|Home | About | Journals | Submit | Contact Us | Français|
Celiac disease is an immune-mediated enteropathy triggered by gliadin, a component of the grain protein gluten. Gliadin induces an MyD88-dependent zonulin release that leads to increased intestinal permeability, a postulated early element in the pathogenesis of celiac disease. We aimed to establish the molecular basis of gliadin interaction with intestinal mucosa leading to intestinal barrier impairment.
α-Gliadin affinity column was loaded with intestinal mucosal membrane lysates to identify the putative gliadin-binding moiety. In vitro experiments with chemokine receptor CXCR3 transfectants were performed to confirm binding of gliadin and/or 26 overlapping 20mer α-gliadin synthetic peptides to the receptor. CXCR3 protein and gene expression were studied in intestinal epithelial cell lines and human biopsy specimens. Gliadin-CXCR3 interaction was further analyzed by immunofluorescence microscopy, laser capture microscopy, real-time reverse-transcription polymerase chain reaction, and immunoprecipitation/Western blot analysis. Ex vivo experiments were performed using C57BL/6 wild-type and CXCR3−/− mouse small intestines to measure intestinal permeability and zonulin release.
Affinity column and colocalization experiments showed that gliadin binds to CXCR3 and that at least 2 α-gliadin 20mer synthetic peptides are involved in this binding. CXCR3 is expressed in mouse and human intestinal epithelia and lamina propria. Mucosal CXCR3 expression was elevated in active celiac disease but returned to baseline levels following implementation of a gluten-free diet. Gliadin induced physical association between CXCR3 and MyD88 in enterocytes. Gliadin increased zonulin release and intestinal permeability in wild-type but not CXCR3−/− mouse small intestine.
Gliadin binds to CXCR3 and leads to MyD88-dependent zonulin release and increased intestinal permeability.
Celiac disease (CD) is an autoimmune enteropathy triggered by ingestion of gluten-containing grains (eg, wheat, rye, and barley). The disease persists in the continued presence of gliadin, the toxic component of gluten.1 Other characteristics of CD include a highly specific autoantibody response against tissue transglutaminase2 and a strong association with specific major histocompatibility complex haplotypes. Greater than 90%–95% of CD patients carry the HLA-DQ2, with the remaining carrying the HLA-DQ8 haplotype; however, non-HLA genes have been implicated in the disease pathogenesis as well.3
Under physiologic conditions, access of gliadin to gutassociated lymphoid tissue is prevented by competent intercellular tight junctions (TJ) that limit passage of macromolecules (including gliadin peptides) across the intestinal epithelial barrier.4 In susceptible individuals, however, the interplay between the initiating stimulus (eg, gliadin) and intestinal cells triggers TJ disassembly. It has been hypothesized that this is an early biologic change that precedes the onset of gliadin-induced immune events that eventually lead to the pathology associated with CD.5
One protein that induces TJ disassembly and therefore is thought to be involved in the early phase of CD is zonulin.6 Increased and persistent production of this protein as determined by Western immunoblotting7 and enzyme-linked immunosorbent assay (ELISA)8 were observed in patients with active CD.6 Furthermore, ex vivo studies, using intestinal biopsy specimens in the microsnapwell system, showed that intestinal biopsy specimens of CD patients mounted a more pronounced response to gliadin when compared with nonceliac controls, including an increased and persistent release of zonulin and a significant increase in intestinal permeability.8 It is noteworthy that epithelial release of zonulin occurs after apical, but not basolateral, exposure to gliadin.9 The latter finding implies that gliadin interacts with an intestinal luminal receptor and prompted us to seek the identity of this moiety.
In this paper, we provide evidence that the chemokine receptor CXCR3 serves as the target receptor for gliadin. Our data demonstrate that, in the intestinal epithelium, CXCR3 colocalizes with gliadin and that this interaction coincides with recruitment of the adapter protein, MyD88, to the receptor. We also demonstrated that binding of gliadin to CXCR3 is crucial for the release of zonulin and subsequent increase of intestinal permeability because CXCR3-deficient mice failed to respond to gliadin challenge in terms of zonulin release and TJ disassembly.
