Gluten is the structural protein component of the grains wheat, rye and barley, which are the basis for a variety of flour- and wheat-derived food products consumed throughout the world. Possibly, the introduction of gluten-containing grains, which occurred about 10,000 years ago with the advent of agriculture, represented a "mistake of evolution" that created the conditions for human diseases related to gluten exposure, the best known of which are mediated by the adaptive immune system: wheat allergy and celiac disease (CD). In both conditions, the reaction to gluten is mediated by T-cell activation in the gastrointestinal mucosa. However, in wheat allergy, it is the cross-linking of immunoglobulin E (IgE) by repeat sequences in gluten peptides (for example, Ser-Gln-Gln-Gln-(Gln-)Pro-Pro-Phe) that triggers the release of chemical mediators, such as histamine, from basophils and mast cells [1]. In contrast, CD, which affects approximately 1% of the general population, is an autoimmune disorder, as heralded by the appreciation of specific serologic markers, most notably serum antitissue transglutaminase (tTG) autoantibodies, by the autoimmune enteropathy that characterizes this condition and by autoimmune comorbidities.

Besides CD and wheat allergy, there are cases of gluten reactions in which neither allergic nor autoimmune mechanisms are involved. These are generally defined as gluten sensitivity (GS) [2–5]. Some individuals who experience distress when eating gluten-containing products and show improvement when following a gluten-free diet may have GS instead of CD. GS patients are unable to tolerate gluten and develop an adverse reaction when eating gluten that usually, and differently from CD, does not lead to small intestinal damage. While the gastrointestinal symptoms in GS may resemble those associated with CD, the overall clinical picture is generally less severe and is not accompanied by the concurrence of tTG autoantibodies or autoimmune disease. Typically, the diagnosis is made by exclusion, and an elimination diet and "open challenge" (that is, the monitored reintroduction of gluten-containing foods) are most often used to evaluate whether the patient's health improves with the elimination or reduction of gluten from the diet.

A number of in vitro studies have confirmed the cytotoxicity of gluten's main antigen, gliadin. Gliadin has agglutinating activity, reduces F-actin content, inhibits cell growth, induces apoptosis, alters redox equilibrium and causes a rearrangement of the cytoskeleton through the zonulin pathway and the loss of tight junction (TJ) competence in the gastrointestinal mucosa [6–9]. The diversity of gluten-induced conditions is in line with the notion that the immune system reacts to and deals with the triggering environmental factor, gliadin, in distinct ways. In the present study, we sought to gain initial knowledge on intestinal barrier function and the immune response to gluten in patients with GS. Specifically, we were interested in understanding to what extent innate and adaptive immune pathways are activated in GS compared to CD. To achieve these aims, we looked at the mucosal expression of genes associated with intestinal barrier function and immune parameters known or implied to be aberrantly regulated in CD. The results provide the first documentation to date of genes and pathways possibly involved in the pathogenesis of GS, and, at the same time, contribute to improving our understanding of the processes leading to CD and other autoimmune phenomena.

CD, an autoimmune enteropathy, results from an inappropriate T-cell-mediated adaptive immune response against ingested gliadin. In the past few years, though, it has become apparent that "classic" CD represents the tip of the iceberg of an overall disease burden [4, 29]. An emerging problem is the clinical characterization of a group of gluten-reactive patients, accounting for roughly 10% of the general population, presenting with symptoms similar to CD but with negative CD serology and histopathology. As in CD, these patients, here and elsewhere referred to as GS [15], experience distress when eating gluten-containing products and show improvement when following a gluten-free diet. Differently from CD, though, in GS the adverse reactions that develop while eating gluten are not followed by the appearance of autoantibodies and by persisting damage to the small intestine. Symptoms in GS may resemble some of the gastrointestinal symptoms that are associated with CD or wheat allergy, but objective diagnostic tests for this condition are currently missing. Therefore, a diagnosis of GS is commonly made by exclusion.

In itself, the absence of autoantibodies and intestinal lesions does not rule out the intrinsic toxicity of gluten, whose intake, even in non-CD individuals, has been associated with damage to other tissues, organs and systems besides the intestine [6, 30, 31]. In the present study, we have sought to identify functional, morphologic and immunologic parameters to help differentiate GS from CD and to preliminarily understand its pathophysiology. We report here for the first time evidence of differential intestinal mucosal responses to gluten in these two conditions.

We have shown that a normal to mild histology in GS is paralleled by a conserved barrier function. Indeed, small intestinal permeability, when tested with a LA/MA double sugar probe, was significantly lower in GS than in CD patients or even DC. Increased intestinal permeability is thought to be an early biological change that precedes the onset of several autoimmune diseases [32–34]. Loss of intestinal barrier function brings with it a continuous aberrant passage of antigens across the intestinal epithelium. This may cause a switch from tolerance to immunity, hence representing an increased risk for autoimmune and allergic diseases in individuals whose other genetic determinants, MHC and non-MHC, give rise to inappropriate antigen processing and presentation. In the intestinal epithelium, paracellular permeability is regulated by intercellular TJ proteins. As recently shown, CLDNs are integral TJ components that are critical for maintaining cell-cell adhesion in epithelial monolayers [35–37]. The overall balance of CLDN species expressed in a particular cell type help to define the characteristics of its TJ. For instance, CLDN1 and CLDN4 are postulated to decrease, whereas CLDN2 is postulated to increase TJ-dependent permeability [35]. In line with this notion, and with the appreciation of reduced small intestinal permeability in GS patients, we have shown here that the GS mucosa expresses significantly higher levels of transcripts for CLDN4 relative to CD or DC. In contrast, other CLDN genes and other genes associated with TJ function measured in this study did not appear to be expressed differently in the GS or CD mucosa compared to controls. Together, these findings suggest that the distinct clinical and serological features between GS and CD patients are associated with marked differences in mucosal barrier function and with apparent differences in the expression of CLDN4, which encodes for a critical TJ component [38]. Further studies are required to compare the distribution and assembly of this and other TJ proteins in these conditions.

