Gluten: a two-edged sword. Immunopathogenesis of celiac disease

Springer Semin Immun (2005) 27:217 232 DOI 10.1007/s00281-005-0203-9 ORIGINAL PAPER Frits Koning. Luud Gilissen. Cisca Wijmenga Gluten: a two-edged sword. Immunopathogenesis of celiac disease Received: 17 January 2005 / Accepted: 15 March 2005 / Published online: 10 August 2005 # Springer-Verlag 2005 Abstract. Celiac disease (CD) is a small intestinal disorder caused by adaptive and innate immune responses triggered by the gluten proteins present in wheat. In the intestine, gluten is partially degraded and modified, which results in gluten peptides that bind with high affinity to HLA-DQ2 or HLA-DQ8 and trigger an inflammatory T cell response. Simultaneously, gluten exposure leads to increased production of IL15, which induces the expression of NKG2D on intraepithelial lymphocytes and its ligand MICA on epithelial cells, leading to epithelial cell destruction. The gluten-specific T cell response results in the production of antibodies against tissue transglutaminase and these are specific indicators of disease. CD is one of the most common inherited diseases, the HLA-DQ locus being the major contributing genetic factor. However, as the inheritance does not follow a Mendelian segregation pattern, multiple other genes, each with relative weak effect, contribute to disease development. An important role for environmental factors, however, can not be ignored as the concordance rate in monozygous twins is considerably less than 100%. The identification of these environmental factors and susceptibility genes may allow a better understanding of disease etiology and provide diagnostic and prognostic markers. The current treatment for CD consists of a life-long gluten-free diet. Although long thought to be impossible, recent results suggest that the development of nontoxic wheat varieties may be feasible, which would aid disease prevention and provide an alternative food source for patients. Keywords: HLA-DQ. Gluten. Tissue transglutaminase. Cereals. Genetic predisposition F. Koning ()) Department of Immunohematology and Blood Transfusion, Leiden University Medical Centre, E3-Q, P.O. Box 9600, 2300 RC Leiden, The Netherlands e-mail: f.koning@lumc.nl. Tel.: +31-71-5263800. Fax: +31-71-5216751 L. Gilissen Plant Research International, Wageningen, The Netherlands C. Wijmenga AMG Department of Biomedical Genetics, Complex Genetics Section, Utrecht University Medical Centre, Utrecht, The Netherlands

218 Springer Semin Immun (2005) 27:217 232 Gluten Over 50 years ago, the crucial role of wheat gluten in the development of celiac disease (CD) was discovered [12]. Gluten is the water-insoluble material from wheat flour. It forms an elastic mass after dough has been washed out. The major components of gluten (or prolamins) are glutenins and gliadins, both representing families of proteins [65, 66]. The gliadins are subdivided into α/β, γ, and ω-gliadins, while the glutenins consist of low molecular weight (LMW) and high molecular weight (HMW) glutenins. Gluten molecules are storage proteins and they have high concentrations of glutamine and proline residues (35 and 15% of the total amino acid content [93]). While the glutamine provides nitrogen to the developing seedling, the proline may be involved in the protection of the seeds against drought. Glutenins contain multiple cysteine residues that allow the formation of covalent bonds between glutenin molecules. These bonds establish the matrix of dough and its elasticity. Moreover, gliadins can link to this matrix and enable water binding, which is important for the viscosity of the dough. Differences in the quality and quantity of the glutenin and gliadin molecules in individual wheat varieties determine whether they are more suitable for, e.g., bread making, baking cookies, or for making pasta [65, 66]. The gluten genes in wheat and the gluten-like genes in barley and rye are members of large gene families: while wheat contains over 200 gliadins, HMW, and LMW genes [65, 66], 60 hordein gene copies have been identified in barley. Gluten and the adaptive immune system HLA-DQ, gluten, and T cells Most CD patients are HLA-DQ2 positive [43, 70, 71, 75], suggesting a direct role for this HLA-class II molecule in disease development. Indeed, there is now compelling evidence that CD4 T cell responses to HLA-DQ2-bound, gluten-derived peptides are the primary mechanism behind disease development [2 4, 35, 69, 88, 89]. Such T cell responses can also be found against HLA-DQ8 bound, gluten-derived peptides [37, 83, 84], which is a likely explanation for the fact that most HLA-DQ2 negative CD patients are usually HLA-DQ8 positive [75]. The identification of the gluten peptides that trigger these T cell responses has not been straightforward as the requirements for peptide binding to HLA-DQ2 and HLA- DQ8 seemed incompatible with the nature of gluten. Although gluten is almost devoid of amino acids with negative charge, such amino acids are highly favored by both HLA-DQ2 and -DQ8 [20, 86, 92]. This puzzle was solved when it became clear that gluten can be modified to fit the requirements for high affinity binding to HLA-DQ2 or -DQ8 (Table 1, Fig. 1). The modification involves the conversion of glutamine into glutamic acid by the enzyme tissue transglutaminase (ttg) [45, 85]. Although it has been not formally proven that ttg activity in vivo modifies gluten, there are very strong arguments in favor of this. First of all, in small intestine biopsies of patients T cells are present that, in most cases, respond only or better to gluten peptides that have been modified by ttg [2 4, 35, 45, 47, 69, 85, 88, 89]. Also, ttg levels are upregulated in biopsies of patients and it is found both at the brush border and just below the epithelium [45, 72], locations where ttg can come

