Shapes TCR Usage in Celiac Disease Posttranslational Modification of Gluten

Posttranslational Modification of Gluten Shapes TCR Usage in Celiac Disease Shuo-Wang Qiao, Melinda Ráki, Kristin S. Gunnarsen, Geir-Åge Løset, Knut E. A. Lundin, Inger Sandlie and This information is current as Ludvig M. Sollid of September 29, 2021. J Immunol 2011; 187:3064-3071; Prepublished online 17 August 2011; doi: 10.4049/jimmunol.1101526

http://www.jimmunol.org/content/187/6/3064 Downloaded from

Supplementary http://www.jimmunol.org/content/suppl/2011/08/18/jimmunol.110152 Material 6.DC1 http://www.jimmunol.org/ References This article cites 31 articles, 12 of which you can access for free at: http://www.jimmunol.org/content/187/6/3064.full#ref-list-1

Why The JI? Submit online.

• Rapid Reviews! 30 days* from submission to initial decision

by guest on September 29, 2021 • No Triage! Every submission reviewed by practicing scientists

• Fast Publication! 4 weeks from acceptance to publication

*average

Subscription Information about subscribing to The Journal of Immunology is online at: http://jimmunol.org/subscription Permissions Submit copyright permission requests at: http://www.aai.org/About/Publications/JI/copyright.html Email Alerts Receive free email-alerts when new articles cite this article. Sign up at: http://jimmunol.org/alerts

The Journal of Immunology is published twice each month by The American Association of Immunologists, Inc., 1451 Rockville Pike, Suite 650, Rockville, MD 20852 Copyright © 2011 by The American Association of Immunologists, Inc. All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. The Journal of Immunology

Posttranslational Modiﬁcation of Gluten Shapes TCR Usage in Celiac Disease

Shuo-Wang Qiao,* Melinda Ra´ki,†,1 Kristin S. Gunnarsen,‡,1 Geir-A˚ ge Løset,‡ Knut E.A. Lundin,†,x Inger Sandlie,‡ and Ludvig M. Sollid*,†

Posttranslational modification of Ag is implicated in several autoimmune diseases. In celiac disease, a cereal gluten-induced en- teropathy with several autoimmune features, T cell recognition of the gluten Ag is heavily dependent on the posttranslational conversion of Gln to Glu residues. Evidence suggests that the enhanced recognition of deamidated gluten peptides results from improved peptide binding to the MHC and TCR interaction with the peptide–MHC complex. In this study, we report that there is a biased usage of TCR Vb6.7 chain among TCRs reactive to the immunodominant DQ2-a-II gliadin epitope. We isolated Vb6.7 and DQ2-aII tetramer-positive CD4+ T cells from peripheral blood of gluten-challenged celiac patients and sequenced the TCRs of a large number of single T cells. TCR sequence analysis revealed in vivo clonal expansion, convergent recombination, semi- Downloaded from public response, and the notable conservation of a non-germline-encoded Arg residue in the CDR3b loop. Functional testing of a prototype DQ2-a-II–reactive TCR by analysis of TCR transfectants and soluble single-chain TCRs indicate that the deamidated residue in the DQ2-a-II peptide poses constraints on the TCR structure in which the conserved Arg residue is a critical element. The findings have implications for understanding T cell responses to posttranslationally modified Ags. The Journal of Immu- nology, 2011, 187: 3064–3071. http://www.jimmunol.org/

osttranslational modification of Ag is a key element of the T cells derived from celiac lesions that recognize the immu- pathogenesis of celiac disease, and posttranslational mod- nodominant DQ2-restricted gliadin DQ2-a-II epitope respond to P ifications of Ags have been similarly implicated in several deamidated gliadin peptides at concentrations that are at least 100- other autoimmune diseases (1–5). Celiac disease is a human in- fold less than concentrations required for similar T cell responses flammatory condition of the small intestine precipitated by the elicited by corresponding native peptides (9). This improved T ingestion of cereal gluten proteins in wheat that consist of glia- cell recognition has been attributed to increased binding affinity din and glutenin subcomponents. The disease has a very strong of deamidated peptides to HLA-DQ2. Indeed, deamidation of the by guest on September 29, 2021 HLA association, and T cells of celiac lesions recognize gluten DQ2-a-II peptide introduces a negative charge in the residue re- epitopes presented by disease-associated HLA-DQ2 (DQA1*05:01/ siding in pocket 4 of the HLA-DQ2 peptide binding groove, DQB1*02:01) or HLA-DQ8 (DQA1*03/DQB1*03:02) molecules a pocket where negatively charged residues are favored (16, 17). (6, 7). These gluten-reactive CD4+ T cells appear to be essential for Consequently, the binding affinity of the DQ2-a-II peptide to the development of the disease, and they can be readily cultured HLA-DQ2 increases by up to 10-fold after deamidation (9). from intestinal lesions of diseased patients but not control subjects However, this is clearly less than the increase in T cell responses (6, 8). Over the years, a number of gluten T cell epitopes, each elicited by deamidation, suggesting that TCR recognition may recognized by distinct T cell clones derived from celiac disease also be involved in the improved recognition of the deamidated patients, have been identified (9–14). For most of these epitopes, gluten peptides. posttranslational conversion of Gln residues to Glu by the enzyme In the current study, we found a biased usage of TCR Vb6.7 chain transglutaminase 2 is required for efficient T cell recognition (15). encoded by the TRBV7-2 gene segment among TCRs reactive to the DQ2-a-II gliadin epitope. Based on this knowledge, we per- *Department of Immunology, Centre for Immune Regulation, University of Oslo, formed single-cell TCR cloning of a large number of Vb6.7-posi- 0027 Oslo, Norway; †Department of Immunology, Centre for Immune Regulation, tive, DQ2-aII tetramer-positive CD4+ T cells sorted from periph- ‡ Oslo University Hospital, Rikshospitalet, 0027 Oslo, Norway; Department of Mo- eral blood of gluten-challenged celiac patients. Analysis of the lecular Biosciences, Centre for Immune Regulation, University of Oslo, 0316 Oslo, Norway; and xDepartment of Medicine, Oslo University Hospital, Rikshospitalet, TCR sequences revealed in vivo clonal expansion, convergent re- 0027 Oslo, Norway combination, semipublic response, and the notable conservation of 1M.R. and K.S.G. contributed equally to this work. a non-germline-encoded Arg residue in the CDR3b loop. Se- Received for publication May 25, 2011. Accepted for publication July 7, 2011. quencing and functional data support the notion that the deamidated This work was supported by the Research Council of Norway and the South-Eastern residue in the DQ2-a-II peptide poses constraints on the TCR Norway Regional Health Authority. structure in which the conserved Arg residue is a critical element. Address correspondence and reprint requests to Dr. Shuo-Wang Qiao, Department of Immunology, Centre for Immune Regulation, Rikshospitalet, 20 Sognsvannsveien, Materials and Methods N-0027 Oslo, Norway. E-mail address: [email protected] T cell culture, tetramer staining, and FACS sorting The online version of this article contains supplemental material. Abbreviations used in this article: IMGT, international ImMunoGeneTics information Biological material was obtained from celiac disease patients according to system; scTCR, single-chain TCR; RU, resonance unit; SPR, surface plasmon reso- protocols approved by the regional ethics committee, and subjects donating nance; TCC, T cell clone. material gave written informed consent. Polyclonal T cell lines and T cell clones (TCCs) were established from intestinal biopsies of celiac disease Copyright Ó 2011 by The American Association of Immunologists, Inc. 0022-1767/11/$16.00 patients as previously described (18). T cells were expanded with anti- www.jimmunol.org/cgi/doi/10.4049/jimmunol.1101526 The Journal of Immunology 3065

