects (marked as I) show little or no IL-10 production, whereas subjects with at least one class III allele (marked as III) show significantly higher responses. This is observed in the memory but not the naive subset.

Close modal
FIG. 3.

Genotype dependence of IL-10 expression by blood lymphocytes stimulated with insulin, or the proinsulin-specific immunodominant epitopes (♦ ⋄, control subjects; ▪ □, Ab+ subjects; • ○, type 1 diabetic patients). There is no difference in whole insulin response, but all epitopes elicit significantly higher responses in class III individuals in memory cells.

FIG. 3.

Genotype dependence of IL-10 expression by blood lymphocytes stimulated with insulin, or the proinsulin-specific immunodominant epitopes (♦ ⋄, control subjects; ▪ □, Ab+ subjects; • ○, type 1 diabetic patients). There is no difference in whole insulin response, but all epitopes elicit significantly higher responses in class III individuals in memory cells.

Close modal
TABLE 1

Subject characteristics

ID numberAge (years)Type 1 diabetes (months)INS-VNTRHLA haplotype (DRB1*, DQB1*)Autoantibodies
ICA (JDF-U)IAA/IA (μU/l)GADA (KU/l)IA-2A (KU/l)
Ab+ subjects 12  I, I 0401, 0302 320 1,074 39 3,328 
 17  I, I 04, 0302 320 73 55 544 
 14  I, I 0401, 0302 209 37 
 19  I, I 0404, 0302 12 4 
 24  I, I 0404, 0302 −55 10 9 
 15  I, I 04, 0302 80 57 11 115 
 10  I, I 0401, 0302 251 
 15  I, I 0402, 0302 54 11 5 
  I, I 04, 0302 40 418 12 11 
 10 10  I, III 0401, 0302 160 479 64 23 
Type 1 diabetic patients 11 I, I 0404, 0302 23,123 4 30 
 12 10 I, I 04, 0302 80 11,663 8 4 
 13 14 I, I 0401, 0302 
 14 22 I, I 0401, 0302 63 110 19 
 15 24 I, I 0402, 0302 21 39 41 
 16 27 I, I 0401, 0302 160 5,332 41 60 
 17 28 11 I, I 0401, 0302 320 12,389 127 2 
 18 30 12 I, I 0401, 0302 80 4,049 87 550 
 19 31 I, I 0401, 0302 59 5 
 20 18 10 I, III 0401, 0302 14,021 50 3 
 21 45 I, III 0401, 0302 23,434 
 22 56 I, III 0401, 0302 1,280 1,249 79 154 
 23 12 I, III 0401, 0302 640 172 65 58 
Control subjects 24  I, I 0404, 0302 −49 
 25 15  I, I 0401, 0302 53 
 26 20  I, I 0401, 0302 63 
 27 20  I, I 0401, 0302 −2 
 28 29  I, I 0407, 0302 26 
 29 29  I, I 0404, 0302 64 
 30 23  III 0401, 0302 
 31 27  I, III 0404, 0302 34 
 32 43  I, III 0401, 0302 203 
 33 20  III 0404, 0302 130 
 34 29  I, III 0403, 0302 17 
ID numberAge (years)Type 1 diabetes (months)INS-VNTRHLA haplotype (DRB1*, DQB1*)Autoantibodies
ICA (JDF-U)IAA/IA (μU/l)GADA (KU/l)IA-2A (KU/l)
Ab+ subjects 12  I, I 0401, 0302 320 1,074 39 3,328 
 17  I, I 04, 0302 320 73 55 544 
 14  I, I 0401, 0302 209 37 
 19  I, I 0404, 0302 12 4 
 24  I, I 0404, 0302 −55 10 9 
 15  I, I 04, 0302 80 57 11 115 
 10  I, I 0401, 0302 251 
 15  I, I 0402, 0302 54 11 5 
  I, I 04, 0302 40 418 12 11 
 10 10  I, III 0401, 0302 160 479 64 23 
Type 1 diabetic patients 11 I, I 0404, 0302 23,123 4 30 
 12 10 I, I 04, 0302 80 11,663 8 4 
 13 14 I, I 0401, 0302 
 14 22 I, I 0401, 0302 63 110 19 
 15 24 I, I 0402, 0302 21 39 41 
 16 27 I, I 0401, 0302 160 5,332 41 60 
 17 28 11 I, I 0401, 0302 320 12,389 127 2 
 18 30 12 I, I 0401, 0302 80 4,049 87 550 
 19 31 I, I 0401, 0302 59 5 
 20 18 10 I, III 0401, 0302 14,021 50 3 
 21 45 I, III 0401, 0302 23,434 
 22 56 I, III 0401, 0302 1,280 1,249 79 154 
 23 12 I, III 0401, 0302 640 172 65 58 
Control subjects 24  I, I 0404, 0302 −49 
 25 15  I, I 0401, 0302 53 
 26 20  I, I 0401, 0302 63 
 27 20  I, I 0401, 0302 −2 
 28 29  I, I 0407, 0302 26 
 29 29  I, I 0404, 0302 64 
 30 23  III 0401, 0302 
 31 27  I, III 0404, 0302 34 
 32 43  I, III 0401, 0302 203 
 33 20  III 0404, 0302 130 
 34 29  I, III 0403, 0302 17 