Springer Semin Immun (2005) 27:217 232 219 Table 1 HLA-DQ2 prefers negative charges in bound peptides at either the p4, p6, or p7 position in the peptide. Likewise, HLA-DQ8 prefers negative charges at position p1 or p9. These negative charges are introduced by the activity of tissue transglutaminase (ttg), which can convert the uncharged amino acid glutamine (Q) into the negatively charged glutamic acid (E) DQ2-binding PQPQLPYPQ + ttg = FRPQQPYPQ + ttg = PFPQPQLPY + ttg = FPQQPQQPF + ttg = PQQSFPQQQ + ttg = DQ8-binding QGSFQPSQQ + ttg = p 123456789 xxx-x--xx PQPELPYPQ FRPEQPYPQ PFPQPELPY FPEQPEQPF PQQSFPEQE -xxxxxxx- EGSFQPSQE The activity of ttg is dependent on the presence of a proline residue (P, underlined) two amino acids C- terminal of the target glutamine or a large hydrophobic amino acid like phenylamaline (F, underlined) three amino acids C-terminal of the target glutamine into contact with gluten. Moreover, it has been observed that gluten can be modified by ttg in biopsy cultures of patients [46]. Thus, gluten modification by ttg is a prerequisite for a full-blown, gluten-specific T cell response. Broad T cell reactivity toward gluten and gluten-like molecules The immune system should ignore food antigens in the gastrointestinal tract. In the case of gluten, this is clearly often not the case and likely caused by several unique features of the Generation of gluten peptides by enzymes in the gastrointestinal tract 1 2 Modification by ttg T Cell IEL Induction of IL15 TCR NKG2D HLA-DQ2(8) MICA Lamina propria APC Enterocyte Epithelium Fig. 1 Gluten triggers adaptive and innate immune responses. Partial degradation of gluten results in the generation of gluten peptides that are modified by tissue transglutaminase, bind to HLA-DQ2 or HLA-DQ8 and trigger a T cell response in the lamina propria. In addition, through an as yet ill-understood mechanism, gluten can induce IL15 that results in the expression of the NKG2D receptor on intraepithelial lymphocytes (IEL) and MICA on enterocytes. As the result of the NKG2D MICA interaction, the enterocytes are killed

220 Springer Semin Immun (2005) 27:217 232 gluten molecules. Gluten is a heterogeneous mixture of related, but distinct, proteins. In a single wheat variety, up to a hundred gluten molecules may be present. They all have a similar and highly distinctive amino acid composition that is dominated by two amino acids, glutamine, and proline [93]. The high proline content renders gluten resistant to enzymatic degradation in the gastrointestinal tract, increasing the chance that gluten fragments will bind to HLA-DQ [64]. Moreover, due to the high amount of glutamine (Q) and proline (P), sequences like QP, QXP (where X is any amino acid), and QXXP are very common in gluten molecules. As the distance between Q and P determines whether or not Q will be converted into glutamic acid by ttg (Table 1), ttg will only modify specific glutamines in gluten [16, 89]. This is linked to gluten toxicity as the modification of all glutamines would generate gluten peptides that would not be able to bind to HLA-DQ2 or HLA-DQ8. Due to the selective modification of glutamines, ttg can generate gluten peptides that have a negatively charged residue at one (or more) of the three positions where HLA-DQ2 and HLA-DQ8 favor such negative charges (Table 1). Consequently, gluten is now known to contain many immunogenic peptides that bind to HLA-DQ2 with relatively high affinity (Table 2)[30, 91]. Such peptides have been identified in α-gliadins, γ-gliadins, and the LMW and HMW glutenins (Table 2)[2 4, 35, 45, 69, 85, 88, 89]. Moreover, many similar peptides are present in the gluten-like molecules in barley, rye, and to a lesser extent, in the gluten-like molecules in oats (Table 2)[5, 90]. Some of those peptides can actually stimulate gluten-reactive T cells [90]. Thus, as soon as a T cell repertoire to gluten peptides has developed, other cereals are equally harmful to patients due to T cell cross-reactivity with the gluten-like molecules in those other cereals. Table 2 Overview of T cell stimulatory peptides in wheat gluten and the hordeins, secalins, and avenins, the gluten-like molecules in barley, rye, and oats respectively Protein source Cereal Peptide sequence HLA-DQ α-gliadin Wheat PQPQLPYPQ DQ2 Hordein Barley PQPQQPFPQ DQ2 Secalin Rye PQPQQPFPQ DQ2 α-gliadin Wheat PFPQPQLPY DQ2 Hordein Barley PFPQPQQPF DQ2 Secalin Rye PFPQPQQPF DQ2 Avenin Oats PYPEQQEPF DQ2 Avenin Oats PYPEQQQPF DQ2 α-gliadin Wheat FRPQQPYPQ DQ2 Hordein Barley FPPQQPFPQ DQ2 α-gliadin Wheat QGSFQPSQQ DQ8 γ-gliadin Wheat FPQQPQQPF DQ2 Avenin Oats FVQQQQQPF DQ2 γ-gliadin Wheat PQQSFPQQQ DQ2 Secalin Rye PQQSFPQQP DQ2 γ-gliadin Wheat IIQPQQPAQ DQ2 LMW-glutenin Wheat FSQQQQSPF DQ2 LMW-glutenin Wheat FSQQQQQPL DQ2 HMW-glutenin Wheat QGYYPTSPQ DQ8 Gluten Wheat QLPQQPQQF DQ2 The Q-residues in bold are modified by ttg

Springer Semin Immun (2005) 27:217 232 221 HLA-DQ gene dose effect It is known that not all HLA-DQ2 positive individuals have an equal risk of disease development [43]. Next to the increased risk in families where one family member is affected (see below), there is a strong impact of the HLA-DQ2 gene dose (Table 3). HLA-DQ molecules are dimers consisting of an alpha and a beta chain. HLA-DQ2 homozygous individuals express one alpha and one beta chain. Al HLA-DQ molecules will thus be HLA- DQ2 dimers (Table 3) and this correlates with a high risk for disease development. In contrast, HLA-DQ2/DQx heterozygotes express two alpha and two beta chains. Consequently, heterozygotes can potentially form four alpha beta combinations, of which only one will be HLA-DQ2 dimers (Table 3) and this results in a much lower risk for disease development. In vitro, the HLA-DQ2 gene dose influences the strength of the gluten-specific T cell response: gluten-specific T cell responses are much stronger when gluten peptides are presented by antigen-presenting cells (APC) homozygous for HLA-DQ2 compared to presentation by APC heterozygous for HLA-DQ2 [91]. Thus, the level of HLA-DQ2 expression influences the magnitude of the gluten-specific T cell response and this correlates with the risk of disease development (Table 3). Gluten and the innate immune system It was recently established that gluten is a two-edged sword; besides its ability to trigger adaptive immune responses, it has an impact on the innate immune system as well. An α- Table 3 Correlation between the HLA-DQ genotype and the risk of disease development HLA-DQ2 genotype α/β Chain combinations Expression Presentation of gluten epitopes HLA-DR3DQ2 α0501,β0201 + All High HLA-DR3/7DQ2 α0501,β0201 + All High α0201,β0202 + Few α0501,β0202 + All α0201,β0201 + Few HLA- DR5/7DQ7/2 HLA-DR3/ XDQ2/X Risk of disease development α0201,β0202 + Few Intermediate α0505,β0301 + None α0201,β0301? None α0505,β0202 + All α0501,β0201 + All Low αxxx,βxxx None α501,βxxx None αxxx,β0201 None HLA-DR7DQ2 α0201,β0202 + Few None The risk of disease development is determined by the quality and quantity of the expressed HLA-DQ2 dimers. Individuals that generate gluten presenting HLA-DQ2 dimers in cis and in trans (HLA-DR3DQ2 homozygous and HLA-DR3/7DQ2 heterozygous) have the highest risk. Individuals that can only generate gluten presenting HLA-DQ2 dimers with some of their cis and in trans combinations (HLA-DR5/7DQ7/2 and HLA-DR3/ XDQ2/X) have a lower risk. Individuals that generate HLA-DQ2 molecules that can not present the full repertoire of gluten peptides (HLA-DR7DQ2) have no risk of disease development. Adapted from Ref. [91]

