Article Text


Getting to grips with gluten
  1. S N McADAM,
  1. Institute of Immunology, Rikshospitalet
  2. University of Oslo, N-0027 Oslo, Norway
  1. L M Sollid. l.m.sollid{at}

Statistics from

Ever since Dicke and coworkers identified gluten as the precipitating agent in coeliac disease (CD),1 2 enormous effort has been invested to identify the elements within gluten that are responsible for eliciting the enteropathy. This has been a frustrating endeavour because of the unusual chemistry and huge complexity of gluten proteins. Gluten consists of the alcohol-water extractable fraction of monomeric gliadins and the alcohol-water unextractable fraction of glutenins cross linked into polymers by intermolecular disulphide bonds. The gliadins can be subdivided on the basis of their amino acid sequences into α/β gliadins, γ gliadins, and ω gliadins, while glutenins can be subdivided into low (LMW) and high (HMW) molecular weight glutenins. Importantly, there are numerous variants of each type of gliadins and glutenins in a single wheat variety. Early feeding studies indicated that both gliadin and glutenin fractions were harmful to coeliac patients.2 3Results with glutenins, however, should be interpreted cautiously as the glutenin fraction also contains subunits with characteristic gliadin sequences but which have odd numbers of cysteine residues that results in the formation of intermolecular disulphide bonds.4 The insolubility of gluten further complicates testing of its toxicity. To bypass this problem, gluten is often fragmented, typically with proteolytic enzymes, prior to testing. Proteolytic fragmentation usually generates a vast array of distinct peptides. From our experience, studies that have obtained results with purified fragments without corroborating results with synthetic analogues should be interpreted with caution. We have often found that a minor peptide constituent can mediate the biological effect of an active fraction rather than the major peptide that is detected by sequencing.

The problem of identifying “toxic” sequences has been further complicated by the lack of reliable and relevant toxicity assays. Assays including culturing of atrophic coeliac mucosa,5culturing of fetal chick or rat intestines,6 7agglutination of K562(S) cells,8 effect on rat liver lysozyme,6 and apoptosis induction in Caco-2 cells9 have been used to test for toxicity. This plethora of assays reflects the diverse set of theories, such as the lectin hypothesis, missed peptidase hypothesis, permeability defect hypothesis, and immunological model, that have been put forward over the years to explain the pathogenesis of CD. There is now convincing evidence that CD is a multifactorial disease driven by an abnormal T cell response to gluten.10 It is therefore timely to question how the majority of the toxicity assays relate to CD.

CD is an immunological disorder

Evidence for the involvement of T cells in CD was first provided by the demonstration that the HLA class II molecules DQ2 and DQ8 confer disease susceptibility.10 These molecules, which are found on the surface of antigen presenting cells, bind peptides (typically consisting of 10–15 amino acid residues) and present them to CD4+ T cells. This raised the possibility that the intestinal T cell infiltration found in CD contained CD4+ T cells specific for gluten peptides complexed with DQ2 or DQ8. This idea was quickly supported by immunohistochemical studies which showed a significant increase in activation of lamina propria CD4+ T cells (CD25+) in coeliac biopsies after an in vitro gluten challenge.11 The presence of gluten reactive intestinal CD4+ T cells was confirmed by isolating CD25+ T cells from gluten challenged biopsies and establishing T cell lines and clones that proliferated in response to gluten.12 Notably, these studies revealed that the gluten specific T cells almost exclusively responded to gluten when presented by DQ2 (or DQ8) and not when presented by any of the other class II molecules found on the antigen presenting cells of patients. The gluten specific T cell lines and clones were positive for the αβ T cell receptor,12with typical Th1 or Th0 phenotypes, secreting large amounts of interferon γ on activation.13 It is likely that activation of these cells is responsible for the intestinal pathology of CD, either directly through secretion of cytokines or indirectly through stimulating the production of metalloproteinases.10 In strong support of T cells being able to cause intestinal pathology, Newberry and colleagues demonstrated that activation of hen egg lysozyme specific CD4+ T cells, by feeding T cell receptor transgenic mice the lysozyme antigen together with a COX-2 inhibitor, was sufficient to cause alterations almost indistinguishable from CD.14