Gliadin (crude wheat), pepsin, and trypsin were purchased from Sigma (St Louis, MO). Gliadin was pepsin/trypsin digested (PT-gliadin) as described previously10 with minor modifications.11 Recombinant α-gliadin was a gift from Dr D. Kasarda (USDA-ARS, Albany, CA). Recombinant interleukin (rIL)-1, monokine induced by interferon (IFN) γ (rMig/CXCL9), IFN-γ-inducible protein 10 (rIP-10/CXCL10), IFN-γ-inducible T-cell α-chemoattractant (rI-TAC/CXCL11), and tumor necrosis factor-α (rTNF-α) were purchased from R&D (Minneapolis, MN) and Calbiochem (San Diego, CA), respectively. Pertussis toxin or inactivated pertussis toxin were kindly provided by Dr N. Carbonetti (University of Maryland, Baltimore, MD).
For the preparation of the affinity column, α-gliadin was dissolved in 70% alcohol, mixed with Affi-Gel 15 Gel, and gently shaken for 4 hours at 4°C. The reaction was terminated by ethanolamine. Soluble total membrane preparations12 from rabbit small intestine were loaded on an Affi-gel 15-α-gliadin affinity column, incubated for 90 minutes at 25°C, washed with phosphate-buffered saline (PBS) containing 0.1% Triton X-100 (Sigma), and eluted with PBS containing 0.1% Triton X-100 with increasing NaCl concentrations. Fractions were collected and subjected to SDS-PAGE. The eluted proteins were characterized by MALDI mass spectroscopy fingerprint analysis (Protein and Nucleic Acid Biotechnology Facility; Stanford University, Palo Alto, CA).
HEK293T cells (2.5 × 106, passages 1–9) were plated in 10-mL culture Petri dishes in complete culture medium (Dulbecco’s modified Eagle medium [DMEM]; Cellgro, Manassas, VA) supplemented with 10% heat-inactivated fetal bovine serum (FBS), 50 U/mL penicillin/ 50 g/mL streptomycin, and 2 mmol/L L-glutamine (Gibco, Carlsbad, CA) and incubated overnight at 37°C in 5% CO2. Cells were transfected with either empty vector (pcDNA3.1; Invitrogen) or CXCR3 construct generated as previously described13 with minor modifications at a concentration of 500 ng/well using Superfect transfection reagent (Qiagen, Valencia, CA). After transfection, fresh complete culture medium was added to the dishes, and cells were incubated overnight at 37°C in 5% CO2.
Overlapping (every 10 amino acids), 20mer peptides were designed based on the amino acid sequence of α-gliadin and synthesized using solid phase synthesis, resulting in a 26 peptide library (see Supplementary data online at www.gastrojournal.org). Peptide synthesis was carried out using standard Fmoc chemistry on Rink resin. Peptides were isolated as tri-fluoro acetate salts at purity levels of greater than 80% by high-performance liquid chromatography.
Both assays were performed at Euroscreen S.A. according to the company protocols (www.euroscreen.com),14 using increasing concentrations of PT-gliadin (for more details, see Supplementary data online at www.gastrojournal.org). To establish the binding affinity of the synthetic gliadin peptides to CXCR3, FITC-labeled CXCR3-binding peptide 4026 was incubated with CXCR3-transfected HEK293T cells and binding kinetic evaluated by flow cytometry analysis (for detailed information, see Supplementary data online at www.gastrojournal.org).
HEK293T cells transiently transfected with either pcDNA empty vector or CXCR3 gene-containing vector were detached by gentle scraping, seeded in Lab-Tek II chamber slides (Nalge Nunc International, Rochester, NY) at a density of 50,000 cells/well and allowed to attach to the wells overnight at 37°C in 5% CO2. A separate small aliquot of detached cells was incubated with 5 µLof allophycocyanin-conjugated anti-human CXCR3 (clone 49801; R&D) or an isotype-matched control (clone 11711; R&D) mouse monoclonal antibody (mAb) and used for flow cytometry analysis to verify the expression of CXCR3 on transfectants. Experimental conditions and staining protocols are described in detail in the Supplementary data online (see Supplementary data online at www.gastrojournal.org).