222 Springer Semin Immun (2005) 27:217 232 gliadin derived peptide, amino acids 31 43, was shown to induce typical disease symptoms in in vivo challenge studies [77]. It also causes characteristic morphological changes when added to biopsy cultures [38, 53, 68]. Moreover, synergistic effects were observed when the 31 43 peptide was added to biopsy cultures together with a second, T cell stimulatory gluten peptide [39]. A recent study indicates that the effect of the 31 43 peptide is linked to the enhanced expression of IL15 (Fig. 1) [25]. Due to the enhanced expression of IL15, MICA expression on epithelial enterocytes in the intestine is induced along with an upregulation of the NKG2D receptor expression on intraepithelial lymphocytes (IEL) [25, 44]. Subsequently, the interaction between NKG2D and its ligand MICA on enterocytes results in enterocyte killing (Fig. 1)[25, 44]. As it is well known that an IEL infiltration is one of the hallmarks of CD [40, 41], these results indicate that gluten exposure can lead to innate immune responses mediated by NKG2D, which is likely to contribute to tissue damage. Antibodies to tissue transglutaminase are specific indicators of disease As antibodies to ttg are only found in CD patients, these have become an important tool for diagnosis of CD [11, 57]. The involvement of ttg in gluten modification and the formation of these antibodies by CD patients is unlikely to be a coincidence. It has been suggested that it is the ability of ttg to crosslink itself to gluten that results in antibody formation [73]. Uptake of such ttg gluten complexes by B cells expressing ttg-specific IgM would lead to internalization and processing of the gluten. Subsequent binding of gluten fragments to HLA-DQ molecules of the B cell would allow gluten-specific T cells to drive ttg-specific antibody formation. This would imply that although the antibodies are specific indicators of disease, they are not initiators of the disease and are unlikely to cause disease symptoms. Additional genetic factors predispose to CD There is strong evidence that multiple genes predispose to CD. Whereas in the general population the incidence of CD is between 0.5 and 1% [8, 10, 14, 42, 61], this is about 10% in families where one member is affected. Also, in monozygotic twins the concordance rate is about 80% [52]. To date, HLA-DQ on chromosome 6p21.3 is the only identified locus that contributes to the development of CD. HLA-DQ, however, is certainly not sufficient as HLA- DQ2 alone can account for about 40% of the genetic variation underlying CD [81]. Hence, other non-hla genes are also important determinants of disease etiology. The following various strategies are currently employed to identify CD susceptibility genes and loci: 1. Genetic linkage studies to identify chromosomal regions likely to contain disease-causing genes 2. Candidate gene association studies to establish if genes suspected to be relevant to disease pathogenesis are indeed causally related 3. Gene expression profiling to identify genes with altered expression due to their involvement in disease etiology and/or pathology.

Springer Semin Immun (2005) 27:217 232 223 Genetic linkage and association studies These studies are usually performed by using affected pairs of siblings. The principle of linkage studies is to establish whether cosegregation of a disease phenotype exists in diseased families. In 1996, Zhong and coworkers [98] reported the first autosomal screen for genes that predispose to CD in the western counties of Ireland. To date, 11 genome-wide screens for linkage regions have been performed in CD, as well as numerous region-specific linkage studies [22, 23, 31, 32, 34, 49, 51, 54, 55, 59, 81, 82, 94, 98]. These studies have identified a number of putative loci across different populations (for an overview, see Table 4). Although most non-hla loci have been reported with weak significance, promising loci are the CELIAC2 locus on 5q31 q33, the CELIAC3 locus on 2q33, the CELIAC4 locus on 19p13.1 and loci on 6q21 q22 and 9p21 p13. The locus on 5q31 q33 (CELIAC2) has been repeatedly identified in various genomewide linkage screens [22, 23, 34, 50, 98], albeit with only suggestive evidence in most of the studies, implying a locus of modest effect. A meta-analysis of 442 celiac families provided convincing evidence for this locus (p=0.000006) [6]. The 5q31 q33 region contains a number of cytokine genes, none of which has so far shown association to CD [60]. Interestingly, this same region has been implicated in inflammatory bowel disease (IBD) (IBD5 [58]), autoimmune thyroid diseases [62], asthma [97], and rheumatoid arthritis [79]. The CELIAC3 locus encompassing the T-lymphocyte regulatory genes CD28, CTLA4, and ICOS in a 300-kb block in chromosome 2q33 has been identified by both linkage and association studies in several different populations [24, 50, 56, 59]. Results for the CELIAC3 locus, however, have been inconsistent, both with respect to genetic association with single nucleotide polymorphisms (SNPs) in CTLA-4 as well as with markers covering the CELIAC2 region (for overviews, see [24, 87]. These results might suggest that the true susceptibility CD gene from this region has not yet been identified, since rather strong association was Table 4 Overview of linkage to non-hla regions in genome-wide screens conducted for CD Population No. of families Study design Suggestive b linkage Significant b linkage Reference (1st/2nd/3rd) a Western Ireland 15/ Affected sibpairs 6p23, 11p11 [98] Italy 39/57/87 Affected sibpairs 5qter (CELIAC2) [22] UK 16/34 Extended families [31, 32] Sweden/Norway 70/36 Affected sibpairs [49] Finland 60/38 Affected sibpairs 4p15 [34] Finland 9/1 Population Isolate 15q12 [94] North Europe 24/ Extended families [55, 56] North America 62/ Extended families 3p26, 5p14, 18q23 [51] Netherlands 67/15 Affected sibpairs 6q21 19p13.1 (CELIAC4) [81] Netherlands 1/ Extended family 9p21 p13 [82] Finland 54/ Affected sibpairs 10p 2q23 q32 (CELIAC3) [59] a Number of families in initial genome-wide screen (1st) and subsequent follow-up studies (2nd and 3rd) b Suggestive linkage, p<7.4 10 4 ; significant linkage, p<2.2 10 5 (according to criteria proposed by Lander and Kruglyak [99]. The symbol denotes that the indicated genome-wide significance threshold was not reached

