PRIMARY PREVENTION OF COELIAC DISEASE THE UTOPIA OF THE NEW MILLENNIUM?

Transcription

1 Perspectives on Coeliac Disease Volume I PRIMARY PREVENTION OF COELIAC DISEASE THE UTOPIA OF THE NEW MILLENNIUM? EDITED BY CARLO CATASSI ALESSIO FASANO GINO ROBERTO CORAZZA AIC Press

2

3 PRIMARY PREVENTION OF COELIAC DISEASE THE UTOPIA OF THE NEW MILLENNIUM?

4 The AIC Meeting was held in the Aula Scarpa of the University of Pavia, Italy, October 12, Meeting participants (left to right, from back to front): P. Ciclitira, F. Koning, L. Sollid, R. Anderson, R. Troncone, M. Mäki, A. Fasano, G. Gasbarrini, F. Cucca, M. Stern, C. Feighery, G.R. Corazza, C. Catassi, P. Howdle, G. Holmes.

5

6

7 Perspectives on Coeliac Disease Volume 1 PRIMARY PREVENTION OF COELIAC DISEASE THE UTOPIA OF THE NEW MILLENNIUM? Editors Carlo Catassi Department of Pediatrics University of Ancona Ancona, Italy Center for Celiac Research, Baltimore, MD, USA Alessio Fasano Division of Pediatric Gastroenterology, Center for Celiac Research, University of Maryland Baltimore, MD, USA Gino Roberto Corazza Department of Gastroenterology, University of Pavia, Pavia, Italy AIC Press

8 Italian Coeliac Society, Via Picotti, Pisa, Italy ã2003 Italian Coeliac Society. All rights reserved. This book is protected by copyright. No part of it may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, or recording, or otherwise, without the written permission of the publisher. Made in Italy The material contained in this volume was submitted as previously unpublished material, except in the instances in which credit has been given to the source from which some of the illustrative material was derived. Great care has been taken to maintain the accuracy of the information contained in the volume. However, neither the AIC nor the Editors can be held responsible for errors or for any consequences arising from the use of the information contained herein.

9 Foreward It is a great honour for me to introduce the Proceedings of the Meeting on Coeliac Disease that was held in Pavia on October 12, First of all, I would like to thank and to remind the memory of Mr. Sergio Spinelli, a man who strongly wanted and appreciated this meeting during its organization, who unfortunately passed just a few months before the event took place. I would also like to thank Prof. Corazza and Dr. Catassi for their hard work and collaboration during the organization of the scientific programme and for selecting the most important scientists who participated as speakers and chairmen. Finally, I cannot forget Mrs. Marina Marengo, Mrs. Katia Pilo and the whole editorial staff, especially Mr. Franco Lucchesi and Mrs. Giusy Cappellotto, for professionally organizing the whole meeting and preparing all the informative material, including this book. The Meeting held in Pavia was one of the most important scientific events on coeliac disease that have ever been organized, not only in Italy but also in Europe. The Italian Coeliac Society (AIC) has always given a strong support to research and felt it was time to organize a meeting for giving an up to date on the most important recent findings on coeliac disease, both in Italy and other countries. The aim of doing this was twofold: (a) to evaluate the level of the current knowledge on the main aspects of coeliac disease by comparing different scientific working groups and (b) to identify the possible topics on which the research should be focused on in the next future. Should these goals be reached, the life of the coeliacs will be easier wherever. I think that the quality level of both the speakers and the audience allowed to entirely achieve these aims. Adriano Pucci President of the AIC viii

10

11 Preface Coeliac disease (CD) is a unique example in medicine of a genetically-based, immune-mediated condition entirely curable. Treatment with the gluten-free diet (GFD) is indeed followed by a full clinical and histological recovery (the true restitutio ad integrum), with the patient shifting from a diseased state to a normal condition of life. In typical cases, the response to treatment is sometimes prodigious, so that a severely malnourished and miserable patient is rapidly transformed into a sturdy and healthy subject. Even in atypical cases, such as those presenting with anaemia, behavioural changes, or infertility, the diet treatment may miraculously be effective. However, there is a price to pay for this success. For many populations glutencontaining foods make a substantial contribution to daily energy intake and are enjoyable to eat. Bread had, and still has, such a basic role in human nutrition to rise to a symbolic value in the Holy Communion. The changes needed to begin and maintain a GFD are substantial and have a major impact on daily life. It is therefore not surprising that patients ask if a cure is on the horizon, wondering whether they will ever be able to tolerate a normal diet. Recently, some gliadin peptides that seem to have a primary role in activating the cascade of immune responses leading to the celiac enteropathy were characterized by different groups worldwide. These findings echoed in the lay media and were enthusiastically interpreted as a clear-cut step toward the development of a coeliac vaccine. Is this science fiction or a realistic hope? In the year 2000 the Board of the Italian Coeliac Society (AIC) ripened the concept that it was time to give members a correct information on this hot topic. Therefore, the AIC Scientific Committee organized an innovative conference on primary prevention of CD, having two basic ideas in mind: (1) invite all the experts working in the field (in order to hear the different opinions); (2) translate the scientific language of the researchers into plain information that could be understood by the lay members of the Society. On October 12, 2001 a high-level scientific conference took place in Pavia, Italy, in the extraordinary scenario of the old Aula Scarpa of the University. On October 13, 2001 the technical concepts were summarized to a large lay audience in a meeting in Milan and society members had plenty of time for asking questions to the international experts. This book contains all the lectures presented at the Pavia meeting in order to give an up-to-date scenario on the recent developments on treatment strategies alternative to the GFD. In the first part of the book leading experts in the field report their results on the identification of toxic gluten peptides. In the second part, the role of environmental xi

12 xii risk factors that can trigger the onset of CD, particularly infant feeding, is discussed in detail. Finally, new strategies of treatment, based on either the introduction of genetically detoxified grains or induction of oral tolerance to gluten, are presented. The reader will appreciate that the holy grail for a possible coeliac cure is still far away. Biochemically speaking, the complexity of gluten toxicity is disarming. Furthermore, the lack of an animal model of disease greatly hampers the investigations on the CD pathophysiology at the mucosal level. In spite of these problems, the route has been traced and primary prevention of CD should not be considered an utopia anymore at the beginning of the third millennium. We hope that this volume will stimulate researchers and clinicians alike to pursue this goal. We express our gratitude to all speakers and chairmen at the Pavia meeting, and to the academic authorities of the University of Pavia. We gratefully acknowledge the support of the AIC National Board, particularly of Ing. Spinelli, who was one of the promoters of this meeting that unfortunately passed away. The publication of this volume would not have been possible without the expertise of the AIC Editorial Board, particularly of Dr. Franco Lucchesi and Ms. Giusy Cappellotto. We also thank Dr. Elisabetta Fabiani for her valuable help with several aspects of this book. C. Catassi A. Fasano G.R. Corazza

13 Contents Current understanding of the basis for coeliac disease... 1 Ludvig M Sollid The identification of toxic T cell stimulatory gluten response peptides early in coeliac disease Frits Koning Toxic gluten peptides in coeliac disease identified by in vivo gluten challenge: a single dominant T cell epitope? Robert P Anderson Role of A-gliadin peptide Paul J Ciclitira The association of the HLA-DQ molecules with coeliac disease in the Saharawi: an evolutionary perspective on coeliac disease Francesco Cucca, Carlo Catassi Primary prevention of coeliac disease by favourable infant feeding practices Anneli Ivarsson, Lars Åke Persson, Olle Hernell Mechanisms of oral tolerance: lessons for coeliac disease? Conleth Feighery Genetically detoxified grains in coeliac disease Federico Biagi, Antonio Di Sabatino, Jonia Campanella, Gino Roberto Corazza Immunotherapy of coeliac disease: where do we stand? Carmen Gianfrani, Mauro Rossi, Giuseppe Mazzarella, Francesco Maurano, VirginiaSalvati, Delia Zanzi, Salvatore Auricchio, Riccardo Troncone The most recent advances on gluten toxicity in coeliac disease Gino Roberto Corazza, Carlo Catassi, Alessio Fasano xiii

14

15 Catassi C, Fasano A, Corazza GR (eds): Primary prevention of coeliac disease. The utopia of the new millennium? Perspectives on Coeliac Disease, vol. 1, AIC Press, pp 1-11 Current understanding of the basis for coeliac disease Ludvig M. Sollid Institute of Immunology, University of Oslo, Rikshospitalet, N-0027 Oslo, Norway Introduction Coeliac disease (CD) has received increased attention in recent years. The disease is more common than previously thought with a prevalence of about 1:130-1:300 in 1 Western societies. It is an acquired disorder; it may be diagnosed in early childhood with classical symptoms like diarrhoea and malabsorption, but it may also be diagnosed 1 later in life often with symptoms that do not directly allude to a gut disease. CD develops because of intolerance to ingested wheat gluten (consisting of the subcomponents gliadin and glutenin) or related proteins from rye and barley. There is a 2 chronic inflammation in the small intestine with resultant flattening of the mucosa. The current treatment is a life-long gluten exclusion diet which often impairs the quality of life of those who are affected. For this reason many coeliacs ask for novel treatment modalities and methods to prevent the disease. CD belongs to the group of chronic inflammatory diseases with multifactorial aetiology where genetic and environmental components are involved. Among these disorders CD stands out as a particularly good model. This paper will briefly cover some of the recent advances in the understanding of this disorder. Genetic factors 3 A high prevalence (10%) among first degree relatives of CD patients and a high 4 concordance rate of % in monozygotic twins indicate that susceptibility to develop CD is strongly influenced by inherited (genetic) factors. Both HLA and non- HLA genes contribute to the genetic predisposition, and assuming a multiplicative model of disease genetics it has been estimated that the overall importance of non-hla 5-6 genes is greater than that of HLA genes. These figures should be interpreted with caution, however, as increased sharing of environmental factors by the sibs would tend to overestimate the role of the non-hla genes. The majority of CD patients carry the DRB1*0301-DQA1*0501-DQB1*0201 1

16 2 THE BASIS OF COELIAC DISEASE haplotype (the DR3, DQ2 haplotype) or are DRB1*11/12-DQA1*0505- DQB1*0301/DRB1*07-DQA1*0201-DQB1*0202 heterozygotes (carry the DR DQ7/DR7-DQ2 haplotypes). The chains encoded by DQA1*0501 and DQA1*0505 differ by one residue in the leader peptide, and the DQbchains encoded by DQB1*0201 and DQB1*0202 differ by one residue in the membrane proximal domain. These substitutions are highly unlikely to have any functional consequence. CD patients with the above mentioned DR-DQ combinations share the same functional DQ molecule on the cell surface, encoded by genes carried in cis (e.g. DQA1*05 and DQB1*02 carried on the same haplotype) or trans position (e.g. DQA1*05 carried on a different 12 haplotype to DQB1*02). Accumulating evidence suggests that the DR3-DQ2, the DR7-DQ2 and the DR5-DQ7 haplotypes have a close evolutionary relationship. Fragments of DNA flanking the DQA1 gene of the DR3-DQ2 haplotype have been identified on the DR5-DQ7 haplotype, and fragments of DNA flanking the DQB1 gene of the DR3-DQ2 haplotype have been identified on the DR7-DQ2 haplotype. The genetic information in the DQ subregion of the DR3-DQ2 haplotype is thus reestablished in DR5-DQ7/DR7-DQ2 heterozygotes, although the sequence information is split between two chromosomes. Susceptibility to CD therefore likely depends on an interaction between at least two genes on the DR3-DQ2 haplotype that are reunited in DR5-DQ7 / DR7-DQ2 heterozygous individuals. The DQA1 and DQB1 genes are the primordial candidates since their products interact to form an HLA class II heterodimer and since they are situated close to a putative recombination site. Almost all the CD patients who are DQA1*05 and DQB1*02 negative bear the DRB1*04, DQA1*03, DQB1*0302 haplotype (i.e. DR4-DQ8 haplotype) and it is likely that these patients have an HLA association which is different to those who are DQ2 positive. Although it is less clear what the primary disease susceptibility determinant of 15 the DR4-DQ8 haplotype is, most data favour DQ8. Overall the existing data suggest that the susceptibility to develop CD is primarily associated with two conventional DQ molecules DQ(a1*05,b1*02) (=DQ2) and to a lesser extent DQ(a1*03,b1*0302) (=DQ8). DQ molecules bind peptides and present these to CD4+ T helper cells carrying the abt cell receptor (TCR). The genetic evidence thus points towards a central role of CD4+ T cells in controlling the disease development in CD. Genome wide linkage studies in CD have indicated numerous susceptibility regions with weak genetic effects, and the indications are strongest for susceptibility genes located at 5qter and 11qter. However, no disease associations have been established so far for any gene of these regions. The only gene for which there are relatively consistent reports on disease association is the CTLA4 gene on chromosome 2q33. CTLA4 is involved in down regulating T-cell responses, and the A allele of the +49 dimorphism is associated with increased CTLA4 expression and enhanced control of T cell proliferation. Thus the +49 dimorphism is a prime suspect for the observed genetic effect, although other polymorphisms in linkage disequilibrium cannot be ruled out. In this respect it is worth noting that the CD28 and ICOS genes whose gene products are central players in T and B cell activation, are located very close to the CTLA4 gene.

17 THE BASIS OF COELIAC DISEASE 3 Environmental factors Gluten is obviously a critical environmental factor in CD. Whether other environmental factors are also involved is still an open question. Gut infections may well be involved. Adenovirus 12 has been in the forefront among the candidate microorganisms. This virus was originally proposed as a candidate because of partial 23 linear homology over 12 amino acids in a virus protein and a-gliadin. Based on our current understanding of the pathogenesis of coeliac disease, the rationale for the candidacy of this virus is weak, and there is in fact very little epidemiological data 24 supporting a role for this virus. Peptide binding to the CD associated DQ2 and DQ8 molecules Both HLA class I and class II molecules bind peptides in a groove located in their 25 membrane distal part. Stable binding is achieved by multiple hydrogen bonds between amino acids of HLA and peptide main chain atoms. Many polymorphic variants of HLA molecules exist. Amino acid residues which differ between the polymorphic variants are clustered around the peptide binding site where they contribute to the formation of specific binding pockets. Side chains of amino acids of the peptide (so-called anchor residues) fit into these pockets and their interaction with HLA contribute to the binding of the peptide. The binding site of HLA class II molecules, in contrast to HLA class I molecules, is open at both ends allowing the bound peptides to protrude. The class II peptide ligands thus vary in length. The interactions with HLA mainly take place in a core region of nine residues. Within this region side chains of amino acids in positions P1, P4, P6, P7 and P9 dock into pockets of the class II binding site. The chemistry and size of the various pockets vary between the different class II alleles so that some amino acids are preferred and some are not preferred. DQ2 has a unique preference for binding peptides with negatively charged side chains at the three middle anchor positions. The binding motif of DQ8 is different from that of DQ2, but DQ8 also displays a preference for binding negatively charged residues at several positions (i.e. P1, P4 and P9). Hence, both the DQ2 and DQ8 molecules share a preference for negatively charged residues at some of their anchor positions. Preferential T cell recognition of gluten peptides presented by DQ2 and Dq8 The observation that gluten reactive CD4+ TCRabT cells can be isolated and propagated from intestinal biopsies of CD patients has been instrumental for recent 32 achievements. Strikingly, such T cells of patients carrying the DR3-DQ2 haplotype were found to recognise gluten fragments presented by the DQ2 molecule rather than by other HLA molecules of the patients. Both DR3-DQ2 positive and DR5- DQ7/DR7-DQ2 positive antigen presenting cells (i.e. carrying the DQA1*05 and DQB1*02 genes in cis or in trans configuration) are able to present the gluten antigen to these patient T cells. Likewise, T cells isolated from small intestinal biopsies of DQ2 negative, DR4-DQ8 positive patients predominantly recognise gluten-derived peptides when presented by the DQ8 molecule. Taken together, these results allude to

18 4 THE BASIS OF COELIAC DISEASE 1 presentation of gluten peptides in the small intestine as the mechanism by which DQ2 1 and DQ8 confer susceptibility to CD. HLA molecules are also important for determining the repertoire of peripheral T cells during maturation in the thymus. A thymic effect of the same DQ molecules on the TCR repertoire selection is, however, not excluded by these results. Importance of gluten deamidation for T cell recognition Wheat gluten is a mixture of a large number of gliadin and glutenin polypeptides. Generally, gluten proteins are rich in proline and glutamine residues while many other amino acids, including glutamic and aspartic acid, are unusually rare. Proteins of the gliadin fraction can be subdivided according to their sequence into the a-, g-, and w- 34 gliadins. Initially it was difficult to reconcile the DQ2 (and DQ8) binding motifs with presentation of gluten peptides, as gluten proteins have an unusual scarcity of negatively charged residues. A clue to help explain this paradox came from the observation that the stimulatory capacity of gliadin preparations for gliadin specific intestinal T cells was significantly enhanced following treatment at high temperatures 35 and low ph. These conditions are known to cause non-specific deamidation of glutamines to glutamic acid and may thus convert gliadin from a protein with very few peptides with the potential to bind to DQ2/DQ8 into one with many. An important and general role for deamidation of gluten for T cell recognition was sustained by analysis of the response pattern of a panel of polyclonal, gliadin specific T cell lines derived 36 from biopsies. All the cell lines responded poorly to a gliadin antigen prepared under conditions of minimal deamidation (chymotrypsin-digestion), when compared to the same antigen that had been further heat-treated in an acidic environment. The characterisation of gluten epitopes recognised by intestinal T cells has extended the knowledge about the importance of deamidation for their T cell recognition. Of the epitopes characterised until now, most of them (DQ2-g-gliadin-I, DQ2-a-gliadin-I and DQ2-a-gliadin-II) fail to stimulate T cells in their native form, but are potent antigens when a single glutamine residue is exchanged with glutamic acid in certain positions. For one DQ8 restricted epitope (DQ8-a-gliadin-I), the T cell recognition is augmented 37 by introduction of negatively charged residues whereas this is not seen for another 38 DQ8 restricted epitope (DQ8-glutenin-I). These data demonstrate that most, but not all gluten specific intestinal T cells from CD patients recognise gluten proteins only after they have undergone deamidation. Tissue transglutaminase deamidates gluten peptides in vivo There is accumulating evidence that the deamidation in vivo is mediated by the enzyme tissue transglutaminase (ttg). ttg is expressed in many different tissues and organs; in the small intestine it is mainly expressed just beneath the epithelium in 39 the gut wall. The activity of ttg in the small intestinal mucosa in CD patients with 41 untreated disease is elevated compared to controls. The enzyme is present both intracellularly and extracellularly, and in the extracellular environment ttg has been demonstrated to play a role in extracellular matrix assembly, cell adhesion and wound

19 THE BASIS OF COELIAC DISEASE 5 42 healing. The calcium dependent transglutaminase activity of ttg catalyses selective 43 crosslinking or deamidation of protein-bound glutamine residues. In contrast to the non-enzymatically mediated deamidation that results in a near random deamidation of the often numerous glutamine residues in gliadin peptides, ttg appears to carry out an ordered deamidation of some few specific glutamines. In all of the known major DQ2 and DQ8 restricted gluten epitopes recognised by gut T cell of adult patients, there are glutamic acid residues modified by ttg which is important for T cell recognition. Interestingly, the deamidation of glutamine residues that are not targeted by ttg (e.g by acid treatment) can be deleterious for T cell recognition. This suggests that deamidation in vivo is mediated by ttg. This idea is further supported by the results of experiments where T cell lines have been established from biopsies challenged with a minimally deamidated gliadin antigen (chymotrypsin-digested). In all but one of 18 adult patients, the established T cell lines only barely responded to the chymotrypsindigested gliadins, but efficiently recognized the in vitro ttg-treated variants of the 46 same gliadins. Normally we do not mount immune responses to edible proteins. Moreover, experimental animal models have demonstrated that oral administration of antigen 47 usually results in systemic hyporesponsiveness to the same antigen. This phenomenon, which is termed oral tolerance, is believed to occur because of active tolerization towards edible proteins. In keeping with this thinking, oral tolerance to gluten in CD patients may not have been established properly or is broken. Given the preferential intestinal T cell response to deamidated gluten fragments in CD patients, it may be that deamidation is involved in the perturbation of the oral gluten tolerance. Deamidation increases the binding affinity of gliadin peptides for DQ2 from poor but 36- significant binders, to epitopes with reasonable, but by no means exceptional affinity 44. The moderate binding affinity of these epitopes concurs with the finding that they do not carry optimal anchors in all the anchor positions. Interestingly, T cell clones specific for the DQ2-g-gliadin-I, DQ2-a-gliadin-I and DQ2-a-gliadin-II epitopes generally fail to recognise native peptides at higher concentrations that should compensate for their lower binding affinity for DQ2. Thus, concurrent with the increase in the binding affinity for DQ2 caused by deamidation of the gliadin peptide, there likely is a change in the conformation1of the gliadin/dq2 complexes. Oral tolerance to antigens ingested in 47 high doses is usually established by T cell anergy or deletion. It is possible that T cells specific for native gluten sequences are usually anergised or deleted, and that phlogistic T cell responses are effectively mounted to novel gluten epitopes being created in an inflamed environment by the help of ttg. How many gluten T cell epitopes? There exist several epitopes in gluten that are recognized by small intestinal T cells 35 of CD patients. Recent results of the author's laboratory and the laboratory of Frits Koning in Leiden indicate that there may be more than ten distinct DQ2 restricted epitopes. The existence of multiple epitopes raises several interesting questions: are only some of the epitopes pathogenic and thereby relevant to explain the HLA association? Are responses towards some of the epitopes generated during the early phases of disease development, while the responses to others are a result of epitope

20 6 THE BASIS OF COELIAC DISEASE spreading? Are different epitopes recognized by distinct groups of patients (e.g. children vs. adults)? Are some epitopes more relevant to disease as responses to them are found in the majority of the patients or because there is a higher precursor frequency of T cells in the lesion specific for these epitopes? The answers to most of these questions must await further investigations. At present we know that for the DQ2-agliadin-I and DQ2-a-gliadin-II epitopes, intestinal T cell reactivity is found in most if 44 not all adult DQ2+ patients, whereas for the DQ2-g-gliadin-I epitope intestinal T cell 36 reactivity is found in fewer DQ2+ patients. Less is known about the DQ8 restricted epitopes because few DQ8 positive patients have been tested so far. However, the DQ8-37 a-gliadin-i appears to be frequently recognized. What causes the variance in responsiveness to the different epitopes and whether this reflects qualitative or quantitative differences between the patients are presently unclear. Formation of the coeliac lesion The evidence discussed above provides strong evidence that CD4+ TCRab+ T cells in the lamina propria are central for controlling the immune response to gluten that produces the immunopathology of CD. The knowledge of the events down-stream of T cell activation is, however, still incomplete. Knowing how the immune system usually utilizes a multitude of effector mechanisms for fighting its opponents, it is reasonable to believe that there may well be multiple effector mechanisms involved in the creation of the coeliac lesion. Adding to the complexity, recent in vitro organ culture studies have indicated that gluten exerts additional immune relevant effects that are independent of T cell activation. Some of these effects have rapid kinetics and conceivably the direct effects of gluten may facilitate subsequent T cell responses. Cytokines produced by lamina propria Cd4+ T cells may be involved in the increased crypt cell proliferation and the increased loss of epithelial cells. IFN-g induces macrophages to produce TNF-a. TNF- activates stromal cells to produce 50 KGF, and KGF causes epithelial proliferation and crypt cell hyperplasia. IFN-gand 51 TNF-acan jointly have a direct cytotoxic effect on intestinal epithelial cells. It is also conceivable that IELs and in particular gdt cells play a role in the epithelial cell 52 destruction by recognizing MIC molecules induced by stress. Alterations of the extracellular matrix can also distort the epithelial arrangement as the extracellular matrix provides the scaffold on which the epithelium lies. Enterocytes adhere to basement membrane through extracellular matrix receptors so that modification or loss of the basement membrane can result in enterocyte shedding. There is evidence for increased extracellular matrix degeneration in CD, and this may 53 be an important mechanism for the mucosal transformation found in CD. The increased production of metalloproteinases by subepithelial fibroblasts and macrophages is likely to be directly or indirectly induced by cytokines that are released from activated T cells. Coeliac patients on a gluten containing diet have increased levels of serum antibodies to a variety of antigens including gluten and to ttg. We do not as yet know whether the autoantibodies play a role in the pathogenesis of CD. The ttg antibodies can inhibit the activity of ttg. This can cause villous atrophy by blocking

21 THE BASIS OF COELIAC DISEASE 7 interactions between mesenchymal cells and epithelial cells during the migration of 56 epithelial cells and fibroblasts from the crypts to the tips of the villi. Moreover, the ttg antibodies may modulate the deamidating activity of ttg either in an inhibiting or 57 promoting fashion. Further research will tell us what role these antibodies play. Translation of the new knowledge into therapy The increasing insight into the molecular and cellular basis of CD should give benefits to the patients. The knowledge on which gluten epitopes are recognized by gut T cells should allow the methods by which gluten free foods are assessed to be improved. Moreover, the new knowledge should uncover novel targets for therapy. There are already some attractive possibilities. Activation of CD4+ gluten specific T cells appears to be a critical checkpoint in the development of CD, and interference with this step in the pathogenesis should be an effective way to control the disease. One possibility, which is basically an extension of today's treatment with a gluten free diet, is to produce wheat that is devoid of T cell epitopes, either by breeding programs or transgenic technology. Success by use of classical breeding already seems unlikely as epitopes are found in both a-gliadins and g-gliadins which are encoded by the Gli-1 and Gli-2 loci located on chromosomes 1 and 6, respectively. The classical breeding approach is further complicated by the hexaploid nature of wheat. ttg is a target for intervention because of its critical role in generating gluten T cell epitopes. Inhibitors of ttg activity exist and likely inhibitors suitable as drugs can be developed. The biggest problem with this approach is that ttg inhibitors may have unacceptable side effects. ttg is involved in many different physiological processes including programmed cell 42 death. Another strategy would be to aim at the gluten specific T cells. If coeliacs have a normal oral tolerance to native gluten proteins, but a broken tolerance to deamidated gluten peptides, exposing the gut immune system to already deamidated peptides may establish oral tolerance to these deamidated gluten peptides as well. This approach will utilize the body's own mechanism to silence T cells. Alternatively, one could try to directly silence gluten specific T cell by using soluble dimers of HLA/ peptide complexes, which have been demonstrated to induce antigen specific apoptosis 58 because of inappropriate T cell stimulation. The central role of DQ2 and DQ8 in presenting gluten peptides offers yet another target for intervention. Blocking the binding-sites of these HLA molecules would prevent presentation of disease inducing gluten peptides. The challenge with this approach will be to find an efficient way to target and block the binding sites of DQ molecules, which are continuously synthesized by antigen presenting cells. This approach, blocking of peptide presentation, has also been suggested as therapy for other HLA associated diseases. CD should be better suited to this approach than many other HLA associated diseases because drug delivery in the gut is easy compared for example to joints in rheumatoid arthritis or islet cells in type 1 diabetes. Whatever new therapeutic modality is introduced in CD, it will have to prove better than the current gluten free diet regime, also when coming to its long-term safety. This fact must be taken into consideration when devising new treatments. Although there are already interesting therapeutic principles with a good, rational basis that can be tested, it may for this reason take some years before a new treatment becomes reality.