CD3/anti-CD28 beads (Invitrogen) for at least 7 d before mRNA isolation cultured in RPMI 1640 plus 10% FCS. We reached .90% transduction and TCR cloning. efﬁciency, which was maintained throughout the experiments. Tetramerized recombinant HLA-DQ2 tethered with gluten-derived In Ag stimulation assays, 50,000 DQ2.2 (IHW09050) or DQ2.5 (patient peptides containing the T cell epitopes DQ2-a-I (QLQPFPQPELPY, CD114) homozygous EBV-transformed cells were incubated with native underlined 9mer core sequence), DQ2-a-II (PQPELPYPQPE), and the or deamidated gliadin 33mer peptides containing the DQ2-a-II epitope control peptide CLIP2 (MATPLLMQALPMGAL) were generated as (33merQ, LQLQPFPQPQLPYPQPQLPYPQPQLPYPQPQPF; 33merE, previously described (19, 20). Likewise, the DQ2-g-II tetramer was gen- LQLQPFPQPELPYPQPELPYPQPELPYPQPQPF) and native or deam- erated with the same protocol where the DQ2-g-II peptide QGIIQPEQ- idated 14-mer DQ2-a-II peptides (aII.Q-PQPQLPYPQPQLPY; aII.E- PAQL was covalently coupled to HLA-DQ2. Circulating tetramer-positive PQPELPYPQPELPY) at 37˚C for 1–4 h before the addition of 25,000 T cells in peripheral blood of celiac patients 6 d after a 3-d gluten chal- TCR-transduced hybridoma cells. In some experiments, 10 mg/ml anti- lenge were stained according to a previously established protocol (21). HLA class II blocking Abs SPV-L3 (anti-DQ), B7/21 (anti-DP), or Single, Vb6.7-positive (FITC–OT145 from Endogen), DQ2-aII tetramer- B8.11 (anti-DR) was added. In anti-CD3 stimulation assays, wells were positive cells were sorted by a FACSAria sorter (BD) into 96-well PCR coated overnight with 0.2 to 5 mg/ml hamster anti-mouse CD3ε Ab (clone plates (Bio-Rad) with 10 ml ice-cold lysis-reverse-transcription buffer 145.2C11, a gift from B. Bogen, Oslo, Norway) and washed before the containing 50 mM Tris-HCl pH 8.2, 75 mM KCl, 3 mM MgCl2, 0.5% addition of 10,000 TCR-transduced hybridoma cells. In Troybody ex- Nonidet P-40, 2 mM deoxyribonucleotide triphosphate 10 mM DTT, periments, recombinant anti–HLA-DP Abs in which 33merQ or 33merE 800 nM TRBC_rev primer 59-TTCACCCACCAGCTCAGCTCC-39, 800 peptides were inserted into loops between b-strands in constant domains nM TRAC_rev primer 59-AGTCAGATTTGTTGCTCCAGGCC-39,12U (25) were used as Ags, and 10,000 DQ2+ monocyte-derived dendritic cells RNasin (Promega), and 30 U SuperScript II reverse transcriptase (Invi- were used as APCs. The dendritic cells were cultured from positively trogen) in each well. selected monocytes (CD14 microbeads; Miltenyi Biotec) in the presence of 1000 U/ml GM-CSF and 500 U/ml IL-4 (both from R&D Systems) for TCR cloning and sequencing 6 d. In some experiments, biotinylated recombinant soluble peptide–DQ2

