Peptide and Major Histocompatibility Complex–Specific Breaking of Humoral Tolerance to Native Insulin With the B9-23 Peptide in Diabetes-Prone and Normal Mice

NOD mice (3–4 weeks of age, all female) were purchased from Taconic Farms (Germantown, NY). BALB/c, C57BL/6, NOR, B10.D2-H2^d, and C.B10-H2^b mice (3–4 weeks of age, all females) were purchased from Jackson Laboratory (Bar Harbor, ME). Table 1 shows MHC haplotype and immunoglobulin alleles of the mice we studied.

Insulin A-chain A1-15 and A7-21; B-chain B1-15, B9-23, B16-30, B9-16, and B13-23; proinsulin P24-33 and P26-34; and tetanus toxin (TT) peptides 830-843 (QYIKANSKFIGITE) were synthesized and purified by reverse-phase high-performance liquid chromatography and identified by mass spectroscopy (Research Genetics, Huntsville, AL). The amino acid sequences of the insulin or proinsulin peptide used are shown in Table 2.

Mice were injected subcutaneously at 4 weeks and again at 8 or 10 weeks of age with 100 μg of peptides emulsified in incomplete Freund’s adjuvant (IFA) (Gibco BRL, Grand Island, NY). For administration of peptide without adjuvant, female BALB/c mice (4 weeks of age) were injected on 5 sequential days and weekly thereafter until 11 weeks of age with B9-23 peptide in saline without adjuvant. The blood glucose levels of the mice after 12 weeks of age were monitored every other week by using a One Touch II blood glucose sensor (Lifescan, Milpitas, CA), and mice were considered to be diabetic after two consecutive blood glucose values >250 mg/dl.

IAA expression of serum was evaluated prospectively beginning at 4 weeks of age until the development of diabetes or until 46 weeks. IAA in nondiabetic-strain mice, including BALB/c, C57BL/6, NOR, B10.D2-H2^d, C.B10-H2^b, and F1 (NOD × C57BL/6) mice, were also evaluated from 4 weeks of age (before first injection of peptide) every week, every other week, or every 4 weeks. IAA was measured with the 96-well filtration-plate micro IAA assay as previously described (8). Briefly, ¹²⁵I-insulin (Amersham) of 20,000 cpm was incubated with 5 μl of serum with and without cold human insulin, respectively, for 3 days at 4°C in buffer A (20 mmol/l Tris-HCl buffer, pH 7.4, containing 150 mmol/l NaCl, 1% bovine serum albumin [BSA], 0.15% Tween-20, and 0.1% sodium azide). A total of 50 μl of 50% protein A/8% protein G-Sepharose (Pharmacia) was added to the incubation in a MultiScreen-NOB 96-well filtration plate (Millipore) that was precoated with buffer A. The plate was shaken for 45 min at 4°C followed by two cycles of four washes, each cycle with cold buffer B (the same buffer as A, except with 0.1% BSA). After washing, 40 μl of scintillation liquid (Microscint-20; Packard, Meriden, CT) was added to each well, and radioactivity was determined directly in the 96-well plate with a TopCount (96-well plate β-counter; Packard) scintillation counter. The result was calculated based on the difference in counts per minute (Δcpm) between the well without cold insulin and the well with cold insulin and expressed as an index:index = (sample Δcpm – human negative control Δcpm)/(human positive control Δcpm – human negative control Δcpm). The limit of normal (0.010) was chosen by the analysis of IAA in nondiabetic-strain mice, including 23 BALB/c mice and C57BL/6 mice (8).

A total of 5 μl of sera was preincubated with 200 ng human insulin (Novo Nordisk, Clayton, NC) and 10,000 ng of B9-23 peptide overnight at 4°C in IAA assay buffer A. ¹²⁵I-insulin of 20,000 cpm was added, and the same IAA assay was performed as previously described. For study of the binding of insulin autoantibodies to rat insulin (Alpco, Windham, NH), 0–50 ng of rat insulin was preincubated with 5 μl of serum of BALB/c mice immunized with B9-23 peptide in IFA. Competition curves are expressed as B/Bo (B – NSB)/(Bo – NSB) (%), where B is the counts binding to ¹²⁵I-insulin with the addition of cold insulin, Bo is the counts without cold insulin, and NSB is the nonspecific binding.