The numbers in bold represent positive antibody titers. IA, insulin antibody; IAA, insulin autoantibody; JDF-U, Juvenile Diabetes Foundation units.

TABLE 2

Cytokine secretion upon stimulation with proinsulin (P-Ins), insulin, and overlapping peptides covering the immunodominant P-Ins64–95 region

INS-VNTRPBMC
CD45RA
CD45RA+
P-InsInsP-Ins64–95P-InsInsP-Ins64–95P-InsInsP-Ins64–95
IFN-γ 9 ± 4 40 ± 27 19 ± 8 15 ± 6 14 ± 6 71 + 28 21 ± 11 20 ± 13 22 ± 9 
 III 0 ± 0 57 ± 48 11 + 9 13 ± 6 25 ± 10 71 + 40 7 ± 5 32 ± 14 37 ± 20 
TNF-α 5 ± 2 12 ± 5 7 ± 3 5 ± 4 7 ± 6 8 ± 6 6 ± 5 4 ± 2 12 ± 6 
 III 7 ± 5 79 ± 65 5 ± 3 0 ± 0 0 ± 0 3 ± 2 0 ± 0 6 ± 1 5 ± 6 
IL-2 10 ± 18 19 ± 14 14 ± 23 9 ± 7 21 ± 37 24 ± 50 37 ± 95 40 ± 91 31 ± 65 
 III 24 ± 26 21 ± 9 19 ± 26 43 ± 7 40 ± 38 48 ± 44 10 ± 10 23 ± 42 18 ± 32 
     P = 0.002      
IL-4 7 ± 4 12 ± 4 8 ± 1 10 ± 4 15 ± 5 23 ± 5 23 ± 13 25 ± 13 14 ± 7 
 III 8 ± 5 9 ± 1 5 ± 3 15 ± 6 25 ± 2 34 ± 5 15 ± 9 4 ± 5 8 ± 5 
IL-5 7 ± 4 9 ± 5 4 ± 2 10 ± 4 10 ± 4 18 ± 4 0 ± 0 3 ± 2 9 ± 4 
 III 23 ± 18 16 ± 8 40 ± 16 0 ± 0 131 ± 119 73 ± 54 104 ± 93 46 ± 41 55 ± 45 
    P = 0.009       
IL-10 4 ± 2 8 ± 11 4 ± 6 5 ± 1 10 ± 3 5 ± 1 7 ± 4 7 ± 4 4 ± 1 
 III 15 ± 6 13 ± 14 11 ± 15 14 ± 6 13 ± 7 18 ± 6 6 ± 5 2 ± 1 7 ± 3 
  P = 0.03   P = 0.03  P = 0.004    
INS-VNTRPBMC
CD45RA
CD45RA+
P-InsInsP-Ins64–95P-InsInsP-Ins64–95P-InsInsP-Ins64–95
IFN-γ 9 ± 4 40 ± 27 19 ± 8 15 ± 6 14 ± 6 71 + 28 21 ± 11 20 ± 13 22 ± 9 
 III 0 ± 0 57 ± 48 11 + 9 13 ± 6 25 ± 10 71 + 40 7 ± 5 32 ± 14 37 ± 20 
TNF-α 5 ± 2 12 ± 5 7 ± 3 5 ± 4 7 ± 6 8 ± 6 6 ± 5 4 ± 2 12 ± 6 
 III 7 ± 5 79 ± 65 5 ± 3 0 ± 0 0 ± 0 3 ± 2 0 ± 0 6 ± 1 5 ± 6 
IL-2 10 ± 18 19 ± 14 14 ± 23 9 ± 7 21 ± 37 24 ± 50 37 ± 95 40 ± 91 31 ± 65 
 III 24 ± 26 21 ± 9 19 ± 26 43 ± 7 40 ± 38 48 ± 44 10 ± 10 23 ± 42 18 ± 32 
     P = 0.002      
IL-4 7 ± 4 12 ± 4 8 ± 1 10 ± 4 15 ± 5 23 ± 5 23 ± 13 25 ± 13 14 ± 7 
 III 8 ± 5 9 ± 1 5 ± 3 15 ± 6 25 ± 2 34 ± 5 15 ± 9 4 ± 5 8 ± 5 
IL-5 7 ± 4 9 ± 5 4 ± 2 10 ± 4 10 ± 4 18 ± 4 0 ± 0 3 ± 2 9 ± 4 
 III 23 ± 18 16 ± 8 40 ± 16 0 ± 0 131 ± 119 73 ± 54 104 ± 93 46 ± 41 55 ± 45 
    P = 0.009       
IL-10 4 ± 2 8 ± 11 4 ± 6 5 ± 1 10 ± 3 5 ± 1 7 ± 4 7 ± 4 4 ± 1 
 III 15 ± 6 13 ± 14 11 ± 15 14 ± 6 13 ± 7 18 ± 6 6 ± 5 2 ± 1 7 ± 3 
  P = 0.03   P = 0.03  P = 0.004    