complexes were immobilized onto streptavidin plates (Roche) and used Downloaded from The TCR was cloned using a seminested PCR approach performed with to stimulate TCR-transduced hybridoma cells. The soluble DQ2 mole- Phusion DNA polymerase (Finnzymes). mRNA of in vitro-cultured TCCs cules contained either the native (PQPQLPYPQPQ) or the deamidated was isolated with RNeasy Mini kit (Qiagen). TRA and TRB gene-specific (PQPELPYPQPQ) DQ2-a-II peptide covalently linked to the N terminus first-strand cDNA was synthesized from mRNA with TRAC_rev and of the DQ2 b-chain spaced by a 6-aa (GSGSGS) linker. Culture super- TRBC_rev primers and SuperScript II reverse transcriptase. After RNase H natant was assayed for murine IL-2 secretion in ELISA 18 h later. digestion (Rnasin Plus; Promega), the cDNA was precipitated with Pellet In murine IL-2 ELISA, wells were coated with 2 mg/ml rat anti-mouse Paint (Novagen), and a poly-C-tail was added to the 39-end with rTerminal IL-2 (clone JES6-5H4; Pharmingen) and detected with 1 mg/ml biotin rat transferase (Roche). The cDNA was reprecipitated, and the TRA and TRB anti-mouse IL-2 (clone JES6-1A12; Pharmingen). The sensitivity of the http://www.jimmunol.org/ genes were PCR amplified with forward poly-G-NotI primer (59-ATA- assay was at least 4 pg/ml. TGCGGCCGCGGGGGGGGGGGGGGGG-39) and reverse TRAC_MluI_ rev (59-ATACGCGTTCTCTCAGCTGGTACACGG-39) or TRBC_MluI_rev Expression and purification of soluble single-chain TCR (59-ATACGCGTAGATCTCTGCTTCTGATGGC-39) primers for 35 cycles 364.14 variants (98˚C for 10 s, 53˚C for 30 s, 72˚C for 15 s). The PCR product was ligated into the pHOG21-phOx cloning vector and subjected to sequencing. The TCR Va and Vb amino acid sequences from TCC364.1.0.14 were From sorted single cells, TCR-specific first-strand cDNA was synthe- fused by a synthetic linker to make single-chain TCR (scTCR). Codon- sized immediately after sorting by incubating the PCR plate at 42˚C for 50 optimized synthetic DNA encoding wild-type or mutant scTCR sequences min followed by 10 min at 72˚C. Subsequently, 2 ml cDNA was used in the (GenScript) were cloned into the pFKPEI vector (26) between the NcoI first PCR with 500 nM TRBV7-2_fwdA (59-ATATGCGGCCGCTCCCAT- and NotI sites and transformed into Escherichia coli Rosetta Blue cells by guest on September 29, 2021 CCTGCCCTGACCCT-39) and TRBC_rev primer in total 15-ml final (Novagen). A Lb103F (amino acid number assigned according to IMGT volume. In the second PCR, 1 ml of the first PCR product was amplified V-gene alignment) mutation was introduced to stabilize the hydrophobic with 500 nM TRBV7-2_fwdB (59-ATATGCGGCCGCCCTTCTGTCTC- interface between the Va and Vb domains (27), resulting in the s.b109R, CTGGGGGC-39) and TRBC_MluI_rev primer in total 20-ml final vol- s.b109A, s.b109K, and s.b109E variants. Protein expression and purifi- ume. Both PCRs were performed for 8 cycles (98˚C for 20 s, 60˚C for 40 s, cation of soluble periplasmic fractions from 400-ml cultures were per- 72˚C for 40 s) followed by 32 (first PCR) or 22 (second PCR) cycles of formed as described (26). Briefly, transformed E. coli Rosetta Blue cells 98˚C for 20 s, 55˚C for 40 s, 72˚C for 40 s, and a final elongation at 72˚C were inoculated from glycerol stocks into 100 ml 23 YT medium sup- for 3 min. Six microliters of the second PCR product was loaded onto a plemented with 100 mg/ml ampicillin and 0.1 M glucose (YTAG medium) 1% agarose gel, and wells with expected bands at around 350 bp were and incubated at 37˚C overnight. The cultures were reinoculated in 400 ml subjected to sequencing. All sequences were analyzed with the inter- YTAG medium to an OD600nm of 0.025. At OD600nm 0.6–0.8, the bacteria national ImMunoGeneTics information system (IMGT)/V-QUEST online were pelleted and resuspended in 400 ml 23 YT medium supplemented resource (22). with 100 mg/ml ampicillin and incubated at 30˚C overnight. The bacteria were pelleted and resuspended in 40 ml ice-cold periplasmic extraction Retroviral transduction and hybridoma functional assays solution (50 mM Tris-HCl, 20% sucrose, 1 mM EDTA, pH 8) supple- The murine T cell hybridoma line BW58a2b2 devoid of endogenous TCR mented with 1 mg/ml lysozyme and 100 mg/ml RnaseA (Sigma-Aldrich) (23) was a kind gift from Dr. Malissen (Centre d’Immunologie de before incubation for 1 h at 4˚C with rotation. The mixture was then Marseille-Luminy, Marseille, France). The cell line was transfected with centrifuged, and the supernatant was collected as the soluble periplasmic human CD4 encoded in the pORF9-hCD4 plasmid (Invivogen), cloned, fraction. After filtration through an 0.2-mm sterile filter, the soluble peri- subcloned, and a stable human CD4-expressing clone was selected for plasmic fractions were then purified on immobilized metal affinity chro- further retroviral TCR transduction. matography spin columns (Qiagen) with a procedure modified from the The TCR Va and Vb sequences of an in vitro-cultured DQ2-a-II– manufacturer’s protocol such that the salt concentration was increased to 1 reactive CD4+ TCC (TCC364.1.0.14) was fused with murine TCR constant M NaCl and the pH was adjusted to 7. The immobilized metal affinity chains, and the chimeric TCR a and b chains were linked with a P2A chromatography fractions were concentrated to 140 ml by Amicon Ultra-4 peptide. Codon-optimized synthetic DNA encoding wild-type or mutant centrifugal filter devices (Millipore) and dialyzed against HBS-EP buffer TCRa-P2A-TCRb sequences (GenScript) were cloned into the pMIG-II (GE Healthcare) supplemented with 0.5 M NaCl. Protein concentrations retroviral plasmid (a generous gift from Dr. Vignali, Denver, CO) (24) were determined at A280 nm on a NanoDrop ND-1000 apparatus (Thermo between the EcoRI and XhoI sites. The pMIG-II–TCR489 construct Scientific) and found to be 12.6 mg/ml for wt.b109R, 16.5 mg/ml for encoding a non–TRBV7-2 TCR derived from an HLA-DQ8–restricted s.b109R, 27 mg/ml for s.b109A, 5.7 mg/ml for s.b109K, and 4.5 mg/ml for s.b109E. Size-exclusion chromatography on a Superdex 200 10/300 gluten-reactive TCC generated in our laboratory was a gift from Dr. Edwin ¨ Liu (University of Colorado, Denver, CO). pMIG-II-TCR plasmids were GL column was performed on an automated AKTA 900 chromatography cotransfected (Lipofectamine 2000; Invitrogen) with pCL-Eco plasmid system (GE Healthcare). The monomeric fractions were pooled, concen- into GP2-293 packaging cells (Clontech). Virus supernatants were col- trated to equal volumes, and protein concentrations were determined as lected 48 and 72 h after transfection and cell debris removed by centri- described earlier. 2 2 fugation and filtration (0.45 mm). Ten thousand BW58a b .hCD4 Surface plasmon resonance binding assays hybridoma cells were incubated with 1 ml virus supernatant supplemented with 10 mg/ml polybrene and subjected to 3000 3 g centrifugation at 32˚C A Biacore T100 instrument was used together with a CM3 chip (GE for 90 min. The virus supernatant was then removed and hybridoma cells Healthcare). All experiments were performed in HBS-EP buffer. Neu- 3066 TCR USAGE IN CELIAC DISEASE

Table I. The usage of TRBV7-2 by in vitro-expanded TCCs sorted DQ2-aII tetramer-positive T cells in several polyclonal lines according to epitope recognition (Supplemental Fig. 2), whereas few or none of the DQ2-aIor DQ2-gII tetramer-positive T cells used Vb6.7. HLA Extended in vitro culture may introduce bias. Therefore, we Epitope Restriction 9-aa Core Sequence TRBV7-2 looked at the Vb6.7 usage in peripheral blood T cells of nine celiac DQ2-a-I DQ2 P F P Q P E L P Y 3 / 4a patients where the T cells were directly stained with tetramers. Oral DQ2-a-II DQ2 P Q P E L P Y P Q 6 / 9 gluten challenge for 3 d of celiac disease patients in remission leads DQ2-a-III DQ2 P Y P Q P E L P Y 0 / 2 DQ2-g-I DQ2 P Q Q S F P E Q Q 0 / 2 to egress of gluten-reactive T cells into the peripheral blood on day DQ2-g-II DQ2 I Q P E Q P A Q L 0 / 1 6, and such cells can be detected by IFN-g ELISPOT assay (30) or DQ2-g-III DQ2 Q Q P E Q P Y P Q 0 / 1 by HLA-DQ2 tetramer staining (21). We found that significantly DQ2-g-IV DQ2 S Q P E Q E F P Q 0 / 2 more DQ2-aII tetramer-positive T cells were Vb6.7 positive com- DQ2-g-VI DQ2 Q Q P F P E Q P Q 0 / 1 + , DQ2-g-VII DQ2 Q Q P E Q P F P Q 0 / 1 pared with all CD4 T cells (p 0.01, paired Wilcoxon rank test), DQ8-a-I DQ8 E G S F Q P S E Q 0 / 4 whereas a slight increase for DQ2-aI tetramer-positive cells did not DQ8-g-I DQ8 E Q P Q Q P Y P E 1 / 2 reach statistical significance (p = 0.10) (Fig. 1). a3 / 4 indicates that of the total of four TCCs cloned, three used the TRBV7-2 In vivo clonal expansion of DQ2-aI and DQ2-aII gene segment. tetramer-positive cells The knowledge of overusage of the TRBV7-2 gene segment among travidin was immobilized by amine coupling to ∼4500 resonance units DQ2-aI– or DQ2-aII–reactive T cells allowed us to sequence Downloaded from (RU) followed by capture of ∼3500 RU biotinylated DQ2-aII or DQ2- TRBV gene segments from sorted single cells that were double CLIP2. Samples of s.b109R and s.b109A at concentrations from 0.32 to 5 positive for Vb6.7 and either tetramer. We successfully sequenced mM were then injected over each surface, as well as a negative reference cell with Neutravidin only, at a flow rate of 30 ml/min at 25˚C. In all TRBV genes of 23–108 sorted single cells from peripheral blood experiments, data were zero-adjusted and the Neutravidin reference cell of each of the four gluten-challenged celiac disease patients value subtracted. (Fig. 2). All cells were Vb6.7-positive and either DQ2-aItetra- mer-positive, DQ2-aII tetramer-positive, or DQ2-aII–tetramer-

