Abstract
In patients with celiac disease, inflammatory T cell responses to HLA-DQ2-bound gluten peptides are thought to cause disease. Two types of HLA-DQ2 molecules exist, termed HLA-DQ2.5 and HLA-DQ2.2. Whereas HLA-DQ2.5 predisposes to celiac disease, HLA-DQ2.2 does not. We now provide evidence that the disease-associated HLA-DQ2.5 molecule presents a large repertoire of gluten peptides, whereas the non-disease-associated HLA-DQ2.2 molecule can present only a subset of these. Moreover, gluten presentation by HLA-DQ2 homozygous antigen-presenting cells was superior to presentation by HLA-DQ2/non-DQ2 heterozygous antigen-presenting cells in terms of T cell proliferation and cytokine secretion. Gluten presentation by HLA-DQ2.5/2.2 heterozygous antigen-presenting cells induced intermediate T cell stimulation. These results correlated with peptide binding to the antigen-presenting cells. Finally, we demonstrate that HLA-DQ trans dimers formed in HLA-DQ2.5/2.2 heterozygous individuals have properties identical with HLA-DQ2.5 dimers. Our findings explain the strongly increased risk of disease development for HLA-DQ2.5 homozygous and HLA-DQ2.2/2.5 heterozygous individuals, and they are indicative of a quantitative model for disease development, where HLA-DQ expression and the available number of T cell-stimulatory gluten peptides are critical limiting factors. This model may have important implications for disease prevention.