T cell epitopes in gluten

Identification of gluten peptides that are recognised by DQ2 and DQ8 restricted T cells from coeliac lesions provides the first rationale basis for the identification of toxic epitopes in CD. Although simple in principle, this has proved to be a demanding task. Biochemical studies revealed that DQ2 preferentially bound peptides containing negatively charged amino acids, which are residues rarely found in gluten. This puzzling observation was explained by Molberget al who demonstrated that the enzyme tissue transglutaminase (tTG) could selectively convert glutamine residues within gluten to glutamic acid, and that this enhanced gluten peptide binding to DQ2 and facilitated their recognition by intestinal T cells.15 van de Wal et aldemonstrated the same phenomenon but this time for a DQ8 restricted epitope.16 Intriguingly, tTG is also the major antigen targeted by IgA and IgG autoantibodies in CD,17 which suggests there is a link between T cell recognition of modified gliadin and antibody production to tTG.18

Initial studies using purified natural gliadin preparations indicated that there were several distinct gluten peptides recognised by gluten specific T cells.19 Five of these epitopes have now been identified. Three epitopes were identified by mass spectrometry analysis of natural gluten fragments: one DQ2 restricted γ gliadin epitope (DQ2-γ-I),20 one DQ8 restricted α gliadin epitope (DQ8-α-I),21 and one DQ8 restricted HMW glutenin epitope (DQ8-HMW-I)22 (table 1). Identifying epitopes from purified natural gliadin is technically demanding and dependent on the limited availability of gluten sequences. This led us to pursue an alternative strategy of identifying DQ2 restricted T cell epitopes using a panel of recombinant gliadins that were cloned and expressed inEscherichia coli. 23Interestingly, only five of 11 different recombinant α gliadins appeared to be stimulatory for intestinal T cells and this reactivity could be accounted for by just two epitopes that share a seven residue sequence (DQ2-α-I and DQ2-α-II).24 These two epitopes are clearly major epitopes, as all adult patients appear to have strong responses to at least one (over 30 tested to date). In contrast, polyclonal T cell lines that respond to the DQ2-γ-I and DQ8-HMW-I epitopes are relatively uncommon,10 22 while little is known about the frequency of responders to the DQ8-α-I epitope.

Table 1

Characteristics of the five epitopes identified

The DQ2-α-I, DQ2-α-II, and DQ2-γ-I epitopes are not recognised by T cells unless a glutamine residue is substituted with glutamic acid. The non-deamidated DQ8-α-I sequence can stimulate T cells but only poorly compared with the optimal deamidated peptide. Dependence on deamidation for efficient T cell recognition of the characterised gluten peptides compares well with an observed preferential recognition of deamidated gluten by polyclonal gluten specific T cell lines (Molberg et al, submitted). Further evidence for the involvement of tTG comes from the observation that it selectively deamidates glutamine residues that are critical for T cell recognition; in fact, deamidation of residues not targeted by tTG often inhibits T cell recognition.16 25 Despite the strong preference for recognition of deamidated gluten peptides, it is also clear that there is occasional recognition of non-deamidated peptides. This is exemplified by the DQ8-HMW-I epitope, recognition of which is inhibited by treatment with tTG.22 Koninget al have questioned whether the epitopes identified to date are involved in disease initiation. They have raised the possibility that non-deamidated epitopes are the first to be recognised in disease and that tTG is then involved in amplifying the T cell response by creating better binding epitopes.26 They have proposed that analysis of T cells isolated from paediatric patients rather than adults would be more appropriate to address this issue. This work will surely give interesting results. However, it should be kept in mind that paediatric biopsies will still be taken some months after initiation of disease and, in immunological terms, this is likely to represent a chronic stage of the T cell response to gluten. Moreover, the preferential binding of DQ2 and DQ8 to peptides with negatively charged residues provides a link between deamidation of gluten and the HLA association in CD. If recognition of non-deamidated peptides is important for initiating disease, the association with DQ2 and DQ8 would be difficult to explain.