Results http://www.jimmunol.org/ Overusage of Vb6.7 (TRBV7-2) in DQ2-a-II–reactive T cells negative. Within both DQ2-aI and DQ2-aII tetramer-positive cells, we found clonal dominance as a result of in vivo clonal Systematic TCR sequencing of 40 in vitro-generated gluten- expansion. This was particularly evident for the DQ2-aI tetramer- reactive TCCs derived from 14 celiac disease patients produced positive cells, in which 32 of 33 (patient CD757) or 36 of 37 29 unique TRBV sequences encoded by 14 different V-gene (patient CD823) sequences obtained represented one clone in each segments. Notably, one of the V-gene segments, TRBV7-2, was of the two patients (Fig. 2). In comparison, the TCR repertoire in used in 10 of the 29 unique TRBV sequences. Closer examination Vb6.7-positive, DQ2-aII tetramer-positive cells was more diverse. revealed that 9 of the 10 TCCs that used this particular V-gene Between 5 and 19 unique TRBV sequences were retrieved from segment were reactive to either the DQ2-a-I or DQ2-a-II epito-

each patient of which more than half of the cells expressed the two by guest on September 29, 2021 pes (Table I). No other V-gene segment was overrepresented. most frequently used sequences (Fig. 3), indicating a considerable To probe how frequently Vb6.7 was used by Ag-specific T cells clonal expansion in the DQ2-aII tetramer-positive cells as well. in in vitro-expanded polyclonal lines, T cell lines cultured from In contrast, we found no identical sequences among tetramer- celiac lesions were costained with various DQ2-tetramers and negative cells. aVb6.7 (28) specific mAb, OT145. Although this Ab was previously reported to specifically bind Vb-chain encoded by the Public response as a result of convergent recombination in b a TRBV7-2*01 allele only (29), we found that CD4+ T cells from V 6.7 and DQ2- II tetramer double-positive cells TRBV7-2*02 homozygous individuals bound OT145 as well In the TRBV7-2 sequences we obtained from both ex vivo DQ2- (Supplemental Fig. 1). We observed overusage of Vb6.7 among aII tetramer-positive single-cell sequencing and in vitro-cultured

FIGURE 1. Overusage of TRBV7-2 encoding the Vb6.7 chain in DQ2-aII tetramer-positive cells analyzed directly ex vivo. Peripheral blood cells from treated celiac disease patients after gluten challenge were stained with Vb6.7 Ab and DQ2 tetramers containing an irrelevant control peptide (CLIP2) or the DQ2-a-I or DQ2-a-II gliadin epitopes. A, CD4-gated T cells are shown. In a representative patient, ∼5% of tetramer-negative cells were Vb6.7-positive compared with 16% among DQ2-aI–positive and 50% among DQ2–aII positive cells. B, The percentage of cells staining positive for Vb6.7 of all CD4+ T cells and of DQ2-aI or DQ2-aII tetramer-positive cells in nine celiac patients is shown. The p values were calculated with Wilcoxon paired rank test. The Journal of Immunology 3067 Downloaded from http://www.jimmunol.org/ by guest on September 29, 2021 FIGURE 2. TRBV7-2 sequences in DQ2-a-II–reactive cells display a conserved Arg in positive 5 of the CDR3b loop. The CDR3b amino acid sequences of all sorted Vb6.7-positive peripheral blood single cells from four different patients (CD757, CD761, CD767, and CD823) and in vitro-cultured DQ2-a-I– and DQ2-a-II–reactive TCCs that use TRBV7-2 are summarized. Numbers to the right of the CDR3b sequences denote the number of cells expressing the sequence in single-cell PCR data or the patient ID for in vitro-cultured TCCs. All patients are TRBV7-2*01/*01 homozygous except CD757 (*) who is TRBV7-2*02/*02 homozygous and CD761 (**) who is TRBV7-2*01/*02 heterozygous. TRBV7-2*01 sequences appear in black, and TRBV7-2*02 sequences appear in blue. The conserved Arg in position 5 appear in red. Two CDR3b amino acid sequences from DQ2-aII–positive cells from CD823 were encoded by two different nucleotide sequences (superscripts a and b).

DQ2-a-II–reactive TCCs, we observed several examples of pub- have led to the assumption of oligoclonality if the cells had lic response in which TCR CDR3b sequences from different beensubjectedtoCDR3lengthanalysisonly.However,single- patients displayed identical amino acid sequences (Fig. 4A–E). cell TCR sequencing revealed a number of different clones on The fact that several CDR3b sequences obtained from single-cell the nucleotide level underscoring the importance of TCR se- sequencing were identical to amino acid sequences from well- quencing. characterized in vitro-generated DQ2-a-II–specific TCCs from The most striking finding from the single-cell TCR data was the duodenum tissue (Fig. 4A,4C,4E) strengthened the assumption conservation of an Arg residue in position 5 of the CDR3b loops that the cells sorted based on DQ2-aII tetramer binding were from Vb6.7-positive DQ2-aII tetramer-binding T cells. Of the indeed Ag specific. On the nucleotide level, identical amino acid total 143 DQ2-a-II cells from which we obtained TRBV7-2 se- sequences were found to be encoded by different recombination quences by single-cell sequencing, 132 (92%) expressed CDR3b events, observed both between individuals and within the same loops with an Arg in position 5. On the clonal level, 35 of 41 individual (CD823) (Fig. 4B,4F), implying that the public re- (85%) unique TRBV7-2 sequences from DQ2-aII tetramer- sponse observed was a result of convergent recombination sup- positive cells contained Arg in this position. In comparison, only porting the notion of an Ag-driven selection process. 1of70(1%)DQ2-aI positive and 4 of 49 (8%) DQ2-aII tetramer- negative cells showed this Arg signature (Fig. 2). TRBV7-2 b b a Conserved Arg in V 6.7-CDR3 from DQ2- II sequences from in vitro-cultured TCCs showed similar results, tetramer-positive cells as a result of selection where 0 of 3 DQ2-a-I and 6 of 6 DQ2-a-II TRBV7-2 sequences The high throughput of ex vivo single-cell sequencing gener- contained Arg in position 5 of CDR3b (Fig. 2). ated a large body of sequence data that revealed interesting Examination of the nucleotide sequences revealed that 73% (30 molecular features of the CDR3b loops in Vb6.7-positive DQ2- of 41) of the Arg residues from DQ2-a-II cells were non-germline aII tetramer-binding T cells. The vast majority (78%) of the encoded (Supplemental Fig. 3). The fact that most of the Arg CDR3b loops were of the same length of 11 aa, a fact that would residues in our data set were non-germline encoded indicated that 3068 TCR USAGE IN CELIAC DISEASE