224 Springer Semin Immun (2005) 27:217 232 observed for a region 2 Mb away from CTLA-4. The chromosome region 2q33 has also been implicated in a variety of autoimmune diseases such as systemic lupus erythematosis, autoimmune thyroid diseases, type 1 diabetes mellitus, and rheumatoid arthritis [80]. More recently, a significant locus on chromosome 19 (CELIAC4) was identified in the Dutch population [81], which awaits formal replication. The EU meta-analysis also suggested this locus, although with modest evidence [6]. This same region may also be of relevance for Crohn s disease [87]. The linkage screen by van Belzen et al. [81] also revealed a locus with suggestive linkage on chromosome 6q21 22, a region that has likewise been implicated in type 1 diabetes mellitus (IDDM15 [9]), multiple sclerosis [1], and rheumatoid arthritis [28]. Finally, a promising CD locus is on chromosome 9p21 p13 with suggestive evidence for linkage in the Dutch population [82] and nominal evidence for linkage in the Swedish/ Norwegian [50] and the Finnish population [34]. Thus, three of the five loci have also been implicated in other autoimmune disorders. It is likely, therefore, that some of the CD genes actually predispose to a general susceptibility to autoimmunity and/or inflammation. Given the high comorbidity of CD with other autoimmune disorders, common pathways may be involved in the destruction of the target tissues. Many functional candidate genes have been tested for association with CD based on knowledge of CD pathogenesis and immunology. Outside the HLA and the 2q33 region, genes studied to date include interferon gamma (IFN-γ), interferon regulatory factor 1 (IRF1), killer cell immunoglobulin-like receptor (KIR) and leukocyte-associated Ig-like receptor-1 (LILR) gene clusters, interleukin 12B (IL12B), and interleukin 6 (IL6). In addition, tissue transglutaminase 2 (TG2) and prolylendopeptidase (PREP) were studied because of their role in gluten modification. Other studied genes include the tumor necrosis factor receptor superfamily, member 6 (FAS), the matrix metalloproteinases 1 and 3 (MMP1/3), fibroblast activation protein, alpha (DPPIV), nitric oxide synthase 2A (NOS2), and elastin (ELN). No convincing disease associations, however, have been found. Gene expression profiling The main aim of these studies is to obtain insight into disease pathogenesis, which is currently limited to mechanisms of inflammation. For this purpose, the expression of genes in mucosal biopsy samples from untreated CD patients with complete villous atrophy was compared to that of normal control biopsies [13, 29]. These studies indicate that in active CD, there is an active Th1 response, enhanced cell proliferation, reduced epithelial differentiation, recruitment of γδt cells, B cells and macrophages, and lack of changes in matrix metalloproteinases. A number of genes mapped to linkage regions (reviewed in [95]), thereby providing candidate genes to be tested in genetic association studies. In conclusion, a small number of interesting genetic regions await further investigation, including the CELIAC2 and CELIAC4 loci. Based on available data, it is expected that the total genetic risk of CD can be attributed to only one or few genes with considerable effect (such as HLA-DQ2), and to many genes with very modest effect (such as CTLA-4). The integration of genetic and microarray technologies may assist in identifying causative disease genes (Fig. 2). More robust strategies may be required to identify all CD susceptibility genes. Such strategies require large numbers of samples (hundreds to thousands) and SNPs that cover the entire genome and capture all haplotype blocks. Due to recent improvements in

Springer Semin Immun (2005) 27:217 232 225 high-throughput technology and statistics, such studies are likely to become feasible. The identification of CD susceptibility genes may lead to the development of easy, applicable, and noninvasive molecular diagnostic tools. The ability to distinguish the healthy from the CD mucosa based on the expression profile opens the possibility to also assemble a panel of marker genes for molecular phenotyping to assist in diagnosis and prognosis. These tools Fig. 2 Outline of a strategy integrating genetic and genomic data. In genetic studies, family material is collected for linkage analysis, which subsequently identifies positional candidate regions and genes. These genes can be further tested for genetic association, as well as functional candidate genes. Genomic studies utilize small intestinal biopsies from CD patients and controls, which are analyzed using microarrays, resulting in lists of significantly differentially expressed genes. These genes might identify putative disease-associated pathways. Both data sources can also be combined as positional candidate genes that show markedly different expression behavior may be candidate genes for genetic association and subsequent sequence analysis to identify disease-causing variants. Furthermore, the different data sources can be integrated to elucidate disease pathways by establishing gene networks