22 8 THE BASIS OF COELIAC DISEASE Acknowledgements Studies in the author's laboratory are funded by grants from the Research Council of Norway, the European Commission (BMH4-CT , QLRT , QLRT ), Medinnova, the Jahre Foundation and EXTRA funds from the Norwegian Foundation for Health and Rehabilitation. References 1. Fasano A, Catassi C. Current approaches to diagnosis and treatment of celiac disease: an evolving spectrum. Gastroenterology 2001; 120: Sollid LM. The molecular basis of celiac disease. Annu Rev Immunol 2000; 18: Ellis A. Coeliac disease: previous family studies. In: McConnell RB, editor. The genetics of coeliac disease. Lancaster: MTP press, 1981: Polanco I, Biemond I, van Leeuwen A, Schreuder I, Meera Khan P, Guerrero J et al. Gluten sensitive enteropathy in Spain: Genetic and environmental factors. In: McConnell RB, editor. The genetics of coeliac disease. Lancaster: MTP Press, 1981: Risch N. Assessing the role of HLA-linked and unlinked determinants of disease. Am J Hum Genet 1987; 40: Petronzelli F, Bonamico M, Ferrante P, Grillo R, Mora B, Mariani P et al. Genetic contribution of the HLA region to the familial clustering of coeliac disease. Ann Hum Genet 1997; 61: Keuning JJ, Pena AS, van Leeuwen A, van Hooff JP, van Rood JJ. HLA-DW3 associated with coeliac disease. Lancet 1976; i: Ek J, Albrechtsen D, Solheim BG, Thorsby E. Strong association between the HLA- Dw3-related B cell alloantigen -DRw3 and coeliac disease. Scand J Gastroenterol 1978; 13: Mearin ML, Biemond I, Pena AS, Polanco I, Vazquez C, Schreuder GT et al. HLA- DR phenotypes in Spanish coeliac children: their contribution to the understanding of the genetics of the disease. Gut 1983; 24: Trabace S, Giunta A, Rosso M, Marzorati D, Cascino I, Tettamanti A et al. HLA- ABC and DR antigens in celiac disease. A study in a pediatric Italian population. Vox Sang 1984; 46: Tosi R, Vismara D, Tanigaki N, Ferrara GB, Cicimarra F, Buffolano W et al. Evidence that celiac disease is primarily associated with a DC locus allelic specificity. Clin Immunol Immunopathol 1983; 28: Sollid LM, Markussen G, Ek J, Gjerde H, Vartdal F, Thorsby E. Evidence for a primary association of celiac disease to a particular HLA-DQ a/bheterodimer. J Exp Med 1989; 169: Lin L, Jin L, Kimura A, Carrington M, Mignot E. DQ microsatellite association studies in three ethnic groups. Tissue Antigens 1997; 50: Lin L, Jin L, Lin X, Voros A, Underhill P, Mignot E. Microsatellite single nucleotide polymorphisms in the HLA-DQ region. Tissue Antigens 1998; 52: Sollid LM, Thorsby E. HLA susceptibility genes in celiac disease: genetic mapping

23 THE BASIS OF COELIAC DISEASE 9 and role in pathogenesis. Gastroenterology 1993; 105: Greco L, Corazza G, Babron MC, Clot F, Fulchignoni-Lataud MC, Percopo S et al. Genome search in celiac disease. Am J Hum Genet 1998; 62: Greco L, Babron MC, Corazza GR, Percopo S, Sica R, Clot F et al. Existence of a genetic risk factor on chromosome 5q in Italian coeliac disease families. Ann Hum Genet 2001; 65: Djilali-Saiah I, Schmitz J, Harfouch-Hammoud E, Mougenot JF, Bach JF, Caillat- Zucman S. CTLA-4 gene polymorphism is associated with predisposition to coeliac disease. Gut 1998; 43: Naluai AT, Nilsson S, Samuelsson L, Gudjonsdottir AH, Ascher H, Ek J et al. The CTLA4/CD28 gene region on chromosome 2q33 confers susceptibility to celiac disease in a way possibly distinct from that of type 1 diabetes and other chronic inflammatory disorders. Tissue Antigens 2000; 56: Holopainen P, Arvas M, Sistonen P, Mustalahti K, Collin P, Maki M et al. CD28/CTLA4 gene region on chromosome 2q33 confers genetic susceptibility to celiac disease. A linkage and family-based association study. Tissue Antigens 1999; 53: Kouki T, Sawai Y, Gardine CA, Fisfalen ME, Alegre ML, DeGroot LJ. CTLA-4 gene polymorphism at position 49 in exon 1 reduces the inhibitory function of CTLA-4 and contributes to the pathogenesis of Graves' disease. J Immunol 2000; 165: Ligers A, Teleshova N, Masterman T, Huang WX, Hillert J. CTLA-4 gene expression is influenced by promoter and exon 1 polymorphisms. Genes Immun 2001; 2: Kagnoff MF, Austin RK, Hubert JJ, Bernardin JE, Kasarda DD. Possible role for a human adenovirus in the pathogenesis of celiac disease. J Exp Med 1984; 160: Mahon J, Blair GE, Wood GM, Scott BB, Losowsky MS, Howdle PD. Is persistent adenovirus 12 infection involved in coeliac disease? A search for viral DNA using the polymerase chain reaction. Gut 1991; 32: Madden DR. The three-dimensional structure of peptide-mhc complexes. Annu Rev Immunol 1995; 13: Johansen BH, Vartdal F, Eriksen JA, Thorsby E, Sollid LM. Identification of a putative motif for binding of peptides to HLA-DQ2. Int Immunol 1996; 8: van de Wal Y, Kooy YMC, Drijfhout JW, Amons R, Koning F. Peptide binding characteristics of the coeliac disease-associated DQa1*0501, b1*0201. molecule. Immunogenetics 1996; 44: Vartdal F, Johansen BH, Friede T, Thorpe C, Stevanovic S, Eriksen JA et al. The peptide binding motif of the disease associated HLA-DQa1*0501, b1*0201. molecule. Eur J Immunol 1996; 26: Godkin A, Friede T, Davenport M, Stevanovic S, Willis A, Jewell D et al. Use of eluted peptide sequence data to identify the binding characteristics of peptides to the insulin-dependent diabetes susceptibility allele HLA-DQ8 (DQ 3.2). Int Immunol 1997; 9: Kwok WW, Domeier ML, Raymond FC, Byers P, Nepom GT. Allele-specific motifs characterize HLA-DQ interactions with a diabetes-associated peptide derived from

24 10 THE BASIS OF COELIAC DISEASE glutamic acid decarboxylase. J Immunol 1996; 156: Lee KH, Wucherpfennig KW, Wiley DC. Structure of a human insulin peptide-hla- DQ8 complex and susceptibility to type 1 diabetes. Nat Immunol 2001; 2: Lundin KEA, Scott H, Hansen T, Paulsen G, Halstensen TS, Fausa O et al. Gliadinspecific, HLA-DQa1*0501,b1*0201. restricted T cells isolated from the small intestinal mucosa of celiac disease patients. J Exp Med 1993; 178: Lundin KEA, Gjertsen HA, Scott H, Sollid LM, Thorsby E. Function of DQ2 and DQ8 as HLA susceptibility molecules in celiac disease. Hum Immunol 1994; 41: Shewry PR, Tatham AS, Kasarda DD. Cereal proteins and coeliac disease. In: Marsh M, editor. Coeliac disease. Oxford: Blackwell Scientific Publications, 1992: Lundin KEA, Sollid LM, Norén O, Anthonsen D, Molberg Ø, Thorsby E et al. Heterogenous reactivity patterns of HLA-DQ-restricted small intestinal T-cell clones from patients with celiac disease. Gastroenterology 1997; 112: Sjöström H, Lundin KEA, Molberg Ø, Körner R, McAdam SN, Anthonsen D et al. Identification of a gliadin T-cell epitope in coeliac disease: general importance of gliadin deamidation for intestinal T-cell recognition. Scand J Immunol 1998; 48: van de Wal Y, Kooy YM, van Veelen PA, Peña SA, Mearin LM, Molberg Ø et al. Small intestinal T cells of celiac disease patients recognize a natural pepsin fragment of gliadin. Proc Natl Acad Sci U S A 1998; 95: van de Wal Y, Kooy YM, van Veelen P, Vader W, August SA, Drijfhout JW et al. Glutenin is involved in the gluten-driven mucosal T cell response. Eur J Immunol 1999; 29: Molberg Ø, McAdam SN, Körner R, Quarsten H, Kristiansen C, Madsen L et al. Tissue transglutaminase selectively modifies gliadin peptides that are recognized by gut-derived T cells. Nat Med 1998; 4: van de Wal Y, Kooy Y, van Veelen P, Peña S, Mearin L, Papadopoulos G et al. Selective deamidation by tissue transglutaminase strongly enhances gliadinspecific T cell reactivity. J Immunol 1998; 161: Bruce SE, Bjarnason I, Peters TJ. Human jejunal transglutaminase: demonstration of activity, enzyme kinetics and substrate specificity with special relation to gliadin and coeliac disease. Clin Sci 1985; 68: Aeschlimann D, Paulsson M. Transglutaminases: protein cross-linking enzymes in tissues and body fluids. Thromb Haemost 1994; 71: Folk JE. Mechanism and basis for specificity of transglutaminase-catalyzed e-g- (glutamyl) lysine bond formation. Adv Enzymol Relat Areas Mol Biol 1983; 54: Arentz-Hansen H, Korner R, Molberg O, Quarsten H, Vader W, Kooy YM et al. The intestinal T cell response to a-gliadin in adult celiac disease is focused on a single deamidated glutamine targeted by tissue transglutaminase. J Exp Med 2000; 191: Quarsten H, Molberg Ø, Fugger L, McAdam SN, Sollid LM. HLA binding and T cell recognition of a tissue transglutaminase - modified gliadin epitope. Eur J Immunol 1999; 29:

25 THE BASIS OF COELIAC DISEASE Molberg Ø, McAdam S, Lundin KEA, Kristiansen C, Arentz-Hansen H, Kett K et al. T cells from celiac disease lesions recognize gliadin epitopes deamidated in situ by endogenous tissue transglutaminase. Eur J Immunol 2001; 31: Komagata Y, Weiner HL. Oral tolerance. Rev Immunogenet 2000; 2: Maiuri L, Picarelli A, Boirivant M, Coletta S, Mazzilli MC, De Vincenzi M et al. Definition of the initial immunologic modifications upon in vitro gliadin challenge in the small intestine of celiac patients. Gastroenterology 1996; 110: Maiuri L, Auricchio S, Coletta S, De Marco G, Picarelli A, Di Tola M et al. Blockage of T-cell costimulation inhibits T-cell action in celiac disease. Gastroenterology 1998; 115: Bajaj-Elliott M, Poulsom R, Pender SL, Wathen NC, MacDonald TT. Interactions between stromal cell--derived keratinocyte growth factor and epithelial transforming growth factor in immune-mediated crypt cell hyperplasia. J Clin Invest 1998; 102: Deem RL, Shanahan F, Targan SR. Triggered human mucosal T cells release tumour necrosis factor - alpha and interferon-gamma which kill human colonic epithelial cells. Clin Exp Immunol 1991; 83: Groh V, Steinle A, Bauer S, Spies T. Recognition of stress-induced MHC molecules by intestinal epithelial gdt cells. Science 1998; 279: Daum S, Bauer U, Foss HD, Schuppan D, Stein H, Riecken EO et al. Increased expression of mrna for matrix metalloproteinases-1 and -3 and tissue inhibitor of metalloproteinases-1 in intestinal biopsy specimens from patients with coeliac disease. Gut 1999; 44: Mäki M. The humoral immune system in coeliac disease. In: Howdle PD, editor. Coeliac disease. London: Bailliére Tindall, 1995: Dieterich W, Ehnis T, Bauer M, Donner P, Volta U, Riecken EO et al. Identification of tissue transglutaminase as the autoantigen of celiac disease. Nat Med 1997; 3: Halttunen T, Mäki M. Serum immunoglobulin A from patients with celiac disease inhibits human T84 intestinal crypt epithelial cell differentiation. Gastroenterology 1999; 116: Sollid LM, Scott H. New tool to predict celiac disease on its way to the clinics. Gastroenterology 1998; 115: Appel H, Seth NP, Gauthier L, Wucherpfennig KW. Anergy induction by dimeric TCR ligands. J Immunol 2001; 166:

26

27 Catassi C, Fasano A, Corazza GR (eds): Primary prevention of coeliac disease. The utopia of the new millennium? Perspectives on Coeliac Disease, vol. 1, AIC Press, pp The identification of toxic T cell stimulatory gluten response peptides early in coeliac disease Frits Koning On behalf of: Willemijn Vader*, Yvonne Kooy*, Peter van Veelen*, Arnoud de Ru*, # Diana Harris*, Willemien Benckhuijsen*, Luisa Mearin, and Jan Wouter Drijfhout* # Departments of *Immunohematology and Blood Transfusion and Paediatrics, Leiden University Medical Centre, Leiden, The Netherlands It is generally accepted that coeliac disease (CD) is caused by uncontrolled T cell responses to gluten peptides that are presented by HLA-DQ2 and/or -DQ8 molecules. In recent years five gluten peptides have been identified that stimulate T cell clones 1-5 derived from small intestinal biopsies of CD patients. An important breakthrough has been the demonstration that deamidation of the gluten peptides by the enzyme tissue transglutaminase (ttg) is either required for, or enhances, T cell recognition of four of 1-8 these peptides. The conversion of glutamine into glutamic acid by deamidation generates negative charges in gluten peptides that facilitate binding to HLA-DQ2 and DQ8 molecules, thus providing a molecular basis for the well established association between CD and HLA-DQ2/8. Two major issues, however, remain unsolved. First, all the studies so far have investigated the gluten specific T cell response in adult patients. It is unclear, therefore, whether the identified gluten peptides are also involved in T cell activation earlier in the disease process. Second, it is not known whether deamidation of gluten peptides is required for the breaking of oral tolerance or that it merely enhances T cell reactivity towards gluten. To investigate these matters we have now carried out an extensive investigation of the gluten specific response in children with recent onset CD. Twenty-two Caucasian DQ2-(DQA1*0501/DQB1*02) positive patients with a confirmed diagnosis of CD were included in the present study. Their age at diagnosis (first small bowel biopsy) was between 1 and 9 years (average age 3.6 years ± 1.8). Biopsies were collected from these patients and used to generate gluten reactive T cell lines. Subsequently, gluten specific T cell clones were generated from T cell lines of nine patients. The T cell lines and clones were tested for reactivity against gluten and gluten that has been treated with ttg (ttg-gluten hereafter). Three patterns of reactivity were observed: 1) T cells that did not respond to gluten but did respond to ttg-gluten; 2) T cells that responded to both gluten and ttg-gluten; 3) T cells that did respond to gluten but not to ttg-gluten. In 8 out of 9 patients ttgdependent clonal T cell responses were found while in 7 patients specific responses to 13

28 14 T CELL RESPONSES TO GLUTEN PEPTIDES non-deamidated gluten were also observed. These results indicate that a large proportion of the gluten specific responses are directed to deamidated gluten but that responses to non-deamidated gluten are also common. Extensive testing of the T cell clones against the 3 known HLA-DQ2 restricted T cell stimulatory gliadin derived peptides indicated that the large majority of the T cell clones did not respond to these peptides, and were thus likely reactive towards yet unidentified gluten peptides. To characterize these novel peptides we have used two different methods. First, we purified and characterized T cell stimulatory gluten epitopes from pepsin/trypsin digests of (ttg-) gluten by rphplc and mass 1-2 spectrometry as described. This method led to the characterization of 3 novel T cell stimulatory peptides. Second, we tested the T cell clones against a set of 250 synthetic gluten peptides, representing gliadin and glutenin sequences. This method led to the identification of 3 additional novel T cell stimulatory peptides. Subsequently we tested the response of the T cell clones to the identified peptides in deamidated and non-deamidated form. The T cell response towards 3 peptides required prior deamidation. In contrast, the response to 2 peptides was found to be largely indifferent to deamidation. Finally, deamidation abolished the response to the sixth peptide. Thus, the effect of deamidation by ttg on gluten specific T cell stimulation is heterogeneous and can be positive, neutral and negative. Subsequently the gluten specific T cell clones of all patients were tested against the 3-5 previously characterized HLA-DQ2 restricted gluten peptides as well as against the peptides reported in the present study. While responses to some peptides were found in one patient only, responses to other peptides were found in various patients and these may thus represent more immunodominant peptides. T cell responses towards the a- gliadin peptides which have been reported to be immunodominant in adult patients were found in three paediatric patients, among whom the identical twins. In these 9 patients we observed 8 distinct reactivity patterns towards the gluten peptides (Table). Altogether, these results indicate a highly diverse response against the gluten peptides. These results indicate a discrepancy between the specificity of adult and paediatric gluten specific T cell responses. While immunodominant responses to a particular a- gliadin peptide are found in adult patients, the response in paediatric patients appears more diverse. Our results also indicate that T cell responses to peptides other than the immunodominant a-gliadin peptide can lead to disease. Moreover, our results demonstrate that responses to non-deamidated peptides are frequently found, suggesting that native gluten peptides are immunogenic in celiac disease patients. The discrepancy between the specificity of the adult and paediatric gluten specific T cell response could be explained by a deamidation driven narrowing of the gluten response towards immunodominant T cell stimulatory peptides after initiation of disease by responses towards a diverse repertoire of gluten peptides. Acknowledgements This study was financially supported by the European Community project no. BMH CT-98, a grant from the Dutch Digestive Disease Foundation and the University of Leiden.

29 T CELL RESPONSES TO GLUTEN PEPTIDES 15 DB JB NB SB JP NP MS NV SV Glia- Glia- Glia- Glia- 2a 20a 9a 1g Glia- 30g Glt- 17 Glt- 156 Glu- 5 Glu- 21 Table. Overview of paediatric T cell responses to gluten peptides. References 1. Van de Wal Y, Kooy YM, Van Veelen PA, Pena AS, Mearin LM, Molberg O, et al. Small intestinal T cells of celiac disease patients recognize a natural pepsin fragment of gliadin. Proc Natl Acad Sci U.S.A 1998; 95: Van de Wal Y. Kooy YM, Van Veelen P, Vader W, August SA, Drijfhout JW, et al. Glutenin is involved in the gluten-driven mucosal T cell response. Eur J Immunol 1999; 29: Sjostrom H, Lundin KE, Molberg O, Korner R, McAdam SN, Anthonsen D, et al. Identification of a gliadin T-cell epitope in coeliac disease: general importance of gliadin deamidation for intestinal T-cell recognition. Scand J Immunol. 1998; 48: Arentz-Hansen H, Korner R, Molberg O, Quarsten H, Vader W, Kooy YM, et al. The intestinal T cell response to alpha-gliadin in adult celiac disease is focused on a single deamidated glutamine targeted by tissue transglutaminase. J Exp Med 2000; 191: Anderson RP, Degano P, Godkin AJ, Jewell DP, Hill AV. In vivo antigen challenge in celiac disease identifies a single transglutaminase-modified peptide as the dominant A- gliadin T-cell epitope. Nat Med 2000; 6: Molberg O, McAdam SN, Korner R, Quarsten H, Kristiansen C, Madsen L, et al. Tissue transglutaminase selectively modifies gliadin peptides that are recognized by gut-derived T cells in celiac disease. Nat Med 1998; 4: Van de Wal Y, Kooy Y, van Veelen P, Pena S, Mearin L, Papadopoulos G, et al. Selective deamidation by tissue transglutaminase strongly enhances gliadin-specific T cell reactivity. J Immunol 1998; 161: Quarsten H, Molberg O, Fugger L, McAdam SN, Sollid LM. HLA binding and T cell recognition of a tissue transglutaminase - modified gliadin epitope. Eur J Immunol 1999; 29: Van de Wal Y, Kooy YMC, Drijfhout JW, Amons R, Koning F. Peptide binding characteristics of the coeliac disease-associated DQ(alpha1*0501, beta1*0201) molecule. Immunogenetics 1996; 44: Vartdal F, Johansen B.H., Friede T, Thorpe CJ, Stevanovic S, Eriksen JE, et al. The peptide binding motif of the disease associated HLA-DQ (alpha 1* 0501, beta 1* 0201) molecule. Eur J Immunol 1996; 26: Kwok WW. Domeier ML, Raymond FC, Byers P, Nepom GT. Allele-specific motifs characterize HLA-DQ interactions with a diabetes - associated peptide derived from glutamic acid decarboxylase. J Immunol. 1996; 156:

30

31 Catassi C, Fasano A, Corazza GR (eds): Primary prevention of coeliac disease. The utopia of the new millennium? Perspectives on Coeliac Disease, vol. 1, AIC Press, pp Toxic gluten peptides in coeliac disease identified by in vivo gluten challenge: A single dominant T cell epitope? Robert P. Anderson The Royal Melbourne Hospital Autoimmunity and Transplantation Division, Walter and Eliza Hall Institute c/o Post Office RMH, Victoria, Australia 3050 b.anderson@wehi.edu.au The realization that as many as 1% of Europeans and North Americans are affected by coeliac disease adds urgency to understanding the immunopathogenesis of coeliac disease and developing a rational therapy without the inconvenience of a gluten free diet. Antigen-specific immunotherapy is highly effective in animal models of antigendriven immune-mediated disease. Because human immune-mediated diseases are usually only available for study when the immune response is well established, it has been impossible to be sure what the initiating antigens are, or whether there are critical dominant T cells epitopes that trigger disease. Coeliac disease is unique among human immune-mediated disease since gluten is known to maintain disease, and treatment is successful when dietary gluten is excluded. If coeliac disease is triggered by one critical gluten component it may be possible to develop antigen-specific therapies or preventive vaccines. This chapter discusses the rationale for in vivo strategies that have allowed the identification of a critical gliadin peptide that is the dominant coeliac-specific T cell epitope in a model alpha-gliadin protein. A molecular model for the interaction of HLA-DQ2, dominant gliadin epitope and T cell receptor is presented, and preliminary data indicating the potential of altered peptide ligands to antagonize this interaction is discussed. CD4 T cells have the potential to initiate immunopathology Gluten and HLA-DQ2 are definitively implicated in the aetiopathogenesis of 1 coeliac disease. Since HLA-DQ2 presents peptides to CD4 T cells, there has been intense interest in defining the specificity and phenotype of T cells in coeliac disease that recognize gluten peptides restricted by HLA-DQ2. CD4 T cells play a pivotal role in coordinating immune responses. CD8 T cell and many B cell responses require CD4 T cell help provided directly or through CD4 T cell activation of antigen presenting cells (particularly dendritic cells). In coeliac disease, CD4 T cells predominantly secreting Th1-like cytokines appear in the small 17