Intestinal epithelial cell lines, IEC6 (rat, passage 36–46) and CaCo-2 (human, passage 30–40), were grown on Lab-Tek I chamber slides and stained for CXCR3 as described above. To localize CXCR3 expression in intestinal tissues, 4-µm sections were prepared, and laser capture microdissection (mouse tissue) or immunohistochemistry (human tissue) were performed as previously described.15,16 Human intestinal mucosa was obtained from non-CD patients who underwent a diagnostic upper endoscopy for dyspepsia (no duodenal damage) and CD patients at the moment of diagnosis (active disease, with a Marsh IIIa–c lesion) during diagnostic endoscopy. RNA extraction and real-time polymerase chain reaction (PCR) protocols are described in the Supplementary data online (see Supplementary data online at www.gastrojournal.org).
IEC6 cells were grown in culture flasks and plated in Petri dishes (1 × 106 cells/mL). Confluent cells were stimulated with PT-gliadin at doses ranging from 100 µg/mL to 1 mg/mL at different time points (15, 45, and 60 minutes). At the end of stimulation, IEC6 cells were lysed in lysis buffer containing a cocktail of protease inhibitors. Total protein content was measured using the Lowry method (Pierce, Rockford, IL). CXCR3 coimmunoprecipitation with the adaptor molecule MyD88 was performed according to the protocol described in the Supplementary data online (see Supplementary data online at www.gastrojournal.org).
Intestinal transepithelial electrical resistance (TEER) and changes in TEER in murine small intestine in response to gliadin exposure were measured using the microsnapwell system.17 Intestinal segments isolated from either CXCR3−/−18 (backcrossed >10 generations onto a C57BL/6 background) or C57BL/6 mice (Jackson Laboratories, Bar Harbor, ME) were incubated with PT-gliadin (1 mg/mL) added to the mucosal side of the tissue, and TEER was monitored every 30 minutes for the duration of the experiment. In selected experiments, medium alone or IP-10/CXCL10 (200 ng/mL), one of the known ligands for CXCR3, was added to the apical side of the tissue. In selected experiments, tissues were preincubated with IP-10/CXCL10 for 30 minutes, after which IP-10/CXCL10 was removed, and PT-gliadin was added to the tissue. In a third series of experiments, intestinal segments were preincubated with medium alone, pertussis toxin (10 ng/mL), or genetically modified (inactivated) pertussis toxin (10 ng/mL) for 30 minutes, followed by addition of PT-gliadin (1 mg/mL). Pertussis toxin or its inactive genetic mutant were present throughout the stimulation. In a fourth series of experiments, intestinal tissues were incubated with 4 different peptides (10 µg/mL) from the α-gliadin synthetic peptide library. TEER data were normalized to the initial value for that specific data set in each animal.
Zonulin was measured in the microsnapwell intestinal culture supernatants by ELISA as previously described.17
Two-tailed Student t tests were used to test differences between 2 groups. Data were paired where appropriate. Values of P < .05 were regarded as significant.
To identify the putative gliadin receptor, membrane fractions were prepared from rabbit small intestine and applied to an Affi-gel α-gliadin affinity column. Three main proteins with estimated molecular weights of 93, 100, and 107 kilodaltons were eluted from the affinity column and subjected to MALDI mass spectrometric fingerprint analysis following digestion with trypsin. The 100-kilodalton band was identified by mass spec/mass spec (MS/MS) as the chemokine receptor CXCR3, based on sequences derived from 21 peptides, whereas the other bands were identified as a heat shock protein (93 kilodaltons) and the glutamate receptor (107 kilodaltons) (data not shown).