Data are means ± SE. Values are rounded to the nearest integer. Ins, insulin; P-Ins, proinsulin.

This article is based on a presentation at a symposium. The symposium and the publication of this article were made possible by an unrestricted educational grant from Servier.

This work was supported by grants from the Juvenile Diabetes Research Foundation (to C.P.), the German Research Foundation (SFB 518 to I.D.-B., B.O.B., and W.K.), the Eli Lilly Foundation International (to I.D.-B.), the German Diabetes Foundation, and the Foundation for the Diabetic Child (to I.D.-B.).

1.
Gottlieb PA, Eisenbarth GS: Diagnosis and treatment of pre-insulin dependent diabetes.
Annu Rev Med
49
:
391
–405,
1998
2.
Graves PM, Eisenbarth GS: Pathogenesis, prediction and trials for the prevention of insulin-dependent (type 1) diabetes mellitus.
Adv Drug Deliv Rev
35
:
143
–156,
1999
3.
Pozzilli P: Prevention of insulin-dependent diabetes mellitus.
Diabete Metab Rev
14
:
69
–84,
1998
4.
Achenbach P, Koczwara K, Knopff A, Naserke H, Ziegler AG, Bonifacio E: Mature high-affinity immune responses to (pro)insulin anticipate the autoimmune cascade that leads to type 1 diabetes.
J Clin Invest
114
:
589
–597,
2004
5.
Lucassen AM, Julier C, Beressi JP, Boitard C, Froguel P, Lathrop M, Bell JI: Susceptibility to insulin dependent diabetes mellitus maps to a 4.1 kb segment of DNA spanning the insulin gene and associated VNTR.
Nat Genet
4
:
305
–310,
1993
6.
Bennett ST, Todd JA: Human type 1 diabetes and the insulin gene: principles of mapping polygenes.
Annu Rev Genet
30
:
343
–370,
1996
7.
Derbinski J, Schulte A, Kyewski B, Klein L: Promiscuous gene expression in medullary thymic epithelial cells mirrors the peripheral self.
Nat Immunol
2
:
1032
–1039,
2001
8.
Kyewski B, Derbinski J, Gotter J, Klein L: Promiscuous gene expression and central T-cell tolerance: more than meets the eye.
Trends Immunol
23
:
364
–371,
2002
9.
Jolicoeur C, Hanahan D, Smith KM: T-cell tolerance toward a transgenic beta-cell antigen and transcription of endogenous pancreatic genes in thymus.
Proc Natl Acad Sci U S A
91
:
6707
–6711,
1994
10.
Heath WR, Allison J, Hoffmann MW, Schonrich G, Hammerling G, Arnold B, Miller JF: Autoimmune diabetes as a consequence of locally produced interleukin-2.
Nature
359
:
547
–549,
1992
11.
Smith KM, Olson DC, Hirose R, Hanahan D: Pancreatic gene expression in rare cells of thymic medulla: evidence for functional contribution to T cell tolerance.
Int Immunol
9
:
1355
–1365,
1997
12.
Pugliese A, Brown D, Garza D, Murchison D, Zeller M, Redondo M, Diez J, Eisenbarth GS, Patel DD, Ricordi C: Self-antigen-presenting cells expressing diabetes-associated autoantigens exist in both thymus and peripheral lymphoid organs.
J Clin Invest
107
:
555
–564,
2001
13.
Chentoufi AA, Palumbo M, Polychronakos C: Proinsulin expression by Hassall’s corpuscles in the mouse thymus.
