JOURNAL OF MOLECULAR RECOGNITIONJ. Mol. Recognit. 2003; 16: 333–336DOI:10.1002/jmr.641

Review

The molecular basis of Celiac disease

Frits Koning*Department of Immunohematology and Blood Transfusion, Leiden University Medical Center, Leiden, The Netherlands

Celiac disease is caused by inflammatory, gluten specific T cell responses in the small intestine. Invariablysuch responses are HLA-DQ2 or HLA-DQ8 restricted, providing an explanation for the strong associationbetween celiac disease and these HLA-class II alleles. It is now clear that some native gluten sequences canbind to HLA-DQ2/8 and induce T cell responses. In addition, modification of gluten peptides by the enzymetissue transglutaminase results in high affinity HLA-DQ2/8 binding peptides that can induce T cellresponses. Thus, gluten molecules contain a large number of immunogenic peptides and this is likely toplay an important role in the breaking of oral tolerance to gluten. Copyright# 2003 JohnWiley& Sons, Ltd.

Keywords: Celiac Disease; tissue transglutaminase; gluten peptides; HLA-DQ

Received 25 April 2003; revised 10 May 2003; accepted 17 May 2003

INTRODUCTION

Celiac disease (CD) is caused by a permanent intolerance togluten. Gluten is a heterogeneous mixture of proteinstermed gliadins and glutenins. Over 90% of the patientsare HLA-DQ2 positive and the remainder are usually HLA-DQ8 positive (Marsh, 1992; Tighe et al., 1992; Sollid et al.,1989; Spurkland et al., 1997). It is now well established thatinflammatory HLA-DQ2 and/or -DQ8-restricted, gluten-specific T cells are present at the site of the lesions in thegut (Lundin et al., 1993, 1997; van de Wal et al., 1998a,b,2000) and these are believed to cause disease. Several glutenpeptides that are recognized by gut-derived T cells from CDpatients have now been identified (van de Wal et al.,1998a,b, 2000; Sjostrom et al., 1998; Molberg et al.,1998; Arentz-Hansen et al., 2000; Vader et al., 2002a,b).Moreover, it has been found that such peptides can bemodified by the enzyme tissue transglutaminase (tTG) thatis either required for or enhances T cell recognition (van deWal et al., 1998b; Molberg et al., 1998). Here a number ofkey observations that have been made in recent years arebriefly described and the way this has reshaped our thinkingabout CD development is discussed.

PEPTIDE BINDING TO HLA-DQ

The (almost) exclusive occurrence of CD in HLA-DQ2-and/or DQ8-positive individuals suggests that these DQ

molecules have unique properties that allow the presentationof gluten-derived peptides to T cells. HLA-DQ2 and -DQ8are HLA-class II molecules that bind peptides of variablelength, usually 12–20 amino acids long (Rammensee et al.,1995). Amino acid side chains in the core of the peptide fitwithin pockets in the class II molecule and mediate specificbinding. Such interactions are unique for a given HLA-classII molecule and this is generally known as a peptide-bindingmotif. When the peptide binding motifs for HLA-DQ2 and-DQ8 were determined it became immediately clear thatboth have a preference for negatively charged residues atseveral positions in the bound peptides (van de Wal et al.,1996; Vartdal et al., 1996; Godkin et al., 1997; Kwok et al.,1996). Strikingly, this prohibited the identification of glutenpeptides that would bind to HLA-DQA2 or -DQ8 sinceamino acids with negative charge are very rare in glutenmolecules.

T CELL STIMULATORY, GLUTENDERIVED PEPTIDES

Subsequently, gluten peptides have been identified thatstimulate gut-derived T cells from CD patients (van deWal et al., 1998a,b, 2000; Sjostrom et al., 1998). We firstidentified an HLA-DQ8 restricted gliadin peptide that isrecognized by DQ8 restricted T cells from all patients (vande Wal et al., 1998a; see Table 1). This peptide containedpepsin cleavage sites at both ends, which suggested that itcould be generated in the stomach, and requires no furtherprocessing for presentation to T cells in the gut. Strikingly,in the minimal nine amino acid core of this gliadin peptideno negatively charged amino acids were present. However,the peptide contained four glutamine residues, includingtwo at positions 1 and 9 of the peptide, the sites where HLA-DQ8 prefers negatively charged residues. On the basis ofthis finding we reasoned that a modified peptide in which theglutamines at these positions were replaced by a glutamic

Copyright # 2003 John Wiley & Sons, Ltd.

*Correspondence to: F. Koning, Department of Immunohematology and Blood

Transfusion, Leiden University Medical Center, PO Box 9600, 2300 RC Leiden,

The Netherlands.

E-mail: f.koning@lumc.nl

Contract/grant sponsor: EC; contract/grant numbers: BHM4-CT98-3087;

QLK1-2000-00657.