32 18 A DOMINANT EPITOPE IN GLUTEN PEPTIDES? intestinal lamina propria within hours of gluten challenge in subjects on gluten free diet. Initial damage to the intestine following gluten challenge in treated coeliac disease is likely to be due to cytokines (for example interferon gamma [IFNg] and tumour necrosis factor [TNF]) secreted by CD4 T cells that activate macrophages causing 2 release of inflammatory mediators and other cytokines that activate other cell types. In addition to immunopathology induced by gluten-specific CD4 T cells, increasing 3 evidence suggests that gluten may also directly activate macrophages, possibly facilitating induction of T cell responses to dietary antigen. Disease chronicity and epitope spreading: relevance to T cell epitope mapping in coeliac disease Experimental autoimmune diseases, such as experimental allergic encepahalitis (EAE), can be initiated by immunization of susceptible mice with adjuvant together with myelin basic protein (MBP) or a specific peptide derived from this protein. This 4 peptide corresponds to the dominant T cell epitope of MBP. Exactly what qualities, in addition to affinity for HLA and resistance to proteolysis during processing, that lead 5 to one peptide in an antigenic protein to be dominant are not fully understood. As EAE progresses, a variety of other peptides (sub-dominant epitopes) derived from MBP and other myelin-associated proteins are recognized by specific T cells in a process termed 4, 6 epitope spreading (also seen in human multiple sclerosis ). Specificity of T cell responses can also be shaped by preferential presentation of peptides by B cells following antigen uptake via B cell receptor or Fc receptor-mediated uptake of 7 antibody-antigen complexes. Post-uptake processing of antigen may also be altered by the cytokine milieu in established immune responses giving rise to a different repertoire 8 of peptides being presented by antigen-presenting cells. Interestingly, when MBP-specific T cells are transferred to a healthy recipient, MBP-specific T cells are initially abundant in spleen and blood. Just before onset of EAE, MBP-specific T cells disappear from spleen and blood, and become abundant at 9 the site of antigen (central nervous system). Consistent with this, oral administration of antigen is initially followed by proliferation of antigen specific T cells in gut- 10 associated as well as systemic lymphoid tissue. Furthermore, T cells with identical 11 specificity are found in murine gut epithelium, lamina propria and the thoracic duct. Taken together, these studies in mice indicate that T cells specific for dominant or subdominant epitopes appear at different time points, and may be located in different anatomical sites according to the chronicity of the immune response. Contrary to the 12 widely held view in coeliac disease research, T cells with identical specificities are present in gut and extra-intestinal sites such as blood in the early phase of immune responses caused by gut antigen. Multiple toxic peptides in coeliac disease identified by multiple methodologies In coeliac disease, a variety of in vivo, ex vivo and in vitro methods have been

33 A DOMINANT EPITOPE IN GLUTEN PEPTIDES? 19 1, 13 exploited to search for toxic gluten peptides. A-gliadin, the first fully sequenced 14 wheat gliadin protein, has been used as the archetypal toxic gluten protein source for most of these peptides (see Fig. 1). More recently, recombinant gliadins and peptides corresponding to parts of cdna-derived gliadin sequences have been studied 15. In view of the diversity of model systems, some with no relationship to coeliac disease (for example fetal intestinal explants), it is not surprising that a variety of different peptides have been defined as toxic. In other studies, non-specific markers of toxicity (for example epithelial cell height) have been used as surrogates for immunological toxicity. If the primary hypothesis is that coeliac disease is a T cell mediated disease, rational identification of toxic gluten peptides in coeliac disease should ideally begin with assays designed to detect HLA-DQ2 restricted CD4 T cells that secrete Th1-associated cytokines (for example interferon gamma). Intestinal T cell clones for identification of toxic gluten peptides More recently, intestinal T cell clones from duodenal biopsies of coeliac patients on gluten free diet pulsed with protease-digested gliadin have been used to search for gliadin-specific T cell epitopes. These studies have revealed that intestinal T cell clones raised from gliadin-pulsed coeliac intestinal biopsies predominantly recognise 12 deamidated gliadin epitopes. T cell epitopes generally correspond to gliadin peptides deamidated by tissue transglutaminase (ttg). ttg is induced with inflammation and apoptosis, and irreversibly cross links proteins and peptides via glu-lys isopeptide 18 bonds, or directly deamidates glutamine residues to glutamate. Intestinal tissue, particularly if inflamed, is rich in ttg. Introduction of glutamate in gliadin T cell 12, 15 epitopes by ttg greatly enhances their binding to HLA-DQ2. T cell epitope mapping in immune-mediated diseases In human immune-mediated diseases, T cell epitopes have been mapped using synthetic 15-20mers overlapping by 10 residues spanning known antigenic proteins in assays of peripheral blood T cells. In coeliac disease, it has been contended that gliadin- 1 specific peripheral blood T cells are qualitatively different from intestinal T cells. Gliadin-specific intestinal T cell clones are generally HLA-DQ2 restricted while 19 peripheral blood T cell clones are HLA-DR, -DQ or -DP restricted. However, these studies were performed before the realization that deamidation of gliadin was important for T cell recognition. Hence, contemporary understanding of toxic gluten peptides in the context of coeliac disease as an HLA-DQ2 associated CD4 T cellmediated disease has relied upon identification of epitopes of intestinal T cell clones using protease-digested gliadin with or without deamidation by ttg. Unfortunately, it is impossible to know whether epitopes of T cell clones are dominant or subdominant, or whether a clone is specific for only a part of a polyclonal T cell response in vivo. Furthermore, recent studies to identify epitopes of gliadin T cell 15 clones have used chymotrypsin-digested gliadins even though chymotrysin selectively cleaves peptide bonds following bulky hydrophobic aminoacids (the same aminoacids known to be anchor residues at the N - and C-terminal end of the HLA-DQ2 1 binding motif ), raising the possibility that bioactive peptides may be artefactually

34 20 A DOMINANT EPITOPE IN GLUTEN PEPTIDES? truncated or destroyed. Hence, a variety of data support the hypothesis that coeliac disease could be initiated by a Th1-like HLA-DQ2 restricted CD4 T cell response focused on particular (probably deamidated) gluten peptide/s. However, current methods relying upon T cell clones, or challenge of coeliac tissue in vivo or ex vivo are incapable of defining whether peptides are dominant or subdominant in the immunopathogenesis of coeliac disease. In vivo gluten challenge and peripheral blood epitope mapping to define T cell 20 epitope hierarchies Antigen challenge of sensitized subjects results in activation of cognate memory T cells. Memory T cells tend to reside at anatomical sites where their cognate antigen was previously encountered. Recently, memory T cells have been divided into effector (cytokine-secreting, CCR7 ) and lymph node-homing subtypes (CCR7 ). We + reasoned that gluten challenge in healthy HLA-DQ2 coeliac subjects following a strict gluten free diet would reactivate gluten-specific memory T cells. The initial intestinal inflammation documented by others may be driven by effector memory T cells, but subsequent appearance of T cells in peripheral blood, perhaps homing back to the gut, may reflect proliferation of gluten-specific T cells in lymphoid tissue during the days after antigen exposure. Hence, the kinetics and frequency of epitope-specific peripheral blood T cells might reveal dominant versus subdominant gluten epitopes. Synthetic 15mer peptides overlapping by 10 aminoacids corresponding to the composite aminoacid sequence (rather than cdna-derived sequence) of A-gliadin (Fig. 1) with or without in vitro deamidation by ttg were studied in overnight ex vivo A 1. VRVPVPQLQP QNPSQQQPQE QVPLVQQQQF PGQQQ QFPPQ QPYPQPQPFP SQQPYLQLQP B P A 61. PQ PQLPYPQ PQ SFPPQQPY PQPQPQYSQP QQPISQQQ AQ QQQQQQQQQQ B [..] P R Q Q[..] A111. QQQILQQILQ QQLIPC MDVV LQQHNIAHAR SQVLQQSTYQ LLQELCCQHL WQIPEQSQCQ B R GS Q Q A171. AIHNVVHAII LHQQQ KQQQQ PSSQVSFQQP L QQYPLGQGS FRPSQQNPQA B [..] [..] Q S A221. QGSVQPQQLP QFEEIRNLAL QTLPAMCNVY I APYC TIAPF GIFGTN B P [..] 14 Fig. 1. Aminoacid sequence of A-gliadin used to derive overlapping 15mer peptides in 20 gluten challenge studies (A), and residues that deviate from the consensus sequence derived from 61 Genbank Triticum aestivum a - and a/b-gliadin cdnas using ClustalW (B) ([..] indicates polymorphic insertion).

35 A DOMINANT EPITOPE IN GLUTEN PEPTIDES? 21 interferon gamma (IFNg) ELISpot assays using peripheral blood mononuclear cells (PBMC). IFNgELISpot assays define the frequency of cells secreting IFNg, and are capable of detecting peptide-specific T cell frequencies as low as 5-10 per million + PBMC. Blood was collected from HLA-DQ2 healthy non-coeliac and coeliac subjects after gluten free diet for at least 4 weeks, and then in the 12 days after commencing gluten challenge (200 g white bread daily for ½ day [n=1], 3 days [n=10], or 10 days [n=1]). Short-term gluten challenge was generally well tolerated by coeliac subjects, 11/12 subjects were able to consume 200g gluten bread for 3 or more days, and 8/12 had only mild symptoms. One subject had abdominal cramps and vomited within 3 hours of the first 2 slices of bread. All symptoms resolved within 1 to 3 days after ceasing gluten challenge. Prior to gluten challenge, no A-gliadin peptide elicited responses in the IFNg ELISpot. However, in 11/12 coeliac subjects and 0/4 healthy control subjects there was induction of IFNgELISpot responses on day 4-8 for one pool of overlapping peptides only when treated with ttg. In all cases, IFNginduction was attributed to 2 peptides overlapping by 10 aminoacids (A-gliadin 56-75). Aminoacid sequencing demonstrated that only one glutamine residue in A-gliadin was susceptible to ttg-mediated deamidation (Q65). Truncations of A-gliadin treated with ttg indicated that residues (PQLPY) were critical for bioactivity, and that the 17mer QLQPFPQPQLPYPQPQS (57-73) was the optimal peptide. Bioactivity of QLQPFPQPELPYPQPQS was identical to ttg-treated QLQPFPQPQLPYPQPQS, demonstrating that bioactivity of this peptide was dependent upon a single deamidated glutamine residue (QE65). Immunomagnetic bead depletion of PBMC prior to addition of peptide showed that this peptide specific immune response was due to CD4 T cells E expressing the b, 7 but not the aintegrin protein, indicating that A-gliadin QE65 specific T cells express the abintegrin 4 7 associated with homing to the intestinal lamina 22 propria. Pre-incubation of PBMC from HLA-DQ2 homozygous subjects with antibody specific for HLA-DQ (but not HLA-DR or HLA-DP) blocks A-gliadin QE65 responses, indicating HLA-DQ2 restriction. In one subject who consumed only 2 slices of bread, IFNg-secreting T cells specific for A-gliadin QE65 were induced on day 6 and persisted until day 12. In another subject who consumed 4 slices of bread daily for 10 days, A-gliadin QE65- specific T cells were also present on day 6-8. IFNgsecretion was only induced by A- gliadin QE65 in subjects who consumed bread for 3 days. However, PBMC collected on day 11 from the subject who consumed bread for 10 days secreted IFNgin response to 6 out of 10 pools of ttg-treated A-gliadin peptides (one pool included A- gliadin 57-73). Hence, T cell epitope spreading occurs as early as 10 days after commencing antigen challenge. These studies indicated: 1. In A-gliadin, there is a hierarchy of T cell epitopes with only one dominant T cell epitope. 2. T cells specific for the dominant epitope are present in peripheral blood only

36 22 A DOMINANT EPITOPE IN GLUTEN PEPTIDES? transiently after gluten challenge in vivo, but have predictable kinetics. 3. Gluten challenge induces T cells specific for the same dominant A-gliadin epitope in all HLA-DQ2 coeliac subjects. Gliadin-specific intestinal T cell clones are focused on two peptides closely related to A-gliadin QE65 15 Arentz-Hansen et al. have shown that two peptides from a panel of 11 recombinant a-gliadins are common epitopes for gliadin-specific intestinal T cell clones. T cell lines specific for one or both of these peptides (QLQPFPQPELPY and + PQPELPYPQPELPY) were raised from intestinal biopsies of 17/17 HLA-DQ2 coeliac subjects on gluten free diet. These peptides closely resemble A-gliadin QE65 and contain the core sequence PELPY, supporting the concept that intestinal and peripheral blood T cell responses induced with gluten challenge are qualitatively similar. More importantly, if intestinal and peripheral blood T cell responses share the same specificity it is likely the ex vivo polyclonal peripheral blood T cell response will be more informative than T cell clones or lines for studies of molecular specificity and definition of T cell epitope hierarchies in coeliac disease. Comparison of HLA-DQ2 restricted epitopes of T cell clones with A-gliadin QE65 We have studied various HLA-DQ2 restricted gliadin epitopes (Tab. 1) in IFNg ELISpot assays using PBMC from HLA-DQ2 coeliac subjects on day 6 of gluten 23 challenge (3 days, 200g gluten-containing bread daily). A-gliadin and GDB2 do not induce IFNgresponses above background levels, while A-gliadin QE65 is generally 25 % and a2 gliadin QE65 QE72 60% as bioactive as A-gliadin QE65 at optimal concentrations (25 µ g/ml). To determine whether T cells specific for a2 gliadin QE65 QE72 are part of the polyclonal T cell response to A-gliadin QE65, IFNgELISpot responses to a2 gliadin QE65 QE72 or A-gliadin QE65 (25 µ g/ml) alone or mixed together were compared. There was no difference between A-gliadin QE65 alone or mixed with a2 gliadin QE65 QE72, suggesting that in vivo T cells specific for a2 gliadin QE65 QE72 and/or A- gliadin QE65 are simply part of the polyclonal T cell response targeting A-gliadin QE65. A-gliadin (QE65) QLQPFPQPELPY a2 gliadin (QE65 QE72) PQPELPYPQPELPY A-gliadin LGQQQPFPPQQPYPQPQPF g-gliadin (GDB2) QQLPQPEQPQQSFPEQERPF Tab. 1. HLA-DQ2 restricted epitopes of gliadin-specific intestinal T cell clones A-gliadin QE65 is the optimal a-, a/b-gliadin polymorphism 23 A-gliadin spans a highly polymorphic region of the a-, a/b-gliadins (Tab.2).

37 A DOMINANT EPITOPE IN GLUTEN PEPTIDES? 23 We have compared IFNgELISpot responses of all the a-, a/b-gliadin polymorphisms of A-gliadin found by SwissProt using the search sequence XXXXXXXPQLPYXXXXX. Amongst these polymorphisms, bioactivity of QLQPFPQPQLPYPQPQ[P,L] is identical to QLQPFPQPQLPYPQPQS after d e a m i d a t i o n b y t T G o r s u b s t i t u t i o n Q E 6 5. t T G - d e a m i d a t e d PQLPYPQPQLPYPQPQ[P,L] is generally 80% as bioactive as QLQPFPQPQLPYPQPQS, but substitution of P69 for serine or leucine reduces bioactivity by 60%. Other polymorphisms are generally less than 20% as bioactive as A-gliadin Hence, ttg-deamidated or QE-substituted at position 9, QLQPFPQPQLPYPQPQ[P,L,S] is the a-, a/b-gliadin polymorphism of A-gliadin with optimal bioactivity. QLQPFPQ PQLPYPQPQP QLQPFPQ PQLPYSQPQP QLQPFPQ PQLPYSQPQQ QLQPFPQ PQLSYSQPQP QLQPFPQ PQLPYLQPQP QLQPFSQ PQLPYSQPQP QLQPFLQ PQLPYSQPQP QLQPFLQPQPFP PQLPYSQPQP QLQPFPQPQLPYPQPQLPYPQLPYPQPQP QLQPFPQPQLPYPQ PQLPYPQPQP QLQPFPQPQPFPPQLPYPQPQLPYPQPQP QLQPFPR PQLPYPQPQP QLQPFPQPQPFP PQLPYPQPPP QLQPFPQPQPFL PQLPYPQPQS QLQPFPQPQPFP PQLPYPQPQS QPQPFP PQLPYPQTQP QPQPFPPQ PQLPYPQTQP Tab. 2. Polymorphisms in the region of A-gliadin among the 61 Triticum aestivum a-, a/b-gliadin cdna-derived protein sequences in Genbank One reason for the selection of QLQPFPQPQLPYPQPQ[L,P,S] and PQLPYPQPQLPYPQPQ[P,L] as potent epitopes may be that these sequences are resistant to proteases. Chymotrypsin and pepsin both cleave peptide bonds after hydrophobic residues. In vitro digestion of these peptides and less potent polymorphisms with tyrosine at position 12 followed by proline, serine or leucine, indicates that proline at position 13 prevents susceptibility to both chymotrypsin and pepsin. Hence, it is possible that the specificity of the T cell response is shaped by the susceptibility of gliadin peptides to proteases in the gut (and presumably in antigen presenting cells). Immune toxicity of wheat, rye and barley due to peptides cross-reactive with A- gliadin 57-73, or epitope spreading initiated by B cells specific for sequences adjacent to A-gliadin QE65 The data we have gathered using PBMC following in vivo gluten challenge in

38 24 A DOMINANT EPITOPE IN GLUTEN PEPTIDES? coeliac disease are consistent with A-gliadin deamidated by ttg being the dominant a-, a/b-wheat gliadin T cell epitope. Searches for peptides including the core sequence PQLPY have not revealed other wheat, rye or barley gluten sequences outside the a-, a/b-gliadins of wheat. One possibility is that epitope spreading initiated by the dominant epitope leads to other epitopes in gluten being recognized. Interestingly, B cell epitopes have been identified that are immediately adjacent or including the C - and N-terminal portions of A-gliadin The other HLA-DQ2 restricted epitopes, g- gliadin GDB2 and A-gliadin are also adjacent to or include sequences identical to or very similar to the same gliadin B cell epitope (QXQPFP). Hence, B cell mediated epitope spreading may be initiated by B cells given help by T cells specific for the dominant epitope. Alternatively, wheat, rye and barley may contain prolamin sequences that are cross-reactive with A-gliadin but do not include PQLPY, or there are other unique dominant epitopes. Fine molecular specificity of A-gliadin QE65-specific peripheral blood T cells Residues in A-gliadin QE65 that determine bioactivity were mapped by comparing bioactivity of A-gliadin QE65 variants substituted with lysine at each 22 aminoacid. Bioactivity was abolished if lysine was substituted at position 8-11 (PELP), and substantially reduced with lysine at positions 4-7 (PFPQ) and (YP). Single aminoacid substituted variants of A-gliadin QE65 with all naturally occurring aminoacids except cysteine at positions 4-13 were synthesized and their bioactivity compared to the parent peptide. Positions 4-7 were highly sensitive to substitution, no more than three aminoacids at each position conveyed bioactivity greater than 50% of the parent peptide. Positions 4-7 and were less sensitive to substitution but certain aminoacids such as proline, lysine and arginine tended to abolish bioactivity. The fine molecular specificity of peripheral blood T cells for A- gliadin QE65 was similar amongst all of the eight coeliac subjects tested. Antigen-specific therapy using A-gliadin Qe65 A-gliadin QE65 is clearly the dominant or one of the dominant gliadin T cell epitopes in HLA-DQ2-associated coeliac disease. Therefore it is reasonable to consider peptide therapeutics based on A-gliadin QE65. Peptide delivered orally or nasally would be simple and easy to formulate. Another possibility is design of altered peptide ligand (APL) antagonists that differ from the parent peptide by one or more residues and subtly alter T cell receptor signaling. APL antagonsists have the 26 potential to switch off or skew TH1 to TH2 responses in vitro. We have shown that at least 5 single aminoacid-substituted variants of A-gliadin QE65 with weak agonist properties also significantly reduce IFNgELISpot responses to A-gliadin QE65 when incubated in 5-fold excess with A-gliadin QE65 (unpublished observations). It is likely that multiple substitutions or cocktails of APL antagonsists will be required for complete blockade of the polyclonal T cell response to A-gliadin QE65.

39 A DOMINANT EPITOPE IN GLUTEN PEPTIDES? 25 Conclusions Gluten challenge in coeliac disease allows gliadin-specific T cell responses to be measured and epitope hierarchies defined. This method has significant advantages over T cell clones. One dominant T cell epitope has been defined in A-gliadin and is likely to be the dominant a-, a/b-wheat gliadin T cell epitope. Whether there are other distinct dominant T cell epitopes in other classes of gliadin in wheat, rye and barley, or whether toxicity of these proteins is due to cross-reactivity with A-gliadin QE65 is crucial to the design of antigen-specific therapeutics. In either case, gluten challenge and testing the bioactivity of peptides and proteins using PBMC will allow the importance of particular epitopes to be determined. The efficacy of altered peptide ligand antagonists targeting A-gliadin QE65 provides proof of principle that peptide therapeutics may be a practical approach to the treatment of coeliac disease without resort to gluten free diet. References 1. Sollid LM. Molecular basis of coeliac disease. Ann Rev Immunol 2000 ; 18: Nilsen EM, Jahnsen FL, Lundin KEA, Johansen F-E, Fausa O, Sollid LM, et al. Gluten induces an intestinal cytokine response strongly dominated by interferon gamma in patients with celiac disease. Gastroenterology 1998; 115: Tuckova L, Flegelova Z, Tlaskalova-Hogenova H, Zidek Z. Activation of macrophages by food antigens: enhancing effect of gluten on nitric oxide and cytokine production. J Leuk Biol 2000; 67: Tuohy VK, Yu M, Ling Y, Kawczak JA, Kinkel RP. Spontaneous regression of primary autoreactivity during chronic progression of experimental autoimmune encephalitis and multiple sclerosis. J Exp Med 1999; 189: Yewdell JW, Bennink JR. Immunodominance in major histocompatibility complex class I-restricted T lymphocyte responses. Ann Rev Immunol 1999 ; 17: Tuohy VK, Yu M, Weinstock-Guttman B, Kinkel RP. Diversity and plasticity of self recognition during the development of multiple sclerosis. J Clin Invest 1997; 99: McCluskey J, Farris AD, Keech CL, Purcell AW, Rischmueller M, Kinoshita G, et al. Determinant spreading: lessons from animal models and human disease. Immunol Rev 1998; 164: Drakesmith H, O'Neill D, Schneider SC, Binks M, Medd P, Sercarz E, et al. In vivo priming of T cells against cryptic determinants by dendritic cells exposed to interleukin 6 and native antigen. Proc Natl Acad Sci USA 1998; 95: Flugel A, Berkowicz T, Ritter T, Labeur M, Jenne DE, Li Z, et al. Migratory activity and functional changes of green fluorescent effector cells before and during experimental autoimmune encephalitis. Immunity 2001; 14: Gutgeman I, Fahrer AM, Altman JD, Davis MM, Chien Y-h. Induction of rapid T cell activation and tolerance by systemic presentation of an orally administered antigen. Immunity 1998; 8: Arstila T, Arstila TP, Calbo S, Selz F, Malassis-Seris M, Vassalli P, et al. Identical T cell clones are located within the mouse gut epithelium and lamina propria and