226 Springer Semin Immun (2005) 27:217 232 may vastly improve the diagnosis of CD particularly as, to date, some 85% of all CD patients go undiagnosed. Insight into the molecular pathways involved in CD etiology may eventually provide new targets for therapeutic intervention. The future of a gluten-free diet Although a gluten-free diet is the best available treatment for CD patients, this diet can be problematic. Gluten-free bread is not appreciated because of unpleasant taste, texture, and storage problems. Moreover, a wide variety of food products such as meat products, ketchups, sauces, sweets, and medication may contain gluten as a hidden ingredient. In addition, safe cereals are often contaminated with unsafe cereals as the result of mixing during transport and/or manufacture of food products. Finally, gluten-free products may originate from gluten-containing raw materials. Selection of nontoxic wheat varieties and adaptations in manufacturing procedures would thus be of benefit to CD patients. Cereals Wheat is one of the most commonly used cereals worldwide. The cultivation of wheat was slowly dispersed by human activity at a rate of a few kilometers a year from its center of origin, Mesopotamia (the Iraq Iran region), one of the places where agriculture started about 10,000 years ago. Wheat reached Western Europe as an agricultural crop about 3,000 years ago. Like many other food crops, wheat is a member of the monocotyledonous Gramineae family (grasses). This family has two subfamilies, the Panicoideae and the Festucoideae, each subdivided into several tribes. Within the Festucoideae, wheat, rye, and barley are members of the Hordeae tribus, oats belong to the Aveneae tribus, and rice to the Oyzeae tribus. Other members of this subfamily are finger millet, teff, and wild rice. Important crops from the Panicoideae subfamily are maize, sorghum, and millet [78]. Whereas wheat, rye, and barley are known to be toxic for patients, oats are generally considered safe for CD patients although recent studies indicate that oats may not be safe for all [5, 27, 36, 76, 90]. In addition, rice and maize are recognized as safe, while millet, finger millet, teff, and Job s tears are experienced as safe (Kasarda 2003, http://www.aaccnet.org/grainbin/kasarda.asp). The safety of sorghum is still unclear. There is a huge variation in the size and chromosome number of different cereals [7]. While rice has the smallest genome (450 Mbp), the genomes of barley, rye, oat, and wheat are at least 10 40 times larger. The expansion of these genomes can be attributed to four different mechanisms: (1) the presence of repetitive sequences that increase the amount of DNA [17]; (2) gene duplications [17]; (3) duplication of the entire genome; and (4) polyploidization, whereby two or more distinct genomes are brought together into the same nucleus. Due to these mechanisms, a huge diversity has developed, especially in wheat [33]. Tetraploid (pasta or durum) wheat was formed by hybridization of the AA and BB genomes. The hexaploid bread wheat (T. aestivum) is the result of a more recent polyploidy event that occurred about 10,000 years ago by hybridization of domesticated tetraploid wheat (AABB) with an Aegilops species bringing in the D-genome to create a new species with the genome AABBDD.

Springer Semin Immun (2005) 27:217 232 227 Another complexing factor is that not all gluten genes are expressed during the development of the wheat kernel. Gene expression is regulated by natural gene silencing, and the degree of silencing differs between the different gene families and loci [19, 67]. In the tetraploid and hexaploid wheat species, gene expression is also regulated by the interaction of the different genomes. Together, these phenomena result in differences of gluten composition among species and varieties. The way out of toxicity Some 10,000 wheat varieties are known and there is great diversity in gluten gene expression. This implies that there may be ways to select safer wheat varieties for patients. To date, no systematic analysis has been carried out to elucidate this aspect. Recently, Spaenij-Dekking et al. (unpublished results) made the first attempt to quantify the toxicity of a range of cultivars, including several diploid ancestor species, tetraploid, and hexaploid cultivars. For this purpose, the presence of toxic gluten peptides was determined with the use of T cells and monoclonal antibodies specific for these peptides [74]. The results demonstrated large quantitative differences in the presence of toxic gluten peptides among the cultivars tested. Some of these cultivars were even found to completely lack particular harmful gluten peptides. Such heterogeneity indicates that through breeding programs, wheat varieties may be generated that are safer for consumption by CD patients. In addition, large-scale genomics wheat research is presently being carried out and expected to elucidate the genetic and allelic diversity of the wheat gluten genes. By linking this genomic data to toxicity data, marker-assisted breeding programs can be established to facilitate the production of nontoxic wheat varieties. Finally, RNA interference is a promising technique to silence those gluten genes that express toxic sequences, as it has been used successfully to silence complete gene families in wheat [18] and apple [21]. In conclusion, the development of less harmful wheat varieties may be established by selecting wheat varieties that lack or contain only low amounts of particular toxic gluten peptides. Marker-assisted breeding programs may subsequently be used to generate wheat that completely lacks such peptides. Finally, RNA interference may be used to eliminate remaining harmful gluten genes. The availability of safe wheat varieties would be particularly advantageous when these could be cultivated on a very large scale and replace the current toxic varieties. This would considerably reduce the chance of unintentional contamination, and would reduce the price of safe products. Such a development, however, can not be expected within the next decade. Conclusions and perspectives Celiac disease has become one of the best-understood HLA-associated diseases. Due to its unique properties, gluten is relatively resistant to degradation in the gastrointestinal tract [64]. Moreover, due to the complex nature of gluten itself and the activity of ttg, many immunogenic peptides can be formed which provide many opportunities for the initiation of an adaptive immune response in the lamina propria (Fig. 1). The likelihood of such immune

228 Springer Semin Immun (2005) 27:217 232 responses is influenced by the availability of HLA-DQ molecules and exposure to gluten: HLA-homozygous individuals have the highest chance of disease development and exposure to high amounts of gluten leads to higher disease incidence [26]. In addition, gluten can act on the innate immune system through the induction of IL15. This leads to the expression of NKG2D on IEL and its ligand MICA on enterocytes. The subsequent NKG2D MICA interaction is sufficient to result in enterocyte killing (Fig. 1). The two processes are likely to influence each other. Inflammation in the lamina propria may contribute to IL15 production and thus enhance enterocyte killing while loss of epithelial integrity may facilitate gluten presentation in the lamina propria. A major unresolved issue is why relatively few individuals develop CD. Over 25% of the population in the Western Hemisphere is HLA-DQ2 positive, eats high amounts of gluten daily, and expresses tissue transglutaminase in the intestine. Immunogenic HLA-DQ peptide complexes would thus be expected to be formed but this does not lead to disease in the majority of cases. Also, the ability of gluten to induce NKG2D/MICA is apparently harmless under normal circumstances. Deciphering the genetic and environmental factors that determine the outcome of the interaction between gluten and the adaptive and innate immune system will thus be the major challenge for the coming years. This may also allow the development of improved diagnostic and prognostic tools. In the meantime, the knowledge already acquired may be applied. The identification of the disease-inducing components in gluten provides opportunities to develop safer foods for patients. A long-term goal would be the generation of wheat that is safe for consumption. In the short-term, food safety may be better guaranteed when methods are used that monitor the presence of the disease-inducing components in food [74]. It may also prove worthwhile to reconsider the way gluten is introduced into the diet. The current policy is ad libitum consumption after the age of 6 months. Considering that there is clear evidence that both the HLA-DQ2 gene dose and the amount of gluten introduced influence the likelihood of disease development, it may be reasonable to introduce gluten more gradually. This may be an effective way to reduce disease incidence and/or delay the age of onset. Finally, can we devise therapy other than a gluten-free diet? More efficient degradation of gluten, for example, through enzyme supplementation [64], is an attractive option. Given that gluten is degradation-resistant due to its high proline content, supplementation with prolylendopeptidases that cleave proteins after a proline would be the best option, provided that these enzymes are effective before the gluten reaches the small intestine [64]. Alternatively, blockers of ttg, HLA-DQ2, and NKG2D would be drugs that specifically target key molecules involved in the adaptive and innate immune response to gluten (Fig. 1). It remains to be established if such drugs would be safe enough to replace a gluten-free diet. References 1. Akesson E, Oturai A, Berg J et al (2002) A genome-wide screen for linkage in Nordic sib-pairs with multiple sclerosis. Genes Immun 3:279 2. Anderson RP, Degano P, Godkin AJ et al (2000) In vivo antigen challenge in celiac disease identifies a single transglutaminase-modified peptide as the dominant A-gliadin T-cell epitope. Nat Med 6:337