Contract/grant sponsor: Stimuleringsfonds Voedingsonderzoek LUMC.

Contract/grant sponsor: Dutch Organization for Scientific Research; contract/

grant number: 912-02-028.

Abbreviations used: CD, celiac disease.

acid would potentially bind better to DQ8 and possesssuperior T cell stimulatory properties. Indeed, both theseamino acid substitutions were found to yield peptides withan up to 100-fold increased T cell stimulatory potential (vande Wal et al., 1998b; Table 1). The substitution of the twoother glutamine residues, however, completely abrogatedthe T cell response (van de Wal et al., 1998b; Table 1). Onlyselective conversion of particular glutamine residues in thispeptide, therefore, results in increased T cell responses.

TISSUE TRANSGLUTAMINASE

Concomitant with this work, Schuppan and colleaguesreported that CD patients make autoantibodies to the en-zyme tissue transglutaminase and that the appearance ofthese antibodies is very specific indicators of disease(Dieterich et al., 1997). Tissue transglutaminase is anubiquitous enzyme found in all organs, including theintestine. It is known to be released upon cellular damageand to crosslink proteins in order to control tissue damage.This crosslinking occurs by forming a covalent bond be-tween the sidechain of a glutamine in one protein with theaminogroup in the sidechain of a lysine in the other protein.It was also known that in the absence of a suitable lysine, theenzyme reactivity could result in deamidation, the conver-sion of a glutamine residue into a glutamic acid. We there-fore determined whether the activity of tissuetransglutaminase would lead to the selective conversion ofglutamine residues. Indeed, tissue transglutaminase wasfound to deamidate the gliadin peptide, selectively at thep1 and p9 positions (van de Wal et al., 1998b).

Similarly, Sollid and colleagues identified an HLA-DQ2restricted gliadin peptide and found that this peptide wasonly recognized after deamidation of particular Q-residuesin the peptide, either as the result of chemical modificationor treatment with tissue transglutaminase (Sjostrom et al.,1998; Molberg et al., 1998). Moreover, evidence wasprovided that the reactivity of many HLA-DQ restricted,gluten specific T cell clones depended on treatment ofgluten with tissue transglutaminase (van de Wal et al.,

1998b; Molberg et al., 1998). Thus, selective deamidationoccurs which favors T cell recognition suggesting a role fortissue transglutaminase in the development of CD.

IDENTIFICATION OF ADDITIONALT CELL STIMULATORY PEPTIDES

Further work quickly revealed that gluten molecules containa large number of T cell stimulatory sequences. Multiple Tcell epitope motifs are present in the alpha- and gamma-gliadin molecules as well as in the glutenins (Arentz-Hansenet al., 2000; Vader et al., 2002a,b; Table 1). While fourpeptides corresponding to these identified motifs do notrequire deamidation for T cell recognition, the majority ofthe peptides do. Importantly, this work also revealed that, inany patient, T cell responses to more than one gluten peptideare found and all give rise to the secretion of inflammatorycytokines and are thus likely to be involved in the diseaseprocess.

SPECIFICITY OF tTG

As mentioned above, selective deamidation of particularglutamine residues in gluten peptides is essential for their Tcell stimulatory properties. This has been observed for allpeptides where T cell recognition is dependent on deamida-tion. Thus, the specificity of tissue transglutaminase appears tocorrelate with gluten toxicity for CD patients but the enzymespecificity had not been determined. We have investigated thisby analyzing the influence of glutamine flanking sequences ontissue transglutaminase activity. For this purpose the DQ8binding gliadin peptide has been taken as a lead compoundsince only two out of the four glutamines in the peptide aredeamidated by tTG. Substitution analogs of the peptide havebeen synthesized in which all amino acids were systematicallyreplaced by amino acids of choice and these analogs weretreated with tissue transglutaminase. Subsequent mass spec-trometry was used to determine the deamidation of theglutamines in these analogs. The results indicate that the

Table 1. Amino acid sequence of T cell stimulatory gluten peptides

Peptide Sequence HLAa tTGb

Glia-( (206–217) SGQGSFQPSQQN DQ8 þGlt (723–735) QQGYYPTSPQQSG DQ8 �Glia-g1 (138–153) QPQQPQQSFPQQQRPF DQ2 þþþGlia-a2 (62–75) PQPQLPYPQPQLPY DQ2 þþþGlia-a9 (57–68) QLQPFPQPQLPY DQ2 þþþGlia-a20 (93–106) PFRPQQPYPQPQPQ DQ2 þþþGlt-156 (40–59) QPPFSQQQQSPFSQ DQ2 þþþGlt-17 (46–60) QQPPFSQQQQQPLPQ DQ2 þþþGlu-5 QQQXPQQPQQF DQ2 �Glia-g30 (222–236) VQGQGIIQPQQPAQL DQ2 �a HLA-restriction.b Dependance of T cell recognition by tissue transglutaminase modification of gluten peptide: (� ) T cell recognition not dependent on tissuetransglutaminase; (þ ) T cell recognition is enhanced by tissue transglutaminase; T cell recognition dependent on tissue transglutaminase.