40 26 A DOMINANT EPITOPE IN GLUTEN PEPTIDES? circulate in the thoracic duct. J Exp Med 2000; 191: Molberg O, McAdam SN, Korner R, Quarsten H, Kristiansen C, Madsen L, et al. Tissue transglutaminase selectively modifies gliadin peptides that are recognized by gut-derived T cells in celiac disease. Nat Med 1998; 4: Weiser H. The precipitating factor in coeliac disease. Balliere's Clin Gastroenterol 1995; 9: Kasarda DD, Okita TW, Bernardin JE, Baeker PA, Nimmo CC, Lew EJ-L, et al. Nucleic acid (cdna) and amino acid sequences of type gliadins from wheat (Triticum aestivum). Proc Natl Acad Sci USA 1984; 81: Arentz-Hansen H, Korner R, Molberg O, Quarsten H, Van der Wal Y, Kooy YMC, et al. The intestinal T cell response to agliadin in adult celiac disease is focused on a single deamidated glutamine targeted by tissue transglutaminase. J Exp Med 2000; 191: Sjostrom H, Lundin KEA, Molberg O, Korner R, McAdam SN, Anthonsen D, et al. Identification of a-gliadin T-cell epitope in coeliac disease: general importance of gliadin deamidation for intestinal T-cell recognition. Scand J Immunol 1998 ; 48 : van de Wal Y, Kooy YM, van Veelen P, Vader W, August SA, Drijfhout JW, et al. Glutenin is involved in the gluten-driven mucosal T cell response. Eur J Immunol 1999 ; 29: Aeschilmann D, Paulsson M. Transglutaminases: protein cross-linking enzymes in tissues and body fluids. Thromb Haemost 1994; 71: Gjertsen HA, Sollid LM, Ek J, Thorsby E, Lundin KEA. T cells from the peripheral blood of coeliac disease patients recognize gluten antigens when presented by HLA- DR, -DQ, or DP molecules. Scand J Immunol 1994 ; 39: Anderson RP, Degano P, Godkin AJ, Jewell DP, Hill AVS. In vivo antigen challenge in celiac disease identifies a single transglutaminase-modified peptide as the dominant A-gliadin T-cell epitope. Nat Med 2000; 6: Sallusto F, Lenig D, Forster R, Lipp M, Lanzavecchia A. Two subsets of memory T lymphocytes with distinct homing potentials and effector functions T cell memory. Nature 1999; 401: Anderson RP, Jewell DP, Hill AVS. Peripheral blood T cells induced by gluten challenge in coeliac disease target a specific molecular motif and express a guthoming integrin. Gastroenterology 2001; 120: A Anderson RP, Jewell DP, Hill AVS. Bioactivity of peptides homologous to the coeliac-specific dominant A-gliadin T cell epitope. Gastroenterology 2011; 120: A ten Dam M, Van de Wal Y, Mearin ML, Kooy Y, Pena S, Drijfhout JW, et al. Antialpha-gliadin antibodies (AGA) in the serum of coeliac children and controls recognize an identical collection of linear epitopes of alpha-gliadin. Clin Exp Immunol 1998; 114: Osman AA, Gunnel T, Dietl A, Uhlig HH, Amin B, Fleckenstein B, et al. B cell epitopes of gliadin. Clin Exp Immunol 2000; 121: Bielekova B, Martin R. Antigen-specific immunomodulation via altered peptide ligands. J Mol Med 2001; 79:

41 Catassi C, Fasano A, Corazza GR (eds): Primary prevention of coeliac disease. The utopia of the new millennium? Perspectives on Coeliac Disease, vol. 1, AIC Press, pp Role of A-gliadin peptide Paul J Ciclitira The Rayne Institute, St. Thomas Hospital, London, UK Coeliac disease was first described in Samuel Gee who gave a classic description of a wasting disease mostly affecting children and causing diarrhoea, concluded at the end of his treaties, "if the patient can be cured at all, it must be by means of the diet". It is unfortunate that his treatment at that time comprised thin gruel which contained wheat flour. It remained until 1953, for Professor Dicke and colleagues to publish their findings that wheat, rye and barley exacerbated the condition. They also noted that oats was toxic, which was probably due to contamination of flour with wheat. Dicke's experiment at that time compromised of making a slurry of the individual flours in water, pouring them down an eight year old child 's throat and noting that the child collapsed with vomiting, diarrhoea and near death. They extended this work to show that the most toxic component of wheat resides in the endosperm or flour fraction. However, the germ and husk were mildly toxic. This was probably due to contamination of these fractions with the gluten proteins. Flour contains proteins and water soluble starch fractions. Purified wheat starch is often used for the production of gluten-free products. Wheat flour proteins can be divided into four groups of which gliadins and glutenins together comprise 90% of the protein. The albumins and globulins are minor protein fractions. The glutenins have molecular weights ranging from 50 kilodaltons to many millions and entrap carbon dioxide in the dough enabling it to prove and rise. It was in early studies that the alcohol soluble or gliadin fraction of gluten was shown to be toxic in coeliac disease. The gliadin proteins were sub-divided into four fractions termed a, b, gand waccording to their relative electrophoretic mobility. More recently, they were classified as a, g, and w fractions according to their N-terminal amino acid sequences. We previously purified, using column chromatography with 20 litres eluent, gliadin into its a,b, g, and w fractions. The complexity of the individual gliadins was shown in further experiments. Having made polyclonal anti-sera against each gliadin fraction, we looked for precipitin lines on Ouchterlony gels which revealed four precipitin lines to gliadin, with an anti-gliadin serum. Precipitin lines were shared between gliadin subfraction suggesting that they share antigenic determinants. We subsequently used an in vitro 27

42 28 ROLE OF A-GLIADIN PEPTIDE organ culture model. Small intestinal biopsies were placed on an organ culture grid and incubated for hours in the presence or absence of a putatively toxic fraction. We measured enterocyte surface cell heights on sections of the biopsies. Using this technique, we reported that each of the gliadin fractions, that is a, b, gand was enterotoxic to coeliac small intestinal mucosa. Dr Kasarda subsequently went on to sequence gliadin. The now classic sequence of A-gliadin is known to be 266 amino 1 acids long. This is an unusual protein in that it is composed by 10-15% proline and 30-35% glutamine residues; this is an efficient method of storing nitrogen for the growing plant. We subsequently went on to develop an in vivo method of assessing the toxicity of gliadin fractions. This involved passing a Quintron multiple biopsy capsule into the proximal small intestine to which was taped an infusion tube. This allowed test fractions to be infused and multiple small intestinal biopsies to be taken over eight hours. These could then be sectioned and assessed blindly for villous height, crypt depth ratio, epithelial surface cell height and intraepithelial lymphocyte count. Using this system, we tested on different days 10 mg, 100 mg followed by 500 mg, and 1g of gliadin. We showed that 10 mg produced no histological relapse, 100 mg minimal and 500 mg moderate histological relapse. With 1 g there was gross flattening of the mucosa which commenced between 1-2 hours, was maximal at 4 hours and started to recover by 6 hours. We observed flattening of the mucosa, with complete loss of the normal villous 2 architecture. We have also used electron microscopy to show increase in the number of lysosome-like bodies below the brush border, which occurs 1-2 hours after commencing gluten challenges, suggesting that these may be relevant to the pathogenesis of the condition. After a matter of hours, there is apoptosis and shedding of the surface enterocytes. We hypothesize that there is absorption of gluten protein between and through the enterocytes. Toxic gliadin peptides are presented by antigen presenting cells to gluten sensitive T cells which exacerbate the condition through the production of TH1 cytokines. We subsequently went on to perform in vivo experiments assessing the toxicity of peptide fractions corresponding to amino acids 31-49, , 3-21 of A-gliadin. We used, as a positive control, unfractionated gliadin. We tested four subjects and showed that a fraction corresponding to amino acids was 3 enterotoxic and caused flattening of the mucosa while the other two peptides did not. In a variety of different experiments we showed that there was an induction of TH1 cytokines and inducible nitric oxide, infiltration of the mucosa with T cells, a change in 4 the villous morphology and enterocyte surface cell height. Since gliadin is presented by antigen presenting cells to gluten sensitive T cells by HLA DQ2 or DQ8, we established a cellular binding assay to study the binding of gliadin peptides to DQ2. We demonstrated that the peptide, corresponding to amino acid residues of A-gliadin which caused in vivo toxicity, bound to DQ2. The two other peptides which we tested in vivo, but which did not cause toxicity, did not bind to 5 DQ2. To date, no group has demonstrated the presence of a small intestinal T cell clone that is sensitive to gliadin However, a peripheral blood clone which responds to 6 this peptide has been demonstrated. Interestingly, the peptide has also been shown to 7 induce CD4 T cell infiltration into the buccal mucosa of coeliac disease patients. We therefore felt that there is good evidence that A-gliadin is a coeliac toxic epitope, although it may be a minor one.

43 ROLE OF A-GLIADIN PEPTIDE 29 References 1. Kasarda DD, Okita TW, Bernadin JE, Backer PA, Nimmo C, Lew E, et al. Nucleic acid (cdna) and amino acid sequences of type gliadins from wheat (Triticum aestivum). Proc Nat Acad Sci 1984; 81: Ciclitira PJ, Evans DJ, Fagg NLK, Lennox ES, Dowling RH. Clinical testing of gliadin fractions in coeliac patients. Clin Sci 1984; 66: Sturgess R, Day P, Ellis HJ, Lundin KE, Gjertsen HA, Konakou M, et al. Wheat peptide challenge in coeliac disease. Lancet 1994; 343: Kontakou M, Przemioslo RT, Sturgess RP, Limb GA, Ellis HJ, Day P, et al. Cytokine mrna expression in the mucosa of treated coeliac patients after wheat peptide challenge. Gut 1995; 37: Shidraw RG, Parnell ND, Ciclitira PJ, Travers P, Evan G, Rosen-Bronson S. Binding of gluten-derived peptides to the HLA-DQ2 (alpha 1*0501, beta 1*0201) molecule, assessed in a cellular assay. Clin Exp Immunol 1998; 111: Gjertsen HA, Lundin KEA, Sollid L, Eriksen JA, Thorsby E.. T cell recognition of gliadin peptides presented by the coeliac disease associated HLA-DQ (a1*0501, B1*0201) heterodimer. Human Immunology 1994; 39: Lahteenoja H, Mäki M, Viander M, Raiha I, Vilja P, Rantala I, et al. Local challenge on oral mucosa with an alpha-gliadin related synthetic peptide in patients with coeliac disease. Am J Gastroenterol 2000; 95:

44

45 Catassi C, Fasano A, Corazza GR (eds): Primary prevention of coeliac disease. The utopia of the new millennium? Perspectives on Coeliac Disease, vol. 1, AIC Press, pp The association of the HLA-DQ molecules with coeliac disease in the Saharawi: an evolutionary perspective 1 2 Francesco Cucca and Carlo Catassi 1 Department of Biomedical and Biotechnologies Sciences, University of Cagliari, 2 Cagliari, Italy. Department of Pediatrics, University of Ancona, Ancona, Italy. The familial clustering of CD Coeliac disease (CD) is an immune-mediated enteropathy caused by the ingestion of wheat and other gluten-containing cereals (rye, barley) in genetically predisposed 1 individuals. CD is strongly clustered in families as illustrated by the sibling lifetime 2-3 risk/population prevalence ratio, lequal s to in different studies. These epidemiological observations indicate the strong influence of genetic factors in the disease predisposition. This is also strongly suggested by the recurrence risk of % observed in monozygotic twins (MZ). Twin studies are particularly important for human geneticists because the risk in MZ twins provides a direct estimate of penetrance (the effect of the genotype on the phenotype) for the whole complement of susceptibility alleles at multiple loci. While it is necessary to extend these preliminary studies in larger cohorts of twins, the fact that the recurrence risk might not be 100%, and thus that the penentrance is incomplete, might indicate that other modifying epigenetic factors, such as the stochastic rearrangements of the T-cell receptor, and/or other environmental factors (in addition to the gluten in the diet) could also be involved in CD. The contribution of genes located in the HLA region to CD predisposition The major histocompatibility complex (MHC) human leukocyte antigen (HLA) region on chromosome 6p21 contains the major and, so far, the only consistently confirmed genetic component for CD predisposition. The HLA/MHC contains a densely packed array of at least 150 genes in 3,500 kb of DNA. Proteins encoded within the HLA region determine the way in which antigens are processed, translocated and presented to T-lymphocytes. During development of the T cell repertoire, HLA class I and class II molecules control positive and negative selection of the T-lymphocytes in the thymus. Not surprisingly, therefore, genes in the HLA region influence susceptibility to a large number of disorders, particularly autoimmune diseases. 31

46 32 COELIAC DISEASE IN THE SAHARAWI A long series of association studies from different Caucasian populations provided a convincing case that within the HLA region, the combined presence of the DQA1*0501 and DQB1*0201 alleles, encoded by genes located in the class II sub-region, is the 5 primary predisposing factor for CD. These alleles are found either in phase within the DR3-DQA1*0501-DQB1*0201 haplotype or in trans in individuals carrying the DR7- HLA-DQA1*0201-DQB1*0201/DR5-HLA-DQA1*0501-DQB1*0301 genotype. Additionally, although to a much lesser extent, another molecule, HLA-DQA1* DQB1*0302, found on DR4 haplotypes, is associated with CD. More recently, linkage analysis in families with multiple cases, in particular tests of allele sharing by affected sibpairs (ASPs), proved that the HLA region encodes the main disease locus predisposing to CD. From these linkage data it was possible to compute that this region contributes about 40 % of the observed familial clustering of the disease assuming a 2-3. multiplicative model of inheritance The role of these DQ molecules is further substantiated by the observation that antigen-presenting-cells positive for the DQA1*0501-DQB1*0201 (DQ2) and DQA1*0301-DQB1*0302 (DQ8) molecules are able to recognize and present gluten derived peptides to T cells obtained from biopsy specimens from the small intestine of 6-7 CD patients. The gluten-reactive T cell clones isolated from biopsies were CD4+, CD8- and inhibition studies with anti-hla antibodies demonstrated predominant 6 antigen presentation by the dimers encoded by the HLA-DQ loci. Non-HLA genes and CD The linkage data from affected sib pair (ASP) families from different populations indicates that the locus specific ls for the whole HLA region is equal to ~3. This value, assuming a multiplicative model of epistasis (which fits with the rapid fall off in recurrence risk from first to second to third degree relatives) is equal to 40% of the familial clustering of the disease (log3/log15). These results imply that 60% of the familial clustering of the disease must be accounted for by non-hla genes. To detect these non-hla loci, some whole genome scans have been performed but, with the exception of the major disease locus in the HLA region on 6p21, they have given weak 8-10 and conflicting results. These results illustrate the difficulties of detecting (or replicating the detection of) genes of low to moderate effects using ASP-based methods. Low penetrance, compounded by small individual genetic effects of most of the non-hla disease loci create major limitations on the power of a study (and all the linkage scans so far performed were severely under-powered). Additionally, locus and allelic heterogeneity further complicate the search. Nevertheless, the generation of primary hypotheses deriving from linkage analysis remains an invaluable step when embarking on an effort to define the genetic risk factors involved in an inherited disease especially since whole genome scans of association are still technically and economically impractical and the presence of a disease model with strong epistasis between the disease loci could severely impair association studies but not linkage 11-5 analysis. Prior evidence of linkage dramatically lowers the threshold (P value to 10 ) required in an association/ld study of the linked region to be confident that the result is 12 not a false positive. Some of the aforementioned factors, in particular those related to the statistical power of the study, could be alleviated by further increasing the number of

47 COELIAC DISEASE IN THE SAHARAWI 33 ASP families, for instance in a consortium type study addressed to analyse all the CD ASP collectable worldwide on the order of 2,000 or more (ASPs). However, this strategy based on the analysis of very large sample sets of ASP families that have to be necessarily collected from different populations cannot provide a significant benefit regarding the confounding effects of genetic heterogeneity. An alternative approach could be to concentrate on an isolated population showing an high disease prevalence, to reconstruct all the relationships between the various affected cases and scan with parametric and non-parametric linkage tests the whole genome to detect non-hla gene effects. It is likely that such a study design, already applied in the case of type 1 13 diabetes, would provide sufficient statistical power to detect significant results across the genome. The high prevalence of CD in the Saharawi population 14 CD is typically found in individuals of Caucasian origin. In Europe and North America the disease prevalence is around % of the general population. Recently a tenfold higher CD frequency (5.6 %) was reported in the Saharawi, a people of Arab and Berber origin with high degree of consanguinity, living in the Sahara desert 17. We analysed the genetic association of the main CD locus in a group of Saharawi children affected with CD and their first-degree relatives. The aims of this study (which is described in details below) were: 1) to evaluate whether the degree of association at the main disease locus was similar to that observed in other populations; 2) to investigate how the frequency of the predisposing HLA molecules in the general Saharawi population relates to the high prevalence observed in this ethnic group. Patients and methods This work was performed on Saharawi refugees living in the camps near Tindouf, Algeria. Blood samples were collected with the informed consent from a group of previously diagnosed subjects with CD and their first-degree relatives. The diagnosis of CD was based on the finding of serum IgA class antiendomysium antibodies (EMA) positivity on at least two consecutive blood samples and, in a subgroup of patients, the 17 typical enteropathy at the small intestinal biopsy, as previously described. EMA were detected by indirect immunofluorescence on serum diluted both 1:5 and 1:50, using monkey oesophagus as the antigenic substrate (Antiendomisio Eurospital, Trieste, Italy). Patients were 79 subjects (33 males and 44 females) deriving from 69 independent families, with a median age of 8 years (range: 2-37) and including 69 probands, 8 siblings and 2 mothers. Overall 136 healthy subjects were also investigated with a median age of 30 years, range 2-70 years. These included 29 individuals from 9, non-cd, unrelated families and 107 first-degree relatives of the CD patients (53 mothers, 17 fathers and 37 siblings). Additionally, in 12 parents of CD patients for which DNA samples were not available, the DQ genotypes were unequivocally deduced based on the available data from the other parent and the offsprings. DNA was extracted using the chelex method starting from dried blood spots 18 (Guthrie cards). Amplification of the polymorphic second exon of the HLA-DRB1,

48 34 COELIAC DISEASE IN THE SAHARAWI DQB1 genes and dot blot analysis of amplified DNA with sequence specific oligonucleotide probes (SSO) were carried out as previously reported. The DQA1 alleles were inferred using the known patterns of linkage disequilibrium from the DRB1-DQB1 haplotypes. Parental haplotypes were reconstructed using a program written by Frank Dudbridge available at ftp://ftp-gene.cimr.cam.ac.uk/pub/software/. A possible drawback of case-control analyses is their sensitivity to problems related to genetic stratification. In order to circumvent this problem, 82 affected family-based control, (AFBAC) DQA1-DQB1 haplotypes have been selected from all the families in 22 which the parents were available, as described for single alleles by Thomson. The AFBAC frequencies are based on the chromosomes that are never transmitted from the parents to affected children and provide a source of controls which is not sensitive to population stratification unless very recent. The control frequencies were also enriched by 18 chromosomes deriving from 9 unaffected and unrelated Saharawi individuals. The pseudo-controls (AFBAC) and control frequencies were nearly identical and therefore were merged (total controls equal to 100 chromosomes). Using these control haplotype frequencies and assuming Hardy-Weinberg equilibrium, we also established 23 the control genotype frequencies as previously described. The frequencies of the HLA class II haplotypes and genotypes obtained in patients and in controls, were then compared using a 2 x 2 chi-squared Pearson test. The odds ratios ORs) were calculated using the Wolf formula [a x d)/(b x c)]; when one element of this equation was 0 we used the Haldane formula: RR=[2a + 1) 2d +1)]/[2b +1) 2c +1)], where a is the number of patients possessing the HLA antigen; d is the number of controls lacking the particular HLA antigen; b the number of patients lacking the particular antigen; c the number of controls possessing the particular HLA antigen. In the association analyses performed in this study, we considered only the probands in the 8 families with more than one affected sibling. A correction for number of tests performed was applied by multiplying the P values for the tested markers present in more than 1 patient or 1 control (respectively 10 haplotypes and 8 genotypes). We also evaluated the evolutionary relationship between the Saharawis and the 24 other human groups. To this aim we performed a multi-dimensional scaling analysis carried out using the correlation matrix of the haplotype frequencies reported in the various populations and determined with a gene counting procedure. Who are the Saharawis? In order to obtain a visual output of the DRB1-DQA1-DQB1 haplotype distribution worldwide and obtain information about the evolutionary relationships between the various human groups we performed a multi-dimensional scaling analysis. The results of this analysis are shown in Fig.1. The first dimension splits Eastern and Western world populations. The Asian and the African appear to be the most differentiated human groups, with the Kogi and Cayapa native American populations as the most differentiated within the Asian group. The Caucasians are closely grouped and are located between the African and the Asian range of variability. The second dimension explains the diversity of the Caucasian group, which includes the North-African and the Saharawi samples. More specifically, the Saharawi population appears to be located in

49 COELIAC DISEASE IN THE SAHARAWI 35 the border between the European and the African range of variability. Thus, the genetic structure of this population is consistent with its geographical location. The association of the DQ molecules with CD in the Saharawi population We first established the sibling lifetime risk/population prevalence ratio, ls in our Saharawi sample set, that was equal to 3 (16.7%/5.6%). Because of the high disease prevalence, the degree of familial clustering was lower in the Saharawi than in Caucasian populations ls = ~30). Next, we established the association of the haplotypes and genotypes at the main disease superlocus HLA-DQB1-DQA1 in this population. The distribution of the various haplotypes among the Saharawi CD patients and controls is shown in Tab. 1. The DQA1*0501-DQB1*0201 haplotype was positively associated with CD (OR: , Pc = 1.4x10 ) while the DQA1*0102-DQB1*0604, haplotype showed some -3 degree of negative association with the disease (OR: 0.4, Pc = 4.7x10 ). To our knowledge the latter finding has been never reported in other populations and therefore needs confirmation in independent sample sets. Tab. 2 shows the distribution of the genotypes among patients and controls. The (DR3)DQA1*0501-DQB1*0201/(DR3)DQA1*0501-DQB1*0201 encoding two putative a* b*0201 heterodimers was positively associated with CD (OR: 6.4, -2 P= 2.4x10 ). The (DR3)DQA1*0501-DQB1*0201/(DR4)DQA1*0301-DQB1*0302 and the (DR3)DQA1*0501-DQB1*0201/(DR7)DQA1*0201-DQB1*0201 genotypes showed a trend toward a positive association (OR: 8.6 and 2.4 respectively) without reaching the level of significance. Overall, 91.3% of the patients, but only 38.9% of the controls were found to be positive for the combined presence of the DQA1*0501 and DQB1*0201 alleles encoding in cis or in trans at least one putative a* b*0201 heterodimer. The patients positive for the DQA1*0501-DQB1*0201 or DQA1* heterodimer HAPLOTYPES Patients Controls DQA1 DQB1 N N OR 95% C.I. P <0.05 Pc < x x x x x x10-2 Tab. 1. Distribution of the DQA1-DQB1 haplotypes in the Saharawi CD patients and controls.

