Bente H. Johansen

Both α and β chain polymorphisms determine the specificity of the disease-associated HLA-DQ2 molecules, with β chain residues being most influential

Abstract We compared the peptide binding specificity of three HLA-DQ molecules; HLA-DQ(α1*0501, β1*0201), HLA-DQ(α1*0201, β1*0202), and HLA-DQ(α1*0501, β1*0301). The first of these molecules confers susceptibility to celiac disease and insulin-dependent diabetes mellitus, while the two latter molecules, which share either the α chain or the nearly identical β chain with HLA-DQ(α1*0501, β1*0201), do not predispose to these disorders. The binding of peptides was detected in biochemical binding assays as inhibition of binding of radiolabeled indicator peptides to affinity-purified HLA-DQ molecules. Binding experiments with several peptides demonstrated a clear difference in peptide binding specificity between the three HLA-DQ molecules. Further, single amino acid substitution analyses indicated that the HLA-DQ molecules have different peptide binding motifs. The experimental data were corroborated by computer modelling analysis. Our data suggest that the three HLA-DQ molecules prefer large hydrophobic residues in P1 of peptides with subtle differences in side-chain preferences. HLA-DQ(α1*0501, β1*0201) and HLA-DQ(α1*0201, β1*0202) both prefer large hydrophobic residues in P9, whereas HLA-DQ(α1*0501, β1*0301) prefers much smaller residues in this position. HLA-DQ(α1*0501, β1*0201) and HLA-DQ(α1*0201, β1*0202), in contrast to HLA-DQ(α1*0501, β1*0301), prefer negatively charged residues in P4 and P7. A less prominent P6 pocket also appears to differ between the three HLA-DQ molecules. Our results indicate that polymorphic residues of both the α and the β chain determine the peptide binding specificity of HLA-DQ(α1*0501, β1*0201), but that the β chain polymorphisms appears to play the most important role. The information on peptide residues which are advantageous and deleterious for binding to these HLA-DQ molecules may make possible the prediction of characteristic features of peptide that bind to HLA-DQ(α1*0501, β1*0201) and precipitate celiac disease.

A peptide-binding assay for the disease-associated HLA-DQ8 molecule.

The study of peptide binding to HLA class II molecules has mostly concentrated on DR molecules. Since many autoimmune diseases show a primary association to particular DQ molecules rather than DR molecules, it is also important to study the peptide‐binding properties of DQ molecules. Here we report a biochemical peptide‐binding assay for the type I diabetes‐associated DQ8, i.e. DQ (α1*0301, β1*0302), molecule. Affinity‐purified DQ8 molecules were tested in peptide‐binding assays using a radiolabelled influenza haemagglutinin (Ha) peptide encompassing positions 255–271(Y) as an indicator peptide. The Ha 255–271(Y) peptide bound to DQ8 in a pH‐dependent fashion showing optimal binding around pH 5. The association kinetics were relatively slow and the resulting complexes were heat labile. The specificity of peptide binding to DQ8 was investigated in competitive inhibition experiments with a panel of 43 peptides of different lengths and sequences. The DQ8 molecules showed a different pattern of peptide binding compared to a previously studied DQ2 molecule. Peptides derived from thyroid peroxidase, HLA‐DQ(α1*0301), HLA‐DQ(α1*0302), retinol receptor and p21ras were among the high‐affinity binders, whereas peptides derived from myelin basic protein were among the low‐affinity binders. The sequence of the high‐affinity peptides conformed with a previously published peptide‐binding motif of DQ8.

Binding of Ras Oncogene Peptides to Purified HLA-DQ(αl*0102, ß1*0602) and -DR(α, ß1*0101) Molecules

Mutated oncogene peptides may be presented to T cells by HLA molecules. To be able to design the optimal peptides for stimulation of T cells in individuals with different HLA molecules, it is important to analyse the binding characteristics of oncogene peptides to HLA. HLA‐DQ6 (DQ(α1*0102, ß1*0602)) and HLA‐DRI (DR(α, ßl*0101)) molecules were purified from lysates of homozygous EBV‐transformed cell lines. Purified HLA molecules were then tested for their ability to bind synthetic peptides in gel filtration assays. A p21 ras oncogene peptide (previously found to stimulate DQ6‐restricted T‐cell clones) and an influenza matrix peptide were labelled with 125I and served as indicator peptides for binding to DQ6 and DR1 respectively. Binding of homologous truncated and mutated p21 ras peptides and unrelated peptides was then evaluated by their capacity to inhibit binding of the indicator peptides. p21 ras‐derived peptides were found to bind to both DQ6 and DR1 molecules indicating the existence of a promiscuous binding motif in these peptides. The binding affinities seemed to vary between the different peptides, but the amino acid substitutions resulting from natural mutations were not critical for binding. Notably, the results obtained for DQ6 in the biochemical peptide binding assay correlated well with results obtained in a functional assay using T‐cell clones as probes.

The Molecular Basis of the HLA Association in Celiac Disease

Celiac disease (CD) is a malabsorptive disorder of the small intestine characterized by villous atrophy, hyperplastic crypts and T cell infiltration in the epithelium and in the lamina propria [1]. The disease is caused by an abnormal immune response to gluten, probably initiated by the activation of T cells in the intestinal mucosa to gluten-derived peptides. CD is strongly associated to specific genes in the HLA complex and is a unique model for studies of type 1 diabetes and other HLA associated diseases since (a) the primary HLA associations have been established, (b) the disease-inducing agent (gluten) is known and (c) gluten-specific T cells from the target organ, the intestinal mucosa, are accessible from biopsies for in vitro studies.