Introduction

Celiac disease—also known as celiac sprue and GLUTEN-sensitive enteropathy—is a prevalent (∼1:100) food hypersensitivity disorder caused by an inflammatory response to wheat gluten and similar proteins of barley and rye.1 The resulting intestinal inflammation often causes symptoms related to malabsorption, but in many patients extra-intestinal symptoms dominate, and in others the disease is clinically silent. Genes encoding HLA-DQ2 and HLA-DQ8 molecules are the single most important predisposing genetic factor; however, although these polymorphisms are necessary, they are not sufficient for disease development. HLA-DQ2 and HLA-DQ8 predispose to disease development by preferential presentation to mucosal CD4+ T cells of proline-rich gluten peptides that have undergone DEAMIDATION by the enzyme TISSUE TRANSGLUTAMINASE (TRANSGLUTAMINASE 2; TG2). Fewer details are known about the effector mechanisms that lead to the development of the typical celiac lesion—villous atrophy, crypt hyperplasia and infiltration of inflammatory cells (Figure 1)—but, once activated, gluten-reactive CD4+ T cells produce cytokines and are likely to control the inflammatory reactions that produce the celiac lesion. This notion is based on the nature of the HLA association and the unique presentation of gluten antigens to T cells by HLA-DQ2 or HLA-DQ8 in the intestine. Recent advances have improved our understanding of the molecular basis for this disorder,2 and new targets for rational therapy have been identified. This paper reviews concepts for new treatments and their current status.

Figure 1: The small-intestinal lesion in patients with celiac disease.
figure 1

Factors that contribute to the development of celiac disease and that can be targeted for new therapies are depicted. Proline-rich fragments of gluten that are resistant to processing by luminal and brush-border enzymes survive digestion5 and can be transported across the mucosal epithelium as polypeptides. CD4+ T cells in the lamina propria recognize predominantly deamidated gluten peptides37 in the context of HLA-DQ2 or HLA-DQ8 molecules on the cell surface of antigen-presenting cells (APCs).38 The deamidation of gluten peptides is mediated by tissue transglutaminase (TG2).39,40,41 The gluten-reactive CD4+ T cells produce interferon (IFN)-γ on activation.27 IFN-γ is also produced by T cells in the epithelium.42 Interleukin (IL)-15, produced by either mononuclear cells in the lamina propria or by enterocytes,30,31 stimulates T cells to migrate to the epithelium and facilitate killing of enterocytes by upregulated expression of MIC by enterocytes and NKG2D by intraepithelial T cells.29,32 IL-15 production is stimulated by gluten.28,29 Gluten can also induce production of the intestinal peptide zonulin, which acts on tight junctions and increases epithelial permeability.43 Adapted with permission from2 ©(2002) Macmillan Publishers Ltd.

Full size image

Current treatment and the need for alternatives

The current treatment for celiac disease is lifelong adherence to a strict gluten-exclusion diet. Gluten is, however, a common (and in many countries unlabeled) ingredient in the human diet, presenting a big challenge for celiac disease patients. Bona fide gluten-free products are not widely available and are usually more expensive than their gluten-containing counterparts. Unsurprisingly, dietary compliance is, at best, imperfect in a large fraction of patients, especially adolescents and adults. There is therefore an urgent need to develop safe and effective therapeutic alternatives.

The quality of life of celiac disease patients would improve if there was a treatment that would allow some gluten to be consumed over a short period of time, for instance during social events or travel. Documentation of the long-term safety for such an agent would probably be less necessary because the indication would be for intermittent treatment. This type of therapy could then pave the way for long-term or permanent treatment.

One goal for a new therapeutic agent would be to enhance the return of full intestinal function in patients who show incomplete recovery in response to a gluten-free diet. This agent might also allow moderate quantities (1–5 g/day) of gluten to be tolerated. Although this is less gluten than is consumed as part of a typical Western diet (∼20 g/day), it could improve quality of life by protecting patients from most forms of 'hidden gluten'. Finally, there is a particular need for treatment alternatives for refractory sprue, which, although rare, is currently treated only with harsh immunosuppressive drugs.