Patients with GS do not present significant autoimmune or allergic comorbidities, and, as we also have shown here, the serology for common autoantibodies, including anti-tTG IgA, is negative. Interestingly, AGA IgA and IgG were positive in almost 50% of cases. Similarly, higher than expected titers of AGAs, signs and symptoms associated with non-CD gluten sensitivity, have also been reported for schizophrenia [39] and autism spectrum disorders [40]. While in CD there is a strong genetic association with the class II MHC haplotype, with about 95% of patients carrying HLA-DQ2 and the remaining 5% carrying HLA-DQ8, we have shown that only about 50% of patients with GS carry HLA-DQ2 and/or HLA-DQ8, a percentage slightly higher than that in the general population. This suggests a reduced level of involvement of MHC-dependent adaptive immune responses in GS relative to CD. We have further shown that the GS mucosa contains increased numbers of CD3⁺ IELs, even though these numbers were significantly lower than those in active CD patients in the context of relatively conserved villous architecture, corresponding to the 0 and 1 stages of the Marsh classification. This is in line with a more limited involvement of the adaptive immune system in GS and may explain why this condition is not accompanied by significant autoimmune phenomena.

In CD, an adaptive response has been shown to be triggered by tTG-deamidated gluten peptides bound to DQ2 or DQ8. This involves the mucosal recruitment and activation of Th1 and Th17 clones and the production of Th1- and Th17-associated cytokines, namely, IFN-γ and IL-17A, which contribute to disrupting barrier function and initiating tissue damage [16, 18–21, 41]. In an earlier report, we showed that IL-17A transcripts are expressed at significantly higher levels in the small intestine mucosa of at least a subgroup of CD patients, but not in GS patients [15]. In this study, we have extended this finding and show that the Th1 signature cytokine, IFN-γ, also is expressed at significantly lower levels in the GS versus CD mucosa. Moreover, in CD but not in GS, we observed a significantly enhanced expression of IL-6, a pleiotropic cytokine that is known to promote the differentiation and function of Th17 cells, as well as a similar trend in the expression of IL-21, consistent with its established role in the pathophysiology of Th1 and Th17 cells.

These findings might indicate that GS is an inflammatory condition mostly supported by innate immune mechanisms. Among these, TLRs represent a family of evolutionarily conserved receptors able to detect microbial invasion via pattern recognition and mediate a rapid inflammatory response which may or may not progress into an antigen-dependent adaptive response. Different combinations of TLRs are expressed in hematopoietic cells and nonhematopoietic cells such as intestinal epithelial cells [42–44]. In this study, we have observed that small intestine expression of TLR2, and to a lesser extent TLR1 but not TLR4, is increased in GS patients. In the absence of markers of adaptive immunity, as we have seen, this suggests a prevalent role of the innate immune system in the pathogenesis of GS.

Taken together, these findings support the idea that the prevalent involvement of innate versus adaptive immune pathways may help explain the clinical and serological differences in GS versus CD patients. Reduced function of Treg cells, and specifically of "adaptive" Treg cells, has been proposed to account for the loss of immune homeostasis and the development of autoimmune responses in CD and related conditions [26]. It might be inferred, then, that Treg could efficiently prevent progression to this response in GS patients. A significantly reduced mucosal expression of the distinctive Treg marker, FOXP3, as appreciated in GS patients in this study, is therefore surprising and counterintuitive in the light of these considerations. However, at least as surprisingly, in several studies FOXP3 and other Treg-expressed molecules, such as TGFB1, have been found to be upregulated in the peripheral blood and intestinal mucosa of patients with CD and related conditions, for example, type 1 diabetes [28, 45]. Analogous findings have been reported in other conditions associated with the extensive involvement of adaptive immunity, such as allergy and asthma, leading to speculation that a compensatory expansion and/or mobilization of Treg might take place concomitantly with the buildup of an adaptive effector response [46]. Paradoxically, then, if this assumption is true, a reduced expression of Treg markers in GS might be interpreted in the context of a generally reduced activation of adaptive immunity relative to CD. While more studies obviously are needed to elucidate this issue, a better understanding of Treg function in CD and related conditions will help characterize the possible pathogenetic role of reduced Treg activation and/or recruitment in GS.

The results of this study suggest that CD and GS are distinct clinical entities caused by different intestinal mucosal responses to gluten. CD results from a complex, and as yet undetermined, interplay of increased intestinal permeability, mucosal damage, environmental factors in addition to gluten, and genetic predisposition, which involves both MHC and non-MHC genes. The typical intestinal lesions in CD are thought to be mediated by both innate and adaptive immune effector pathways. Our findings suggest that, in a different way, GS is associated with prevalent activation of an innate immune response. Although the mechanisms responsible for the loss of intestinal barrier function in CD have been delineated in part, the factors responsible for the loss of gluten tolerance and the development of autoimmunity in this condition are still incompletely understood. We believe that this study could contribute to the clinical characterization of GS as a condition associated with prevalent gluten-induced activation of innate immunity in the absence of detectable changes in mucosal barrier function, and that it provides additional clues to the definition of the complex gluten-induced changes in TJ regulation and immune processes underlying CD pathogenesis. Double-blind, placebo-controlled studies are necessary to further solidify the definition of GS patients and to search for specific biomarkers for a proper diagnosis.