The heterozygous DQ2/8 (DQA1*05:01-DQB1*02:01/DQA1*03:01-DQB1*03:02) genotype confers the highest risk in
type 1 diabetes (T1D), whereas the DQ6/8 (DQA1*02:01-DQB1*06:02/DQA1*03:01-DQB1*03:02) genotype is protective. The mechanism of dominant protection by DQ6 (DQB1*06:02) is unknown. We tested the hypothesis that DQ6 interferes with
peptide binding to DQ8 by competition for islet
epitope ("
epitope stealing") by analysis of the islet ligandome presented by
HLA-DQ6/8 and -DQ8/8 on dendritic cells pulsed with islet
autoantigens preproinsulin (PPI), GAD65, and IA-2, followed by competition assays using a newly established "
epitope-stealing" HLA/
peptide-binding assay.
HLA-DQ ligandome analysis revealed a distinct DQ6
peptide-binding motif compared with the susceptible DQ2/8 molecules. PPI and IA-2
peptides were identified from DQ6, of DQ6/8 heterozygous dendritic cells, but no DQ8 islet
peptides were retrieved.
Insulin B6-23, a highly immunogenic CD4
T-cell epitope in patients with T1D, bound to both DQ6 and DQ8. Yet, binding of InsB6-23 to DQ8 was prevented by DQ6. We obtained first functional evidence of a mechanism of dominant protection from disease, in which HLA molecules associated with protection bind islet
epitopes in a different, competing, HLA-binding register, leading to "
epitope stealing" and conceivably diverting the immune response from islet
epitopes presented by disease-susceptible HLA molecules in the absence of protective HLA.