50 36 COELIAC DISEASE IN THE SAHARAWI were instead 95.6% in comparison with 41.6% of the controls. Finally, all but one patient (98.6%) and 42.5% of the controls were found to be positive for the DQB1*0201 allele. The CD associations observed in the Saharawi population were then contrasted with 5 those observed in other populations. While the (DR3)DQA1*0501- DQB1*0201/(DR3) DQA1*0501-DQB1*0201 genotype seems to be strongly and GENOTYPES DQA1 DQB1 DQA1 DQB1 Patients Controls N N OR 95% C.I. P <0.05 Pc < x x x Tab. 2. Distribution of the DQA1-DQB1 genotypes in the Saharawi CD patients and controls. consistently predisposing in all the ethnic groups so far studied, it should be noted that in the northern European and in the Sardinian populations, but not in the Saharawis, the (DR3)DQA1*0501-DQB1*0201/(DR7)DQA1*0201-DQB1*0201 confers a stronger risk than the (DR3)DQA1*0501-DQB1*0201/(DR4)DQA1*0301-DQB1*0302 genotype and shows similar OR to those conferred by the (DR3)DQA1*0501- DQB1*0201/(DR3) DQA1*0501-DQB1*0201 genotype.. Comment With a mean prevalence of 5.6 % in the general pediatric population, CD is much more common in the Saharawi than in any other population thus far studied in the 17 world. However, the role of genetic and environmental factors in causing such an impressive disease frequency is unknown. In this work, we established the degree of association with CD of the HLA class II DQ haplotypes and genotypes present in the Saharawi population. As in other human groups, the main predisposing molecule was represented by the combined presence of the DQA1*0501 and DQB1*0201 alleles (DQw2) encoding the putative predisposing heterodimer a* b*0201. Given the many factors that could cause a chronic enteropathy in a developing country (e.g. intestinal infections and parasites, primary malnutrition), concerns had been raised on the definition of the celiac phenotype in such a difficult environment. The presence of a strong association with the established HLA predisposing molecules provides indirect

51 COELIAC DISEASE IN THE SAHARAWI 37 confirmation of the reliability of the CD diagnosis (based on both the serum EMA testing and the small intestinal biopsy) in these North African children. Most importantly, these results are also clear evidence that CD has a similar HLA association in distantly related populations. The aetiological mechanism of the class II association involves binding and T cell recognition of gluten derived peptides deaminated in the 25 small intestine by the enzyme tissue transglutaminase, and these properties of class II molecules are very likely conserved among different ethnic groups. In a manner consistent with the high disease prevalence, the main CD susceptibility haplotype DQA1*0501-DQB1*0201, detected nearly always within a DR3 haplotype, exhibited one of the highest frequencies in the world in the general background Saharawi population. This could partially explain the very high CD prevalence observed in the Saharawi population. However, it should be noted that a similar frequency of DQA1*0501-DQB1*0201 has been found also in other populations, such 26 as the Sardinians where the disease is around 5 times less frequent (1%). Our results therefore suggest that other non-dq genes or environmental factors are also operative in determining the high CD prevalence observed in the Saharawi. From an evolutionary perspective, such disease frequency might be related to the high prevalence of genes such as the Dq2, that have been selectively neutral or even beneficial for several hundred generations, that became disadvantageous following quick and dramatic changes of the environmental context in which the Saharawis lived. In the traditional Saharawi diet, the intake of gluten was very low and its introduction usually occurred after the second or third year of life. During the last century, in a span of few generations, wheat-based products, especially bread, have become the staple food after the Spanish colonization. Nowadays, gluten is introduced early in the Saharawi infant's diet, occasionally as early as the first month of life. Since in CD the 27 degree of intestinal mucosal damage is related to the amount of ingested gluten, the enteropathy of the ancestral CD-prone individuals was presumably milder. As such, in terms of genetic fitness, it did not represent a major selective disadvantage or might have even exerted some protective effect against intestinal infections because a moderately atrophic jejunal mucosa partially lacks the membrane receptors required 28 for microorganisms adhesion. Moreover, a mild CD and the accompanying small degree of inflammation of the jejunal mucosa, can also provide a source of reactive lymphomonocytes, thus further increasing the protective effect against intestinal infections. In contrast, the severe celiac enteropathy currently found in Saharawi children is associated with an early and high gluten intake and it is no longer neutral or protective but instead harmful, as it causes early intestinal malabsorption and 29 malnutrition, potentially leading to death. From these data a more general theory can be proposed. High disease frequency of a complex trait in a given population might be related to the high prevalence of genes that have been selectively neutral or even beneficial for several hundred generations, but became disadvantageous following quick changes of the environment. According to this interpretation, complex traits can be regarded as residual traits caused by dramatic transformations of the environmental context in which a given population used to live across evolutionary time periods. Some environmental transformations can occur too quickly to be accompanied by changes in the genetic structure of the human populations. Indeed, the evolutionary time periods necessary to modify the genetic

52 38 COELIAC DISEASE IN THE SAHARAWI Dimension Sub Saharian Africa SEN FIL Asia 1 + Europe and N. Africa + + Gre Sard Bulg HONG KONG Cre ETH Rom + CHI JAP + + Ita + Tun -0.6 Tur POL + Mor + + Fra Spa + -1 S AFR + GAB + Cze Alg+ Nw Rus Saha + + Nor + Gb Dimension 1 Fig. 1. Multidimentional scaling of 26 populations tested for HLA DRB1-DQA1-DQB1 haplotypes. The first dimension splits Eastern and Western world populations. The Caucasians are closely grouped and located in the African range of variability. The second dimension explains the diversity of the Sub-Saharan African and Caucasian groups, which include the North-African and Saharawi samples. On the right box, European and North African patterns were enlarged Population samples: ALG, Algerian ; BULG, Bulgarian ; CHI, Chinese ; CRE, Cretan ; CZE, Czech ; ETH, Ethiopian ; FIL, Filipino ; FRA, French; ITA, Italian; RUM, Rumanian; S-AFR, South-African; SEN, Senegalese; SPA, Spanish ; GAB, Gabonese ; GB, British ; GRE, Greek ; HONG KONG, Hong-Kong ; JAP, Japanese ; MOR, Moroccan ; NOR, Norwegian ; NW-RUS, North-Western Russian ; POL, Polynesian ; TUN, Tunisian ; TUR, Turkish ;SARD, Sardinian and in circle font the Saharawi (current work). asset of a given population are on a much larger scale in comparison to those required to change environmental factors, i.e. the socalled gene-environment evolutionary tempo mismatch. Finally there is another important lesson coming from CD to understand the bases of other multifactorial traits. CD tells us that some of the environmental factors involved in triggering these complex traits could be nearly ubiquitous (gluten). It might be extremely difficult to identify them in diseases in which the target organ is not able to regenerate when the trigger factor is removed. Acknowledgements We wish to thank Stefano De Virgiliis, M.Doloretta Macis, Mauro Congia, Michael Whalen, Laura Morelli, Patrizia Zavattari and Antonio Cao for advice and support. References 1. Mäki M, Collin P. Coeliac disease. Lancet 1997; 349: Risch N. Assessing the role of HLA-linked and unlinked determinants of disease. Am

53 COELIAC DISEASE IN THE SAHARAWI 39 J Hum Genet 1987; 401: Petronzelli F, Bonamico M, Ferrante P, Grillo R, Mora B, Mariani P, et al. Genetic contribution of the HLA region to the familial clustering of coeliac disease. Ann Hum Genet 1997; 61: Polanco I, Biemond I, van Leeuven A, Shreuder I, Meera Khan P, Guerrero J, et al. In: mcconnel RB, ed. The genetics of coeliac disease. Lancaster 1981: MTP, Sollid LM, Thorsby E. HLA susceptibility genes in celiac disease: genetic mapping and role in pathogenesis. Gastroenterology 1993; 105: Lundin KE, Scott H, Hansen T, Paulsen G, Halstensen TS, Fausa O, et al. Gliadinspecific, HLA-DQ alpha 1*0501,beta 1*0201 restricted T cells isolated from the small intestinal mucosa of celiac disease patients. J Exp Med 1993; 178: Lundin KE, Scott H, Fausa O, Thorsby E, Sollid LM. T cells from the small intestinal mucosa of a DR4, DQ7/DR4, DQ8 celiac disease patient preferentially recognize gliadin when presented by DQ8. Hum Immunol 1994; 41: Greco L, Corazza G, Babron MC, Clot F, Fulchignoni-Lataud MC, Percopo S, et al. Genome search in celiac disease. Am J Hum Genet 1998; 62: Zhong F, McCombs CC, Olson JM, Elston RC, Stevens FM, McCarthy CF, et al.an autosomal screen for genes that predispose to celiac disease in the western counties of Ireland. Nat Genet 1996; 14: Houlston RS, Tomlinson IP, Ford D, Seal S, Marossy AM, Ferguson A, et al. Linkage analysis of candidate regions for coeliac disease genes. Hum Mol Genet 1997; 6: Culverhouse R, Suarez BK, Lin J, Reich T. A perspective on epistasis: limits of models displaying no main effect. Am J Hum Genet 2002; 70: Dahlman I, Eaves IA, Kosoy R, Morrison VA, Heward J, Gough SC, et al. Parameters for reliable results in genetic association studies in common disease. Nat Genet 2002; 30: Verge CF, Vardi P, Babu S, Bao F, Erlich HA, Bugawan T, et al. Evidence for oligogenic inheritance of type 1 diabetes in a large Bedouin Arab family. J Clin Invest 1998; 15: Trier JS. Celiac sprue. N Engl J Med 1991; 325: Catassi C, Fabiani E, Rätsch IM, Coppa GV, Giorgi PL, Pierdomenico R, et al. The coeliac iceberg in Italy. A multicentre antigliadin antibodies screening for coeliac disease in school-age subjects. Acta Paediatr 1996; Suppl 412: Not T, Horvath K, Hill ID, Partanen J, Hammed A, Magazzù G, et al. Celiac disease risk in the USA: high prevalence of antiendomysium antibodies in healthy blood donors. Scand J Gastroenterol 1998; 33: Catassi C, Rätsch IM, Gandolfi L, Pratesi R, Fabiani E, El Asmar R, et al. Why is coeliac disease endemic in the people of the Sahara? Lancet 1999; 354: Singer S, Tanguay R. Use of chelex to improve the PCR signal for a small number of cells. Amplifications 1989; 3: Horn GT, Bugawan TL, Long CM, Erlich HA. Allelic sequence variation of the HLA-DQ loci, relationship to serology and to insulin-dependent diabetes susceptibility. Proc Natl Acad Sci 1988; 85: Wordsworth BP, Allsopp CEM, Youg RP, Bell JI. HLA-DR typing using DNA

54 40 COELIAC DISEASE IN THE SAHARAWI amplification by the polymerase chain reaction and sequential hybridization to sequence-specific oligonucleotide probes. Immunogenetics 1990; 32: Bugawan TL, Erlich HA. Rapid typing of HLA-DQB1 DNA polymorphism using nonradiactive oligonucleotide probes and amplified DNA. Immunogenetics 1991; 33: Thomson G. Mapping disease genes: family-based association studies. Am J Hum Genet 1995; 57: Noble JA, Valdes AM, Cook M, Klitz W, Thomson G, Erlich HA. The role of HLA class II genes in insulin-dependent diabetes mellitus: molecular analysis of 180 Caucasian, multiplex families. Am J Hum Genet 1996; 59: Kruscal, J.B. Multidimensional Scaling by Optimizing Goodness of fit to a Nonmetric Hypothesis. Psychometrika 1964; 29: Schuppan D. Current concepts of celiac disease pathogenesis. Gastroenterology 2000; 119: Meloni G, Dore A, Fanciulli G, Tanda F, Bottazzo GF. Subclinical coeliac disease in schoolchildren from northern Sardinia. Lancet 1999; 353: Marsh MN. Gluten, major histocompatibility complex, and the small intestine: a molecular and immunobiologic approach to the spectrum of gluten-sensitivity "celiac sprue". Gastroenterology 1992;102: Kerneis S, Chauviere G, Darfeuille-Michaud A, et al. Expression of receptors for enterotoxigenic Escherichia coli during enterocytic differentiation of human polarized intestinal epithelial cells in culture. Infect Immun 1992;60: Rätsch IM, Catassi C. Coeliac disease is a primary, potentially treatable, health problem of Saharawi refugee children. Bulletin of the World Health Organization 2001; 79: Arnaiz-Villena A, Benmamar D, Alvarez M, Diaz-Campos N, Varela P, Gomez- Casado, et al. HLA allele and haplotype frequencies in Algerians. Relatedness in Spaniards and Basques. Hum Immunol 1995; 43: Ivanova R, Naoumova E, Lepage V, Djoulah S, Yordanov Y, Loste MN et al. HLA- DRB1, DQA1, DQB1 DNA polymorphism in the Bulgarian population. Tissue Antigens 1996; 47: Chang YW, Hawkins BR. HLA class I and class II frequencies of a Hong Kong Chinese population based on bone marrow donor registry data. Hum Immunol 2002; 56: Fernandez-Viña M, Gao X, Moraes ME, Moraes JR, Salatiel I, Miller et al. Alleles at four HLA class II loci determined by oligonucleotide hybridization and their associations in five ethnic groups. Immunogenetics 1991; 34: Arnaiz-Villena, A., Iliakis, P., Gonzalez-Hevilla, M., Longas, J., Gomez-Casado, E., Sfyridaki, K., et al. The origin of Cretan populations as determined by characterization of HLA alleles. Tissue Antigens 1999; 53: Cernà,M, Fernandez-Vina M, Ivaskova E. and Stastny P. Comparison of HLA class II alleles in Gypsy and Czech populations by DNA typing with oligonucleotide probes. Tissue Antigens 1992; 39: Fort M, de Stefano, G.F., Cambon-Thomsen, A., Giraldo-Alvarez, P., Dugoujon, J.M., Ohayon, E., et al. HLA class II allele and haplotype frequencies in Ethiopian Amhara and Oromo populations. Tissue Antigens 1998; 51:

55 COELIAC DISEASE IN THE SAHARAWI Mack SJ, Bugawan TL, Moonsamy PV, Erlich JA, Trachtenberg EA, Paik YK, et al. Evolution of Pacific/Asian populations inferred from HLA class II allele frequency distributions. Tissue Antigens 2000; 55: Imanishi T, Akaza T, Kimura A, Tokunaga K, Gojobori T. Allele and haplotype frequencies for HLA and complement loci in various ethnic groups. In, Tsuji, K., Aizawa, M. and Sasazuki, T. eds.:, HLA Oxford University Press, Oxford, UK, Vol. 1, pp Migot-Nabias F, Fajardy I, Danze PM, Everaere S, Mayombo Y, Minh TN, et al. HLA class II polymorphism in a Gabonese Banzabi population. Tissue Antigens 1999; 53: Doherty DG, Vaughan RW, Donaldson PT, Mowat AP. HLA DQA, DQB, and DRB genotyping by oligonucleotide analysis, distribution of alleles and haplotypes in British Caucasoids. Hum Immunol 1992; 34: Papassavas EC, Spyropoulou-Vlachou M, Papassavas AC, Schipper RF, Doxiadis IN, Stavropoulos-Giokas C. MHC Class I and II phenotype, gene, and haplotype frequencies in Greeks using molecular typing data. Hum Immunol; 61: Hashimoto M, Kinoshita T, Yamasaki M, Tanaka H, Imanishi T, Ihara, et al. Gene frequencies and haplotypic associations within the HLA region in 916 unrelated Japanese individuals. Tissue Antigens 2000;44: Izaabel H, Garchon HJ, Caillat-Zucman S, Beaurain G, Akhayat O, Bach JF et al. HLA class II DNA polymorphism in a Moroccan population from the Souss, Agadir area. Tissue Antigens 1998; 51: Rønningen KS, Spurkland A, Markussen G, Iwe T, Vartdal F, Thorsby E. Distribution of HLA class II alleles among Norwegian Caucasians. Hum Immunol 1990; 29: Kapustin S, Lyshchov A, Alexandrova J, Imyanitov E, Blinov M. HLA class II molecular polymorphisms in healthy Slavic individuals from North-Western Russia. Tissue Antigens 1999; 54: Gao X, Zimmet P, Serjeantson SW. HLA-DR,DQ sequence polymorphisms in Polynesians, Micronesians, and Javanese. Hum Immunol 1992; 34: Hmida S, Gauthier A, Dridi A, Quillivic F, Genetet B, Boukef K. HLA class II gene polymorphism in Tunisians. Tissue Antigens 1995; 45: Saruhan-Direskeneli G, Esin S, Baykan-Kurt B, Ornek I, Vaughan R. Eraksoy M. HLA-DR and -DQ associations with multiple sclerosis in Turkey. Hum. Immunol. 1997; 55: Lampis R, Morelli L, Congia M, Macis MD, Mulargia A, Loddo M, et al. The interregional distribution of HLA class II haplotypes indicates the suitability of the Sardinian population for case-control association studies in complex diseases. Hum Mol Genet 2000; 9:

56

57 Catassi C, Fasano A, Corazza GR (eds): Primary prevention of coeliac disease. The utopia of the new millennium? Perspectives on Coeliac Disease, vol. 1, AIC Press, pp Primary prevention of coeliac disease by favourable infant feeding practices 1,2 1,3 Anneli Ivarsson, Lars Åke Persson and Olle Hernell Department of Public Health and Clinical Medicine, Epidemiology and Department 3 of Clinical Sciences, Pediatrics, both at Umeå University, Umeå, Sweden, and ICDDR,B: Centre for Health and Population Research, Dhaka, Bangladesh. Coeliac disease is now recognised as a common health problem throughout a large 1-12 part of the world. An effective treatment is available through adherence to a strict gluten-free diet. Screening studies have revealed that most cases are undiagnosed, which indicates the need for active case recognition, and possibly also a need for screening efforts. However, even when coeliac disease has been diagnosed, the widespread use of gluten-containing foods makes compliance with the treatment 15 difficult. Clearly, then, it is desirable to explore additional strategies such as primary prevention. Family clustering The risk for coeliac disease is higher in first degree relatives compared to the general population. In family members of coeliac patients, a prevalence of 10% is often 16 mentioned, and in monozygous twins the concordance rate has been estimated at %, or even higher. Family clustering is generally considered as evidence for a strong genetic influence in coeliac disease susceptibility. However, the fact that family members share not only genetics but, in many respects, also the same environment, should be taken into consideration. This is especially true for monozygous twins, who have a similar environment not only during childhood, as is the case for most siblings, but also have a similar environment in utero. A complex aetiology The aetiology of coeliac disease is not fully understood, largely as a consequence of its complexity, although our understanding of different aspects of the disease process is rapidly increasing. Based on present knowledge, an immunological pathogenesis for the disease seems most likely. In some genetically susceptible individuals, exposure to gluten, or related proteins in rye and barley, triggers an errant immune response, which 43

58 44 INFANT FEEDING PRACTICES AND COELIAC DISEASE by means of a cascade of events results in a chronic enteropathy. The human leukocyte antigen (HLA)-DQ2 molecule is expressed by more than 90% of coeliac disease patients, compared to 20-30% of healthy controls, and the 22 majority of the remaining patients express the DQ8 molecule. It has been estimated that the HLA-complex confers 40% of the sibling family risk, and that non-hla genes are the most important. However, efforts to identify these contributing non-hla genes have thus far not resulted in conclusive evidence. In assessing the impact of genetics, the ratio lis crucial. This ratio is constructed by the prevalence in relatives of 28 an affected individual over the prevalence within the general population. Thus, if causal environmental exposures are aggregated in the family, the impact of genetics 29 will be overestimated. Most common diseases have a multifactorial aetiology, and they thus develop through an interaction between an individual's genetic predisposition and various 30 environmental exposures. This has been shown for insulin dependent diabetes mellitus (IDDM), which is an autoimmune disease with many similarities to coeliac disease. The immune system has a key function in the pathogenesis of both diseases, 31 and there is an increased risk for coeliac disease in persons with IDDM. It has recently been suggested that each genetic risk factor, taken separately, may frequently be present in the general population, and it is the combination of some of 26 these and their interaction with environmental factors that induces coeliac disease. Thus, in addition to the mere presence of gluten in the diet, environmental exposures may also be expected to be part of the causal pattern responsible for coeliac disease. Failure of oral tolerance Environmental exposures, including infant feeding, influence the immunological process resulting in oral tolerance - or intolerance - to a food constituent. Considering recent knowledge of the aetiology of coeliac disease, this may be viewed at least hypothetically as a failure in the development of, or the later loss of, oral 35 tolerance. In most individuals, oral tolerance towards gluten develops and prevails throughout life. If oral tolerance fails to develop, however, or is later broken down, then gluten may act like a dangerous foreign antigen, resulting in the development of coeliac disease. The Swedish epidemic Sweden has experienced an unusual epidemic of symptomatic coeliac disease in children. The incidence reached levels higher than ever reported, and the decline that followed was amazingly abrupt (Fig. 1). Only children below two years of age were affected, and most of the cases had symptoms as severe as those that had been observed earlier. These findings are based on our population-based prospective incidence register established in 1991, covering 40% of the Swedish child population, and 36 retrospective data collected back to 1973 from 15% of the child population. Recently it was shown that this was indeed an epidemic of coeliac disease enteropathy, and not only a shift in the proportion of cases with symptomatic

59 INFANT FEEDING PRACTICES AND COELIAC DISEASE 45 Cases per person-years years years years Year of diagnosis Fig. 1. Annual incidence rate of coeliac disease in different age groups of children from to From with permission. enteropathy that were thus more readily diagnosed. A screening of 2½-year-old Swedish children including birth cohorts both from the epidemic and post-epidemic years showed this. In the cohort of (epidemic years) with 3004 children, there were 22 cases diagnosed due to symptomatic coeliac disease, and out of screened children, as many as 9 had previously unrecognised coeliac disease. In the cohort of (post-epidemic years), only about half as many coeliac disease cases were identified, and this was true for symptomatic cases as well as cases detected by screening, with the studies being comparable in all other respects (Carlsson A, personal communication). Obviously, the epidemic cannot be explained by genetic changes in the population, as it occurred over such a short time period. This epidemic of coeliac disease indicates an abrupt increase and decrease, respectively, of one or a few causal factors influencing a large proportion of Swedish infants over the period in question. As only children below two years of age were affected, changes in infant feeding practices were suspected to have contributed. If these causal exposures could be solidly identified, this might open doors to the development of a primary prevention strategy. Infant feeding The infant feeding pattern is one of the few environmental exposures other than dependence on gluten proteins in the diet that has been approached as possibly contributing to coeliac disease aetiology. Taking into consideration established guidelines for causation (table), some aspects of infant feeding will be discussed as potential risk determinants for development of coeliac disease.

60 46 INFANT FEEDING PRACTICES AND COELIAC DISEASE Temporal relationship Plausibility Consistency Strength Dose-response relationship Reversibility Study design Coherence of evidence 41 Adapted from. Breast feeding Table. Guidelines for the evaluation of causality. Does the possible cause precede the effect? Is the association consistent with other knowledge? Have similar results been shown in other studies? What is the strength of the association between the possible cause and the effect? Is increased exposure to the possible cause associated with increased effect? Does the removal of a possible cause lead to reduction of disease risk? Is the evidence based on a strong study design? How many lines of evidence support the conclusion? Based on increasing knowledge about the immunological impact of breast milk, it seems plausible that introduction of a dietary antigen while the child is still breast-fed might increase the likelihood of developing oral tolerance to that antigen. Whether or not this relationship holds true for dietary gluten and the risk for development of coeliac disease has not been conclusively established. Based on observations of coeliac disease patients, it was suggested as early as the s that breast-feeding delays onset of the disease, a view supported by later similar studies. Furthermore, an increase in breast-feeding was suggested as a possible factor contributing to the declining incidence of coeliac disease in the early 1970s in England, Scotland and Ireland. 52 In the 1980s, the Italian case-referent studies by Auricchio et al (216 cases/ siblings) and Greco et al (201 cases/1949 referents) demonstrated that coeliac disease cases were breast-fed for a shorter duration than the referents. This was confirmed in a 54 Swedish case-referent study by Fälth-Magnusson et al (72 cases/288 referents), and 55 recently in a German study by Peters et al (143 cases/137 referents). In contrast, in a 56 family study Ascher et al did not find a difference in breast-feeding duration when comparing screening-detected cases with their siblings (8 cases/73 siblings). This study was comparatively small, and the design involved overmatching with respect to dietary factors, constraints discussed by the authors. Taken together, the case-referent studies demonstrate that coeliac disease cases in general have been breast-fed for a shorter period than other children. However, these studies could not clarify whether breast-feeding had a direct causal effect, or if the protective effect was indirect as a consequence of the postponed introduction of infant formula (more specifically cow's milk protein), or if it occurred through reduction of the amount of dietary gluten ingested at an early age. We have recently reported results from a case-referent study (491 cases/ referents), which could eliminate some of the above-mentioned constraints through a somewhat different design in comparison with the previously mentioned studies. Our

61 OR (95%CI) INFANT FEEDING PRACTICES AND COELIAC DISEASE 47 study was population-based, with a high participation rate, and the results should thus be representative for Sweden at large. Since our findings are consequences of biological phenomena, they should be valid for infants in general. Only incident cases, i.e. newly diagnosed cases, were included, which reduced the recall period. A comprehensive questionnaire concerning children's diet and health in general was mailed to the families; it did not reveal our special interest in coeliac disease. A semiquantitative food frequency questionnaire was used to assess both the consumption of gluten-containing cereals at the time when these cereals were introduced into the diet and the total amount consumed by seven months of age. The amount of glutencontaining flour in homemade foods was calculated based on standard Swedish recipes and the amount in industrially produced foods was obtained from the manufacturers. Multivariate analyses were used to adjust risk estimates for confounding and to suggest 57 causal relationships. Our main finding was that the risk for coeliac disease was reduced if the child was breast-fed during the time period when gluten-containing foods were introduced [Odds 57 Ratio (OR) = 0.59, 95% Confidence Interval (CI) ]. This protective effect was even more pronounced if the child was also breast-fed beyond the period of gluten introduction (OR=0.36, 95% CI ) (Fig. 2). These risk estimates were adjusted for the age of the infant when gluten was introduced into the diet and the amount of gluten that was given ½ ¼ Discontinued Continued Continued beyond Breast feeding status at introduction of gluten Fig. 2. Breast-feeding status (BF) at introduction of gluten-containing flour into the diet and risk (OR, 95% CI) for coeliac disease before two years of age. A protective effect of breast-feeding was supported by our ecological study of the Swedish epidemic, i.e. using aggregated data to explore any temporal relationship between the changes in incidence rate and changes in infant dietary patterns. Both the rise and later fall in incidence had a temporal relationship to a change in the proportion 36 of infants introduced to gluten while still being breast-fed. It is important to note that at the time of these studies the majority of Swedish

