Abstract
IL-23 has been implicated in the pathogenesis of several tissue-specific autoimmune diseases. Currently, celiac disease (CD) is the only autoimmune disease in which both the major genetic (95% HLA-DQ2+) and etiologic factors (dietary glutens) for susceptibility are known. We demonstrate that wheat gliadin induces significantly greater production of IL-23, IL-1β, and TNF-α in PBMC from CD patients compared with HLA-DQ2+ healthy controls, strongly advocating a role for IL-23 in the pathogenesis of CD. Moreover, IL-1β alone triggered IL-23 secretion and the IL-1R antagonist inhibited this response in PBMC and purified monocytes. This sequence of events was replicated by β-glucan, another substance known to induce IL-23 production. Our results suggest that gliadin and β-glucan stimulate IL-23 secretion through induction of the IL-1 signaling pathway and reveal for the first time that the IL-1 system regulates IL-23 production. These findings may provide therapeutic targets for this disease and other inflammatory conditions mediated by IL-23.
Interleukin-23 is a recently described cytokine produced by APC in response to several activating stimuli (1). IL-23 has been implicated in the pathogenesis of several tissue-specific autoimmune diseases and is currently thought to perpetuate inflammation by activating cells from both arms of the immune system (1). Although the downstream effects of IL-23 have been the primary focus of autoimmune research, the agents that initiate production of this cytokine in the context of autoimmunity are not well understood and warrant further investigation.
Celiac disease (CD)3 is an autoimmune disease that primarily affects the small intestine. Like other types of autoimmunity, CD has been considered a typical Th1 disease due to the increased levels of IFN-γ associated with the disease (2). Emergence of the IL23-Th17 pathway has prompted reanalysis of cell-mediated tissue damage previously attributed to the IL12-Th1 axis and highlighted the fundamental role of the innate arm in shaping the adaptive immune response (3). Although cytokines associated with Th17-mediated pathology such as IL-1β, IL-6, IL-15, and TNF-α have been implicated in the pathogenesis of CD, an association with IL-23 production has not yet been reported (4, 5, 6).
CD provides a unique model for investigating autoimmunity because both the major genetic (95% HLA-DQ2+) and etiologic factors (dietary glutens) for susceptibility are known (7). Although the HLA-DQ2/DQ8-restricted T cell response to gluten-derived gliadin peptides has been extensively documented in CD, little is known regarding the innate immune response to dietary glutens in these patients (7, 8). Because the majority of individuals with these alleles are exposed to the etiologic agent but never develop CD, we set out to test the hypothesis that disease susceptibility might be related to differences in IL-23 responses to dietary glutens in HLA-DQ2+ individuals with CD compared to those without disease. The results of these studies led to findings that both support a role for IL-23 in the pathogenesis of CD and provide convincing evidence that the IL-1 system regulates IL-23 production from human monocytes.
Materials and Methods
Cells
PBMC from patients and healthy individuals were isolated by density gradient centrifugation in lymphocyte separation medium (ICN Biomedicals) according to the manufacturer’s instructions. Highly purified lymphocytes and monocytes (95% purity) were obtained from healthy donors as described above followed by countercurrent centrifugal elutriation. All individuals gave informed consent. The study protocol was approved by the Institutional Review Board at the University of Maryland School of Medicine (Baltimore, MD).
DNA extraction and HLA typing
DNA was extracted from a portion of the PBMC using the QIAamp DNA mini kit (Qiagen) and the alleles of genes encoding HLA-DQ were determined by the DQA1 and DQB1 SSP UniTray kit (Dynal Biotech) per the manufacturers’ instructions.
Reagents
A pepsin-trypsin digest of gliadin (PTG) was prepared by enzymatic digestion as described previously (5). β-Glucan from barley (Sigma-Aldrich) was dissolved in 95% ethanol (EtOH) followed by distilled water, stirred vigorously at 100°C for 3 min, allowed to cool, and stored at 10 mg/ml at 4°C until used. The p31-43 epitope and 25 overlapping 20-mers of α-gliadin were synthesized to >95% purity at the University of Maryland Biopolymer Laboratory and stored at −20°C until used. Human recombinant IL-1β and recombinant IL-1R antagonist (IL-1ra) were purchased from R & D Systems.