Gluten specific T cell responses in gut and peripheral blood

Studies using T cells extracted from intestinal biopsies have confirmed the importance of cell populations taken from the disease affected organ, and demonstrated how, despite its convenience, use of cells from peripheral blood can be misleading. One of the most striking observations on the intestinal T cell response to gluten was the bias in DQ2 or DQ8 restriction,12 a bias that is not reflected when peripheral blood is used to generate the T cell lines.27 Moreover, in contrast with intestinal T cells, peripheral blood derived T cell lines and clones generated from patients on a gluten free diet failed to show enhanced responses to tTG treated gluten.15 It appears that the repertoire of epitopes recognised by gluten specific T cell lines and clones differ depending on whether the T cells are taken from peripheral blood or intestinal biopsies. In an early study, Gjertsen et al used a DQ2 restricted T cell clone derived from peripheral blood to identify residues 31–47 of A-gliadin as one epitope.28 More recently, Godkin et al found that a peptide corresponding to residues 150–164 of A-gliadin was positive in an interferon γ ELISPOT assay for peripheral blood cells from CD patients but not from DQ2+controls.29 However, screening of intestinal T cell lines failed to demonstrate recognition of these peptides (see Arentz-Hansen and colleagues24 and unpublished results) which raises doubts as to their relevance in CD. Conversely, we were unable to detect responses to the immunodominant DQ2-α-I and DQ2-α-II intestinal T cell epitopes when gluten specific lines were made from peripheral blood taken from coeliac patients on a gluten free diet (Arentz-Hansen et al, unpublished results). It has also become apparent that the T cell repertoire present within the periphery is different depending on whether or not CD patients are consuming gluten. We have generated a peripheral blood derived gluten specific T cell line from an untreated patient. Notably, unlike similar lines from patients on gluten free diets, this line responded significantly better to gluten preparations that had been treated with tTG (Arentz-Hansen et al, unpublished observation). More convincingly, Anderson et al analysed peripheral blood responses using ELISPOT and confirmed the identity of the DQ2-α-I peptide.30Interestingly however, this response was not detectable unless patients, who had been on a gluten free diet, were first challenged with bread. Surprisingly, the response to non-deamidated A-gliadin peptide (150–164) that this group recently identified also using ELISPOT29 was not found in this study, regardless of whether or not patients were fed bread.

It is notable that the T cell epitopes identified do not correspond to the toxic sequences identified using other in vitro assays. Indeed, by culturing atrophic intestinal biopsies, de Ritis et al classified peptides purified from cynagen bromide and chymotrypsin cleaved A-gliadin as toxic and non-toxic.31We have been unable to demonstrate intestinal T cell responses to peptides corresponding to those they have described as toxic, and more importantly a peptide corresponding to the immunodominant DQ2-α-I epitope was classified as non-toxic in their system. This raises the question of which effect of gluten is detected in the atrophic intestinal biopsy culture assay, as it appears not to be associated with T cell recognition per se. There is evidence from the in vitro organ culture system that gluten may exert a rapid T cell independent effect.32 This effect could possibly facilitate a subsequent T cell response to gluten.

Future prospects

Identification of toxic epitopes raises several exciting possibilities that may be of practical benefit to patients. Based on our current knowledge, coeliac patients should be advised to avoid consuming foods containing cereals that are known to contain these epitopes. For example, the DQ2-α-I sequence is found in an α gliadin gene from Spelt wheat,33 and the DQ2-γ-I sequence is present within γ gliadin genes isolated from Durum wheat. The potential of this approach will be enhanced as more cereal genes are sequenced and as further T cell epitopes are identified. Longer term goals could be the development of therapeutic strategies for modulating the T cell response to deamidated gluten or the development of non-toxic wheat strains. However, these goals may be difficult to achieve for two reasons. Firstly, the response to different gluten epitopes is complex; while some epitopes are clearly dominant, most patients have responses to a number of epitopes. Secondly, bread wheat (Triticum aestivum) is hexaploid in nature which makes breeding out “toxic” alleles complicated. This will be even more difficult if genes located at several different gliadin/glutenin loci around the wheat genome encode for the T cell epitopes. Although some of the objectives mentioned here may remain wishful thinking, it is clear that characterisation of the T cell response to gluten has begun to provide a conceptual framework that explains the pathogenesis of CD.


Work in the authors' laboratory is supported by grants from the Research Council of Norway and the European Commission (project BMH4-CT98–3087).


8th United European Gastroenterology Week

The UEGW abstract book (Gut 2000;47(suppl III)) has again been produced as a CD-ROM and can be found attached to the inside back cover of this issue.

View Abstract


  • Leading articles express the views of the author and not those of the editor and editorial board.

  • Abbreviations used in this paper:
    coeliac disease
    low molecular weight
    high molecular weight
    tissue transglutaminase

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.