CXCR3, a 7-transmembrane G-protein-coupled receptor, 19 is involved in cellular activation and cell migration (cytoskeleton rearrangement) into inflamed tissues, in particular of γ/δ T lymphocytes (as observed in CD).13,20 In contrast, the glutamate receptor21 and heat shock proteins22 are involved in other cellular functions. Based on the observation that CXCR3 activities are potentially relevant to CD pathogenesis, we pursued the possible role of CXCR3 as a receptor for PT-gliadin.
To establish whether CXCR3 is the receptor responsible for PT-gliadin-induced mucosal events leading to increased intestinal permeability, the following series of experiments were performed:
Immunofluorescence microscopy experiments were performed to determine whether PT-gliadin and CXCR3 colocalize. Transiently transfected, CXCR3-expressing HEK293T cells were incubated with PT-gliadin and stained for CXCR3 and gliadin. FACS analysis revealed that >60% of transfected cells expressed CXCR3 (data not shown). After immunofluorescence staining, colocalization of CXCR3 and gliadin was observed in CXCR3-transfected (Figure 1A–C) but not in pcDNA3.1-transfected cells (Figure 1D). As additional controls for the specificity of the observed staining, PT-gliadin-treated CXCR3-transfected cells were stained with isotype control or secondary Ab alone (Figure 1E). Furthermore, CXCR3-transfected cells were incubated with the irrelevant protein bovine serum albumin (BSA) (1 mg/mL) and stained with a specific anti-BSA Ab (Figure 1F). None of the control stainings showed colocalization.
To further demonstrate direct and specific gliadin binding to CXCR3, a competitive binding assay was performed. Our results showed that PT-gliadin caused a concentration-dependent displacement of the radiolabeled CXCR3 ligand I-TAC from its target receptor on CHO-K1 host cells (Figure 2). However, contrary to the other CXCR3 ligands,14 PT-gliadin binding to CXCR3 did not activate Ca2+ signaling (see Supplementary data online at www.gastrojournal.org). To define whether α-gliadin domain(s) are involved in CXCR3 binding, a synthetic peptide library consisting of 26, 10 AA overlapping, 20mer peptides was subjected to the binding assay. The results show that 2 of these peptides displaced radiolabeled I-TAC from CXCR3-expressing cells (Table 1). The specificity of this binding was confirmed by kinetic experiments performed on HEK293T cells transfected with human CXCR3 that showed a dissociation constant of peptide 4026 of 32 µmol/L (see Supplementary data online at www.gastrojournal.org).
To study the receptor expression in intestinal epithelial cells, we measured CXCR3 steady-state messenger RNA (mRNA) and protein expression in various human and murine intestinal cell lines and small intestinal tissues. Immunofluorescence analysis of human CaCo-2 cells showed constitutive CXCR3 expression (Figure 3A and B). Cross-reactivity of the antihuman CXCR3 mAb with rat CXCR3 permitted visualization of CXCR3 expression on the surface of rat IEC6 cells (Figure 3C and D). Real-time reverse-transcription (RT)-PCR analysis of Caco-2 cells confirmed CXCR3 mRNA expression (Figure 3E).
To confirm intestinal epithelial expression of CXCR3 in vivo, both murine and human intestinal tissues were analyzed. Murine small intestine was subjected to laser capture microdissection followed by real-time RT-PCR analysis. Although CXCR3 mRNA expression was more abundant in the lamina propria (probably because of the large number of CXCR3-positive immune cells present at this site), measurable expression of the receptor was detected also in murine intestinal epithelial cells (Figure 4). Immunohistochemical analysis of human small intestinal tissue stained both for CD3+ cells and CXCR3 confirmed that the receptor is expressed not only by immune cells but also by enterocytes (Figure 5A–C).