Potential new therapies

Enzyme therapy

Prolyl endopeptidases (PEPs) are endoproteolytic enzymes. In contrast to human gastrointestinal proteases, PEPs can readily cleave proline-rich immunostimulatory gluten peptides.3 The potential to use PEPs as a treatment for celiac disease is appealing because their specificity complements gastrointestinal proteolytic processes. Not only does every PEP-catalyzed cleavage generate one new amino and carboxyl terminus, but it also truncates the long-peptide end products of gastric and duodenal gluten metabolism. Both of these outcomes provide residual smaller peptides that are suitable substrates for the intestinal brush-border aminopeptidases and carboxypeptidases. Microorganisms and plants express various PEPs. There is also a human PEP, but this is expressed only in the cytosol,4 and is therefore unlikely to have a physiological role in the degradation of gluten peptides.

It has been proposed that oral administration of a therapeutic dose of a suitably formulated PEP might counter the toxic effects of moderate quantities of ingested gluten.3 This hypothesis is supported by extensive in vitro, in vivo (in rats) and ex vivo (using biopsy-derived T cells) studies on synthetic gluten peptides, recombinant gliadin molecules and whole gluten as obtained in a grocery store.3,5,6,7,8 In each case, dose-dependent breakdown of immunotoxic gluten peptides by a Flavobacterium meningosepticum PEP was observed. This proteolytic action was also reflected in a concomitant reduction of immunogenicity, as judged by reduced stimulation of most polyclonal T-cell lines from patients with celiac disease.8 Together, these results set the stage for controlled studies in patients with celiac disease. The primary near-term clinical challenge is to identify a threshold dose of gluten that is well tolerated by most patients when consumed in conjunction with a specified PEP dose.

Until recently, most studies directed towards enzyme therapy have utilized the F. meningosepticum PEP. This enzyme meets most criteria for a therapeutically effective PEP, with the possible exception of sluggish kinetics against a few T-cell EPITOPES in gluten. Its primary limitation lies in the high production costs, which are presumably due to its relatively poor level of heterologous expression. A homologous PEP from Myxococcus xanthus seems to be comparable to the F. meningosepticum enzyme with respect to gluten detoxification, but can be expressed at much higher levels in Escherichia coli.7 In addition, Lactobacillus helveticus has a zinc-dependent PEP that can also cleave long substrates with relatively broad subsite specificity.9 Further studies should help to clarify the relative pros and cons of the available PEPs and show which should be used for human clinical trials.

A pharmacologically useful oral PEP formulation might also include one or more complementary enzymes. For example, the cysteine endoprotease EPB derived from barley is a glutamine-specific enzyme that rapidly hydrolyzes intact gliadin polypeptides (Box 1) into short peptides under acidic conditions.10 Although it is unable to cleave some of the more proline-rich inflammatory peptides, such as the 33-mer from α-gliadin, it could enhance the efficacy of a duodenally active PEP by simplifying gluten structure and texture while food is still in the stomach.

The long-term safety of a candidate therapeutic enzyme (or enzyme cocktail) designed to act in the upper small intestine must also be evaluated, initially in animals and subsequently also in humans. Primary safety concerns must ensure that the exogenous enzyme is non-allergenic, and is not assimilated intact into the bloodstream in appreciable quantities. Other potential risks include structural damage to intestinal mucosa and the consequent loss of nutrient absorptive capacity or alterations in regulation of gut hormones. In this regard, it must be noted that some proteases of pharmacological interest (e.g. barley EPB) have been components of the human diet for a long time, and are therefore unlikely to present a health risk.

Proteases can also detoxify gluten-containing products before they are ingested. This is particularly relevant for products that have minor, but significant, gluten content. Analogously, proteases of certain lactobacilli present in sourdough are able to proteolyze proline-rich gluten peptides,11 and a challenge experiment indicates that celiac disease patients, at least in the short term, tolerate a carefully prepared sourdough bread.12

Tissue transglutaminase inhibitors

The importance of gluten-peptide deamidation for the recognition of T-cell epitopes implicates TG2 inhibitors as a candidate therapy for celiac disease. Indeed, ex vivo gluten challenge of patient-derived small-intestine biopsies showed that the proliferative capacity of gluten-responsive T cells could be blocked by co-treatment with cystamine, a TG2 inhibitor.13