We used five B9-23–reactive CD4 T-cell clones that were established as previously described (4,5). All of these clones are reactive to insulin peptide B9-23 and restricted by NOD MHC class II I-A^g7 (6). All clones utilize the dominant T-cell receptor (TCR) α-chain (AV13S3 and AJ53) (12), and one (BDC12.2-40) of five reacted with peptide B9-16, and four reacted with B13-23 (13). Assays of T-cell clones were performed by co-culture of 2.5 × 10⁴ T-cells with 1.0 × 10⁶ irradiated (3,500 rad) NOD, BALB/c, or C57BL/6 spleen cells in the presence of antigen at a concentration of 100 μg/ml. The cells were incubated for 72 h in flat-bottom 96-well plates, pulsed with 0.5 μCi [³H]thymidine per well for 6 h, and harvested into 96-well filter plates (Packard). After washing and drying, 40 μl of scintillation liquid was added to each well, and radioactivity was determined with a TopCount 96-well plate β-counter (Packard). Proliferation is represented as a stimulation index (defined as counts per minute in cells stimulated with antigen/counts per minute in cells without antigen).

Statistical analysis included χ², Fisher’s exact, and Mann-Whitney (Prism, GraphPad Software, San Diego, CA) tests.

A single subcutaneous administration of insulin B9-23 peptide in IFA decreases the incidence of diabetes relative to mice given a control TT-peptide (830-843) (14). Administration of B9-23 peptide in IFA at 4 weeks of age significantly enhanced IAA expression at 8 weeks of age (10 of 10 positive) compared with the unmanipulated NOD mice (6 of 15 positive, P < 0.005) and the NOD mice given a TT-peptide (3 of 9, P < 0.005). The levels of IAAs at 8 weeks (0.80 ± 0.22) were increased by B9-23 peptide administration compared with the levels of IAAs in unmanipulated NOD mice (index 0.10 ± 0.06, P < 0.005) relative to NOD mice given the TT-peptide (0.02 ± 0.01, P < 0.005). With the B9-23 peptide immunization at 4 and 10 weeks of age, IAA expression was prolonged. All B9-23–treated mice were positive at 30 weeks of age, and six of eight were positive at 46 weeks (the end of follow-up). The mean levels of IAAs in nondiabetic NOD mice given B9-23 peptide in IFA did not significantly change from 12 (0.93 ± 0.35) to 30 (0.67 ± 0.33) weeks of age. In contrast, most of the IAA expression disappeared by 30 weeks of age in unmanipulated NOD mice and NOD mice given TT-peptide in IFA (Fig. 1).

Induction of autoantibodies by administration of B9-23 peptide in IFA to NOD mice may relate to previous sensitization of the autoimmune-prone NOD mice. To test whether such autoantibodies were only induced in NOD mice, we evaluated BALB/c mice and C57BL/6 mice. Of note, induction of IAA with the B9-23 peptide was observed in non–diabetes-prone BALB/c mice and NOR mice but not in C57BL/6 mice (Fig. 4). To study the peptide specificity for the induction of IAAs, we evaluated a series of A-chain peptides, including A1-15 and A7-21, and B-chain peptides, including B1-15 and B17-30 (Table 2). In contrast to the B9-23 peptide, these other peptides did not induce IAAs in BALB/c mice (Fig. 2). None of the peptides induced IAAs in C57BL/6 mice (Fig. 2). In addition, at 4 weeks of age, we administered three times our usual amount of B9-23 peptide to five C57BL/6 mice (300 μg in IFA); and again, none developed IAAs measured at 6 and 8 weeks. In contrast, six of six BALB/c mice were IAA positive at 8 weeks with 100 μg B9-23 (P < 0.01) whereas one-tenth our usual amount of B9-23 (10 μg in IFA) induced insulin autoantibodies by 8 weeks in one of five BALB/c mice.

To test how early the peptide immunization can induce IAAs, we followed the time course of IAA expression after immunization with the B9-23 peptide in IFA. Two of six mice expressed IAAs at 2 weeks, and all mice expressed IAAs at 3 weeks after the first immunization (Fig. 2).

Previous studies indicate that spontaneous NOD anti-insulin NOD T-cell clones that recognize insulin peptide B9-23 recognize either a B9-16 peptide or a B13-23 peptide but not both, and a subset of B13-23–reactive NOD T-cells can be stimulated with B13-23 peptide presented by BALB/c antigen-presenting cells.

Given the ability of the B9-23 peptide to induce autoantibodies, we evaluated the B9-16 and B13-23 peptides, identified as two distinct epitopes for NOD B9-23–reactive T-cell clones utilizing the same TCR α-chain: AV13S3 (13). We also studied proinsulin peptides described by Rudy et al. (15). The induction of autoantibodies was observed in BALB/c mice with B13-23 peptide but not with the other peptides. IAA were induced in fewer mice with the B13-23 peptide compared with the mice with B9-23 peptide (B9-23 6 of 6 vs. B13-23 4 of 12, P < 0.01) (Fig. 3).

