HLA-B7-restricted islet epitopes are differentially recognized in type 1 diabetic children and adults and form weak peptide-HLA complexes

Abstract

The cartography of β-cell epitopes targeted by CD8(+) T cells in type 1 diabetic (T1D) patients remains largely confined to the common HLA-A2 restriction. We aimed to identify β-cell epitopes restricted by the HLA-B7 (B*07:02) molecule, which is associated with mild T1D protection. Using DNA immunization on HLA-B7-transgenic mice and prediction algorithms, we identified GAD and preproinsulin candidate epitopes. Interferon-γ (IFN-γ) enzyme-linked immunospot assays on peripheral blood mononuclear cells showed that most candidates were recognized by new-onset T1D patients, but not by type 2 diabetic and healthy subjects. Some epitopes were highly immunodominant and specific to either T1D children (GAD(530-538); 44% T cell-positive patients) or adults (GAD(311-320); 38%). All epitopes displayed weak binding affinity and stability for HLA-B7 compared with HLA-A2-restricted ones, a general feature of HLA-B7. Single-cell PCR analysis on β-cell-specific (HLA-B7 tetramer-positive) T cells revealed uniform IFN-γ and transforming growth factor-β (TGF-β) mRNA expression, different from HLA-A2-restricted T cells. We conclude that HLA-B7-restricted islet epitopes display weak HLA-binding profiles, are different in T1D children and adults, and are recognized by IFN-γ(+)TGF-β(+)CD8(+) T cells. These features may explain the T1D-protective effect of HLA-B7. The novel epitopes identified should find valuable applications for immune staging of HLA-B7(+) individuals.

FIG. 1.

HLA-B7–restricted GAD candidate epitopes were identified. HLA-B7–transgenic mice (n = 9) were immunized with plasmid DNA encoding full-length human GAD. Their splenocytes were subsequently recalled with the 37 GAD peptides listed in Table 1 in IFN-γ ELISpot plates. Each symbol represents an individual mouse, and the number of IFN-γ SFCs/10⁶ splenocytes is given for each peptide. The dotted line represents the positive cutoff (15 SFCs/10⁶ splenocytes). Peptides recognized in at least one mouse were selected and are indicated by arrows. The DMSO-negative control and concanavalin A (ConA)–positive control stimulations are also shown.

FIG. 2.

Validation of HLA-B7–restricted candidate GAD and PPI epitopes in HLA-B7⁺, new-onset T1D children and adults. Whole PBMCs were tested using an IFN-γ ELISpot format, as detailed in RESEARCH DESIGN AND METHODS. A: Number of T1D children (square symbols) and adults (round symbols) testing positive to each peptide. Reactivities are ranked as negative (<3 SDs above basal; white symbols) or positive (≥3 SDs; black symbols). When positive, responses are further ranked as low (3–4 SDs), medium (4–5 SDs), and high (≥5 SDs). The percentage of patients positive to each epitope are indicated. Responses to a viral epitope mix and to PHA are included as positive controls. B: Frequencies (IFN-γ SFCs/10⁶ PBMCs) of T cells reactive to each individual epitope. Each symbol represents an individual patient.

FIG. 3.

Recognition of HLA-B7–restricted islet epitopes in different study subjects. A: Percentage of new-onset T1D children (white bars), new-onset T1D adults (back bars), T2D subjects (gray bars), and healthy subjects (hatched bars) responding to each individual epitope. B: Relative distribution of epitope specificities among total β-cell epitope-reactive CD8⁺ T cells. The percent prevalence of each epitope out of all epitopes recognized among new-onset T1D children (n = 10) (left) and adults (n = 9) (right) is shown.

FIG. 4.

Longitudinal follow-up of IFN-γ ELISpot responses in T1D patients. Patients previously tested by ELISpot close to diagnosis (t = 0) (Supplementary Fig. 1) were reassayed under identical conditions after 7 days to 14 months of follow-up, as indicated. Reactivities testing positive at either time point are depicted and ranked as absent (<3 SDs above basal), low (3–4 SDs), medium (4–5 SDs), and high (≥5 SDs) as in Fig. 2A. Responses against a pool of viral epitopes are included as positive controls.

FIG. 5.

HLA affinity and stability measurements. A: HLA-B7 and -A2 Kd binding affinity for T-cell epitopes derived from islet Ags or from infectious and tumor Ags identified here and in previous reports, as detailed in RESEARCH DESIGN AND METHODS and Supplementary Table 1. B: HLA-B7 and -A2 binding stability (half-life) measured for the same epitopes. Each symbol represents an individual epitope, and bars show median and interquartile range values for each distribution. Each symbol depicts the mean value of at least two separate measurements.

FIG. 6.

mRNA expression profiles of epitope-specific HLA-B7– and HLA-A2–restricted TMr⁺CD8⁺ T cells. A: Comparison of IFN-γ ELISpot and HLA-B7 TMr readouts for different epitopes in patients C09 and A07, as indicated at the top of each column. Top row displays representative ELISpot wells out of triplicate measurements. Bottom row shows the corresponding fraction of TMr⁺CD8⁺ cells detected in the same individual. B: TMr⁺CD8⁺ cells were single-cell sorted, analyzed by reverse-transcriptase PCR, and compared with similar measurements previously performed on HLA-A2⁺, new-onset T1D patients (16). Percent numbers of HLA-B7–restricted PPI_8–16- (black round symbols), GAD_530–538- (black diamond symbols), GAD_311–320- (back triangle symbols), and CMV_417–426–specific (gray round symbols) T cells (n = 20/each) expressing each mRNA are depicted. White and gray squares show frequencies of gene-expressing single HLA-A2–restricted PPI_6–14- and CMV_495–503–specific T cells sorted from 7 HLA-A2⁺, new-onset T1D patients. neg, negative; +++, positive response ≥5 SDs above background.