62 48 INFANT FEEDING PRACTICES AND COELIAC DISEASE infants were breast-fed for six months or longer. As a result, most of the infants were introduced to cow's milk products and other food products while still being breast-fed. Also, for most infants the termination of breast-feeding did not coincide with introduction of infant formula, but rather with increased ingestion of other foods. Moreover, in the case-referent study we could show that a reduced risk for coeliac disease still remained when the age of the infants at introduction of gluten into the diet was considered, along with the amount of gluten given at that time (fig. 2). Thus, our findings strongly supported breast-feeding as directly reducing the risk for coeliac disease, and not merely influencing the risk indirectly through changes in other exposures. It has recently been suggested that the intestinal lesion in coeliac disease might be caused by an uncontrolled production of pro-inflammatory IFN-gby intraepithelial T cells that is not sufficiently counteracted by an increased production of downregulating TGF-b1 (Forsberg G, personal communication). It is thus tempting to 35 speculate that the TGF-b1 in breast-milk compensates for this although this is only one of several possible mechanisms for a protective effect of human milk. Thus, a protective effect of being breast-fed when dietary gluten is introduced is supported by several epidemiological studies of different design. Moreover, the protective effect is biologically plausible when taking into account our present knowledge of breast-milk composition and the impact on immune responses, and the aetiology of coeliac disease. Amount of dietary gluten The dose of dietary antigen ingested may influence whether or not oral tolerance develops. It is not yet clear if this applies to the amount of gluten given in relation to the risk of developing coeliac disease. Interestingly, a larger consumption of wheat gluten was reported for healthy infants in Sweden and Italy as compared to Finland, Denmark and Estonia, and the former countries also reported a higher occurrence of coeliac disease. The study designs, involving aggregated data, did not, however, allow adjustments for differences in other exposures. 54 The Swedish case-referent study by Fälth-Magnusson et al indicated that the coeliac disease cases, more often than the referents, were introduced to gluten by means of gluten-containing follow-on formula. In infancy it is clear that bottle-feeding, compared to feeding by cup and spoon, more readily contributes a larger amount of food, and thus also a larger amount of gluten. In the Fälth-Magnusson study, however, it was not possible to estimate the amount of gluten given to the infants. In our case-referent study it was possible, for the first time, to assess the consumption of gluten-containing cereals on an individual basis by use of a semi- 57 quantitative food frequency questionnaire. We could show that introduction of glutencontaining foods in larger amounts, as compared to small or medium amounts, was an independent risk factor for coeliac disease development (adjusted OR=1.5, 95% CI ). Also, at seven months of age the cases consumed larger amounts of glutencontaining flour than the referents. It is important to note that through the use of multivariate analyses we adjusted for differences in breast-feeding practices and the

63 INFANT FEEDING PRACTICES AND COELIAC DISEASE 49 age of the infant when gluten was introduced into the diet. Thus, our findings clearly indicate that introduction of gluten in larger amounts increases the risk for coeliac disease. Furthermore, we found that the type of food used as the source of gluten, i.e. 57 solid foods or follow-on formula, was not important as an independent risk factor. The relevance of our observations is therefore not only applicable to Sweden with its custom of using gluten-containing follow-on formula from the age of six months. The daily amount of gluten consumed during infancy as a risk factor for coeliac disease is further supported by our ecological study of the Swedish epidemic, where we used aggregated data to explore any temporal relationship between changes in 36 incidence rate and changes in infant dietary patterns. The rise in incidence was preceded by a twofold increase in the average daily consumption of gluten through the use of follow-on formula, and later, the fall in incidence coincided with a consumption 36 that was decreased by one-third. 62 Marsh et al concluded that gluten-sensitised individuals respond in a time-related and dose-dependent fashion to gliadin, which is an observation supported by several other studies. These experimental studies do not, however, clarify whether the amount of gluten is also crucial when infants are introduced to this antigen for the first time. Taken together, there is evidence to suggest that consumption of a large amount of gluten-containing flour (increased antigen dose) during infancy increases the risk for coeliac disease. It is, however, not clear whether there is a direct dose-response effect or a threshold effect. Furthermore, it seems likely that the amount of gluten tolerated varies with the genetic predisposition of the individual, other environmental exposures, and the age of the individual. Age at introduction of gluten There might be an age interval during which humans have decreased ability to develop oral tolerance to a newly introduced dietary antigen. Hypothetically, the age of the infant upon introduction of gluten into the diet might thus influence the risk for coeliac disease. In a comparison of English coeliac disease patients in the 1950s and 1960s, it was suggested that earlier introduction of dietary gluten resulted in earlier presentation of 67 the disease. Some clinical studies in which differences in breast-feeding duration have been taken into account did not show such a relationship. In fact, a delayed introduction of gluten into the diet of infants was suggested as contributing to the decline in incidence of coeliac disease in England, Scotland and Ireland in the 1970s. However, at that time comparable dietary changes occurred in Sweden without any observed change in incidence. Furthermore, the increased incidence in Swedish children in the middle of the 1980s was preceded by a delayed introduction of dietary 36,38-39 gluten from four until six months of age. Thus, these ecological observations resulted in contradictory findings. However, a study design based on aggregated data cannot by itself provide conclusive evidence. However, the case-referent design based on individual data allows for adjustments for differences in other exposures. Such studies in which adjustments have been made for differences in breast-feeding duration have indicated that the age of the infant at

64 50 INFANT FEEDING PRACTICES AND COELIAC DISEASE introduction of dietary gluten is of no importance with respect to coeliac disease risk. In our case-referent study we could move one step further by also adjusting for the differences in amount of gluten-containing foods given at various ages of introduction. Our results also suggested that the age of the infant at introduction of dietary gluten was 57 of no importance for coeliac disease risk, although with the limitation that only the first year of life was evaluated. Thus, present evidence does not support age of the infant at the time of gluten introduction as an independent risk factor for coeliac disease development. However, even when considered together these studies do not encompass all possible ages for introduction of gluten, e.g. after one year of age or even later. Thus, this possibility also needs to be evaluated. Other possible contributing factors Infections It has been proposed that adenovirus type 12, and also other adenoviruses, could initiate coeliac disease as a consequence of sequence similarities between a protein produced in conjunction with the viral infection and A-gliadin. This has been questioned, but not convincingly excluded. Gastrointestinal infections cause a 74 disruption of the barrier function of the small intestinal mucosa, which theoretically could result in an increased antigen penetration and unfavourable immune responses. We found a higher risk for coeliac disease in children born during summer as 75 compared to winter, but only in children below two years of age. These findings indicate that environmental exposure(s) with a seasonal pattern may have a causal effect. A temporal relationship suggests that this might be due to a causal effect of infections during foetal life and/or an interaction between infections and introduction of gluten into the diet. However, any exposure with a seasonal pattern might be the explanation, and non-infectious exposures should also be explored. In our case-referent study we found that children who experienced three or more infectious episodes before six months of age had an increased risk for coeliac disease 76 before two years of age (adjusted OR=1.4, 95% CI ). This was true even when episodes of gastroenteritis were excluded, and after adjustments were made for differences in infant feeding patterns and family socio-economic group. Infectious episodes later in infancy were not included, as these could be secondary to the disease itself. The risk for coeliac disease increased considerably if in addition to having many infections the child was also introduced to gluten in large amounts, as compared to 76 small and medium amounts. Socio-economic background In our case-referent study children in families belonging to the lower as compared to middle and upper socio-economic strata of Swedish society had an increased risk for 76 coeliac disease (adjusted OR=1.4, 95% CI ), but again only in the age group below two years. Our reported estimate was adjusted for differences in infant feeding practices and infectious episodes. This indicates that there are also other as yet unidentified environmental exposures contributing to the causal pattern of coeliac disease.

65 INFANT FEEDING PRACTICES AND COELIAC DISEASE 51 A multifactorial aetiology A simplified model of the multifactorial aetiology of coeliac disease is outlined below (Fig. 3). It is based on the results of our own studies put into the context of present knowledge of coeliac disease aetiology. The emphasis in the model is on the influence of environmental exposures on coeliac disease risk. Throughout life, including foetal life, an individual's genetic makeup interacts with the continuous and varying exposures of the environment. Theoretically, both genes and the environment, as well as the interaction between them, confer either increased or reduced disease risk. The exposures of causal importance most likely vary throughout life, and the range of potential risk determinants is wide. More specifically, with regard to coeliac disease the immunological response to dietary gluten is shaped. Genetic susceptibility and presence of dietary gluten are considered necessary casual factors, i.e. without these factors coeliac disease will not develop. However, component causal factors also contribute to whether or not coeliac disease develops. Combined, the necessary causes and one or several component causes produce a 77 sufficient cause, i.e. development of the disease is unavoidable. As component causal factors in coeliac disease aetiology we suggest: i) duration of breast-feeding, ii) whether breast-feeding is ongoing or not when gluten is introduced into the diet, iii) amount of gluten given to infants when introduced into the diet, and iv) repeated infectious episodes early in life. Associated factors are not considered to have a causal effect by themselves, but act through other directly causal factors. When identified, however, they may be used as markers for an increased disease risk, and thereby focus the search for causal factors. Such a factor is seasonality in births, with an increased risk for coeliac disease in Season Dietary recommendations Food contents Socioeconomic conditions Structural factors Associated factors Infections Breast feeding??? Component causal factors Amount DIETARY GLUTEN NECESSARY CAUSAL FACTORS GENETICS sex Immunopathogenesis Fetal life Infancy Childhood Adulthood Fig. 3. Causal model on the multifactorial aetiology of coeliac disease. 37 From, with permission.

66 Age (years) 52 INFANT FEEDING PRACTICES AND COELIAC DISEASE children born during summer as compared to winter. It might be that seasonality reflects a varying exposure to infectious episodes throughout the year, but other possible explanations should be explored. Another associated factor is the increased risk for coeliac disease in Swedish children from the lower as compared to middle and upper socio-economic strata, suggesting that there are further component causes yet to be identified. Structural factors are any change on a societal level influencing the risk for coeliac disease. However, these must of course exhibit their effect through component causes close to the individual, as illustrated in the depicted model (Fig. 3). These could include dietary recommendations influencing how gluten is introduced into the infant diet, and changes in the gluten content in industrially produced infant foods. What can be learnt from the Swedish epidemic? What caused the epidemic? Changes in infant feeding practices have been suspected to contribute to the Swedish epidemic of coeliac disease in children. To clarify whether or not this was the case, we used an ecological approach comparing estimated yearly changes in infant feeding practices with the yearly incidence rate of coeliac disease in children below two 36 years of age. We collected national data for the years 1980 to 1997 on duration of breast-feeding 36 and intake of gluten-containing cereals in Swedish infants. The latter was estimated by changes in gluten intake by means of industrially produced follow-on formulas as an estimate of changes in total gluten intake. These follow-on formulas, combined with 59 porridge, provide about half of the total intake of gluten proteins, and of that half, follow-on formulas account for about 90%, according to the manufacturers. It should be noted that the incidence rate for a particular year in children below two years of age might also be influenced by the exposure pattern of the two preceding years. The reason for this, as illustrated by the Lexis diagram, is that the incidence rate in children below two years of age in 1997, for instance, is based on cases diagnosed that year, and these children were born during the period 1995 to 1997, which is thus the period during which they might have been exposed (Fig. 4). Incidence rate in children 0-2 years of age in Exposure period Fig. 4. A Lexis diagram illustrating the population and exposure periods on which the incidence rate by age is based.

67 INFANT FEEDING PRACTICES AND COELIAC DISEASE Our results showed rapidly increasing incidence rates between 1985 and The period of interest with regard to exposure was characterised by i) about half of the infants being breast-fed at six months of age, ii) doubling of the average daily consumption of flour through use of follow-on formula from 1981 to 1983 with regard to the total amount of wheat, rye and barley, while the amount of oats decreased, and iii) a national recommendation at the end of 1982 to postpone introduction of gluten from four to six months of age (Fig. 5).This latter change was in accordance with European Incidence rate Breastfeeding index Wheat/Rye/Barley index Changed national recommendations Fig. 5. Annual incidence rate of coeliac disease per person years for children years of age from 1980 to 1997, in relation to breast-feeding habits, flour consumption and c h a n g e d n a t i o n a l re c o m m e n d a t i o n s. T h e a r ro w s i n d i c a t e c h a n g e d recommendation on introduction of gluten to infants. Breast-feeding index based on the proportion of all children breast-fed at six months of age (1980; index 100 = 37%). Wheat/Rye/Barley index based on the estimated average daily consumption of these cereals -1-1 provided by follow-on formulas (1980; index 100 = 16 gram x child x day ). 36 From with permission. 78 recommendations. A rapid decline in the incidence rate began in 1995, and was still ongoing through The period of interest with regard to exposure was characterised by i) a continuous increase (from 54% to 76%) in the proportion of infants still breast-fed at six months of age, ii) the average daily consumption of flour through the use of follow-on formula decreased by one third starting in 1995 with regard to the total amount of wheat, rye and barley, while the amount of oats increased, and iii) the national recommendation was changed in the autumn of 1996 to introduction of gluten into the diet in smaller amounts from four months of age, and preferably while the child is still breast-fed (Fig. 5). However, starting at the end of the 1980s, and particularly during a one-year period starting in the autumn of 1995, the media focused much attention on the increased incidence of coeliac disease in relation to the current dietary pattern. This most likely promoted a change in dietary patterns that was more extensive than revealed

68 54 INFANT FEEDING PRACTICES AND COELIAC DISEASE by our study, and that took place before the new national recommendation was launched. A certain delay between changes in exposure and disease occurrence would be expected. Indeed, in the 1980s the change in gluten intake with regard to both amounts consumed during infancy and age at introduction preceded the increase in incidence. In contrast, the decrease in amounts of gluten consumed was first noted in 1995, i.e. the same year as the decline in incidence started. Further, in 1997 the estimated amount of gluten consumed had decreased by one third, while the incidence rate had already returned to the level of the early 1980s. It should also be noted that the proportion of infants breast-fed at six months of age continuously increased during the 1990s without any sharp increase before, or coinciding with, the decrease in incidence. Hence, the changes in exposure caused by each of these factors alone cannot explain the entire epidemic. Considering the combined effects of changes in infant feeding practices over time, it is clear that both the rise and fall in incidence were accompanied by a change in the proportion of infants introduced to gluten in small amounts while still being breast-fed. Thus, Swedish infant feeding practices have shifted over time from a favourable to an unfavourable and then back again to a favourable pattern with respect to coeliac disease risk. Public health impact The public health impact of a change in causal exposures can be estimated by the population attributable fraction (AF), which takes into account the prevalence of different exposures among the cases (Pc), and the adjusted odds ratios (OR) of these exposures [AF = Pc (OR-1)/OR]. We used our case-referent study, performed during the peak of the epidemic, for such estimates. Our analysis revealed that about half of the coeliac disease cases during the epidemic might have been avoided if all infants had been introduced to gluten in small amounts while still being breast-fed (Fig. 6) = Preventable cases No of cases A B C D Continued Small-medium Continued Large Discontinued Small-medium Discontinued Large Breast-feeding status at introduction of flourand amount of flour given Fig. 6. Preventable coeliac disease cases below two years of age with respect to breastfeeding status at introduction of gluten-containing flour into the diet and the amount of flour given. Risk estimates were based on conditional logistic regression with 392 matched sets of cases and referents, and adjusted for age of the infant when flour was introduced. 37 Odds ratio (95% CI); A) 1.0, B) 2.0 (1.4, 3.0), C) 2.8 (1.9, 4.0), D) 3.3 (2.3, 4.8). From with permission.

69 INFANT FEEDING PRACTICES AND COELIAC DISEASE 55 Thus, other casual exposures changing over time must also have contributed to the epidemic. However, it is likely that the proportion of coeliac disease cases that might be avoided by appropriate measures varies among different populations, being higher in populations with unusually unfavourable environmental exposures, as was apparently the situation for Swedish infants during the high incidence years of the epidemic. What will follow after the epidemic? In a Swedish study based on symptomatic adult coeliac disease patients, the highest 79 age-specific prevalence was reported for people born between 1927 and 1936, and interestingly, the majority of cases in our population-based adult screening study were 5 also born during that time period. Thus, these studies indicate a cohort effect, i.e. the life span of certain cohorts coincides with environmental exposures that have resulted in an excess risk for coeliac disease throughout life. Accordingly, it might also be that as a consequence of an unfavourable exposure during their first years of life, the birth cohorts of the epidemic years will carry an excess risk for coeliac disease throughout their lives. If so, the cohorts of the postepidemic period might have a decreased lifetime risk for coeliac disease. Thus far, at comparable ages the cohorts of this later period actually have a lower cumulative Cases per 1000 births Age (years) Fig. 7. Cumulative incidence of coeliac disease by age for the birth cohorts from 1973 to To reduce the graph complexity, the cohorts of 1973 to 1982 are aggregated in groups of five, the cohorts of 1983 to 1990 in groups of two, while the cohorts from 1991 to 1998 are reported separately.

70 56 INFANT FEEDING PRACTICES AND COELIAC DISEASE incidence than the cohorts of the epidemic period (Fig. 7). A longer follow-up will reveal to what extent this lower risk continues, and to what extent new cases develop later in life. Large screening studies of cohorts from both the epidemic and post-epidemic periods, with repeated screening for a considerable number of years, would increase our understanding of the natural history of coeliac disease development in relation to exposures early in life. These exposures would be complicated, of course, by the addition of exposures later in life. An option for primary prevention Primary prevention of coeliac disease is a clearly desirable option, as it would contribute substantially to the health of the general population. By definition, it aims at intervening before the disease has developed. Genetic modification of the gluten-containing cereals so that they lose their ability to trigger celiac disease might someday become an alternative, although this is something for the distant future. Another option, at least theoretically, would be induction of immune tolerance to gluten peptides by giving a vaccine. Even today, coeliac disease could be effectively prevented if the use of gluten-containing cereals was abandoned, but in most countries such a suggestion would be considered unacceptable. However, if coeliac disease has a multifactorial aetiology, which is likely, then primary prevention would be possible, at least in some individuals, without completely abandoning the use of dietary gluten. This might be attained through a change in component causal exposures, thereby increasing the chance for infants to develop oral tolerance to gluten, and possibly also promoting the maintenance of tolerance throughout life (Fig. 3). When knowledge is consolidated from the clinical, epidemiological and basic sciences, it seems likely that infant feeding practices, in addition to the mere presence of gluten in the diet, have an important role in coeliac disease development. Thus, a significant contribution to primary prevention seems possible through a gradual introduction of gluten-containing foods into the infant diet before breast-feeding is discontinued. The duration of breast-feeding in itself seems to reduce the risk even further, and a long duration should thus be promoted. The risk for coeliac disease is then clearly reduced, at least up to two years of age. This infant feeding pattern most likely also contributes to a decreased lifetime risk of coeliac disease, although this remains to be established. Only a few of all the potential causal environmental exposures have thus far been explored, and the search for such factors, which exhibit their effect during different periods in the life span, should be intensified. This approach most likely will lead to the identification of other entry-points for primary prevention. Acknowledgements We thank Don Kasarda, US Department of Agriculture, Albany, California, for valuable comments on the manuscript.

71 INFANT FEEDING PRACTICES AND COELIAC DISEASE 57 References 1. Grodzinsky E, Franzen L, Hed J, Ström M. High prevalence of celiac disease in healthy adults revealed by antigliadin antibodies. Ann Allergy 1992; 69: Catassi C, Rätsch I-M, Fabiani E, Rossini M, Bordicchia F, Candela F, et al. Coeliac disease in the year 2000: exploring the iceberg. Lancet 1994; 343: Not T, Horvath K, Hill ID, Partanen J, Hammed A, Magazzu G, et al. Celiac disease risk in the USA: High prevalence of antiendomysium antibodies in healthy blood donors. Scand J Gastroenterol 1998; 33: Johnston SD, Watson RGP, McMillan SA, Sloan J, Love AHG. Coeliac disease detected by screening is not silent--simply unrecognized. QJM 1998; 91: Ivarsson A, Persson LÅ, Juto P, Peltonen M, Suhr O, Hernell O. High prevalence of undiagnosed coeliac disease in adults: a Swedish population-based study. J Intern Med 1999; 245: Catassi C, Rätsch IM, Gandolfi L, Pratesi R, Fabiani E, El Asmar R, et al. Why is coeliac disease endemic in the people of the Sahara? [letter]. Lancet 1999; 354: Csizmadia CGDS, Mearin ML, von Blomberg BME, Brand R, Verloove-Vanhorick SP. An iceberg of childhood coeliac disease in the Netherlands [letter]. Lancet 1999; 353: Gandolfi L, Pratesi R, Cordoba JC, Tauil PL, Gasparin M, Catassi C. Prevalence of celiac disease among blood donors in Brazil. Am J Gastroenterol 2000; 95: Rabassa EB, Sagaró E, Fragoso T, Castañeda C, Gra B. Coeliac disease in Cuban children. Arch Dis Child 1981; 56: Sher KS, Fraser RC, Wicks AC, Mayberry JF. High risk of coeliac disease in Punjabis. Epidemiological study in the south Asian and European populations of Leicestershire. Digestion 1993; 54: Rawashdeh MO, Khalil B, Raweily E. Celiac disease in Arabs. J Pediatr Gastroenterol Nutr 1996; 23: Demir H, Yuce A, Kocak N, Ozen H, Gurakan F. Celiac disease in Turkish children: presentation of 104 cases. Pediatr Int 2000; 42: Marsh MN. Screening for latent gluten sensitivity: questions many, but answers few [editorial; comment]. Eur J Gastroenterolog Hepatol 1996; 8: Catassi C, Fabiani E, Ratsch IM, Coppa GV, Giorgi PL. Celiac disease in the general population: should we treat asymptomatic cases? J Pediatr Gastroenterol Nutr 1997; 24: S Lohiniemi S. Tricky to find, hard to treat, impossible to cure. Lancet 2001; 358: Ellis A. Coeliac disease: Previous family studies. In: McConnell RB, ed. The genetics of coeliac disease: International symposium on the genetics of coeliac disease, 1979, Liverpool, England. MTP, Lancaster: 1981; Polanco I, Biemond I, van Leeuwen A, et al. Gluten sensitive enteropathy in Spain: Genetic and environmental factors. In: McConnell RB, ed. The genetics of coeliac disease: International symposium on the genetics of coeliac disease, 1979, Liverpool, England. MTP, Lancaster: 1981; Bardella MT, Fredella C, Prampolini L, Marino R, Conte D, Giunta AM. Gluten sensitivity in monozygous twins: a long-term follow-up of five pairs. Am J Gastroenterol 2000; 95: Hervonen K, Karell K, Holopainen P, Collin P, Partanen J, Reunala T. Concordance of Dermatitis Herpetiformis and Celiac Disease in Monozygous Twins. J Invest Dermatol 2000; 115: Ferguson A. Coeliac disease research and clinical practice: maintaining momentum into the twenty-first century. Baillieres Clin Gastroenterol 1995; 9: Walker-Smith J, Murch S. Coeliac disease. Diseases of the small intestine in childhood. 4 ed. Isis Medical Media; Oxford: 1999;

72 58 INFANT FEEDING PRACTICES AND COELIAC DISEASE 22. Sollid LM. Molecular basis of celiac disease. Annu Rev Immunol 2000; 18: Sollid LM, Lundin KEA. An inappropriate immune response. Lancet 2001; 358: s Petruzzelli F, Bonamico M, Ferrante P, Grillo R, Mora B, Mariani P, et al. Genetic contribution of the HLA region to the familial clustering of coeliac disease. Ann Hum Genet 1997; 61: Bevan S, Popat S, Braegger CP, Busch A, O'Donoghue D, Fälth-Magnusson K, et al. Contribution of the MHC region to the familial risk of coeliac disease. J Med Genet 1999; 36: Clot F, Babron MC. Genetics of celiac disease. Molecular Genetics and Metabolism 2000; 71: King AL, Ciclitira PJ. Celiac disease: strongly heritable, oligogenic, but genetically complex. Molecular Genetics and Metabolism 2000; 71: Risch N. Assessing the role of HLA-linked and unlinked determinants of disease. Am J Hum Genet 1987; 40: Salvetti M, Ristori G, Bomprezzi R, Pozzilli P, Leslie RDG. Twins: mirrors of the immune system. Immunol Today 2000; 21: Sing CF, Haviland MB, Reilly SL. Genetic architecture of common multifactorial diseases. Ciba Found Symp 1996; 197: Cronin CC, Shanahan F. Insulin-dependent diabetes mellitus and coeliac disease [see comments]. Lancet 1997; 349: Strobel S, Mowat AM. Immune responses to dietary antigens: oral tolerance. Immunol Today 1998; 19: Constant SL, Bottomly K. Induction of Th1 and Th2 CD4+ T cell responses: the alternative approaches. Annu Rev Immunol 1997; 15: Kay AB. Allergy and allergic diseases. First of two parts. N Engl J Med 2001; 344: Hernell O, Forsberg G, Hammarström S, Ivarsson A, Hammarström M-L. Coeliac disease: A model to study oral tolerance. In: Hart AL, Stagg AJ, Graffner H, Glise H, Falk P, Kamm MA, eds. Gut Ecology. Martin Dunitz, London: 2002; Ivarsson A, Persson LÅ, Nyström L, Ascher H, Cavell B, Danielsson L, et al. Epidemic of coeliac disease in Swedish children. Acta Paediatr 2000; 89: Ivarsson A. On the multifactorial aetiology of coeliac disease. Scand J Nutr 2001; 45: Ascher H, Krantz I, Kristiansson B. Increasing incidence of coeliac disease in Sweden. Arch Dis Child 1991; 66: Cavell B, Stenhammar L, Ascher H, Danielsson L, Dannaeus A, Lindberg T, et al. Increasing incidence of childhood coeliac disease in Sweden. Results of a national study. Acta Paediatr 1992; 81: Carlsson AK, Axelsson IEM, Borulf SK, Bredberg ACA, Ivarsson SA. Serological screening for celiac disease in healthy 2. 5-year-old children in Sweden. Pediatrics 2001; 107: Beaglehole R, Bonita R, Kjellström T. Basic epidemiology. World Health Organization; Geneva, 1993; Cummins AG, Thompson FM. Postnatal changes in mucosal immune response: A physiological perspective of breast feeding and weaning. Immunol Cell Biol 1997; 75: Hanson LA. Breastfeeding provides passive and likely long-lasting active immunity. Ann Allergy Asthma Immunol 1998; 81: Garofalo RP, Goldman AS. Expression of functional immunomodulatory and antiinflammatory factors in human milk. Clin Perinatol 1999; 26: Goldman AS. Modulation of the gastrointestinal tract of infants by human milk. Interfaces and interactions. An evolutionary perspective. J Nutr 2000; 426S-31S. 46. Andersen DH, Di Sant'Agnese PA. Idiopathic celiac disease: I. Mode of onset and diagnosis. Pediatrics 1953; 11:

