Evidence that HLA-DQ9 confers risk to celiac disease by presence of DQ9-restricted gluten-specific T cells Michael Bodd a, *, Stig Tollefsen a , Elin Bergseng b , Knut E.A. Lundin a , Ludvig M. Sollid a,b a Centre for Immune Regulation and Department of Immunology, Oslo University Hospital-Rikshospitalet, Oslo, Norway b Centre for Immune Regulation and Department of Immunology, University of Oslo, Oslo, Norway ARTICLE INFO Article history: Received 2 November 2011 Accepted 24 January 2012 Available online 31 January 2012 Keywords: HLA-DQ9 Celiac disease MHC HLA-DQ 57 DQ8-glut-1 epitope ABSTRACT We describe the gluten T-cell response of a DR7DQ2/DR9DQ9 heterozygous celiac disease patient (CD555). Interestingly, this patient had T cells recognizing gluten in the context of human leukocyte antigen (HLA) molecules of both haplotypes. For the DR9DQ9 haplotype, DQ9 was identified as the antigen-presenting molecule. As DQ9 carries aspartate at DQ 57 but is otherwise identical to DQ8 and not considered associated with celiac disease, we aimed to characterize this DQ9-restricted T-cell response in detail. By fractionation of pepsin-trypsin digested gliadin we identified an epitope stimulatory for several T-cell clones. This epitope was identical to an epitope (DQ8-glut-1) previously identified in DQ8 patients. In CD555, this was the dominant DQ9-restricted epitope, whereas no T-cell response was found toward two other DQ8-restricted epitopes. These findings correlated with peptide binding data demonstrating that this epitope bound better to DQ9 than the two other DQ8-restricted epitopes. Although glutamine to glutamate exchange at P9 improved binding of all three epitopes to DQ8, no such effect was observed for DQ9. The differential ability of DQ8 and DQ9 to harness a negatively charged anchor at P9 may result in fewer potential gluten epitopes in DQ9 patients. Our data further indicate that DQ9 is a susceptibility factor for celiac disease.  2012 American Society for Histocompatibility and Immunogenetics. Published by Elsevier Inc. All rights reserved. 1. Introduction The absence of aspartate at position 57 is considered a particularly important functional feature of the type 1 diabetes and celiac disease (CD) associated HLA-DQ8 molecule [1–3]. HLA-DQ9 (DQA1*03/DQB1*03:03) differs from HLA-DQ8 (DQA1*03/DQB1*03:02) only in position 57, where DQ8 has ala- nine and DQ9 has aspartate. This difference leads to the loss of a salt bridge between arginine 76 and aspartate 57 of DQ8 and hence to a preference of negatively charged peptides in P9 of the peptide binding groove of DQ8, but not of DQ9 [3–5]. In support of the importance of the 57 polymorphism, it has been stated that DQ9 is not associated with CD [5]. It has recently been proposed that the remarkable association of DQ8 to CD is related to the ability to recruit negatively charged TCR and that this is linked to the 57 polymorphism [2]. However the effect of this polymorphism on peptide–major histocompatibility complex (MHC) interaction has not been fully investigated. Gluten-reactive CD4 + T cells are found in the intestinal mucosa of CD patients but not of healthy controls [6]. Most intestinal glu- ten-reactive T cells respond to gluten peptides only after conver- sion by the enzyme transglutaminase 2 (TG2) of certain glutamine residues to glutamic acid [7]. In the present study, we found that DQ9-restricted gluten-reactive CD4 + T cells could be isolated from the small intestine of a CD patient expressing DQ9 and DQ2.2. We established T-cell clones (TCC) and identified the epitope recog- nized. We further investigated the binding of gluten peptides to DQ8 and DQ9. Our findings suggest that DQ9 can be implicated in CD, and the findings shed light on the importance of the HLA-DQ 57 polymorphism for disease initiation. 2. Subjects and methods 2.1. Subject Intestinal biopsy samples were obtained on two separate occa- sions from an adult female patient (CD555). She carried the sero- logic human leukocyte antigen (HLA) type DR7DR9/DQ2DQ3 and genomic DQ type DQA1*02:01/03:02 and DQB1*02/03:03. Based on known linkage disequilibria, she is very likely to carry the DR7DQ2 and DR9DQ9 haplotypes. She had a diagnosis of CD from 1999 which was in accordance with the American Gastroenterol- ogy Association guidelines [8] with a positive endomysium anti- body test and histology showing Marsh 3C. Upon accidental expo- sure to gluten, the patient experienced abdominal pain and diarrhea. As a child she experienced abdominal pain and failure to thrive. Biopsy samples were taken both in 2004 as part of routine follow-up and in 2007 as part of a study examining the gluten response in patients with DQ2.2 (Bodd et al., Gastroenterology, in press). She was on a gluten-free diet and was well-treated on both * Corresponding author. E-mail address: michael.bodd@rr-research.no. (M. Bodd). Human Immunology 73 (2012) 376-381 Contents lists available at SciVerse ScienceDirect 0198-8859/$36.00  2012 American Society for Histocompatibility and Immunogenetics. Published by Elsevier Inc. All rights reserved. doi:10.1016/j.humimm.2012.01.016