The efficacy and side effects of TG2 inhibitors as a treatment of celiac disease are unknown. A few gluten T-cell epitopes are recognized without being modified by TG2.14,15 Activation of such cells might be sufficient to drive the gut inflammation. Although TG2 knockout mice have no overt spontaneous abnormalities,16,17 they develop splenomegaly, autoantibodies and immune complex glomerulonephritis by systemic triggering of apoptosis.18 Thus, the optimal TG2 inhibitor should have activity limited to the gut mucosa. Local side effects related to extracellular matrix formation and wound healing can be envisaged. Of course, a prerequisite for investigating safety and efficacy is the availability of one or more selective TG2 inhibitors.

Although several TG2 inhibitors have been described in the literature, many of these compounds (e.g. monodansyl cadaverine) contain primary amines in addition to potential inhibitory motifs, and it remains unclear whether the observed effects are due to excess competing amines or blockage of TG2 substrate turnover. A few studies have utilized a suicide inhibitor, L-682777, which inhibits human TG2, however, L-682777 is a specific inhibitor of Factor XIIIa, and is therefore unsuitable for evaluating TG2 biology in animals and humans.

More recently, mechanism-based active-site inhibitors of guinea pig and human TG2 have been reported. Synthetic compounds containing halo-dihydroisoxazole are particularly promising. One such compound, KCC009, inhibits intestinal TG2 when dosed orally, and is well tolerated by rodents at pharmacologically effective doses.19 It also has a short serum half-life, which limits the exposure of other organs to this compound. Such an inhibitor could be used to validate TG2 as an appropriate therapeutic target for treating celiac disease.

Blocking of HLA-DQ peptide presentation

The crucial role of the HLA in celiac disease development makes it an obvious target for therapeutic intervention. Blocking the binding sites of HLA-DQ2 and, to a lesser extent, HLA-DQ8, would prevent the presentation of disease-inducing gluten peptides.

The concept of HLA blockade is not new and was developed without much success for the treatment of type 1 diabetes and rheumatoid arthritis. The lack of success was due partly to difficulties in obtaining effective drug delivery. This should be less of a problem in celiac disease because the blocking compound can be administered before or in parallel with the antigen (i.e. gluten). In our opinion, HLA blockade as a therapy for celiac disease should be pursued.

Side effects, such as immunosuppression, are unlikely because many healthy individuals are homozygous for HLA alleles. The recently solved X-ray crystal structure of HLA-DQ2 complexed with a deamidated gluten peptide provides important information for the development of an HLA-DQ2-blocking compound.20 A blocking compound should prevent presentation of gluten peptides to T cells and be unrecognizable by any T cell, to avoid hypersensitivity responses to the blocking compound itself. This might be achieved by making the compound either small enough to avoid forming interactions with the T-cell receptor, or so big that no T-cell receptor can dock onto HLA-DQ2 with the bound blocker.

Silencing of gluten-reactive T cells

Celiac disease is exceptional in that the disease can be put in remission and induced in a controlled fashion. This could be utilized for therapeutic purposes. Conceivably, gluten-reactive T cells could be eliminated or made unresponsive by oral gluten challenge concomitant with the administration of agents that alter the outcome of the T-cell activation. Antibodies to CD321 and CD154 (CD40L),22 for example, can induce T-cell silencing, but they produce unwanted side effects such as toxic cytokine syndrome (anti-CD3) and thromboembolic events (anti-CD154). Alternatively, gluten-reactive T cells might be silenced by soluble dimers of HLA–peptide complexes, because such dimers induce the apoptosis of antigen-specific T cells as a result of inappropriate stimulation.23 This approach is complicated by an increasing number of characterized gluten epitopes. Induction of tolerance by intranasal administration of gluten or gluten T-cell epitopes has also been suggested as a possible treatment modality.24 Moreover, tolerance can be achieved by targeting gluten epitopes to dendritic cells that induce T-cell tolerance, but this approach is further complicated by the presence of multiple T-cell epitopes. In addition, it is unclear which marker should be used to target tolerogenic dentritic cells in the intestinal mucosa.