334 F. KONING

Copyright # 2003 John Wiley & Sons, Ltd. J. Mol. Recognit. 2003; 16: 333–336

spacing between glutamine and proline residues in glutenpeptides has a major influence on the deamidation pattern(Vader et al., 2002b). In the sequences QP and QXXP, the Q isnot a target for tissue transglutaminase. In contrast, in thesequences QXP, QXXF and QXPF the Q is deamidated bytissue transglutaminase (Vader et al., 2002b). This is highlysignificant since glutamine and proline are the two mostabundant amino acids in gluten. Consequently, this observa-tion could be used for the design of a predictive algorithm toidentify which gluten peptides are recognized by glutenspecific T cells of patients (Vader et al., 2002b).

MODEL FOR DISEASEDEVELOPMENT

These results indicate that there are many immunogenicgluten peptides. Importantly, some of these can stimulate Tcells in their native form while the majority is only recog-nized after deamidation by tissue transglutaminase. T cellreactivity towards native gluten peptides in the intestinewould result in tissue damage due to the release of inflam-matory cytokines by these T cells. It is well established thattissue damage results in the release of cytoplasmic tissuetransglutaminase, which can subsequently modify glutenpeptides and generate of a whole series of potent T cellstimulatory peptides. This would attract more gluten spe-cific T cells into the small intestine, leading to more tissuedamage, more tissue transglutaminase, more T cell stimu-latory gluten peptides, etc. A vicious circle has been setin motion and oral tolerance is broken. CD is the result(Fig. 1).

While this model incorporates the currently availabledata, it does not explain why only a minor proportion of

HLA-DQ2 and/or -DQ8 positive individuals develop dis-ease. It is important to note that gluten specific T cellresponses may arise in all HLA-DQ2 and/or -DQ8 positiveindividuals since they are exposed to gluten and tissuetransglutaminase is present in the intestine. Apparently,some control these gluten specific responses better as others.It is unknown which factors are involved but there is nowoverwhelming evidence that additional genetic factors playa role in disease development (Papadopoulos et al., 2001).The influence of currently unknown environmental factorscannot be ruled out. It will be the challenge of future studiesto determine the nature of these factors.

PROSPECTS

We have gained insight into the nature of the gluten specificT cell response that causes CD. The characterization of thegluten peptides involved may allow novel approaches to thedevelopment of safer food products for CD patients, forexample by selective removal of such peptide sequencesfrom gluten proteins. The insight into the nature of thegluten-specific T cell response and the involvement of tissuetransglutaminase in the generation of the majority of T cellstimulatory gluten peptides offers new targets for interfer-ence in disease development and thus the potential foralternative treatment protocols for patients.

Acknowledgements

This work was supported by grants from the EC (BHM4-CT98-3087 andQLK1-2000-00657), The ‘Stimuleringsfonds Voedingsonderzoek LUMC’,and the Dutch Organization for Scientific Research (ZonMW grant 912-02-028). The author thanks Willemijn Vader for the design of Fig. 1.

Figure 1. Model for celiac disease development. For explanation see text.

MOLECULAR BASIS OF CELIAC DISEASE 335

Copyright # 2003 John Wiley & Sons, Ltd. J. Mol. Recognit. 2003; 16: 333–336

REFERENCES

Arentz-Hansen H, Korner R, Molberg Ø, Quarsten H, Vader W,Kooy YMC, Lundin KEA, Koning F, Roepstorff P, Sollid LM,McAdam S. 2000. The intestinal T cell response to �-gliadinin adult celiac disease is focused on a single deamidatedglutamine targeted by tissue transglutaminase. J. Exp. Med.191: 603–612.

Dieterich W, Ehnis T, Bauer M, Donner P, Volta U, Riecken EO,Schuppan D. 1997. Identification of tissue transglutaminaseas the autoantigen of celiac disease. Nat. Med. 3: 797–801.

Godkin A, Friede T, Davenport M, Stevanovic S, Willis A, JewellD, Hill A, Rammensee HG. 1997. Use of eluted peptidesequence data to identify the binding characteristics ofpeptides to the insulin-dependent diabtes susceptibility al-lele HLA-DQ8 (DQ3.2). Int. Immunol. 9: 905.

Kwok WW, Domeier ME, Raymond FC, Byers P, Nepom GT.1996. Allele-specific motifs characterize HLA-DQ interactionswith a diabetes-associated peptide derived from glutamicacid decarboxylase. J. Immunol. 156: 2171–2177.