In vitro cell cultures
PBMC were incubated in RPMI 1640 supplemented with 10% heat-inactivated human AB serum, 1% l-glutamine, 1% penicillin-streptomycin, and 20 mM HEPES buffer (complete RPMI medium) with and without 25–500 μg/ml PTG, β-glucan, 0.5–50 ng/ml rhIL-1β, or 25 μg/ml synthetic peptides of α-gliadin in 96-well U-bottom plates (Denville Scientific) at 37°C in 5% CO2 for 6–72 h. PBMC were also treated with 0.5 μg/ml recombinant human IL-1ra at 37°C in 5% CO2 for 1 h and then stimulated with 100 μg/ml PTG or 500 μg/ml β-glucan for 20h. Elutriated lymphocytes, monocytes, or immature monocyte-derived dendritic cells (immDC) (cultured in the presence of GM-CSF and IL-4 for 72 h) were incubated under the same culture conditions.
Cytokine and chemokine analysis
Cell-free culture supernatants were analyzed for IL-1β, IL-1ra, IL-6, IL-12p70, TNF-α (Bio-Plex cytokine assay kit; Bio-Rad), or IL-1β and IL-23 (ELISA kit; eBioscience), IL-1ra, and CCL20 (Quantikine ELISA kit; R & D Systems) following the manufacturers’ protocols.
Statistical analyses
Data are presented as mean values + SD. Paired two-tailed Student’s t tests were used to calculate p values within the same individuals and unpaired two-tailed Student’s t tests were used to calculate p values between study groups. p < 0.05 was considered statistically significant.
Results and Discussion
IL-23 has been implicated in the pathogenesis of several other HLA class II-associated autoimmune diseases but has not yet been reported to be involved in CD. To explore this possibility, we exposed PBMC from CD patients to increasing doses of PTG or β-glucan, an agent known to stimulate IL-23 and IL-1β secretion, and analyzed production of these cytokines in the cell-free culture supernatants. Both stimuli induced dose-dependent production of IL-23 and IL-1β (Fig. 1 A). The stimulatory effects of PTG were not due to endotoxin contamination, because the presence of LPS in this preparation of PTG was ruled out in earlier studies (5).
Gliadin induces robust production of IL-23 and related proinflammatory cytokines from CD patients. A, PBMC from CD patients were cultured with 0, 25, 100, 250, or 500 μg/ml PTG or β-glucan (b-glucan) for 24 h. Secreted IL-23 and IL-1β were quantified by ELISA. Data represent mean values from three independent experiments. B, PBMC from seven CD patients and six HLA-DQ2+ healthy individuals (HD) were incubated with (filled bars) or without (open bars) 100 μg/ml PTG for 48 h, and culture supernatants were analyzed for production of IL-1β, IL-1ra, IL-6, IL-12p70, IL-23, and TNF-α. Error bars indicate + SD. ∗, p < 0.05; ∗∗, p < 0.005.
Gliadin induces robust production of IL-23 and related proinflammatory cytokines from CD patients. A, PBMC from CD patients were cultured with 0, 25, 100, 250, or 500 μg/ml PTG or β-glucan (b-glucan) for 24 h. Secreted IL-23 and IL-1β were quantified by ELISA. Data represent mean values from three independent experiments. B, PBMC from seven CD patients and six HLA-DQ2+ healthy individuals (HD) were incubated with (filled bars) or without (open bars) 100 μg/ml PTG for 48 h, and culture supernatants were analyzed for production of IL-1β, IL-1ra, IL-6, IL-12p70, IL-23, and TNF-α. Error bars indicate + SD. ∗, p < 0.05; ∗∗, p < 0.005.
Having established that PTG induces IL-23 in PBMC from CD patients, we predicted that the IL-23 response to PTG would be different (quantitatively or qualitatively) in PBMC from HLA-DQ2+ healthy individuals, HLA-DQ2 being the HLA class II allele that is the major genetic requirement in CD. To test this, PBMC from HLA-DQ2+ individuals with and without CD were cultured with PTG and the supernatants were analyzed for the production of IL-23 and IL-1β as well as IL-1ra, IL-6, IL-12p70, and TNF-α. We found that PTG stimulated the production of IL-23, IL-1β, and TNF-α in all donors tested; however, production of these cytokines was significantly higher from CD patients compared with healthy controls (Fig. 1,B). PTG increased IL-6 from basal levels in PBMC from CD patients but did not have the same effect in PBMC from all healthy controls. Moreover, PTG significantly reduced secretion of the anti-inflammatory mediator IL-1ra from CD patients, but not from all healthy controls, and did not induce IL-12p70 from any of the PBMC tested (Fig. 1 B). These results advocate a role for IL-23-mediated inflammation in the pathogenesis of CD and also demonstrate that PTG induced robust production of proinflammatory cytokines associated with innate immune responses from patients with CD.