hybridoma cells by stable transfection for coreceptor binding to HLA-DQ2. Despite similar levels of CD3ε and TCR surface expression in the wild-type and mutant transfectants (Supplemental Fig. 4), only hybridoma cells expressing the wild-type TCR were reactive to the DQ2-a-II ligand (Fig. 5A). None of the mutants in which the Arg residue in position 5 of the CDR3b (position 109 of the TCR b-chain according to IMGT V-gene alignment) was mutated to Ala (b109A), Lys (b109K), or Glu (b109E), responded to deamidated (Fig. 5A) or native peptides containing the gliadin DQ2-a-II epitope (Table II). The cell signaling machineries in the transfectants were intact as anti-CD3ε stimulation elicited similar levels of murine IL-2 secretion in both wild-type and mutants (Fig. 5B). Additional experiments conﬁrmed that the wild-type transfectant responded to Ag in the same way as the corresponding TCC from which the TCR sequence originated. This included DQ2 restriction (Fig. 5C), deamidation dependence (Fig. 5D), epitope speciﬁcity, as well as ability to recognize the DQ2- a-II peptide presented by the closely related DQ2.2 molecule.

FIGURE 3. In vivo clonal expansion of Vb6.7 and DQ2-aII tetramer The importance of the Arg residue in DQ2-a-II recognition was Downloaded from double-positive cells. The clonal distribution based on CDR3b nucleotide independently conﬁrmed in a set of experiments using soluble sequences of Vb6.7 and DQ2-aII tetramer double-positive cells from recombinant TCRs. The variable domains of 364.14 TCR were single-cell TRBV sequencing in four patients is shown. Patient ID and the expressed recombinantly as single-chain constructs either with the total number of cells analyzed in each individual are shown in the center of wild-type sequence (wt.b109R), a stabilized version of the wild each distribution scheme. Number of cells sharing the same CDR3b nu- type (s.b109R), or with mutants where the Arg residue was mu- cleotide sequence is shown in the slices.

tated to Ala (s.b109A), Lys (s.b109K), or Glu (s.b109E) residues. http://www.jimmunol.org/ Initial surface plasmon resonance (SPR) binding experiments this trait had been subjected to selection, presumably driven by the demonstrated poor functional yields of wt.b109R, and therefore, cognate Ag, the DQ2-a-II peptide. point mutations were introduced at the V domain interface to increase intrinsic molecular stability as described (27). The mu- b The conserved Arg in position 5 of CDR3 is crucial for tants s.b109K and s.b109E could not be produced in functional a DQ2- -II epitope recognition yields that allowed SPR measurements. A dramatic reduction in To probe the functional importance of Arg in position 5 of CDR3b, binding to DQ2-aII was observed in the s.b109A mutant com- we made transfectants with the TCR of the TCC TCC364.1.0.14 pared with s.b109R, which expressed the wild-type CDR3b se- (i.e., TCR 364.14). The wild-type 364.14 TCR expressing the quence (Fig. 6A,6B). Neither s.b109R nor s.b109A bound the by guest on September 29, 2021 conserved Arg residue in a prototypic CDR3b loop, ASSI- DQ2-CLIP2 molecule that displayed an irrelevant peptide, dem- RSTDTQY, and mutant 364.14 TCRs were retrovirally transduced onstrating the speciﬁcity of binding to DQ2-aII (Fig. 6C,6D). into a murine hybridoma cell line, BW58a2b2, which was devoid Combined, consistent functional data from 364.14 TCR trans- of endogenous TCR (23). Human CD4 was introduced into the fectants and binding data from soluble 364.14 TCRs showed that

FIGURE 4. Convergent recombination and public responses in TRBV7-2 sequences from DQ2-a-II–reactive cells. Amino acid and nucleotide sequences of some DQ2-a-II–reactive TRBV7-2–CDR3b loops are grouped A–F according to shared CDR3b amino acid sequences. All sequences derived from blood were obtained with TRBV7-2–speciﬁc single-cell PCR, and sequences derived from duodenum were obtained from in vitro-generated DQ2-a-II– reactive TCCs. The conserved Arg residue and the corresponding encoding nucleotides are underlined. Column 3 shows the IDs of patients from whom the respective sequences are derived, and column 5 shows the number of cells found to express the given CDR3b nucleotide sequence. All sequences use TRBJ2-3*01 but different TRBD gene segments. The junctional analysis was performed by the IMGT/V-QUEST online resource (22). Dots represent germline nucleotides that are removed during the VDJ recombination. The Journal of Immunology 3069