To investigate whether CXCR3 expression is altered in CD, human small intestinal biopsy specimens obtained from both non-CD and CD patients were subjected to immunohistochemical and real-time RT-PCR analysis. CXCR3 staining was detected at higher levels in the lamina propria and the epithelium of CD patients (Figure 5E) as compared with non-CD controls (Figure 5D). CXCR3 mRNA expression in biopsy specimens revealed a 9.6-fold increase in CXCR3 mRNA expression in CD patients with active disease compared with CXCR3 gene expression in non-CD patients (P = .004). This disease-associated enhanced mRNA expression returned to levels seen in non-CD intestinal tissue in CD patients in remission after implementation of a gluten-free diet (Figure 5G).
We next established whether CXCR3 is required for the PT-gliadin-induced increase in zonulin release and subsequent changes in intestinal permeability previously described.8 Intestinal tissues of wild-type C57BL/6 and CXCR3−/− mice were mounted in microsnapwell chambers, and PT-gliadin was added to the mucosal (eg, apical) side of the tissue. No differences in intestinal mucosal morphology or baseline TEER were noted between C57BL/6 and CXCR3−/− mice (data not shown). An initial series of experiments was designed to evaluate whether the effect of PT-gliadin on zonulin release and intestinal permeability is CXCR3 dependent. Intestinal segments from wild-type mice showed a significant 30% drop in TEER (Figure 6A). These TEER changes were preceded temporally by the release of zonulin following mucosal PT-gliadin challenge (Figure 6B). Conversely, CXCR3−/− mice did not exhibit changes in either intestinal TEER or zonulin release in response to gliadin (Figure 6A and B). To establish whether the zonulin pathway is operative in CXCR3−/− mice, we repeated the permeability experiments using the zonulin agonist AT1002. A significant drop in TEER was observed when CXCR3−/− tissues were challenged with the AT1002 compared with baseline (Figure 6C).
In addition, we chose 2 CXCR3-binding peptides and 2 peptides that did not show binding to CXCR3 from the α-gliadin synthetic peptide library and applied them to the luminal side of wild-type intestinal segments. Only the 2 CXCR3-binding peptides A (4026) and B (4022) induced a significant decrease in TEER, whereas the nonbinding peptides C (4018) and D (4030) did not alter intestinal permeability (Figure 6D).
A second set of experiments was performed on intestinal tissue from wild-type mice to assess whether the effects after gliadin binding to CXCR3 could be induced by other CXCR3 ligands. One of 3 previously described CXCR3 ligands, IP-10/CXCL10, was applied to the mucosal side of the intestinal tissue, and TEER w as measured. IP-10/CXCL10 did not cause significant changes in either TEER or zonulin release compared with medium alone (data not shown).
From these experiments emerges that IP-10/CXCL10 and PT-gliadin induce different cellular activation patterns after binding to CXCR3. We next evaluated whether IP-10/CXCL10 binding to CXCR3 affects PT-gliadin-induced changes in intestinal permeability as a result of the receptor tachyphylaxis. Pretreatment of wild-type intestinal segments with IP-10/CXCL10 for 30 minutes did not overall prevent the effects of PT-gliadin on TEER. However, the time in which the TEER started to drop following PT-gliadin exposure was delayed by 30 minutes in tissues pretreated with IP-10/CXCL10 (Figure 6E). These results suggest that CXCR3 receptors could be temporally unavailable secondary to IP-10/CXCL10 and needed to shuttle back to the cell surface before PT-gliadin could bind and exert its effects on TEER.
A third series of experiments was performed to examine whether gliadin binding to CXCR3, a G-protein-coupled receptor, requires G-protein signaling. For these experiments, intestinal tissue of wild-type mice was mounted in microsnapwells, and PT-gliadin was added to the mucosal side after pretreatment with medium alone, pertussis toxin (a G protein-coupled receptor inhibitor), or an inactive genetic mutant of pertussis toxin. PT-gliadin induced the expected drop in TEER in tissues preincubated with medium alone, and this was prevented by preincubation with pertussis toxin but not with its inactive mutant (Figure 6F).