73 INFANT FEEDING PRACTICES AND COELIAC DISEASE Greco L, Mayer M, Grimaldi M, Follo D, De Ritis G, Auricchio S. The effect of early feeding on the onset of symptoms in celiac disease. J Pediatr Gastroenterol Nutr 1985; 4: Mäki M, Kallonen K, Lähdeaho ML, Visakorpi JK. Changing pattern of childhood coeliac disease in Finland. Acta Paediatr Scand 1988; 77: Kelly DA, Phillips AD, Elliott EJ, Dias JA, Walker-Smith JA. Rise and fall of coeliac disease Arch Dis Child 1989; 64: Logan RFA, Rifkind EA, Busuttil A, Gilmour HM, Ferguson A. Prevalence and "incidence" of celiac disease in Edinburgh and the Lothian region of Scotland. Gastroenterology 1986; 90: Stevens FM, Egan-Mitchell B, Cryan E, McCarthy CF, McNicholl B. Decreasing incidence of coeliac disease. Arch Dis Child 1987; 62: Auricchio S, Follo D, De Ritis G, Giunta A, Marzorati D, Prampolini L, et al. Does breast feeding protect against the development of clinical symptoms of celiac disease in children? J Pediatr Gastroenterol Nutr 1983; 2: Greco L, Auricchio S, Mayer M, Grimaldi M. Case control study on nutritional risk factors in celiac disease. J Pediatr Gastroenterol Nutr 1988; 7: Fälth-Magnusson K, Franzén L, Jansson G, Laurin P, Stenhammar L. Infant feeding history shows distinct differences between Swedish celiac and reference children. Pediatr Allergy Immunol 1996; 7: Peters U, Schneeweiss S, Trautwein EA, Erbersdobler HF. A case-control study of the effect of infant feeding on celiac disease. Ann Nutr Metab 2001; 45: Ascher H, Krantz I, Rydberg L, Nordin P, Kristiansson B. Influence of infant feeding and gluten intake on coeliac disease. Arch Dis Child 1997; 76: Ivarsson A, Hernell O, Stenlund H, Persson LÅ. Breast-feeding protects against celiac disease. Am J Clin Nutr 2002; 75: Mäki M, Holm K, Ascher H, Greco L. Factors affecting clinical presentation of coeliac disease: Role of type and amount of gluten-containing cereals in the diet. In: Auricchio S, Visakorpi JK, eds. Common food intolerances 1: Epidemiology of coeliac disease. Karger, Basel: 1992; Ascher H, Holm K, Kristiansson B, Mäki M. Different features of coeliac disease in two neighbouring countries. Arch Dis Child 1993; 69: Weile B, Cavell B, Nivenius K, Krasilnikoff PA. Striking differences in the incidence of childhood celiac disease between Denmark and Sweden: A plausible explanation. J Pediatr Gastroenterol Nutr 1995; 21: Mitt K, Uibo O. Low cereal intake in Estonian infants: the possible explanation for the low frequency of coeliac disease in Estonia. Eur J Clin Nutr 1998; 52: Marsh MN. Gluten, major histocompatibility complex, and the small intestine. A molecular and immunobiologic approach to the spectrum of gluten sensitivity ('celiac sprue'). Gastroenterology 1992; 102: Doherty M, Barry RE. Gluten-induced mucosal changes in subjects without overt smallbowel disease. Lancet 1981; i: Leigh RJ, Marsh MN, Crowe P, Kelly C, Garner V, Gordon D. Studies of intestinal lymphoid tissue. IX. Dose-dependent, gluten-induced lymphoid infiltration of coeliac jejunal epithelium. Scand J Gastroenterol 1985; 20: Ferguson A, Blackwell JN, Barnetson RS. Effects of additional dietary gluten on the small intestinal mucosa of volunteers and of patients with dermatitis herpetiformis. Scand J Gastroenterol 1987; 22: Catassi C, Rossini M, Ratsch IM, Bearzi I, Santinelli A, Castagnani R, et al. Dose dependent effects of protracted ingestion of small amounts of gliadin in coeliac disease children: a clinical and jejunal morphometric study. Gut 1993; 34: Anderson CM, Gracey M, Burke V. Coeliac disease. Some still controversial aspects. Arch

74 60 INFANT FEEDING PRACTICES AND COELIAC DISEASE Dis Child 1972; 47: Lindberg T. Coeliac disease and infant feeding practices [letter]. Lancet 1981; i: Stenhammar L, Ansved P, Jansson G, Jansson U. The incidence of childhood celiac disease in Sweden. J Pediatr Gastroenterol Nutr 1987; 6: Kagnoff MF, Paterson YJ, Kumar PJ, Kasarda DD, Carbone FR, Unsworth DJ, et al. Evidence for the role of a human intestinal adenovirus in the pathogenesis of coeliac disease. Gut 1987; 28: Lähdeaho ML, Parkkonen P, Reunala T, Mäki M, Lehtinen M. Antibodies to E1b proteinderived peptides of enteric adenovirus type 40 are associated with celiac disease and dermatitis herpetiformis. Clin Immunol Immunopathol 1993; 69: Howdle PD, Blair Zajdel ME, Smart CJ, Trejdosiewicz LK, Blair GE, Losowky MS. Lack of a serologic response to an E1B protein of adenovirus 12 in coeliac disease. Scand J Gastroenterol 1989; 24: Vesy CJ, Greenson JK, Papp AC, Snyder PJ, Qualman SJ, Prior TW. Evaluation of celiac disease biopsies for adenovirus 12 DNA using a multiplex polymerase chain reaction. Mod Pathol 1993; 6: Holm S, Andersson Y, Gothefors L, Lindberg T. Increased protein absorption after acute gastroenteritis in children. Acta Paediatr 1992; 81: Ivarsson A, Hernell O, Nyström L, Persson LÅ. Children born in the summer have increased risk for coeliac disease. J Epidemiol Community Health [in press]. 76. Ivarsson A. On the multifactorial etiology of celiac disease: An epidemiological approach to the Swedish epidemic. Department of Clinical Sciences, Pedatrics, Umeå University, Sweden. [Dissertation] Rothman KJ, Greenland S. Causation and causal inference. In: Rothman KJ, Greenland S, eds. Modern epidemiology: 2nd ed. Lippincott - Raven, Philadelphia: 1998; ESPGAN committee on nutrition. Guidelines on infant nutrition III. Recommendations for infant feeding. Acta Paediatr Scand Suppl 1982; 302: Hallert C, Gotthard R, Jansson G, Norrby K, Walan A. Similar prevalence of coeliac disease in children and middle-aged adults in a district of Sweden. Gut 1983; 24:

75 Catassi C, Fasano A, Corazza GR (eds): Primary prevention of coeliac disease. The utopia of the new millennium? Perspectives on Coeliac Disease, vol. 1, AIC Press, pp Mechanisms of oral tolerance - lessons for coeliac disease? Conleth Feighery Department of Immunology, Trinity College Dublin and St. James's Hospital, Dublin, Ireland Introduction Coeliac disease (CD) is an inflammatory disease of the upper small intestine and 1 results from gluten ingestion in genetically susceptible individuals. Gluten is the essential environmental factor in the development of CD: this is demonstrated by the clinical and mucosal recovery which follows the institution of a gluten-free diet. Genetic factors also play a critical role in the development of CD and a strong association with certain Major Histocompatibility Complex (MHC) class II genes 2 (HLA-DQ2 and HLA-DQ8) is well established. However, CD is a polygenic disorder and many additional genes, yet to be identified, also contribute to the pathogenesis of CD. There is no evidence to suggest that these genes are in any sense abnormal and similar genetic polymorphisms are well represented in the normal population. Since gluten is part of the normal stable diet, and the genes which permit the development of CD are present in the normal population, the question can be reasonably posed - what makes a particular individual susceptible to gluten induced damage? It is speculated that additional factors, for example infectious agents and sex hormones, may be involved. Other environmental factors may also contribute but currently there are few clues as to what these are. This leads to the broader issue how does the gut immune system view foods? Foods of their essence are foreign proteins and as such might be targeted for elimination, as is the case with any other foreign antigen. CD is a common disorder. According to current evidence, it may affect 0.5 to 1% of 3-4 the population. This suggests that gluten has unique, but ill-understood, properties which frequently results in a T cell, delayed-type hypersensitivity response to this food antigen. There is little evidence that other food proteins cause a similar reaction. In contrast, many foods stimulate food allergy, a type I hypersensitivity reaction with 5 specific IgE production to the offending antigen. In food allergy, a systemic reaction to 61

76 62 MECHANISMS OF ORAL TOLERANCE the food is quite common and the reaction remits promptly after the food has been eliminated from the diet. Food allergy often develops in early childhood but may no 6 longer be clinically problematic in later years. Concept of immune tolerance The absent or low immune reactivity of the gut to food antigens raises the concept of 7 immune tolerance. Essentially, it would appear that a limited, local reaction develops to ingested foods and this response is confined to the bowel mucosa. This permits the antigenicity of food to be recognised but does not allow a damaging immune response to develop. In this situation, immune tolerance of the food is said to have taken place. From the viewpoint of an immunologist, tolerance is defined as the inability or failure of lymphocytes to respond to an antigen. The importance of tolerance is best observed in the concept of self-tolerance: in this, an individual's lymphocytes fail to react with self-antigen and as a consequence the development of autoimmune disease is prevented. Self-tolerance is a central paradigm in immunology, first proposed some years ago by Paul Erlich who predicted that horror autotoxicus might develop if such a control system was not in place. Central and peripheral tolerance 7 Immune tolerance can be divided into central and peripheral forms. Central tolerance relates to events which take place in organs such as the thymus and bone marrow. In this, it is envisaged that as T cells mature in the thymus, those cells which might be capable of self reactivity (by binding strongly to self-antigens presented by self-mhc molecules) undergo apoptosis and are thereby deleted. These events prevent the development of T cell mediated autoimmunity. There is considerable experimental evidence in support of such thymic events. It is also proposed that a similar mechanism 9 may cause deletion of B cells in the bone marrow. Despite the evidence for lymphocyte deletion in the thymus, it is not precisely known how universal this process is. Indeed, it is increasingly being accepted that a low level of autoreactivity is physiological and may even be important for normal immune 10 function. Since it is improbable that all T lymphocytes capable of self-reactivity are removed by this process, a second mechanism has been invoked: this is referred to as peripheral tolerance (Fig. 1). In this, the potential self-reactivity of circulating lymphocytes is curbed and these cells are rendered tolerant in the periphery. The mechanisms involved in peripheral tolerance are not well understood. 7-9 There is evidence that some lymphocytes are deleted in the periphery. A more popular concept is that lymphocyte anergy develops. One suggested basis for the development of anergy is that when lymphocytes encounter self-antigen, they do so in the absence of the appropriate additional signals provided by co-stimulatory molecules. These molecules include in particular the B7/CD28 interaction, which markedly enhances interleukin (IL)-2 production and hence full T 9 cell activation. If such co-stimulation is absent, the lymphocyte enters a dormant state,

77 MECHANISMS OF ORAL TOLERANCE 63 Peripheral tolerance Clonal deletion T cell apoptosis Clonal anergy T cell Inactive / asleep? Regulatory clone T reg T reg cytokines Fig. 1. Mechanisms of peripheral tolerance. T reg = regulatory T cell incapable of reacting with a specific antigen. It is possible that anergic T cells can be rescued from this state, thereby permitting such a cell to display auto-reactivity. The presence of circulating, regulatory T cells is a further mechanism that has been proposed to explain the phenomenon of peripheral tolerance. Following appropriate activation, these cells release cytokines which inhibit the auto-reactivity of the immune system. In the 1970s, considerable interest surrounded the concept of T suppressor cells performing this task. However, since no specific cell markers or specific suppressor factors could be reliably identified, the concept became disreputable and it was damaging to a research grant application or publication to mention the S word! However, fashions change and there is a saying that what goes around, comes around and there is now a growing belief that regulatory T cells play a central role in the 11 prevention of auto-immunity. According to this concept, regulatory T cells can be 12 considered to conduct the orchestra of the immune response. Finally, a fourth mechanism whereby peripheral tolerance may operate is the 7 concept of antigen ignorance. According to this concept, antigen internalised within a tissue or present in only very small quantities, is unable to interact productively with lymphocytes. The immune control mechanisms described above do not develop for such antigens. If tissue damage develops, antigen is released and a potentially damaging immune response may develop. Mucosal immune system Exposure to external antigens in humans commonly takes place at two different sites: the skin and the mucosal surfaces. Although antigen can enter via the skin, the

78 64 MECHANISMS OF ORAL TOLERANCE potential for antigen interaction at mucosal surfaces is even greater. The nature of skin, a multi-layered relatively impermeable structure, inhibits the penetration of harmful antigenic structures such as bacteria. In contrast, mucosal surfaces are typically protected by a single layer of epithelial cells. This makes the task of antigen access all the easier. Furthermore, in sites such as the small intestine which have a major role in absorption of nutrients, constant sampling of luminal material takes place. Thus, the mucosal immune system needs to differentiate between harmful antigens (such as pathogenic bacteria) and innocuous antigens (such as food). The organisation of the mucosal immune response is structured to permit a local response to the majority of antigens which penetrate to the lamina propria cells of the mucosa. However, this immune response is typically of low intensity and remains local. Systemic reaction to such antigen is either minimal or totally absent. Precisely how this type of response is achieved is not fully understood. Secretory IgA produced in the mucosa may play an important role but other immune components are doubtless also 13 involved. This controlled mucosal immune response could be considered as a form of peripheral tolerance and the term oral tolerance is frequently used. Oral tolerance The phenomenon of oral tolerance has been extensively investigated in various animal model systems over the past 30 years. As with all experimental models, they may not precisely replicate the normal in vivo situation but may nonetheless be informative about how local mucosal immune responses are regulated. The essential feature of experimental oral tolerance is that by feeding an antigen orally, the induction of a systemic immune response to that antigen is prevented. A typical experiment examining oral tolerance is to study what happens to an animal after oral administration of ovalbumin: when the animal is later exposed to ovalbumin systemically (by injection), circulating lymphocyte responsiveness to that antigen is 16 either absent or significantly diminished. Animals not pre-fed ovalbumin have a normal systemic response to the antigen. This effect of oral feeding causing systemic tolerance could be regarded as a form of induced peripheral tolerance. In oral tolerance, although hyporesponsiveness of both specific T and B lymphocytes is noted, T cells seem to be more profoundly affected. There is no well validated equivalent experiment confirming the induction of oral tolerance in man. Relevance of oral tolerance to coeliac and other gastrointestinal diseases The essence of the concept of oral tolerance is that local gut mucosal immune responses are regulated and prevent systemic reactions to orally ingested antigens. If, for whatever reason, a breakdown in oral tolerance develops, an excessive inflammatory mucosal reaction may take place. Thus, such a breakdown may be regarded as central to the development of inflammatory gastrointestinal disorders. Examples of these include coeliac disease, Crohn's disease and ulcerative colitis. In the latter two diseases the responsible gut antigens are not known, although bacterial antigens are suspected. In the case of CD, it is well established since the 1950s that 17 gluten is the causative agent. It is postulated that immune tolerance of gluten, found in normal subjects, is lost in CD patients and the consequence is inflammation at the major

79 MECHANISMS OF ORAL TOLERANCE 65 site of gluten exposure, the small intestine. Experimental models of oral tolerance differ from CD in certain respects. In CD, the most prominent evidence of inflammation occurs locally, where antigen exposure takes place. This is not the case in animals in which the effect of abrogation of oral tolerance predominantly affects systemic immune responses. However, in some animal models, 18 oral challenge with the antigen was shown to cause a mild mucosal lesion. Of interest, there is also evidence that gluten intolerance is associated with systemic damage in some individuals, as represented by diseases such as dermatitis herpetiformis, epilepsy and cerebellar ataxia. Moreover, there is some evidence that CD is also associated with inflammation in the distal intestine and rectal histological changes have been 22 reported. Finally, it is worth noting that systemic antibody responses to gluten are found in a range of disorders, in the absence of local gut damage or evidence that these 23 antibodies have a systemic pathogenic significance. Interest in oral tolerance Over the decades, an interest in the topic of oral tolerance has persisted, in the belief that its study could lead to a more fundamental understanding of homeostasis of the immune system. Furthermore, it is postulated that systemic inflammatory diseases 14 might be controlled by inducing oral tolerance to the stimulating antigen. According to this concept, the administration of oral antigen (such as the autoantigenic target of systemic disease) could lead to a reduced or aborted systemic response to this antigen. Thus various feeding studies with myelin basic protein (in patients with multiple sclerosis ) and collagen (in patients with rheumatoid arthritis ) have been described. Little therapeutic benefit has been noted in these studies to-date. Finally, an understanding of oral tolerance mechanisms is pertinent to the development of oral vaccines, since this route of vaccination is designed to give not only a local protective immune response, but also systemic protection. Factors involved in control of the mucosal immune response A clear understanding of the homeostatic mechanisms responsible for controlling the mucosal immune response has yet to be reached. Undoubtedly, many interrelated factors play a role (Fig. 2) and these will be considered briefly here. The epithelial lining layer outpost of mucosal immunity The epithelial cell sitting on its basement membrane, in a sense guards the mucosal tissue from full, free exposure to antigen. However, the epithelial cell is not an absolute barrier and it permits rapid absorption of soluble antigen. Thus, epithelial cells, like many other cells in the body, constantly sample the external milieu. We know that food antigens can rapidly penetrate this epithelial layer, since these antigens are found in the 26 circulation even within minutes of food ingestion. In addition to antigen absorption, epithelial cells have also been shown to process and load antigen into MHC II molecules. This was demonstrated in the case of gliadin, 27 using immuno-electron microscopy. Furthermore, it is known that MHC class II 28 molecules are expressed or can be induced on intestinal epithelial cells. These findings raise the possibility that epithelial cells might function as antigen presenting

80 66 MECHANISMS OF ORAL TOLERANCE Epithelial cells IELs Anatomical structure Antigen structure Lamina propria T cells Environment (cytokines) Antigen presenting cells Fig. 2. Components involved in control of mucosal immune response. IELs =intraepithelial lymphocytes. cells. However, there is little evidence that conventional antigen stimulation takes 29 place and it may be that epithelial cells interact with T lymphocytes using non- 30 classical MHC molecules such as CD1d. Since epithelial cells do not express the costimulatory molecules CD80 and CD86, it might be expected that these cells play a role in inducing anergy of T cells rather than an active T cell response. The importance of the epithelial cell barrier is emphasised by what happens when barrier function is disrupted. This may occur during the course of local inflammation in 33 which tight junction function is impeded. This in turn would allow ready access of antigens to immunocompetent cells in the lamina propria. This may explain the presence of high levels of antibodies to dietary antigens (including gliadin) in Crohn's 34 and peptic ulcer disease. A further demonstration of the importance of the epithelial cell barrier is found in certain gene knockout mouse models in which epithelial cell adherence or their cytoskeletal structure is disrupted: in these models, genes for either E-cadherin or keratin 8 (respectively) are knocked out, with the spontaneous development of inflammatory bowel disease as a consequence. Antigen factors The size, structure and nature of antigens affects the response of the mucosal immune system. Thus in health, soluble dietary antigens induce a limited, local immune response and often a barely discernible response in the systemic circulation. Likewise, the gut flora induces a local immune response only. In contrast, particulate antigen is 37 absorbed by specialised epithelial cells called microfold or M cells. M cells are found

81 MECHANISMS OF ORAL TOLERANCE 67 interposed between the lining epithelial cells and following uptake of antigen, particulate antigen is processed and presented to local T helper cells. This in turn results in a systemic immune response to the antigen. As discussed earlier, in contrast to other food antigens, gluten appears to have an unique capacity to stimulate a local T cell mediated immune response in CD patients, leading to damage of the local mucosa. The chemical properties of gluten involved in eliciting such a response are unknown, since no essential structural difference is 10 thought to exist between self-antigens and foreign antigens. Intraepithelial lymphocytes Intraepithelial lymphocytes (IELs) are T cells located within the surface epithelium of the mucosa. Despite intense research of these cells over the past two decades, their 38 principal function remains enigmatic. The majority of IELs express the CD8 molecule and employ the abt cell receptor (TCR). It is well established that increased numbers of IELs are found in certain intestinal inflammatory states such as coeliac disease. Moreover, in coeliac disease the number of IELs expressing the gdtcr 39 increases, with as many as 30% belonging to this phenotype. According to recent evidence, IELs may include heterogeneous populations of T cells, natural killer T cells 40 and also natural killer cells. It has been speculated that IELs play a role in oral tolerance and there is some experimental evidence to support a role for gdt cells in 41 tolerance induction. Lamina propria T cells The major population of lamina propria T cells express the CD4 molecule and employ the abtcr. These T helper cells are thought to have initially encountered mucosal antigen in organised lymphoid structures such as Peyer's patches and after passage through the systemic circulation have specifically homed back to the intestine. In this location, lamina propria T cells play a major role in orchestrating other cells of 12 the immune system and act as conductors of this orchestra. These T cells possess a full range of surface molecules to allow productive interaction with many other cell types. Moreover, individual populations manufacture the variety of cytokines required to influence B cell production of specific antibody isotypes, to induce specific populations 13 of antigen presenting cells and to activate other populations of T cells. Thus, regulation of immune events in the intestine is largely a function of these cells, and hence maintenance of oral tolerance or mucosal immune homeostasis is likely to be achieved by these cells. Dendritic and other antigen presenting cells The activation of intestinal T cells requires the presentation of antigen by local populations of antigen presenting cells (APCs). This task may be performed by a variety of APCs including dendritic cells, macrophages, B cells and, as discussed earlier, possibly by enterocytes. Populations of all these cell types have been identified in the mucosal tissue. In individual situations, it is likely that the presentation of antigen is principally performed by specific cell types. Dendritic cells, in particular, may play a central role in presenting antigen and influencing the evolution of specific

82 68 MECHANISMS OF ORAL TOLERANCE types of T cell populations in the intestine. Much has been learned about dendritic cells in recent years: some types particularly influence the development of Th1 cells and others Th2 cells. These dendritic populations can be identified on the basis of expression or absence of the adhesion molecule CD11c and also by the type of cytokines (such as IL-12) which they produce. The heterogeneity of dendritic cells is increasingly being recognised and it is now appreciated that the major function of some populations is to induce T cell tolerance, 44 rather than a productive immune response. As with all events in the immune system, a circular feedback system (so-called inter-cellular cross-talk) operates between dendritic cells and the T cells which they help stimulate (fig. 3). Thus, the cytokine products of dendritic cells influence the T cell populations which they generate and in turn, the cytokines produced by T cells 42 influence the type of dendritic cells which occupy that locale. A further population of cells, likely to be involved in antigen presentation, which express the macrophage scavenger receptor (CD163) has been identified by 45 immunohistochemistry in the central core of individual villi. The cytokine products of this population of macrophage-like cells, including secreted CD163, may play a role in 46 down-regulation of the local immune response. Cytokine environment of the mucosa Virtually all cells in the mucosa, including enterocytes, produce a wide range of cytokines and the profile of cytokines determines the net effect on the type of immune response elicited. It is presumed that a major, final determining influence is the type of cytokines produced by the local T helper populations. Since the principal default setting of the mucosal immune response is towards tolerance of encountered antigen, the cytokines which could determine tolerance have been investigated. The three cytokines which have received most interest for their potential involvement are IL-4, IL-10 and TGF-b. For some time it was considered that T cell tolerance in the mucosa was achieved by a predominant Th2 response to local antigens. Major products of Th2 cells include the cytokines IL-4 and IL-10. However, in some animal model experiments, it was shown that these cytokines are not required for tolerance induction. Moreover, in experimental oral tolerance, inhibition of IgE production is a characteristic feature: the reverse would 14 be predicted in an IL-4 dominant environment. Hence, it is now proposed that tolerance in the intestine may be achieved by a further T cell population, referred to as T 47 regulatory cells. The major cytokine products of these cells are IL-10 and transforming growth factor-beta (TGF-b). Although IL-10 is not essential in some models of oral tolerance, the importance of this cytokine in maintaining tolerance is emphasized by the spontaneous development 48 of colitis in mice when the gene for this cytokine is knocked out. Of great interest, 49 TGF-bcan reverse the colitis. It is also known that IL-10 may be required for the production of TGF-band that the latter cytokine inhibits the IL-12/interferon-gamma 50 (IFN-g) pathway, likely to play a central role in many inflammatory diseases. It is currently hypothesised that control of the mucosal immune system is maintained by regulatory T cells, but at present, other than studying their cytokine