As previously described, we found that induction of IAAs was observed in non–diabetes-prone BALB/c (H-2^d) and NOR (H-2^g7) mice after the administration of B9-23 peptide at 4 weeks. C57BL/6 mice (H-2^b) were resistant to induction of IAA with any insulin peptide. To determine the genetic region associated with induction of IAAs by peptide immunization, we examined the effect of administration of B9-23 peptide on congenic inbred mice, including B10.D2-H2^d and C.B10-H2^b mice (Table 1). IAAs were induced in B10.D2-H2^d mice having C57BL/10 background genes with H-2^d but not in C.B10-H2^b mice having BALB/c background genes with H-2^b. To eliminate the possibility of an inhibitory effect of the H-2^b MHC, we also evaluated F1 (NOD × C57BL/6) mice. Induction of IAAs observed in the F1 (NOD × C57BL/6) mice was similar to that observed in the NOD mice (Fig. 4).

To determine whether the induced IAAs recognized the B9-23 peptide or native insulin, we evaluated IAA absorption by human insulin and B9-23 peptide. IAAs expressed by NOD mice after B9-23 peptide immunization and the induced IAAs in BALB/c, NOR, and B10.D2(H-2^d) mice were completely absorbed by human insulin. In contrast, competition was not observed with the B9-23 peptide alone, even when added in concentrations 500 times greater than that which gave maximal inhibition with human insulin (Fig. 5).

The amino-acid sequence of mouse insulin is identical to rat insulin. To confirm that the insulin antibodies induced by the B9-23 peptide in BALB/c mice are autoantibodies, we evaluated competition with rat insulin. Rat insulin absorbed the IAAs with 50% inhibition of binding at 2 × 10⁻⁷ M and complete inhibition at the highest concentration tested (Fig. 6).

I-A^g7 is the only class II molecule expressed by NOD mice caused by a deletion in the I-Eα promoter (16). NOD and BALB/c mice have an identical α-chains (I-Aα^d) but differ in their β-chains (NOD I-Aβ^g7 and BALB/c I-Aβ^d). To evaluate potential cross-restriction of NOD B9-23 reactive clones, we assessed BALB/c (I-A^d and I-E^d) and C57BL/6 (I-A^b and I-E⁰) spleen cells as antigen-presenting cells (APCs) to NOD mouse T-cell clones (e.g., BDC12-2.40). The BDC12-2.40 clone, which recognized the B9-16 epitope, was entirely specific for I-A^g7 and was stimulated only by B9-23 peptide presented by NOD spleen cells. The other four clones that recognized the B13-23 epitope showed cross-restriction with stimulation in the presence of BALB/c or NOD spleen cells. These B13-23–reactive clones were not stimulated with C57BL/6 spleen cells (Fig. 7).

The relative ease of autoantibody induction and specificity of induction suggests that even the immune system of a normal BALB/c mouse may be primed to respond to insulin peptide B9-23. To further evaluate the ease of IAA induction, the B9-23 peptide was administered without adjuvant in saline to BALB/c mice. As shown in Fig. 8, IAAs were readily induced (Fig. 8).

Insulin is an important target of the autoimmunity of type 1 diabetes in humans and NOD mice (17,18,19,20,21). IAAs are detected before the clinical onset of diabetes (18,22). Early expression of anti-insulin autoantibodies is associated with early development of diabetes (8). Despite the strong association of autoantibody expression with disease development, the mechanisms underlying production of anti-insulin autoantibodies are poorly defined. The B-chain peptide of insulin has been found to contain a dominant epitope recognized by the majority of T-cells of islet-infiltrating cells from young NOD mice (6,7). Insulin, given subcutaneously, intranasally, or orally, can prevent diabetes in NOD mice (23,24). Administration of insulin B-chain or B-chain peptide B9-23 as well as insulin can prevent diabetes in NOD mice (14,25), and soluble insulin B-chain administration boosts the anti-insulin autoantibody response in the NOD mouse (26). IAAs in humans with type 1 diabetes fail to bind isolated insulin B-chain (18). In this study, the induction of IAAs was evaluated in vivo after administration of insulin peptides.

Insulin B9-23 peptide administered in IFA induced a strong anti-insulin autoantibody response. In addition, the administration of the B9-23 peptide in IFA at 4 and 10 weeks led to IAAs persisting at least 46 weeks. This may correlate with the observation that a single injection of the peptide in IFA at 4 weeks of age can inhibit the development of diabetes for months. As described in previous reports (26), IAAs in both unmanipulated NOD mice and NOD mice treated with B9-23 peptides in IFA were of the IgG1 subclass (data not shown), whereas for BALB/c mice after immunization with B9-23 peptide, antibodies were predominantly IgG1, but a lower level of IgG2a was also induced.