Lundin KE, Scott H, Hansen T, Paulsen G, Halstensen TS,Fausa O, Thorsby E, Sollid LM. 1993. Gliadin-specific, HLA-DQ(�1*0501,�1*0201) restricted T cells isolated from thesmall intestinal mucosa of celiac disease patients. J. Exp.Med. 178: 187–196.

Lundin KEA, Sollid LM, Anthonson D, Noren O, Molberg Ø,Thorsby E, Sjostrom H. 1997. Heterogeneous reactivity pat-terns of HLA-DQ-restricted, small intestinal T cell clonesfrom patients with celiac disease.Gastroenterology 112: 752.

Marsh MN. 1992. Gluten, major histocompatibility complex, andthe small intestine. Gastroenterology 102: 330.

Molberg Ø, McAdam S, Korner R, Quarsten H, Kristiansen C,Madsen L, Fugger L, Scott H, Noren O, Roepstorff P,Lundin KEA, Sjostrom H, Sollid LM. 1998. Tissue transglu-taminase selectively modifies gliadin peptides that are re-cognized by gut derived T cells in celiac disease. Nat. Med. 4:713–717.

Papadopoulos G, Wijmenga C, Koning F. 2001. Interplay be-tween genetics and the environment in the development ofceliac disease: perspectives for a healthy life. J. Clin. Invest.108: 1261–1266.

Rammensee HG, Friede T, Stevanovic S. 1995. MHC ligands andpeptide motifs: first listing. Imunogenetics 41: 178–228.

Sjostrom H, Lundin KEA, Molberg Ø, Korner R, McAdam S,Anthonsen D, Quarsten H, Noren O, Roepstorff P,Thorsby E, Sollid LM. 1998. Identification of a gliadin T cell

epitope in coeliac disease: general importance of gliadindeamidation for intestinal T cell recognition. Scand. J. Im-mnol. 48: 111–115.

Sollid LM, Markussen G, Ek J. 1989. Evidence for a primaryassociation of coeliac disease to a particular HLA-DQ alpha/beta heterodimer. J. Exp. Med. 169: 345–350.

Spurkland A, Ingvarsson G, Falk ES, Knutsen I, Sollid LM,Thorsby E. 1997. Dermatitis herpetiformis and celiac diseaseare both primarily associated with the HLA-DQ(�1*0501,�1*02) or the HLA-DQ(�1*03, �1*0302) heterodimers. TissueAntigens 49: 29.

Tighe MR, Hall MA, Barbado M. 1992. HLA class II allelesassociated with celiac disease susceptibility in a southernEuropean population. Tissue Antigens 40: 90–97.

Vader W, Kooy Y, van Veelen P, de Ru A, Harris D, BenckhuijsenW, Pena S, Mearin L, Drijfhout JW, Koning F. 2002a. Thegluten response in children with recent onset celiac disease. Ahighly diverse response towards multiple gliadin and glute-nin derived peptides. Gastroenterology 122: 1729–1737.

Vader W, de Ru A, Wal Y, van de Kooy Y, Benckhuijsen W,Mearin L, Drijfhout JW, van Veelen P, Koning F. 2002b.Specificity of tissue transglutaminase explains cereal toxi-city in celiac disease. J. Exp. Med. 195: 643–649.

Vartdal F, Johansen BH, Friede T, Thorpe CJ, Stevanovic S,Eriksen JE, Sletten K, Thorsby E, Rammensee HG,Sollid LM. 1996. The peptide binding motif of the disease-associated HLA-DQ(�1*0501,�1*0201) molecule. Eur. J. Im-munol. 26: 2764–2774.

van de Wal Y, Kooy YMC, Drijfhout JW, Amons R, Koning F.1996. Peptide binding characteristics of the coeliac disease-associated DQ(�1*0501,�1*0201) molecule. Immunoge-netics 44: 246–253.

van de Wal Y, Kooy Y, Veelen van P, Pena S, Mearin L, Molberg Ø,Lundin L, Mutis T, Benckhuijsen W, Drijfhout JW, Koning F.1998a. Small intestinal cells of celiac disease patients recog-nize a natural pepsin fragment of gliadin. Proc. Natl Acad.Sci. USA 95: 10050–10054.

van de Wal Y, Kooy YMC, Veelen van P, Pena AS, Mearin ML,Papadopoulos GK, Koning F. 1998b. Selective deamidationby tissue transglutaminase strongly enhances gliadin-spe-cific T cell reactivity. J. Immunol. 161: 1585–1588.

van de Wal Y, Kooy YMC, Veelen van P, August SA, Drijfhout JW,Koning F. 2000. Glutenin is involved in the gluten-drivenmucosal T cell response. Eur. J. Immunol. 29: 3133–3139.

336 F. KONING

Copyright # 2003 John Wiley & Sons, Ltd. J. Mol. Recognit. 2003; 16: 333–336