FIGURE 5. Arginine residue in position 5 of CDR3b is important for DQ2-a-II recognition by TCR 364.14. A,BW58a2b2.hCD4 cells transfected with wild-type 364.14 TCR (solid circle), b109A (diamond), b109K (+) or b109E (3) mutant TCRs were tested in assays where deamidated 33mer peptide (33merE) containing gliadin DQ2-a-II epitope was presented by DQ2.5 APC. B, The same TCR transfectants were stimulated with plate-bound anti-CD3ε Abs. Untransfected cells (*). C, Activation of wild- type 364.14 TCR transfectant by 1 mM 33merE peptide was blocked by 10 mg/ml anti-DQ Ab but not by anti-DP or anti-DR Abs. D, The wild-type 364.14 TCR transfectant was tested with either deamidated 33mer (solid circle), native 33mer (open circle), deamidated 14mer DQ2-a-II peptide (solid triangle), or native 14mer (open triangle). The arrows indicate peptide concentrations needed to elicit 18 pg/ml mouse IL-2 secretion. None of the mutant transfectants were activated by native 33mer peptide. E, Downloaded from Two-log stronger activation of wild-type 364.14 TCR transfectant by anti–HLA-DP Abs (Troybodies) engineered to contain deamidated (solid circle) 33mer compared with those containing native 33mer (open circle). F, Two-log stronger activation of wild- type 364.14 TCR transfectant by streptavidin- http://www.jimmunol.org/ immobilized biotinylated soluble deamidated DQ2- aII molecules (solid triangle) compared with native DQ2-aII (open triangle) molecules. For all panels, the experiments were repeated at least two times, and the error bars indicate SEM of each triplicate. The dotted lines indicate the detection limit of mouse IL- 2 ELISA. WT, wild-type. by guest on September 29, 2021 the conserved Arg in position 5 of the CDR3b loop was crucial for Although peptide binding to DQ2 increased after deamidation, it TCR recognition of the cognate peptide–MHC ligand DQ2:DQ2- alone could not account for the dramatic improvement in T cell a-II peptide complex. Mutation of this Arg residue to the small recognition. In standard T cell assays as that shown in Fig. 5D, the amino acid residue Ala, to a positively charged Lys residue, or to peptide Ag was present during the entire assay time such that in a negatively charged Glu residue had detrimental effects on theory, an apparent 10-fold lower binding affinity to HLA-DQ2 of 364.14 TCR recognition of DQ2-a-II. the native gliadin a-II peptide (9) could be compensated by a 10- fold higher peptide concentration in T cell assay. However, as The TCR interacts with the posttranslationally modified Glu shown in Fig. 5D, a 2-log higher concentration of the native residue of the DQ2-a-II epitope peptide was necessary to elicit T cell responses comparable to that Data thus far showed in vivo Ag-driven TCR selection of DQ2- observed with the deamidated peptide. This difference in recog- a-II–reactive T cells in celiac patients that resulted in biased Vb nition was confirmed when the native and deamidated gliadin usage, in vivo clonal expansion, convergent recombination, semi- epitopes were offered as part of Ab molecules (Troybodies) (25) public response, and the notable conservation of an Arg residue that targeted cell surface receptors of the APC (Fig. 5E). The at position 5 of CDR3b. We next asked which part of the DQ2: efficient presentation of the Troybody reagents allowed more ac- DQ2-a-II complex was interacting with the Arg residue and thus curate assessment of Ag recognition, and importantly the con- participated in the selection of Vb6.7 chains containing the sig- version of Gln to Glu that to some extent occurred in the workup nature Arg in the CDR3b loop. of synthetic peptides could be excluded as a confounding factor. Finally, to remove HLA-DQ2 binding affinity differences in the assay, we engineered soluble recombinant HLA-DQ2 molecules in Table II. Recognition of variants of the DQ2-a-II epitope by TCR which the native and the deamidated DQ2-a-II epitope peptides transfectants expressing wild-type or mutant 364.14 TCR were tethered covalently to the peptide binding groove via a 6-aa peptide linker. Biotinylated DQ2-aII.Q molecules, either con- Peptide Ag TCR Transfectant taining a peptide with a native sequence (PQPQLPYPQPE) or 1 23456789 WTb109A b109K b109E equivalent amounts of biotinylated DQ2-aII.E molecules containing the deamidated epitope (PQPELPYPQPE), were immo- 222 PQPELPYPQPE +++ bilized on streptavidin plates and presented to hybridoma cells 222D 2222222 +++ 222 222Q 2222222L P Y (+) 222 expressing the DQ2-a-II–reactive TCR. In this assay, where T cell 222K 2222222 2 2 2 2 recognition of titrated amounts of stable DQ2-aII complexes were +++, strong T cell response; (+), weak T cell response at the highest Ag concen- tested, a near 2-log difference in T cell recognition elicited by the tration only; 2, no T cell response; and WT, wild-type. deamidated and native Ag was observed (Fig. 5F). This showed 3070 TCR USAGE IN CELIAC DISEASE

FIGURE 6. SPR analyses of scTCRs s.b109R and s.b109A with soluble DQ2-aII and DQ2-CLIP2. Representative sensograms are shown. A–D,In all experiments, Neutravidin was immobilized to ∼4500 RU, followed by capture of ∼3500 RU biotinylated DQ2-aII (A, B) or DQ2-CLIP2 (C, D) molecules. Soluble scTCR s.b109R (A, C) and mutant s.b109A (B, D) were then injected at various concentrations ranging from 0.32 to 5 mM at 25˚C. Downloaded from

clearly that the DQ2-a-II–reactive TCR was able to differentiate generated TCCs or directly isolated from peripheral blood of ce- between the native Gln residue and the deamidated Glu residue liac disease patients, we provide a comprehensive characterization