We found that IAA expression was induced by B9-23 peptide administration (even in non-diabetes-prone BALB/c mice and in NOD-derived NOR mice, the latter of which have an MHC identical to that in NOD mice, with 12% of other genes of C57BL/6 mouse origin) (27). The induction of IAAs was observed specifically with B9-23 peptide instead of A-chain, other B-chain, or proinsulin peptides. This specificity suggested that induced IAAs with B9-23 peptide were not a reflection of cross-reactivity of antibody to the peptide as “a part of insulin.” The antibodies induced by the B9-23 peptide were not absorbed with the peptide alone, but recognized rat insulin, which is identical in sequence to mouse insulin. Anti-IAAs induced by the B9-23 peptide in mice could not recognize the peptide alone, suggesting that spontaneously occurring anti-insulin B-cells are activated with the help of activated T-cells specific to B9-23 peptide. We also found that the IAAs induced by B9-23 peptide administration appeared as early as 2 weeks in BALB/c mice. Such an early appearance of autoantibodies suggested that even normal mice contain sufficient numbers of anti-insulin B-cells to rapidly produce IAAs.

The induction of IAAs mapped to the MHC rather than to the background genes (determined through analysis of congenic inbred mice that have C57/B10 [B10.D2-H2^d] and BALB/c background [C.B10-H2^b]). Although I-E expression in NOD mice provides strong resistance to the development of diabetes (28), the I-E molecule in the BALB/c mice did not inhibit the induction of IAAs. The I-A^b molecule in F1 (NOD × C57BL/6) mice also did not have an inhibitory effect on the induction of IAAs.

To test B9-23 peptide presentation by BALB/c and NOD APCs, we analyzed the reactivity of several NOD-derived CD4 T-cell clones with BALB/c and C57BL/6 APCs. Four of five T-cell clones were activated by B9-23 peptide with either BALB/c APCs or NOD APCs but not with C57BL/6 APCs. Kanagawa et al. (29) reported that several I-A^g7–restricted clones (including B9-23–reactive clones) show cross-restriction by NOD or BALB/c spleen cells. The study of APC transfectants expressing I-A^g7 and/or a mutated I-A^g7 molecule, which has similar sequence to I-A^d, shows the same cross-restriction observed in the study of splenocytes, suggesting greater importance of I-A rather than I-E for this phenomenon (29). Of note, the α-chains of I-A^d and I-A^g7 are identical (Table 1). The cross-restriction between NOD and BALB/c APCs was observed in the response to B13-23–reactive clones but was not found with the clones reacting with the B9-16 peptide. IAAs were induced in BALB/c mice with B13-23 peptide but not with the B9-16 peptide. With current BALB/c studies, we suspect that B9-23 and B13-23 peptides are presented by I-A^d, but presentation by I-E^d is not excluded.

The finding of MHC restriction for the induction of IAAs by B9-23 peptide suggests that peptide in IFA-activated T-cells can break B-cell tolerance to insulin. Anti-insulin autoantibody–producing B-cells probably take up and process native insulin and present insulin peptide B9-23 on their MHC class II molecules. Several studies provide evidence for the importance of B-lymphocytes in the pathogenesis of type 1 diabetes in NOD mice. NOD mice genetically deficient in B-cells are resistant to diabetes (10,11). Noorchashm et al. (20) developed NOD mice with an MHC class II (I-A^g7) deficiency confined to the B-cell compartment. NOD mice with I-A^g7–deficient B-cells were resistant to the development of diabetes, suggesting that B-cells may be critical antigen-presenting cells (30).

In general, insulin peptide administration protects NOD mice, although it induces anti-insulin autoantibodies. We suspect that T-cells activated by B9-23 peptide drive both B-cell production of IAAs and a protective immune response. This is likely to be a Th2-dominated response (31). Determination of IAAs after peptide immunization directed at T-cells should provide in vivo insight into the dynamics of induction of anti-insulin autoimmunity. Aside from insulin molecules, peptides (whether endogenous or exogenous) that may spontaneously induce such anti-insulin autoimmunity are currently unknown. It is possible that in the spontaneous disease of humans and mice, an insulin peptide or a mimitope peptide is presented to the immune system and rapidly induces IAAs. Such a peptide, as evidenced by the current study, can be presented peripherally and induce a long-lasting humoral response to intact insulin. Most remarkably, the response is peptide-specific (B9-23 versus all other insulin peptides studied) and occurs after peptide administration without adjuvant. This strongly suggests that mice are not only not tolerant to this self-peptide, but also readily sensitized to it. The lack of tolerance to both B9-23 and insulin, as evidenced by the induction of IAAs, probably accounts for the dominant B9-23 response in the spontaneous disease.

This study was supported by grants from the National Institutes of Health (DK-55969 and P30-DK-57516), the Juvenile Diabetes Foundation, and the Children’s Diabetes Foundation.

Norio Abiru has a mentor-based fellowship from the American Diabetes Association.

We are grateful to Alpco (Windham, NH) for providing rat insulin. We thank David T. Robles for carefully reviewing the manuscript and Duane Walborn for technical assistance.