Cytokine therapy

Various cytokine therapies are being developed for the treatment of chronic inflammatory diseases. Celiac disease is not a pioneering disorder for the testing of such therapeutics, owing to the low acceptance of side effects by patients and the presumed lower consumer market. Nevertheless, this approach might yield a compound that is well tolerated with a mode of action relevant to celiac disease. There are already some compounds that merit attention.

Interleukin (IL)-10 counterbalances the TH1 IMMUNE RESPONSE that is typical of celiac disease. It can suppress gluten-dependent T-cell activation in ex-vivo cultured celiac intestinal mucosa,25 but, in a pilot study of refractory celiac disease, recombinant IL-10 had little effect.26 IL-10 has also been a candidate therapy for Crohn's disease but, notably, a Phase II/III trial with recombinant IL-10 was discontinued owing to a lack of effect (Table 1).

Table 1 Potential therapies for celiac disease.
Full size table

Interferon (IFN)-γ is the dominant cytokine produced by gluten-reactive T cells,27 and antibodies neutralizing IFN-γ should have a good chance of curtailing the inflammatory effects of the anti-gluten T-cell response. Antibodies of this kind are being tested in Phase I/II trials for Crohn's disease (Table 1). Depending on the results of these trials, anti-IFN-γ agents might become candidates for testing in celiac disease.

IL-15 is thought to be the central mediator of the innate effects of gluten in celiac disease,28,29,30,31 making IL-15 neutralizing agents therapeutic candidates. Two such compounds are in clinical trials; a humanized anti-IL-15 antibody, HuMax-IL-15, is in Phase II trials for rheumatoid arthritis and possibly other inflammatory conditions, and an IL-15/Fc chimeric protein, CRB-15, is in preclinical testing (Table 1).29 Earlier studies suggest that HuMax-IL-15 has acceptable side effects, and might therefore be a candidate for testing in celiac disease.

Selective adhesion molecule inhibition

Another novel class of therapeutic agents under development for the treatment of chronic inflammatory diseases acts by selective inhibition of leukocyte adhesion. This inhibition prevents leukocytes from migrating into inflamed tissues. For example, a humanized antibody against INTEGRIN-α4, natalizumab, is being used for the treatment of multiple sclerosis, and is under evaluation for inflammatory bowel disease (IBD) (Table 1). A small-molecule integrin-α4 antagonist, T0047, has also been developed and is undergoing clinical testing. Preliminary reports indicate that natalizumab has beneficial effects in IBD with moderate side effects.

Another humanized monoclonal antibody, MLN02, which targets the adhesion molecule integrin-α4β7 that is expressed by gut T cells, is being tested in Phase II clinical trials for the treatment of IBD. This agent is intriguing with respect to celiac disease, because it aims to prevent migration of T cells to the lamina propria. Conceivably, this agent could interfere with the action of the HLA-DQ-restricted T cells in the lamina-propria-recognizing gluten peptides. Possible side effects include increased susceptibility to gastrointestinal infections and a potential for disturbances in oral tolerance to food proteins.

Other therapeutic options

The recent finding that NKG2D is important for the killing of enterocytes by intraepithelial lymphocytes in celiac disease led to the suggestion that NKG2D antagonists might be used for therapy.29,32 This concept has gained support from a mouse model of type 1 diabetes. Treatment with a nondepleting anti-NKG2D monoclonal antibody during the prediabetic stage in NOD MICE prevented disease completely by impairing the expansion and function of autoreactive CD8+ T cells.33 ZONULIN antagonists have also been suggested as therapy for celiac disease aiming to prevent gluten-induced disruption of the epithelial barrier.34

Finally, the development of grains that have low or no content of immunotoxic sequences, but with reasonable baking quality, should be mentioned. Such grains can potentially be developed by selective breeding of ancient wheat varieties,35 by transgenic technology involving mutation of sequences giving rise to immunostimulatory sequences,36 by small interfering RNA (siRNA) technology, or by incorporation of nontoxic gluten genes into harmless organisms such as rice. Although these grains would be technically challenging to engineer, and there is a possibility that cross-pollination with gluten-containing grains might lead to reintroduction of immunotoxic sequences, the availability of such grains could give patients with celiac disease a nutritionally better diet. If used for the nutritional needs of the general population, this could eventually prevent celiac disease.