83 MECHANISMS OF ORAL TOLERANCE 69 IEL Lamina propria M cell CD4 DC Cytokines of each cell type influence each other Epithelial lining Fig. 3. Interaction between CD4 + T cells and dendritic cells in lamina propria of the mucosa. DC = dendritic cell; IEL = intraepithelial lymphocyte. products, we do not have reliable, specific markers for these cells. It is probable that TGF-band IL-10 are major products of these cells. However, it should be emphasised, that even when a cytokine is shown not to be essential for experimental oral tolerance, this does not mean that in vivo, this cytokine has no role to play, since a level of redundancy is a common feature of the immune system. Mucosal immune tolerance and coeliac disease The preceding discussion of mucosal immune tolerance raises issues which may be pertinent to the pathogenesis of coeliac disease. Is the essential defect in coeliac disease a failure of the normal tolerance mechanisms, which allow gluten exposure but no mucosal damage in normal individuals? If so, might tolerance of gluten be restored and thereby permit CD patients to eat a normal diet? If this were possible, how might we go about achieving it? Obviously, these are major issues and it could be argued that restoring tolerance to gluten is a fanciful solution, perhaps unlikely ever to be achieved. Nonetheless, with the gathering pace of modern knowledge and technology, it is a goal which should at least be examined. Adolescent coeliac disease Some 30 years ago, one of the most frequent inquiries about the nature of CD, was whether gluten intolerance was permanent. Of course, the dogma now is that permanent

84 70 MECHANISMS OF ORAL TOLERANCE 51 intolerance is a central feature of the disorder. However, in the past, cases of possible 52 transient gluten sensitivity were described. Is it possible that this phenomenon of transiency is more common than currently appreciated?. The events that happen to pediatric cases of CD, when they reach adolescence, may be informative concerning the permanency of gluten intolerance. Detailed information about these events is absent. Nonetheless, it is recognized that many children adhere less well to a strict gluten free diet when they reach adolescence. This may represent the general rebellion of teenagers! Despite eating gluten containing foods, few symptoms may develop. This in turn may lead to increasing carelessness about their diets. Some of these patients eventually have intestinal biopsies performed and the mucosa can appear surprisingly normal or even completely normal. Doubts about the accuracy of the initial diagnosis of CD may then be raised. Some of these patients are then given a gluten challenge and it is reported that it can take many months or even years for histological evidence of CD to return. This contrasts with gluten challenge in patients with an adult diagnosis of CD when symptomatic and histological evidence 56 of disease relapse often only requires one or two months of gluten intake. Although categorical data about adolescent remission of CD is lacking, the information available strongly suggests that this really does occur. If this is the case, it may indicate that restoration of tolerance to gluten is possible. The majority, if not all of these patients, eventually develop symptomatic gluten intolerance in adulthood, demonstrating the essential permanency of CD. However, it is possible that investigation of adolescent CD patients, during their period of remission, may give insight to T cell tolerance mechanisms towards gluten. Conclusions The gut mucosal immune response to the majority of food and other antigens is typically of low intensity and remains confined to the local environment. The term ²oral tolerance²may be applied to this controlled immune response. However, in the case of a single food type, gluten, an enhanced T cell immune response develops in a substantial number of individuals and features of coeliac disease develop. In this disorder, immune tolerance of gluten is apparently lost. With an improved understanding of the mechanisms of oral tolerance, it is possible that manipulation of the immune response to gluten in coeliac patients could restore the normal physiological state of low reactivity to this food antigen. This would remove the requirement of gluten-free diet treatment of coeliac disease. References 1. Feighery C. Coeliac disease Fortnightly review. BMJ 1999; 319: Kagnoff MF. Genetic basis of Coeliac Disease: role of HLA genes. In: MN Marsh (Ed) Coeliac Disease, Oxford, Blackwell Scientific Publications, 1992; pp Catassi C, Fabiani E, Rätsch IM, Coppa GV, Giorgi PL, Pierdomenico R, et al. The coeliac iceberg in Italy. A multicentre antigliadin antibodies screening for coeliac disease in school-age subjects. Acta Paediatr Suppl 1996; 412: McMillan SA, Watson RPG, McCrum EE, Evans AE. Factors associated with serum

85 MECHANISMS OF ORAL TOLERANCE 71 antibodies to reticulin, endomysium, and gliadin in an adult population. Gut 1996; 39: Furukawa CT. Adverse Reactions to Food. In: Allergy in Primary Care. Eds. LC. Altman, JW. Becker, PV Williams. Publishers. W.B. Saunders Company 2000; 18: Sampson HA. What should we be doing for children with peanut allergy? J Pediatr 2000; 137: Kamradt T, Mitchison NA. Tolerance and Autoimmunity. N Engl J Med 2001; 344: Bibel DJ. Milestones in Immunology: a historical exploration. Science Tech Publishers, Wisconsin, USA Janeway CA, Travers P, Walport M, Capra JD. Immunobiology: the immune system in health and disease. Churchill Livingstone, Davidson A, Diamond B. Advances in Immunology: Autoimmune Diseases. N Engl J Med 2001; 345: Shevach EM. Certified Professionals: CD4+ CD25+ Suppressor T Cells. J Exp Med 2001; 193: F Feighery C. Intestinal Immune System - A Hidden Symphony. In: Gastrointestinal Immunology and Gluten-Sensitive Disease. Proceedings of the Sixth International Symposium on Coeliac Disease held at Trinity College, Dublin in July Eds. C. Feighery, C. O'Farrelly, Publishers Oak Tree Press, Dublin. 13. Brandtzaeg P, Baekkevold ES, Farstad IN, Jahnsen FL, Johansen FE, Nilsen EM, et al. Regional specialization in the mucosal immune system: what happens in the microcompartments? Immunol Today 1999; 3: Garside P, Mowat AM, Khoruts A. Oral tolerance in disease. Gut 1999; 44: Weiner HL. Induction and mechanism of action of transforming growth factor-betasecreting Th3 regulatory cells. Immunol Rev 2001; 182: Mowat McI A. The regulation of immune responses to dietary protein antigens. Immunology Today 1987; 8: Dicke WK. Coeliakie. PhD thesis. University of Utreacht, Utreacht, The Netherlands, Lamont AG, Mowat AM, Parrott DM. Priming of systemic and local delayed-type hypersensitivity responses by feeding low doses of ovalbumin to mice. Immunology 1989; 66: Fry L. Dermatitis herpetiformis: from gut to skin, contributions to the nature of coeliac disease. In: Coeliac Disease, Mäki M, Collin P, Visakorpi JK (eds). Proceedings of the Seventh International Symposium on Coeliac Disease, Finland. Publishers. Coeliac Disease Study Group, Tampere, Finland,1996; pp Cronin CC, Jackson LM, Feighery C, Shanahan F, Abuzakouk M, Ryder DQ, et al. Coeliac disease and epilepsy. Q J Med 1998; 91: Hadjivassiliou M, Grunewald RA, Chattopadhyay AK, Davies-Jones GA, Gibson A, Jarratt JA et al. Clinical, radiological, neurophysiological, and neuropathological characteristics of gluten ataxia. Lancet 1998; 352: Ensari A, Marsh MN, Loft DE, Morgan S, Moriarty K. Morphometric analysis of intestinal mucosa. V. Quantitative histological and immunocytochemical studies of rectal mucosae in gluten sensitivity. Gut 1993; 34:

86 72 MECHANISMS OF ORAL TOLERANCE 23. McCormick PA, Feighery C, Dolan C, O'Farrelly C, Kelleher P, Graeme-Cook F, et al. Altered gastrointestinal immune response in sarcoidosis. Gut 1988; 29: Weiner HL, Mackin GA, Matsui M, Orav EJ, Khoury SJ, Dawson DM, et al. Doubleblind pilot trial of oral tolerization with myelin antigens in multiple sclerosis. Science 1993; 259: Barnett ML, Combitchi D, Trentham DE. A pilot trial of oral type II collagen in the treatment of juvenile rheumatoid arthritis. Arthritis Rheum 1996; 39: Husby S, Jensenius JC, Svehag SE. Passage of undergraded dietary antigen into the blood of healthy adults. Further characterization of the kinetics of uptake and the size distribution of the antigen. Scand J Immunol 1986; 24: Zimmer KP, Poremba C, Weber P, Ciclitira PJ, Harms E. Translocation of gliadin into HLA-DR antigen containing lysosomes in coeliac disease enterocytes. Gut 1995; 36: Kelly J, Weir DG, Feighery CF. Differential expression of HLA-D gene products in the normal and coeliac small bowel. Tissue Antigens 1987; 31: Mayer L, Schlien R. Evidence for function of Ia molecules on gut epithelial cells in man. J Exp Med 1987; 166: Canchis PW, Bhan AK, Landau SB, Yang L, Balk SP, Blumberg RS. Tissue distribution of the non-polymorphic major histocompatibility complex class I-like molecule, CD1d. Immunology 1993 ; 80: Byrne B, Madrigal-Estebas L, McEvoy A, Carton J, Doherty DG, O'Donoghue D, et al. Human duodenal epithelial cells constitutively express molecular components of antigen presentation but not costimulatory molecules. (2002, submitted). 32. Mowat AM. Regulation of Immune Responses to Dietary Protein Antigens. In: Gastrointestinal Immunology and Gluten-Sensitive Disease. Proceedings of the Sixth International Symposium on Coeliac Disease held at Trinity College Dublin, July Eds. Feighery C, O'Farrelly C. Publishers: Oak Tree Press, Dublin. 33. Madara JL. Regulation of the movement of solutes across tight junctions. Annu Rev Physiol 1998; 60: Feighery C, Weir DG, Whelan A, Willoughby R, Youngprapakorn S, Lynch S, et al. Diagnosis of gluten-sensitive enteropathy: is exclusive reliance on histology appropriate? Eur J Gastroenterol Hepatol 1998; 10: Hermiston ML, Gordon JI. Inflammatory bowel disease and adenomas in mice expressing a dominant negative N-cadherin. Science 1995; 270: Baribault H, Penner J, Iozzo RV, Wilson-Heiner M. Colorectal hyperplasia and inflammation in keratin 8-deficient FVB/N mice. Genes Dev 1994; 8: Trier, J. S. Structure and function of intestinal M cells. Gastroenterol Clin North Am. 1991; 20: Hayday A, Theodoridis E, Ramsburg E, Shires J. Intraepithelial lymphocytes: exploring the Third Way in immunology. Nat Immunol 2001; 2: Viney J, MacDonald T, Spencer J. Gamma delta T cells in the gut epithelium. Gut 1990: 31: Eiras P, Leon F, Camarero C, Lombardia M, Roldan E, Bootello A, et al. Intestinal intraepithelial lymphocytes contain a CD3 - CD7+ subset expressing natural killer markers and a singular pattern of adhesion molecules. Scand J Immunol 2000; 52:

87 MECHANISMS OF ORAL TOLERANCE Ke Y, Kapp JA. Oral antigen inhibits priming of CD8+ CTL, CD4+ T cells and antibody responses which activating CD8+ suppressor T cells. J Immunol 1996; 156: Banchereau J, Briére F, Caux C, Davoust J, Lebecque S, Liu Y-J, et al. Immunobiology of dendritic cells. Annu Rev Immunol 2000; 18: Lanzavecchia A, Sallusto F. Regulation of T cell immunity by dendritic cells. Cell 2001; 106: Jonuleit H, Schmitt E, Steinbrink K, Enk AH. Dendritic cells as a tool to induce anergic and regulatory T cells. Trends in Immunol 2001; 22: Whelan CA. (personal communication). 46. Hogger P, Sorg C. Soluble CD163 inhibits phorbol ester-induced lymphocyte proliferation. Biochem Biophys Res Commun 2001; 288: Groux H, O'Garra A, Bigler M, Rouleau M, Antonenko S, de Vries JE, et al. A CD4+ T-cell subset inhibits antigen-specific T-cell responses and prevents colitis. Nature 1997; 389: Kuhn R, Lohler J, Rennick D, Rajewsky K, Muller W. Interleukin-10-deficient mice develop chronic enterocolitis. Cell 1993; 75: Neurath MF, Fuss I, Kelsall BL, Presky DH, Waegell W, Strober W. Experimental granulomatous colitis in mice is abrogated by induction of TGF-b-mediated oral tolerance. J Exp Med 1996; 183: Fishman-Lobell J, Friedman A, Weiner HL. Different kinetic patterns of cytokine gene expression in vivo in orally tolerant mice. Eur J Immunol 1994; 24: Weijers HA, Lindquist B, Anderson CM, et al. Round table discussions. Diagnostic criteria in celiac disease. Acta Paediatr Scand 1970: 59: Walker-Smith JA. Transient gluten intolerance. Arch Dis Child 1970; 45: Schmitz J. Is celiac disease a lifelong disorder? Clin Invest Med 1996; 19: Ansaldi N, Tavassoli K, Faussone D, Forni M, Oderda G. Clinical-histological behavior of celiac patients after gluten load following the definitive diagnosis. Pediatr Med Chir 1988; 10: McNicholl B, Egan-Mitchell B, Fottrell PF. Variability of gluten intolerance in treated childhood coeliac disease. Gut 1979; 20: O'Mahony C. Immunological and Clinical Studies in Gluten Sensitive Disease. University College Dublin. MD Thesis 1986.

88

89 Catassi C, Fasano A, Corazza GR (eds): Primary prevention of coeliac disease. The utopia of the new millennium? Perspectives on Coeliac Disease, vol. 1, AIC Press, pp Genetically detoxified grains in coeliac disease Federico Biagi, Antonio Di Sabatino, Jonia Campanella, Gino Roberto Corazza Department of Gastroenterology, University of Pavia, IRCCS Policlinico San Matteo, Pavia, Italy Coeliac disease (CD) is a chronic enteropathy triggered in genetically predisposed individuals by gluten, the storage proteins of wheat, barley, and rye. So, from a coeliacologist point of view gluten represents the coeliac offending agent. More precisely, gluten can be defined as the rubbery mass that remains when wheat dough is 1-2 washed to remove starch granules and other soluble constituents. Classification of gluten proteins According to plant taxonomy, all the species (i.e. wheat, barley, and rye) toxic for patients affected by CD are members of the Gramineae grass family. However, it should be noted that other Gramineae grass family species, such as rice, sorghum, maize, and probably oats, are safe for CD patients. Since wheat, barley, and rye all fall in the tribe Hordae/Triticeae, it is possible that only species found in that tribe have proteins active in CD. Moreover, Hordae proteins are all rich in proline and glutamine, which is the reason why they were named prolamins. Although non toxic prolamins, such as avenins and zeins are also proline - and glutamine-rich, they have a lower 3 proline content compared to Hordae tribe grains. On the basis of their different solubility in a 70% ethanol solution, wheat proteins 1-2 can be divided into gliadins and glutenins. Gliadins are soluble single-chain polypeptides. On the basis of different electrophoretic mobility, gliadins component were classified into a-, b-, g-, and w-gliadins. Glutenins are insoluble proteins crosslinked one to another by intrachain disulphide bonds. Components were obtained after the reduction of the disulphide bonds, and grouped into low - (LMW) and highmolecular weight (HMW) subunits. However, it was shown that the extraction of gluten with aqueous solution does not lead to clear-cut fractions. Very often, glutenin subunits can be found in the soluble gliadin fraction and gliadins are also present in the insoluble glutenin fraction. So, it has recently emerged that the most important criteria to classify 4-5 gluten components is the primary structure. Using analysis of amino acid sequence and molecular mass, gluten proteins have been divided into high-, medium - and low- 75

90 76 GENETICALLY DETOXIFIED GRAINS molecular weight proteins. High-molecular weight proteins (x - and y-type subunits) comprise the HMW subunits of glutenins; glutamine, glycine, and proline represent 70% of total residues. Medium-molecular weight group consists of w-gliadins; 80% of the amino acids is due to glutamine, proline, and phenylalanine, but cysteine and methionine, i.e. sulphur containing amino acids, are almost absent. Low-molecular weight proteins consist of a-, b-, g-gliadins and LMW subunits of glutenins; they are rich in sulphur containing amino acids. Functional properties of gluten proteins Although the only known function of these proteins is to act as storage proteins, by releasing nitrogen for the germinating seed, their biochemistry structure has a crucial 2,6 role in providing the unique baking quality of wheat. Both LMW and HMW have not only intrachain but also interchain disulphide bonds allowing them to crosslink other glutenin and gliadin peptides, which results in the formation of beta-spirals. During the baking process, CO 2 is produced by yeast fermentation and is trapped in the beta-spiral three dimensional net. Glutenins spirals become extended but maintain the capacity to return to the colloid state. So, elasticity and cohesivity of the dough system are due to the cross-linked nature of the glutenin subunits, mainly the high molecular weight glutenins. On the other hand, soluble monomeric gliadins guarantee extensibility and viscosity to the dough (Fig. 1). Coeliac toxicity of gliadins and glutenins Since the demolition of these proteins to single amino acids results on loss of coeliac C C N N N TECHNOLOGICAL PROPERTIES N N C C C HMWx HMWy LMW g a Elasticity & Cohesivity Extensibility & Viscosity Fig. 1. Schematic representation of gluten protein components. Black segments unique sequence; white segments repeating sequence; ] intrachain disulphide bond; - interchain disulphide bond; C C-terminal sequence; N N-terminal sequence; HMWx high-molecular weight glutenins x-type subunit; HMWy high-molecular weight glutenins y-type subunit; LMW low-molecular weight glutenins; g-gliadins; a-gliadins.

91 GENETICALLY DETOXIFIED GRAINS 77 toxicity, several studies investigated which one of the several gluten components is responsible for CD intolerance. Briefly, a-, b-, g- and w-gliadins are all toxic for CD patients. Although glutenins were considered to be non dangerous for coeliac patients, 1,7 their safety is at least controversial. First of all alcohol extraction does not lead to clear-cut fractions. So, gliadins have actually been found within the insoluble glutenins, and that is an obstacle in understanding pure glutenin toxicity. Moreover, it has very recently been shown that glutenin peptides can induce an in-vitro specific activation and proliferation of T cell from coeliac small bowel. Such a response is virtually identical to that seen with gliadin peptides and is nowadays considered to be an indirect 8-9 evidence of coeliac toxicity. Although a major effort needs to be done to understand which gluten components are toxic in CD, this knowledge would be extremely important in order to develop genetically detoxified grains (GDG). The development of wheat strains lacking those genes coding for toxic peptides is hopefully regarded as a new and profitable tool for the treatment of CD. GDG could provide the solution to problems such as the poor palatability and high costs of commercially available gluten-free foods. GDG may be theoretically achieved with the modern techniques of genetic plant engineering which makes the deleting or silencing of one or more genes a possible approach. Genetics of wheat proteins Bread wheat is a hexaploid species containing three different but related genomas (A, B, and D), each consisting of seven chromosome pairs (Fig. 2). Information for storage proteins is encoded by clustered or dispersed gene families located on chromosomes 1 and 6 of the three different genomes. In particular, genes for a-gliadins are located on the short arm of chromosomes 6A, 6B, and 6D, tightly clustered at the Gli-A2, Gli-B2, and Gli-D2 loci. Genes coding for gw -and -gliadins are on the short arm 1A 1B 1D 6A Glu-1 (HMW) Glu-B2 Gli-A3 Gli-B3 Glu-3 (LMW) Gli-1 (gw) Gli-2 (ab) Gli-A5 Gli-B5 Gli-D5 6B 6D Long Arm Short Arm Fig. 2. Chromosomal position of the genes coding for storage proteins of bread wheat endosperm.

92 78 GENETICALLY DETOXIFIED GRAINS of chromosome 1 at three homologous loci, named Gli-1. However, some w-gliadin peptides have been found to be encoded by additional dispersed genes located on the same arm of the same set of chromosomes. Genes controlling LMW glutenin subunits are clustered at Glu-A3, B3, and D3 loci and only the genes coding for HMW glutenin subunits occur on the long arm of the group 1 chromosomes, at a major locus named Glu-1. So, the multigene family nature of the wheat genes and the spread of these genes over six different chromosomes entail a very high degree of genetic polymorphism. By applying a Southern analysis to the common cultivar Cheyenne, Anderson et al. found that up to 150 genes code for a-gliadin, although only half of them are expressed An Italian study showed that there is a high degree of variability in the genomic DNA between different cultivars. A genetic polymorphism at the Gli-2 locus encoding 14 for a-gliadin was evident among different subspecies of Triticum aestivum. Naturally occurring detoxified grains This huge genetic variability let people hope to individuate a naturally occurring GDG. This was the case of the nullisomic 6A-tetrasomic 6D variety of Chinese Spring wheat, in which chromosome 6A is missing but compensated for by two extradoses of chromosome 6D. Preliminary results were encouraging. The administration for two weeks of 65 grams of bread made with this wheat variety did not deteriorate absorptive 15 function. However, it was later proven that these variants do contain a-gliadin, coded by B and D chromosomes, and that they are toxic, on both histological and clinical 16 ground, for patients affected by CD. More recently, the toxicity of two lines of cultivar Raeder, one lacking the Gli-A2 encoded gliadin components and the other lacking both Gli-A2 and Gli-D1 gliadin components was tested with an organ culture system based on coeliac duodenal biopsy 17 specimens. Although the Gli-A2 components were not compensated for by extradoses of chromosomes B and D, those lines were deficient but not devoid of gliadins. Although the toxicity of these two mutant lines was reduced compared to the original cultivar Raeder, yet it was still present. In conclusion, all these studies show that a naturally-occurring non-toxic line has not been found so far. Technologies to develop genetically detoxified grains Technologies allowing genetic plant transformation do exist. For example, a fragment of foreign DNA cut with a restriction enzyme, and plasmid, cut open with the same enzyme, are mixed together. DNA ligase is added to stitch the complementary base sequences so that the recombinant plasmid contains foreign DNA that, under appropriate conditions, is taken up by the Agrobacterium, a naturally occurring plant pathogen. When host plant cells are exposed to Agrobacterium, plasmid DNA is transferred into plant chromosomes. At the end of the process transgenic plants are 18 regenerated from single transformed cells (Fig. 3A). An alternative method is the biolistic gene gun technique. A suspension of thousands of tiny gold or tungsten particles coated with transgenic DNA are fired into the target tissue, using compressed helium as propellant. Particles 19 penetrate the nucleus and incorporate into the host genoma (Fig. 3B).

93 GENETICALLY DETOXIFIED GRAINS 79 Foreign DNA EcoR1 DNA ligase Plasmid from Agrobacterium EcoR1 Agrobacterium Recombinant plasmid containing foreign DNA Gene suspension Sheath fluid Carbon dioxide 3A Cultured cells from host plant Whole plants regenerated from single cells Plasmid DNA transferred into plant chromosomes A) agrobacterium technique B) gene gun technique Fig. 3. Genetic plant transformation technologies. Additionally, it may be possible to switch off specific plant genes to eliminate or reduce natural toxins or allergens. This can take place both at posttranscriptional and transcriptional level. In the first case, which is the most common, injected doublestranded RNA is cut into small fragments of nucleotides, amplified and then dissociated into single strands complementary, but in reverse to the target messenger RNA which is thus inactivated. In other words, silenced genes are still active but 20 messenger RNA is degraded before it can be translated into proteins. It should be kept in mind that these techniques have been applied to wheat only with the aim of improving dough rheological properties by increasing the level of high molecular weight glutenins, resulting in improved functional properties. Moreover, transformed plants show Mendelian segregation of the transgenes and strategies have been developed to ensure transgene stability through generations. However, these techniques have not been directed to the treatment of CD so far. Genetically detoxified grains for coeliac disease Petri dish with receiving cells Power source Electric field A project aiming at the production of a genetically detoxified wheat still suitable for mixing and baking should be set up taking into account a few points: (a) a minimal amount of gluten can be toxic to the coeliac small intestine; (b) all the different classes of gliadins have been shown to be harmful for coeliac patients; (c) upwards 150 genes codify for the a-gliadins; (d) identification of all possible toxic sequences is still lacking; (e) bakery properties are mainly due to HMW glutenins. Obviously, the first thing to be checked is whether HMW glutenins are really safe 8 for CD patients. Despite the most recent findings on T cell assays, it should be noted that pure glutenins have not been tested on coeliac small bowel mucosa so far. So, once pure glutenins would be obtained, their safety should be tested first in-vitro and then in- 3B