despite that these residues were positioned in the P4 pocket and of the TCR repertoire of T cells specific for immunodominant http://www.jimmunol.org/ should not be solvent exposed. gluten epitopes. Our results provide novel insights into how CD4+ Given the conspicuous presence of a positively charged Arg T cells that are centrally involved in the development of celiac residue at the tip of the CDR3b loop, one can envision that this may disease recognize posttranslationally modified gluten Ags and be the TCR residue that “reads” the Glu residue of the DQ2-a-II reveal that a non-germline-encoded part of TCR is central in complex through a charge-to-charge interaction. This notion was recognition of a posttranslationally modified residue. supported by a vigorous proliferative response elicited by the DQ2- Analysis of TRBV7-2 sequences obtained from single-cell se- a-IIE4D peptide where the Glu residue in P4 was replaced by an- quencing of DQ2-aII tetramer-positive T cells revealed in vivo other negatively charged residue, Asp (Table II). To note, we never clonal expansion, convergent recombination, semipublic response, observed Lys, another positively charged residue, in position 5 of the and the notable conservation of an Arg residue at position 5 of by guest on September 29, 2021 CDR3b loops of DQ2-a-II–reactive, Vb6.7-positive TCRs. Simi- CDR3b. Importantly, the observations were made from direct se- larly, a TCR displaying Lys that was introduced into this position by quencing of Ag-specific cells that have not been subjected to culture in vitro mutagenesis was not reactive to the DQ2-a-II ligand (Fig. and thus any bias introduced by in vitro expansion. Moreover, 5A). This demonstrates the presence of additional structural con- similar results were found in all four celiac disease patients studied, straints that may modify the charge-to-charge interaction between and identical Vb-chain amino acid sequences were found in direct Arg in the CDR3b and Glu in the DQ2-a-II peptide. sequencing of tetramer-sorted peripheral blood T cells and TCR In addition to the highly conserved Arg residue in position 5, the sequencing of in vitro-generated TCCs derived from intestinal bi- neighboring residues, which were often non-germline encoded, opsies of a different set of celiac disease patients. also exhibited certain degrees of conservation. The residue in The fact that most of the Arg residues in our data set are non- position 4 of the canonical 11-aa CDR3b loop was most often Ile germline encoded indicates that this trait has been subjected to or Leu, whereas the position 6 C-terminally to the conserved Arg selection, presumably driven by the cognate Ag, the DQ2-a-II residue was most often occupied by Ser, His, Tyr, or Ala residues. epitope. This epitope has undergone a posttranslational modifica- Collectively, the canonical CDR3b loop sequence ASS(I/L)R(S/ tion (i.e., a deamidation from Gln to Glu). Our data suggest that H/Y/A)TDTQY was present in 67% of the Vb6.7-positive TCRs the in vivo-selected TCRs specifically sense this posttranslational (96 of total 143 single cells) reactive to DQ2-a-II and represent modification at the P4 position of the epitope. The amino acid more than 86% (96 of 111) of the CDR3b loops with a length residue positioned in the P4 pocket should not be much solvent of 11 aa. Thus, the positively charged Arg residue in the TCR exposed so it is not obvious how this sensing takes place. The exact is probably the primary residue interacting with the negatively determination will require a crystal structure, but obtaining this is charged Glu residue of the peptide, demonstrated by its highly unfortunately not trivial. We, as well as others (31), have previously conserved occurrence, with additional contributions from neigh- tried to solve the crystal structure of the DQ2:DQ2-a-II complex boring residues that showed clear nonrandom distribution, but without success. Our functional data demonstrate that the Glu res- nevertheless less conservation. idue of the DQ2-a-II peptide can be replaced by Asp suggesting that the negative charge is important for TCR recognition. Further, Discussion our mutational analysis of a prototype TCR indicates that the Arg Gluten-reactive CD4+ T cells from celiac disease patients repre- residue in the CDR3b loop is critical for the TCR recognition. It is sent a unique source of disease-relevant human T cells. The TCR hence conceivable that there is direct interaction between the neg- repertoire of these cells is the result of natural Ag-driven selection atively charged P4 Glu residue of the peptide and the positively and expansion in a human HLA-associated inflammatory disease. charged Arg residue of the CDR3b loop in the TCRs. By use of highly specific MHC tetramers and sequencing of TCRs In our data set, we observed semipublic TCR response (i.e., from a large number of single T cells, either derived from in vitro- similar Vb sequences were found to be dominant in different The Journal of Immunology 3071 individuals). Altogether, the CDR3b sequence ASS(I/L)R(S/H/Y/ 9. Arentz-Hansen, H., R. Ko¨rner, O. Molberg, H. Quarsten, W. Vader, Y.M. C. Kooy, K. E. A. Lundin, F. Koning, P. Roepstorff, L. M. Sollid, and S. N. McAdam. 2000. A)TDTQY was observed in roughly two thirds of all sorted Vb6.7 The intestinal T cell response to a-gliadin in adult celiac disease is focused on and DQ2-aII double-positive cells from four different patients. a single deamidated glutamine targeted by tissue transglutaminase. J. Exp. Med. Public TCR response arises from either highly recurrent TCRs that 191: 603–612. 10. Sjo¨stro¨m, H., K. E. A. Lundin, O. Molberg, R. Ko¨rner, S. N. McAdam, are exported from the thymus to the periphery in relatively high D. Anthonsen, H. Quarsten, O. Noreń, P. Roepstorff, E. Thorsby, and frequency or as a result of high-avidity interaction with the pep- L. M. Sollid. 1998. Identification of a gliadin T-cell epitope in coeliac disease: tide–MHC complex followed by in vivo expansion. Our data general importance of gliadin deamidation for intestinal T-cell recognition. Scand. J. Immunol. 48: 111–115. cannot differentiate the relative contribution of these two not 11. Arentz-Hansen, H., S. N. McAdam, O. Molberg, B. Fleckenstein, K. E. Lundin, mutually exclusive mechanisms. Affinity studies of a larger panel T. J. Jørgensen, G. Jung, P. Roepstorff, and L. M. Sollid. 2002. Celiac lesion of prototypical versus non-prototypical DQ2-a-II–reactive TCRs T cells recognize epitopes that cluster in regions of gliadins rich in proline residues. Gastroenterology 123: 803–809. or analysis of the TCR repertoire of naive DQ2-a-II–reactive 12. Vader, W., Y. Kooy, P. Van Veelen, A. De Ru, D. Harris, W. Benckhuijsen, T cells will give answers to this question. S. Penã, L. Mearin, J. W. Drijfhout, and F. Koning. 2002. The gluten response in children with celiac disease is directed toward multiple gliadin and glutenin In several human autoimmune diseases, increasing evidence peptides. Gastroenterology 122: 1729–1737. suggests that T cells bypass tolerance by recognizing autoantigens 13. Qiao, S. W., E. Bergseng, O. Molberg, G. Jung, B. Fleckenstein, and that have been posttranslationally modified. This is observed in L. M. Sollid. 2005. Refining the rules of gliadin T cell epitope binding to the disease-associated DQ2 molecule in celiac disease: importance of proline anticitrulline response (2) or glycosylation-dependent T cell re- spacing and glutamine deamidation. J. Immunol. 