Springer Semin Immun (2005) 27:217 232 229 3. Arentz-Hansen H, Körner R, Molberg Ø et al (2000) The intestinal T cell response to α-gliadin in adult celiac disease is focused on a single deamidated glutamine targeted by tissue transglutaminase. J Exp Med 191:603 4. Arentz-Hansen H, McAdam SN, Molberg Ø et al (2002) Celiac lesion T cells recognize epitopes that cluster in regions of gliadins rich in proline residues. Gastroenterology 123:803 5. Arentz-Hansen H, Fleckenstein B, Molberg O et al (2004) The molecular basis for oat intolerance in patients with celiac disease. PLoS Med 1:84 6. Babron MC, Nilsson S, Adamovic S et al (2003) Meta and pooled analysis of European coeliac disease data. Eur J Hum Genet 11:828 7. Bennet MD, Smith JB (1991) Nuclear DNA amounts in angiosperms. Philos Trans R Soc Lond Ser B 334:309 8. Catassi C, Ratsch IM, Fabiani E et al (1994) Coeliac disease in the year 2000: exploring the iceberg. Lancet 343:200 9. Cox NJ, Wapelhorst B, Morrison VA et al (2001) Seven regions of the genome show evidence of linkage to type 1 diabetes in a consensus analysis of 767 multiplex families. Am J Hum Genet 69:820 10. Csizmadia CGDS, Mearin MKL, von Blomberg BMW et al (1999) An iceberg of childhood coeliac disease in the Netherlands. Lancet 353:813 11. Dieterich W, Ehnis T, Bauer M et al (1997) Identification of tissue transglutaminase as the autoantigen of celiac disease. Nat Med 3:797 12. Dicke WK, Weijers HA, van de Kamer JH (1953) Coeliac disease. II. The presence in wheat of a factor having a deleterious effect in cases of coeliac disease. Acta Paediatr 42:34 13. Diosdado B, Wapenaar MC, Franke L et al (2004) A microarray screen for novel candidate genes in celiac disease pathogenesis. Gut 53:944 14. Fasano A, Berti I, Gerarduzzi T et al (2003) Prevalence of celiac disease in at-risk and not-at-risk groups in the United States: a large multicenter study. Arch Intern Med 163:286 15. Feldman M, Lupton FGH, Miller TE (1995) Wheats. In: Smartt J, Simmonds NW (eds) Evolution of crop plants, vol 2. Longman, Harlow, Essex, pp 184 192 16. Fleckenstein B, Molberg O, Qiao SW et al (2002) Gliadin T cell epitope selection by tissue transglutaminase in celiac disease. Role of enzyme specificity and ph influence on the transamidation versus deamidation process. J Biol Chem 277:34109 17. Feuillet C, Keller B (2002) Comparative genomics in the grass family: molecular characterization of grass genome structure and evolution. Ann Bot 89:3 10 18. Folck A, Wieser H, Lörz H et al (2003) Silencing the alpha-gliadins in wheat. In: Book of abstracts, ISPMB 7th international congress of plant molecular biology. ISPMB, Barcelona, pp 403 (S28 28) 19. Gianibelli MC, Larroque OR, MacRitchie F et al (2001) Biochemical, genetic, and molecular characterization of wheat glutenin and its component subunits. Cereal Chem 78:635 646 20. Godkin A, Friede T, Davenport M et al (1997) Use of eluted peptide sequence data to identify the binding characteristics of peptides to the insulin-dependent diabetes susceptibility allele HLA-DQ8 (DQ3.2). Int Immunol 9:905 21. Gilissen LJWJ, Bolhaar STHP, Matos CI et al. J Allergy Clin Immunol (in press) 22. Greco L, Corazza G, Babron MC et al (1998) Genome search in celiac disease. Am J Hum Genet 62:669 23. Greco L, Babron MC, Corazza GR et al (2001) Existence of a genetic risk factor on chromosome 5q in Italian coeliac disease families. Ann Hum Genet 65:35 24. Holopainen P, Arvas M, Sistonen P et al (1999) CD28/CTLA4 gene region on chromosome 2q33 confers genetic susceptibility to celiac disease. A linkage and family-based association study. Tissue Antigens 53:470 25. Hüe S, Mention J-J, Monteiro RC et al (2004) A direct role for NKG2D/MICA interaction in villous atrophy during celiac disease. Immunity 21:367 26. Ivarsson A, Persson LÅ, Nyström L et al (2000) Epidemic of coeliac disease in Swedish children. Acta Paediatr 89:165 27. Janatuinen EK, Pikkarainen PH, Kemppainen TA et al (1995) A comparison of diets with and without oats in adults with celiac disease. N Engl J Med 333:1033 28. Jawaheer D, Seldin MF, Amos CI et al, North American Rheumatoid Arthritis Consortium (2003) Screening the genome for rheumatoid arthritis susceptibility genes: a replication study and combined analysis of 512 multicase families. Arthritis Rheum 48:906 29. Juuti-Uusitalo K, Maki M, Kaukinen K et al (2004) cdna microarray analysis of gene expression in coeliac disease jejunal biopsy samples. J Autoimmun 22:249