Diabetes
53
:
354
–359,
2004
14.
Risch N: Assessing the role of HLA-linked and unlinked determinants of disease.
Am J Hum Genet
40
:
1
–14,
1987
15.
Stead JD, Buard J, Todd JA, Jeffreys AJ: Influence of allele lineage on the role of the insulin minisatellite in susceptibility to type 1 diabetes.
Hum Mol Genet
9
:
2929
–2935,
2000
16.
Vafiadis P, Grabs R, Goodyer CG, Colle E, Polychronakos C: A functional analysis of the role of IGF2 in IDDM2-encoded susceptibility to type 1 diabetes.
Diabetes
47
:
831
–836,
1998
17.
Vafiadis P, Bennett ST, Todd JA, Grabs R, Polychronakos C: Divergence between genetic determinants of IGF2 transcription levels in leukocytes and of IDDM2-encoded susceptibility to type 1 diabetes.
J Clin Endocrinol Metab
83
:
2933
–2339,
1998
18.
Vafiadis P, Bennett ST, Colle E, Grabs R, Goodyer CG, Polychronakos C: Imprinted and genotype-specific expression of genes at the IDDM2 locus in pancreas and leucocytes.
J Autoimmun
9
:
397
–403,
1996
19.
Bennett ST, Wilson AJ, Cucca F, Nerup J, Pociot F, McKinney PA, Barnett AH, Bain SC, Todd JA: IDDM2-VNTR-encoded susceptibility to type 1 diabetes: dominant protection and parental transmission of alleles of the insulin gene-linked minisatellite locus.
J Autoimmun
9
:
415
–421,
1996
20.
Vafiadis P, Bennett ST, Todd JA, Nadeau J, Grabs R, Goodyer CG, Wickramasinghe S, Colle E, Polychronakos C: Insulin expression in human thymus is modulated by INS VNTR alleles at the IDDM2 locus.
Nat Genet
15
:
289
–292,
1997
21.
Pugliese A, Zeller M, Fernandez A Jr, Zalcberg LJ, Bartlett RJ, Ricordi C, Pietropaolo M, Eisenbarth GS, Bennett ST, Patel DD: The insulin gene is transcribed in the human thymus and transcription levels correlated with allelic variation at the INS VNTR-IDDM2 susceptibility locus for type 1 diabetes.
Nat Genet
15
:
293
–297,
1997
22.
Chentoufi AA, Polychronakos C: Insulin expression levels in the thymus modulate insulin-specific autoreactive T-cell tolerance: the mechanism by which the IDDM2 locus may predispose to diabetes.
Diabetes
51
:
1383
–1390,
2002
23.
Leroux L, Desbois P, Lamotte L, Duvillie B, Cordonnier N, Jackerott M, Jami J, Bucchini D, Joshi RL: Compensatory responses in mice carrying a null mutation for Ins1 or Ins2.
Diabetes
50 (Suppl. 1)
:
S150
–S153,
2001
24.
Faideau B, Briand JP, Lotton C, Tardivel I, Halbout P, Jami J, Elliott JF, Krief P, Muller S, Boitard C, Carel JC: Expression of preproinsulin-2 gene shapes the immune response to preproinsulin in normal mice.
J Immunol
172
:
25
–33,
2004
25.
Thebault-Baumont K, Dubois-Laforgue D, Krief P, Briand JP, Halbout P, Vallon-Geoffroy K, Morin J, Laloux V, Lehuen A, Carel JC, Jami J, Muller S, Boitard C: Acceleration of type 1 diabetes mellitus in proinsulin 2-deficient NOD mice.
J Clin Invest
111
:
851
–857,
2003
26.