175: 254–261. sponse (32) in rheumatoid arthritis, disulfide bond formation in 14. Tye-Din, J. A., J. A. Stewart, J. A. Dromey, T. Beissbarth, D. A. van Heel, proinsulin in type 1 diabetes (5), and anticitrulline response in A. Tatham, K. Henderson, S. I. Mannering, C. Gianfrani, D. P. Jewell, et al. 2010. Comprehensive, quantitative mapping of T cell epitopes in gluten in celiac multiple sclerosis (3). In this article, we show that a posttransla- disease. Sci. Transl. Med. 2: 41ra51. Downloaded from tionally modified residue in the antigenic peptide, the negatively 15. Sollid, L. M. 2002. Coeliac disease: dissecting a complex inflammatory disorder. charged Glu residue in P4, interacts with the TCR and conse- Nat. Rev. Immunol. 2: 647–655. 16. van de Wal, Y., Y. M. C. Kooy, J. W. Drijfhout, R. Amons, and F. Koning. 1996. quently selects TCRs with a conserved Arg residue in a non- Peptide binding characteristics of the coeliac disease-associated DQ(a1*0501, germline-encoded part of the CDR3b loop. This represents an b1*0201) molecule. Immunogenetics 44: 246–253. 17. Johansen, B. H., F. Vartdal, J. A. Eriksen, E. Thorsby, and L. M. Sollid. 1996. interesting example of in vivo TCR selection exerted by a post- Identification of a putative motif for binding of peptides to HLA-DQ2. Int. translationally modified antigenic peptide in a human disease. Immunol. 8: 177–182. http://www.jimmunol.org/ Future studies will reveal if posttranslationally modified Ags of 18. Molberg, Ø., S. N. McAdam, K. E. A. Lundin, and L. M. Sollid. 2000. Studies of gliadin-specific T cells in celiac disease. In Celiac Disease—Methods and other autoimmune diseases impose similar selection of particular Protocols. M. N. Marsh, ed. Humana Press, Totowa, NJ, p. 105. TCR repertoires. 19. Quarsten, H., S. N. McAdam, T. Jensen, H. Arentz-Hansen, Molberg Ø, K. E. A. Lundin, and L. M. Sollid. 2001. Staining of celiac disease-relevant T cells by peptide-DQ2 multimers. J. Immunol. 167: 4861–4868. Acknowledgments 20. Fallang, L. E., S. Roh, A. Holm, E. Bergseng, T. Yoon, B. Fleckenstein, We thank the celiac disease patients for donating biological material, A. Bandyopadhyay, E. D. Mellins, and L. M. Sollid. 2008. Complexes of two cohorts of CLIP peptides and HLA-DQ2 of the autoimmune DR3-DQ2 haplo- Dr. Kjetil Taskeń (Biotechnology Centre of Oslo, Oslo University) for type are poor substrates for HLA-DM. J. Immunol. 181: 5451–5461. access to the Biacore T100 instrument, and Marie Kongshaug Johannesen 21. Ra´ki, M., L. E. Fallang, M. Brottveit, E. Bergseng, H. Quarsten, K. E. A. Lundin, and Bjørg Simonsen for excellent technical assistance. and L. M. Sollid. 2007. Tetramer visualization of gut-homing gluten-specific by guest on September 29, 2021 T cells in the peripheral blood of celiac disease patients. Proc. Natl. Acad. Sci. USA 104: 2831–2836. Disclosures 22. Brochet, X., M. P. Lefranc, and V. Giudicelli. 2008. IMGT/V-QUEST: the highly The authors have no financial conflicts of interest. customized and integrated system for IG and TR standardized V-J and V-D-J sequence analysis. Nucleic Acids Res. 36(Web Server issue): W503-8. 23. Letourneur, F., and B. Malissen. 1989. Derivation of a T cell hybridoma variant deprived of functional T cell receptor a and b chain transcripts reveals a non- References functional a-mRNA of BW5147 origin. Eur. J. Immunol. 19: 2269–2274. 1. Corthay, A., J. Ba¨cklund, J. Broddefalk, E. Michae¨lsson, T. J. Goldschmidt, 24.Holst,J.,K.M.Vignali,A.R.Burton,andD.A.Vignali.2006.RapidanalysisofT- J. Kihlberg, and R. Holmdahl. 1998. Epitope glycosylation plays a critical role cell selection in vivo using T cell-receptor retrogenic mice. Nat. Methods 3: 191–197. for T cell recognition of type II collagen in collagen-induced arthritis. Eur. J. 25. Tunheim, G., K. W. Schjetne, I. B. Rasmussen, L. M. Sollid, I. Sandlie, and Immunol. 28: 2580–2590. B. Bogen. 2008. Recombinant antibodies for delivery of antigen: a single loop 2. Masson-Bessie`re, C., M. Sebbag, E. Girbal-Neuhauser, L. Nogueira, C. Vincent, between b-strands in the constant region can accommodate long, complex and T. Senshu, and G. Serre. 2001. The major synovial targets of the rheumatoid tandem T cell epitopes. Int. Immunol. 20: 295–306. arthritis-specific antifilaggrin autoantibodies are deiminated forms of the a- and 26. Gunnarsen, K. S., E. Lunde, P. E. Kristiansen, B. Bogen, I. Sandlie, and b-chains of fibrin. J. Immunol. 166: 4177–4184. G. A. Loset. 2010. Periplasmic expression of soluble single chain T cell 3. Martin, R., J. N. Whitaker, L. Rhame, R. R. Goodin, and H. F. McFarland. 1994. receptors is rescued by the chaperone FkpA. BMC Biotechnol. 10: 8. Citrulline-containing myelin basic protein is recognized by T-cell lines derived 27. Richman, S. A., D. H. Aggen, M. L. Dossett, D. L. Donermeyer, P. M. Allen, from multiple sclerosis patients and healthy individuals. Neurology 44: 123–129. P. D. Greenberg, and D. M. Kranz. 2009. Structural features of T cell receptor 4. Schellekens, G. A., B. A. de Jong, F. H. van den Hoogen, L. B. van de Putte, and variable regions that enhance domain stability and enable expression as single- W. J. van Venrooij. 1998. Citrulline is an essential constituent of antigenic chain ValphaVbeta fragments. Mol. Immunol. 46: 902–916. determinants recognized by rheumatoid arthritis-specific autoantibodies. J. Clin. 28. Arden, B., S. P. Clark, D. Kabelitz, and T. W. Mak. 1995. Human T-cell receptor Invest. 101: 273–281. variable gene segment families. Immunogenetics 42: 455–500. 5. Mannering, S. I., L. C. Harrison, N. A. Williamson, J. S. Morris, D. J. Thearle, 29. Li, Y., P. Szabo, M. A. Robinson, B. Dong, and D. N. Posnett. 1990. Allelic K. P. Jensen, T. W. H. Kay, J. Rossjohn, B. A. Falk, G. T. Nepom, and variations in the human T cell receptor V b 6.7 gene products. J. Exp. Med. 171: A. W. Purcell. 2005. The insulin A-chain epitope recognized by human T cells is 221–230. posttranslationally modified. J. Exp. Med. 202: 1191–1197. 30. Anderson, R. P., P. Degano, A. J. Godkin, D. P. Jewell, and A. V. S. Hill. 2000. 6. Lundin, K. E., H. Scott, T. Hansen, G. Paulsen, T. S. Halstensen, O. Fausa, In vivo antigen challenge in celiac disease identifies a single transglutaminase- E. Thorsby, and L. M. Sollid. 1993. Gliadin-specific, HLA-DQ(a1*0501,b1*0201) modified peptide as the dominant A-gliadin T-cell epitope. Nat. Med. 6: 337– restricted T cells isolated from the small intestinal mucosa of celiac disease 342. patients. J. Exp. Med. 178: 187–196. 31. Henderson, K. N., H. H. Reid, N. A. Borg, S. E. Broughton, T. Huyton, 7. Lundin, K. E., H. Scott, O. Fausa, E. Thorsby, and L. M. Sollid. 1994. T cells R. P. Anderson, J. McCluskey, and J. Rossjohn. 2007. The production and from the small intestinal mucosa of a DR4, DQ7/DR4, DQ8 celiac disease pa- crystallization of the human leukocyte antigen class II molecules HLA-DQ2 and tient preferentially recognize gliadin when presented by DQ8. Hum. Immunol. HLA-DQ8 complexed with deamidated gliadin peptides implicated in coeliac 41: 285–291. disease. Acta Crystallogr. Sect. F Struct. Biol. Cryst. Commun. 63: 1021–1025. 8. Molberg, O., K. Kett, H. Scott, E. Thorsby, L. M. Sollid, and K. E. A. Lundin. 32. Ba¨cklund, J., S. Carlsen, T. Ho¨ger, B. Holm, L. Fugger, J. Kihlberg, 1997. Gliadin specific, HLA DQ2-restricted T cells are commonly found in H. Burkhardt, and R. Holmdahl. 2002. Predominant selection of T cells specific small intestinal biopsies from coeliac disease patients, but not from controls. for the glycosylated collagen type II epitope (263-270) in humanized transgenic Scand. J. Immunol. 46(3): 103–109. mice and in rheumatoid arthritis. Proc. Natl. Acad. Sci. USA 99: 9960–9965.