230 Springer Semin Immun (2005) 27:217 232 30. Kim CY, Quarsten H, Bergseng E et al (2004) Structural basis for HLA-DQ2-mediated presentation of gluten epitopes in celiac disease. Proc Natl Acad Sci U S A 101:4175 31. King AL, Yiannakou JY, Brett PM et al (2000) A genome-wide family-based linkage study of coeliac disease. Ann Hum Genet 64:479 32. King AL, Fraser JS, Moodie SJ et al (2001) Coeliac disease: follow-up linkage study provides further support for existence of a susceptibility locus on chromosome 11p11. Ann Hum Genet 65:377 33. Levy AA, Feldman M (2002) The impact of polyploidy on grass genome evolution. Plant Physiol 130:1587 1593 34. Liu J, Juo SH, Holopainen P et al (2002) Genomewide linkage analysis of celiac disease in Finnish families. Am J Hum Genet 70:51 35. Lundin KE, Scott H, Hansen T et al (1993) Gliadin-specific, HLA-DQ(α1*0501,β1*0201) restricted T cells isolated from the small intestinal mucosa of celiac disease patients. J Exp Med 178:187 36. Lundin KEA, Nilsen EM, Scott HG et al (2003) Oats induced villous atrophy in coeliac disease. Gut 52:1149 37. Lundin KA, Scott H, Fausa O et al (1994) 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 41:285 38. Maiuri L, Troncone R, Mayer M et al (1996) In vitro activities of A-gliadin-related synthetic peptides: damaging effect on the atrophic coeliac mucosa and activation of mucosal immune response in the treated coeliac mucosa. Scand J Gastroenterol 31:247 39. Maiuri L, Ciacci C, Ricciardelli I et al (2003) Association between innate response to gliadin and activation of pathogenic T cells in coeliac disease. Lancet 362:30 40. Maki M, Holm K, Collin P et al (1991) Increase in/t cell receptor bearing lymphocytes in normal small bowel mucosa in latent celiac disease. Gut 32:1412 41. Maki M, Collin P (1997) Coeliac disease. Lancet 349:1755 42. Maki M, Mustalahti K, Kokkonen J et al (2003) Prevalence of celiac disease among children in Finland. N Engl J Med 348:2517 43. Mearin ML, Biemond I, Pena A et al (1983) HLA-DR phenotypes in Spanish coeliac children: their contribution to the understanding of the genetics of the disease. Gut 24:532 44. Meresse B, Chen Z, Ciszewski C et al (2004) Coordinated induction by IL15 of a TCR-independent NKG2D signaling pathway converts CTL into lymphokine-activated killer cells in celiac disease. Immunity 21:357 45. Molberg Ø, McAdam S, Körner R et al (1998) Tissue transglutaminase selectively modifies gliadin peptides that are recognized by gut derived T cells in celiac disease. Nat Med 4:713 46. Molberg Ø, McAdam S, Lundin KEA et al (2001) T cells from celiac-disease lesions recognize gliadin epitopes deamidated in situ by endogenous tissue transglutaminase. Eur J Immunol 31:1317 47. Molberg Ø, Flaete NS, Jensen T et al (2003) Intestinal T-cell responses to high-molecular-weight glutenins in celiac disease. Gastroenterology 125:337 48. Morley M, Molony CM, Weber TM et al (2004) Genetic analysis of genome-wide variation in human gene expression. Nature 430:743 49. Naluai AT, Nilsson S, Samuelsson L et al (2000) 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 56:350 50. Naluai AT, Nilsson S, Gudjonsdottir AH et al (2001) Genome-wide linkage analysis of Scandinavian affected sib-pairs supports presence of susceptibility loci for celiac disease on chromosomes 5 and 11. Eur J Hum Genet 9:938 51. Neuhausen SL, Feolo M, Camp NJ et al (2002) Genome-wide linkage analysis for celiac disease in North American families. Am J Med Genet 111:1 52. Papadopoulos G, Wijmenga C, Koning F (2001) Interplay between genetics and the environment in the development of celiac disease: perspectives for a healthy life. J Clin Invest 108:1261 53. Picarelli A, Di Tola M, Sabbatella L et al (1999) 31 43 amino acid sequence of the alpha-gliadin induces anti-endomysial antibody production during in vitro challenge. Scand J Gastroenterol 34:1099 54. Percopo S, Babron MC, Whalen M et al (2003) Saturation of the 5q31 q33 candidate region for coeliac disease. Ann Hum Genet 67:265 55. Popat S, Bevan S, Braegger CP et al (2002a) Genome screening of coeliac disease. J Med Genet 39:328 56. Popat S, Hearle N, Hogberg L et al (2002) Variation in the CTLA4/CD28 gene region confers an increased risk of coeliac disease. Ann Hum Genet 66:125