Moriyama H, Abiru N, Paronen J, Sikora K, Liu E, Miao D, Devendra D, Beilke J, Gianani R, Gill RG, Eisenbarth GS: Evidence for a primary islet autoantigen (preproinsulin 1) for insulitis and diabetes in the nonobese diabetic mouse.
Proc Natl Acad Sci U S A
100
:
10376
–10381,
2003
27.
DE Nanclares GP, Bilbao JR, Castano L: No association of INS-VNTR genotype and IAA autoantibodies.
Ann N Y Acad Sci
1037
:
127
–130,
2004
28.
Walter M, Albert E, Conrad M, Keller E, Hummel M, Ferber K, Barratt BJ, Todd JA, Ziegler AG, Bonifacio E: IDDM2/insulin VNTR modifies risk conferred by IDDM1/HLA for development of type 1 diabetes and associated autoimmunity.
Diabetologia
46
:
712
–720,
2003
29.
Douglas Petersen F-L, van der Keur M, de Vries RRP, Roep BO: Autoreactive and immunoregulatory T-cell subsets in insulin-dependent diabetes mellitus.
Diabetologia
42
:
443
–449,
1999
30.
Durinovic-Belló I, Schlosser M, Riedl M, Maisel N, Rosinger S, Kalbacher H, Deeg M, Ziegler M, Elliott J, Roep B, Karges W, Boehm B: Pro- and anti-inflammatory cytokine production by autoimmune T cells against preproinsulin in HLA-DRB1*04, DQ8 type 1 diabetes.
Diabetologia
47
:
439
–450,
2004
31.
Durinovic-Belló I, Maisel N, Schlosser M, Kalbacher H, Deeg M, Eiermann T, Karges W, Boehm B-O: Relationship between T and B cell responses to proinsulin in human type 1 diabetes.
Ann N Y Acad Sci
1005
:
288
–294,
2003
32.
Durinovic-Belló I, Riedl M, Rosinger S, Maisel N, Kalbacher H, Deeg M, Schreckling H-J, Schlosser M, Ziegler M, Kuehnl P, Boehm B-O: Th2 dominance of T helper cell response to preproinsulin in individuals with preclinical type 1 diabetes.
Ann N Y Acad Sci
958
:
209
–213,
2002
33.
Arif S, Tree TI, Astill TP, Tremble JM, Bishop AJ, Dayan CM, Roep BO, Peakman M: Autoreactive T cell responses show proinflammatory polarization in diabetes but a regulatory phenotype in health.
J Clin Invest
113
:
451
–463,
2004
34.
Herzog B, Ott P, Dietrich M, Quast S, Karulin A, Kalbacher H, Karges W, Tary-Lehmann M, Lehmann P, Boehm B, Durinovic-Belló I: Increased in vivo frequency of IA-2 peptide-reactive, IFNγ+/IL-4- T cells in type 1 diabetic subjects.
J Autoimmun
23
:
45
–54,
2004
35.
Ott P, Dittrich M, Herzog B, Guerkov R, Gottlieb P, Putnam A, Boehm B, Durinovic-Belló I, Tary-Lehmann M, Lehmann P: T cells recognize multiple GAD65 and proinsulin epitopes in human type 1 diabetes, suggesting determinant spreading.
J Clin Immunol
24
:
327
–339,
2004
36.
Groux H, O’Garra A, Bigler M, Rouleau M, Antonenko S, de-Vries JE, Roncarolo MG: A CD4+ T-cell subset inhibits antigen-specific T-cell responses and prevents colitis.
Nature
389
:
737
–742,
1997
37.
Levings MK, Bacchetta R, Schulz U, Roncarolo MG: The role of IL-10 and TGF-beta in the differentiation and effector function of T regulatory cells.
Int Arch Allergy Immunol
129
:
263
–276,
2002
38.