We recently reported that PT-gliadin-induced zonulin release is MyD88-dependent.11 To investigate whether PT-gliadin binding to CXCR3 induces recruitment of the adapter protein MyD88, IEC6 intestinal epithelial cells were stimulated with PT-gliadin and subjected to coimmunoprecipitation assays. These assays revealed an association of CXCR3 and MyD88 after PT-gliadin challenge that was concentration and time dependent (Figure 7). This association was optimal when PT-gliadin was present at a concentration of 1 mg/mL (Figure 7A) and reached a plateau after 45 minutes of incubation (Figure 7B).
TJs are central to the regulation of intestinal permeability because they maintain the contiguity of intestinal epithelial cells and are capable of prompt and coordinated responses to the many physiologic challenges to the intestinal epithelial barrier.4 Increased intestinal permeability appears to be an early biologic change that precedes the onset of autoimmune diseases, including CD and type I diabetes.8,23,24 The peculiarity of CD is that it is the only autoimmune disease for which the triggering environmental factor gliadin is known. This offers a unique opportunity to study the cellular and molecular basis of the autoimmune process using enzymatically digested gliadin as a stimulus in experimental assays.
We showed a direct effect of gliadin on intestinal barrier function,9 which was confirmed by others.25 This effect of gliadin is polarized, eg, gliadin increases intestinal permeability only when administered on the luminal side of the intestinal tissue.9 These data formed the basis for the present study because a missing link has been the identification of the luminal structure to which gliadin binds and through which gliadin induces epithelial zonulin release and TJ disassembly.
Our MS/MS data identified the chemokine receptor CXCR3 and 2 other proteins, a glutamate receptor and a heat shock protein, as the proteins that bound to α-gliadin. We chose to investigate the possible role of CXCR3 as a receptor for gliadin because of its function in recruiting γ/δ lymphocytes, a marker of early stage in CD pathogenesis.26 In contrast, the glutamate receptor is an intrinsic transmembrane ion channel that is opened in response to binding of a chemical messenger but has not been described to be involved in cell activation and rearrangement of the cytoskeleton.21 Heat shock proteins are cytoplasmic proteins involved in intracellular processes including protein folding and protein conformation and are found extracellularly only as shed contents from necrotic cells providing a strong danger signal to the immune system.22
The identification of CXCR3 as a receptor for gliadin is important for several reasons. The chemokine receptor CXCR3 is involved in various pathophysiologic conditions. Its biologic role is to provide a mechanism for cells that express this receptor to migrate to its ligands, the chemokines Mig/CXCL9, IP-10/CXCL10, and I-TAC/CXCL11, which share the receptor, but exert different and nonredundant effects.27 CXCR3 is associated with leukocyte recruitment to target organs and subsequent T helper cell 1 immune-mediated tissue damage in viral and bacterial infections28,29 and autoimmune disease states.30,31 CXCR3 is predominantly expressed on different T-cell subsets, including activated T helper cell 1 (Th1) cells,32 T lymphocytes,33 a newly identified E-cadherin-bearing CD8+ T-cell subset that specifically homes to the gut,34 and natural killer cells,35 but its expression has been reported on other cell types as well.36,37
The phenomenon that ligands other than chemokines can bind to chemokine receptors has been reported previously; for example, human immunodeficiency virus uses the CCR5 chemokine receptor for cell entry,38 and PGP, a peptide derived from the extracellular matrix, signals through the CXCR2 receptor on neutrophils causing neutrophil recruitment into the lungs and production of superoxide.39
With this paper, we report for the first time CXCR3 expression in intestinal epithelium. CXCR3 expression showed the same qualitative distribution in both CD and non-CD intestinal tissues, but its expression was higher in CD. These differences were paralleled by higher CXCR3 gene transcription in CD patients with active disease that returned to baseline levels when the disease was in remission following the implementation of a gluten-free diet. The enhanced CXCR3 mRNA expression in intestinal tissue from active CD patients reflects the importance of CXCR3 expression on both intestinal epithelial cells and intraepithelial lymphocytes and its distinct regulation in CD. Our immunohistochemical staining studies show that epithelial CXCR3 expression is predominantly related to enterocytes and not to the large number of intraepithelial CXCR3-expressing γ/δ-positive T lymphocytes that typically infiltrate the intestinal mucosa during the acute phase of CD.20
The role of CXCR3 in mediating the PT-gliadin-induced zonulin release and subsequent increase in intestinal permeability was confirmed using CXCR3−/− mice in which PT-gliadin failed to release zonulin and, consequently, to reduce TEER. Our observation that pretreatment of C57BL/6 wild-type intestinal tissue with the G-protein inhibitor pertussis toxin prevented the effect of PT-gliadin on intestinal permeability is consistent with activation through CXCR3 that leads to subsequent TJ disassembly.
