To identify the binding motifs of peptides which bind to the celiac disease and insulin-dependent-diabetes-mellitus (IDDM)-associated DQ2 molecule, peptides were eluted from affinity-purified DQ2 molecules. The eluted peptides were separated by reverse-phase HPLC. Prominent peptide peaks and the remaining pool of peptides were sequenced by Edman degradation. Truncated variants of eight different peptides with a length of 9-19 amino acids were identified; among them class II-associated invariant chain peptides (CLIP) and peptides that stem from HLA class I alpha, HLA-DQ alpha 1*0501, Ig and CD20 molecules. Data from the pool sequencing and the biochemical binding analyses of synthetic variants of an eluted high-affinity ligand (HLA class I alpha 46-60), indicate that the side chains of amino acid residues at relative position P1 (bulky hydrophobic), P4 (negatively charged or aliphatic), P6 (Pro or negatively charged), P7 (negatively charged) and P9 (bulky hydrophobic) are important for binding of peptides to DQ2. Computer modeling of the DQ2 with variants of the high-affinity ligand in the groove suggests that peptides bind to DQ2 through the primary anchors P1, P7 and P9 and making additional advantageous interactions using the P4 and P6 positions.