Schlosser M, Strebelow M, Wassmuth R, Arnold M-L, Breunig I, Rjasanowski I, Ziegler B, Ziegler M: The Karlsburg type 1 diabetes risk study of a normal schoolchild population: association of beta-cell autoantibodies and human leukocyte antigen-DQB1 alleles in antibody-positive individuals.
J Clin Endocrinol Metab
87
:
2254
–2261,
2002
39.
Bingley PJ, Christie MR, Bonifacio E, Bonfanti R, Shattock M, Fonte MT, Bottazzo GF, Gale EA: Combined analysis of autoantibodies improves prediction of IDDM in islet cell antibody-positive relatives.
Diabetes
43
:
1304
–1310,
1994
40.
Verge CF, Gianani R, Kawasaki E, Yu L, Pietropaolo M, Jackson RA, Chase HP, Eisenbarth GS: Prediction of type I diabetes in first-degree relatives using a combination of insulin, GAD, and ICA512bdc/IA-2 autoantibodies.
Diabetes
45
:
926
–933,
1996
41.
Strebelow M, Schlosser M, Ziegler B, Rjasanowski I, Ziegler M: Karlsburg Type I diabetes risk study of a general population: frequencies and interactions of the four major type I diabetes-associated autoantibodies studied in 9419 schoolchildren.
Diabetologia
42
:
661
–670,
1999
42.
Erlich H, Bugawan T, Begovich AB, Scharf S, Griffith R, Saiki R, Higuchi R, Walsh PS: HLA-DR, DQ and DP typing using PCR amplification and immobilized probes.
Eur J Immunogenet
18
:
33
–55,
1991
43.
Durinovic-Belló I, Steinle A, Ziegler AG, Schendel DJ: HLA-DQ-restricted, islet-specific T-cell clones of a type I diabetic patient: T-cell receptor sequence similarities to insulitis-inducing T-cells of nonobese diabetic mice.
Diabetes
43
:
1318
–1325,
1994
44.
Durinovic-Belló I, Boehm B-O, Ziegler A-G: Predominantly recognized proinsulin T helper cell epitopes in individuals with and without islet cell autoimmunity.
J Autoimmun
18
:
55
–66,
2002
45.
Alleva DG, Crowe PD, Jin L, Kwok WW, Ling N, Gottschalk M, Conlon PJ, Gottlieb PA, Putnam AL, Gaur A: A disease-associated cellular immune response in type 1 diabetics to an immunodominant epitope of insulin.
J Clin Invest
107
:
173
–180,
2001
46.
Polychronakos C: Animal models of spontaneous autoimmune diabetes: notes on their relevance to the human disease.
Curr Diabetes Rep
4
:
151
–154,
2004
47.
Chiu HK, Palmer JP: Autoimmune diabetes: more than just one flavor?
J Endocrinol Invest
27
:
480
–484,
2004
48.
Itoh M, Takahashi T, Sakaguchi N, Kuniyasu Y, Shimizu J, Otsuka F, Sakaguchi S: Thymus and autoimmunity: production of CD25+CD4+ naturally anergic and suppressive T cells as a key function of the thymus in maintaining immunologic self-tolerance.
J Immunol
162
:
5317
–5326,
1999
49.
Jordan MS, Boesteanu A, Reed AJ, Petrone AL, Holenbeck AE, Lerman MA, Naji A, Caton AJ: Thymic selection of CD4+CD25+ regulatory T cells induced by an agonist self-peptide.
Nat Immunol
2
:
301
–306,
2001
50.
Baecher-Allan C, Viglietta V, Hafler DA: Human CD4+CD25+ regulatory T cells.
Semin Immunol
16
:
89
–98,
2004
51.
Bach JF: Regulatory T cells under scrutiny.
Nat Rev Immunol
3
:
189
–98,
2003