Springer Semin Immun (2005) 27:217 232 231 57. Reif S, Lerner A (2004) Tissue transglutaminase the key player in celiac disease: a review. Autoimmun Rev 3:40 58. Rioux JD, Daly MJ, Silverberg MS et al (2001) Genetic variation in the 5q31 cytokine gene cluster confers susceptibility to Crohn disease. Nat Genet 29:223 59. Rioux JD, Karinen H, Kocher K et al (2004) Genomewide search and association studies in a Finnish celiac disease population: identification of a novel locus and replication of the HLA and CTLA4 loci. Am J Med Genet 130:345 60. Ryan AW, Thornton JM, Brophy K et al (2004) A directed candidate gene analysis of coeliac disease susceptibility on chromosome 5. In: Abstract book of 11th ISCD, Belfast, April 2004 (abstract no. P41) 61. Rostami K, Mulder CJ, Werre JM et al (1999) High prevalence of celiac disease in apparently healthy blood donors suggests a high prevalence of undiagnosed celiac disease in the Dutch population. Scand J Gastroenterol 34:276 62. Sakai K, Shirasawa S, Ishikawa N et al (2001) Identification of susceptibility loci for autoimmune thyroid disease to 5q31 q33 and Hashimoto s thyroiditis to 8q23 q24 by multipoint affected sib-pair linkage analysis in Japanese. Hum Mol Genet 10:1379 63. Schadt EE, Monks SA, Drake TA et al (2003) Genetics of gene expression surveyed in maize, mouse and man. Nature 422:297 64. Shan L, Molberg Ø, Parrot I et al (2002) Structural basis for gluten intolerance in celiac sprue. Science 297:2275 65. Shewry PR, Tatham AS (1990) The prolamin storage proteins of cereal seeds: structure and evolution. Biochem J 267:1 66. Shewry PR, Halford NG (2002) Cereal seed storage proteins: structures, properties and role in grain utilization. J Exp Bot 53:947 67. Shewry PR, Lookhart GL (2003) Wheat gluten protein analysis. American Association of Cereal Chemists, St. Paul, MN, USA 68. Shidrawi RG, Day P, Przemioslo R et al (1995) In vitro toxicity of gluten peptides in coeliac disease assessed by organ culture. Scand J Gastroenterol 30:758 69. Sjostrom H, Lundin KEA, Molberg Ø et al (1998) Identification of a gliadin T cell epitope in coeliac disease: general importance of gliadin deamidation for intestinal T cell recognition. Scand J Immunol 48:111 70. Sollid LM, Markussen G, Ek J et al (1989) Evidence for a primary association of coeliac disease to a particular HLA-DQ alpha/beta heterodimer. J Exp Med 169:345 71. Sollid LM, Thorsby E (1993) HLA susceptibility genes in celiac disease: genetic mapping and role in pathogenesis. Gastroenterology 105:910 72. Sollid LM (2002) Coeliac disease: dissecting a complex inflammatory disorder. Nat Rev Immunol 2:647 73. Sollid LM, Molberg Ø, McAdam S et al (1997) Autoantibodies in coeliac disease: tissue transglutaminase guilt by association? Gut 41:851 74. Spaenij-Dekking EHA, Kooy-Winkelaar EMC, Drijfhout JW et al (2004) A novel and sensitive method for the detection of T cell stimulatory epitopes of α/β- and γ-gliadin. Gut 53:1267 75. Spurkland A, Ingvarsson G, Falk ES et al (1997) Dermatitis herpetiformis and celiac disease are both primarily associated with the HLA-DQ (alpha 1*0501, beta 1*02) or the HLA-DQ (alpha 1*03, beta 1*0302) heterodimers. Tissue Antigens 49:29 76. Srinivasan U, Leonard N, Jones E et al (1996) Absence of oats toxicity in adult coeliac disease. BMJ 313:1300 77. Sturgess R, Day P, Ellis HJ et al (1994) Wheat peptide challenge in coeliac disease. Lancet 343:758 78. Thompson T (2000) Questionable foods and the gluten-free diet: survey of current recommendations. J Am Diet Assoc 100:463 79. Tokuhiro S, Yamada R, Chang X et al (2003) An intronic SNP in a RUNX1 binding site of SLC22A4, encoding an organic cation transporter, is associated with rheumatoid arthritis. Nat Genet 35:341 80. Vaidya B, Pearce S (2004) The emerging role of the CTLA-4 gene in autoimmune endocrinopathies. Eur J Endocrinol 150:619 81. van Belzen MJ, Meijer JWR, Sandkuijl LA et al (2003) A major non-hla locus in celiac disease maps to chromosome 19. Gastroenterology 125:1032 82. Van Belzen MJ, Vrolijk M, Meijer JWR et al (2004) A genomewide screen in a four-generation Dutch family with celiac disease: evidence for linkage to chromosomes 6 and 9. Am J Gastroenterol 99:461 83. van de Wal Y, Kooy Y, van Veelen P et al (1998) Small intestinal cells of celiac disease patients recognize a natural pepsin fragment of gliadin. Proc Natl Acad Sci U S A 95:10050

232 Springer Semin Immun (2005) 27:217 232 84. van de Wal Y, Kooy YMC, van Veelen P et al (2000) Glutenin is involved in the gluten-driven mucosal T cell response. Eur J Immunol 29:3133 85. van de Wal Y, Kooy YMC, van Veelen P et al (1998) Selective deamidation by tissue transglutaminase strongly enhances gliadin-specific T cell reactivity. J Immunol 161:1585 86. van de Wal Y, Kooy YMC, Drijfhout JW et al (1996) Peptide binding characteristics of the coeliac disease-associated DQ(α1*0501,β1*0201) molecule. Immunogenetics 44:246 87. van Heel DA, Fisher SA, Kirby A et al (2004) Genome scan meta-analysis group of the IBD International Genetics Consortium. Inflammatory bowel disease susceptibility loci defined by genome scan metaanalysis of 1952 affected relative pairs. Hum Mol Genet 13:763 88. Vader W, Kooy Y, van Veelen P et al (2002) The gluten response in children with recent onset celiac disease. A highly diverse response towards multiple gliadin and glutenin derived peptides. Gastroenterology 122:1729 89. Vader W, de Ru A, van de Wal Y et al (2002) Specificity of tissue transglutaminase explains cereal toxicity in celiac disease. J Exp Med 195:643 90. Vader W, Stepniak D, Bunnik EM et al (2003) Characterization of cereal toxicity for celiac disease patients based on protein homology in grains. Gastroenterology 125:1105 91. Vader W, Stepniak D, Kooy Y et al (2003) The HLA-DQ2 gene dose effect in celiac disease is directly related to the magnitude and breadth of gluten-specific T-cell responses. Proc Nat Acad Sci U S A 100:12390 92. Vartdal F, Johansen BH, Friede T et al (1996) The peptide binding motif of the disease-associated HLA- DQ(α1*0501,β1*0201) molecule. Eur J Immunol 26:2764 93. Wieser H, Seilmeier W, Eggert M et al (1983) Tryptophan content of cereal proteins. Z Lebensm-Unters- Forsch 177:457 94. Woolley N, Holopainen P, Ollikainen V et al (2002) A new locus for coeliac disease mapped to chromosome 15 in a population isolate. Hum Genet 111:40 45 95. Wijmenga C, Wapenaar MC (in press) A combined genetics and genomics approach to unravelling molecular pathways in coeliac disease 96. Yvert G, Brem RB, Whittle J et al (2003) trans-acting regulatory variation in Saccharomyces cerevisiae and the role of transcription factors. Nat Genet 35:57 97. Xu J, Meyers DA, Ober C et al (2001) Genomewide screen and identification of gene gene interactions for asthma-susceptibility loci in three U.S. populations: collaborative study on the genetics of asthma. Am J Hum Genet 68:1437 98. Zhong F, McCombs CC, Olson JM et al (1996) An autosomal screen for genes that predispose to celiac disease in the western counties of Ireland. Nat Genet 14:329 99. Lander E, Kruglyak L (1995) Genetic dissection of complex traits: guidelines for interpreting and reporting linkage results. Nat Genet 11(3):241 247