Interestingly, PT-gliadin, as well as 2 α-gliadin synthetic peptides, bound to CXCR3 but did not cause Ca2+ release (see Supplementary data online at www.gastrojournal.org) as reported for the 3 known natural CXCR3 ligands, Mig/CXCL9, IP-10/CXCL10, and I-TAC/CXCL11.14,27 The fact that IP-10/CXCL10 failed to cause TEER changes suggests that other intracellular signaling pathway(s) could be responsible for the PT-gliadin-induced zonulin release and TJ disassembly. Pretreatment with IP-10/CXCL10 caused a delay of TJ disassembly but was not able to inhibit the effects of PT-gliadin on intestinal permeability, indicating that both proteins act via binding to CXCR3 but exert different effects with regard to TJ disassembly and zonulin release. Zonulin characterization revealed that it belongs to a family of serine proteases with structure similarities with a series of growth hormones, including epidermal growth factor. One can hypothesize that the effect of gliadin on epidermal growth factor-related signaling as was recently reported40 could eventually be mediated by zonulin.
Our data suggest that recruitment of the adapter protein MyD88 to CXCR3 is involved. Until recently, MyD88 has been described to be associated uniquely with signalin via Toll-like receptors (TLR) and the interleukin (IL)-1R family. TLRs are a family of pattern recognition receptors that recognize evolutionary highly conserved structures on microorganisms and give rise to nuclear factor-κB activation and proinflammatory gene transcription.41 This knowledge was extended recently with the finding that MyD88 can associate with the IFN-γ receptor, providing an alternative way by which IFN-γ can enhance proinflammatory gene expression.42 In our experiments, CXCR3 activation by PT-gliadin failed to activate nuclear factor-κB, IRF-3, or p38 (data not shown). This observation could indicate that CXCR3 associates with another receptor that, in turn, leads to recruitment of MyD88 “by proxy.” This concept would exclude both TLR2 and TLR4 as coreceptors because our previous studies ruled out the involvement of these 2 TLRs in zonulin signalling and increased permeability.11 Alternately, PT-gliadin-dependent CXCR3 activation signals leading to zonulin release may be mediated by a yet undefined pathway downstream of the recruitment of MyD88. Support for a direct interaction was suggested by our Clustal W analysis that identified a TIR-like region within the C-terminus of CXCR3 (Quan Nhu, unpublished observation).
In conclusion, using biochemical, genetic, and physiologic approaches, we identified the chemokine receptor CXCR3 as the receptor that binds gliadin. Our data suggest that gliadin binds to CXCR3 on epithelial cells to initiate an increase in intestinal permeability through an MyD88-dependent release of zonulin that enables the paracellular passage of gliadin (and possibly other non-self antigens) from the intestinal lumen to the gut mucosa. In genetically predisposed individuals, gliadin may attract and stimulate other CXCR3-expressing cells, including T cells, CD3+CD8+ T cells, and natural killer cells,33,34,43 leading to the early activation of the innate immune arm of the CD inflammatory response.44
Supported in part by National Institutes of Health grants DK-48373 (to A.F.) and AI-18797 (to S.N.V.).
The authors thank Dr Anna Sapone, and Rex Sun for their technical assistance with some of the snapwell experiments, and Manjusha Thakar for her assistance with the immunoprecipitation experiments.
Conflicts of interest: S.N.V. and A.F. have financial relationship with Alba Therapeutics.