The way forward

Criteria for a clinical candidate

Notwithstanding promising progress, an animal model for celiac disease remains elusive. In its absence, the evaluation of preclinical efficacy of an experimental therapeutic agent for celiac disease is limited to in vitro studies and, in some cases, pharmacodynamic measurements in healthy humans. In addition, regardless of its mode of action, a clinical candidate for treating celiac disease must satisfy certain safety criteria that can be tested in animals. These include good oral and systemic tolerance of the agent, absence of antigenicity, and localization of the drug to the gut. In particular, structural integrity should be confirmed via post-mortem histological evaluations of the entire gastrointestinal tract of treated animals.

As objective endpoints for early clinical trials in patients with celiac disease can probably be achieved in less than 1 month (see below), a 3–6 month initial toxicological evaluation of an experimental therapeutic would be appropriate. Downstream clinical studies would require long-term safety testing in animals; the criteria for such studies would be different when drugs are intended for intermittent therapy to when intended for regular, lifelong use.

Monitoring the treatment effect

The strategies used for early clinical studies (phase I and II) with an experimental therapeutic agent for celiac disease are of particular interest. If well designed, these studies can yield pivotal efficacy data, while concomitantly identifying the subpopulation of patients that stand to benefit the most from the new drug. At the same time, because health risks are usually at their highest when studying a new chemical entity, such studies must be conducted with particular caution.

No efforts have been made thus far to evaluate the efficacy of an investigational drug for celiac disease, so the question of how best to perform early proof-of-concept clinical trials remains unaddressed. Clearly the most effective format for such clinical trials would be a crossover mode in which patients with celiac disease are exposed alternately to gluten plus placebo and gluten plus drug, for 2–4 weeks each. The variables in such a format would be gluten dose, duration of gluten exposure, and evaluation methods at the beginning and end of each stage of the trial.

Ordinarily, gastroenterologists use serum antibody levels and endoscopic and intestinal histological analyses for evaluating patients with celiac disease. Although their utility for initial diagnosis as well as follow-up care of patients is well documented, endoscopic and histological methods have serious limitations in the context of evaluating investigational drugs.

The conversion from normal to abnormal antibody titers requires exposure to high levels of gluten for several months, which might produce disturbing symptoms and be a health risk. By contrast, whereas endoscopic and histological procedures are more sensitive and specific, repeated endoscopic evaluations in the context of a crossover clinical trial involve some risk, are expensive, and require experienced evaluators to extract relatively subjective data. An alternative non-invasive approach is the use of functional tests of gut malfunction (e.g. xylose absorption tests, fecal fat analysis, lactulose/mannitol permeability tests), which are sensitive but non-specific tests for celiac disease. Finally, the radiotelemetry video capsule might also provide useful information for assessment of overall small-intestinal structure in the context of a clinical trial.

For some therapeutic approaches, such as enzyme therapy, it might be possible to assess pharmacodynamic efficacy in healthy adult volunteers, because the incremental benefit of an oral enzyme formulation on gastric and/or duodenal proteolysis of gluten can be safely monitored by intestinal intubation with small-diameter plastic tubes. Similarly, the efficacy of certain other therapeutic strategies, such as cytokine or integrin therapy, is likely to be validated in patients with more serious inflammatory bowel disorders before initiating clinical trials for their use as a treatment for celiac disease.

Conclusions

The need for alternatives to the gluten-exclusion diet for the treatment of celiac disease has been made even more urgent by the increasing number of patients diagnosed with this disease. Fundamental studies have revealed several attractive targets for therapy, some of which are already showing promise in the context of other medical conditions. It will be interesting to see whether any of these will become reality in the coming years.

Review criteria

PubMed was searched in November 2004 using the terms “celiac disease”, “therapy”, “prolyl AND peptidase”, “transglutaminase”, “integrin”, “interferon” and “interleukin”. In addition, searches were done using SciFinder Scholar, Google, Google Scholar and the Investigational Drugs Database. Corporate websites of companies known to be developing relevant anti-inflammatory drugs were also searched. Citations were chosen based on their direct relevance to statements in the text. Some information cited in the text has been presented only at meetings and symposia; in such situations the meeting abstract is cited.