Several immunodominant epitopes of α-gliadin that preferentially bind HLA-DQ2 and DQ8 molecules as well as “innate” peptides have been implicated in the immunopathogenesis of CD (2, 5, 6, 9). To determine whether any of these epitopes stimulated IL-23 production, we incubated PBMC from HLA-DQ2+ individuals with and without CD with the p31–43 peptide (6) or with 25 overlapping peptides spanning the entire sequence of α-gliadin. None of the synthetic peptides recapitulated the IL-23 response induced by PTG from any of the individuals tested (negative data not shown). Therefore, the observed IL-23 response may be due to other subtypes of gliadin (β-, γ-, or ω-gliadin) contained in PTG or additional properties or epitopes of α-gliadin created by pepsin trypsin digestion of whole gliadin not contained in the recombinant products. Differences could also be due to the distinct culture conditions including cell source, pretreatment with IFN-γ, Ag concentration, incubation time, or a different readout for the determination of cytokine responses (mRNA vs secreted protein).
While conducting these studies, we detected a strong positive correlation between levels of IL-1β and IL-23 and also found that IL-1β secretion always preceded that of IL-23 by several hours (data not shown). These observations, together with a recent report demonstrating that IL-1β was essential and IL-23 dispensable for induction of human Th17 cells (10), led us to hypothesize that IL-23 production is dependent on IL-1 signaling. To test this hypothesis, we treated PBMC from CD patients with IL-1ra, the endogenous IL-1 receptor antagonist, before stimulation with PTG or the positive control, β-glucan. IL-1ra completely inhibited IL-23 responses to both PTG and β-glucan, illustrating a fundamental role for IL-1 signaling in the induction of IL-23 in human PBMC (Fig. 2,A). IL-1ra also markedly reduced the levels of IL-1β (Fig. 2 A) and TNF-α and IL-6 (data not shown) in these culture fluids.
IL-1 cytokines regulate the IL-23 response in vitro. PBMC from CD patients were incubated with or without 0.5 μg/ml IL-1ra for 1 h and exposed to 100 μg/ml PTG or 500 μg/ml β-glucan (b-glucan) (A) or cultured with 5 ng/ml IL-1β alone for 20 h (B). Secretion of IL-23 and IL-1β were determined by ELISA. These data are mean values from 10 independent experiments. Error bars indicate + SD. ∗∗, p < 0.005; ∗∗∗, p < 0.0005.
IL-1 cytokines regulate the IL-23 response in vitro. PBMC from CD patients were incubated with or without 0.5 μg/ml IL-1ra for 1 h and exposed to 100 μg/ml PTG or 500 μg/ml β-glucan (b-glucan) (A) or cultured with 5 ng/ml IL-1β alone for 20 h (B). Secretion of IL-23 and IL-1β were determined by ELISA. These data are mean values from 10 independent experiments. Error bars indicate + SD. ∗∗, p < 0.005; ∗∗∗, p < 0.0005.
To determine whether IL-1β alone could trigger IL-23 secretion, we incubated PBMC with and without exogenous IL-1β at concentrations similar to those produced by PBMC exposed to PTG and analyzed culture supernatants for IL-23. We found that IL-1β alone stimulated production of IL-23, albeit at much lower levels than PTG or β-glucan, suggesting that additional signaling pathways and/or cytokines triggered by these Ags enhance the production of IL-23 (Fig. 2 B). Moreover, the in vitro effects of these IL-1 cytokines were not unique to CD, as similar results were obtained using PBMC from healthy individuals (data not shown). Taken together, these data provide the first evidence that the IL-1 system regulates IL-23 production in human PBMC and indicate that IL-1ra may be used therapeutically to inhibit IL-23 mediated inflammatory diseases.
Novel studies in humans have identified circulating monocytes as the major source of IL-23 and also demonstrated this population’s capacity to produce high levels of IL-1β and activate Th17 cells (10). We therefore predicted that circulating monocytes were the source of IL-23 being produced in PBMC stimulated with PTG. Because PTG and β-glucan induced IL-23 responses in PBMC from all individuals tested, and these responses depended on IL-1 signaling similarly, we used elutriated lymphocyte and monocyte fractions from healthy individuals to further examine these cytokine responses. Highly purified lymphocytes, monocytes, or immDC were cultured with PTG overnight and the supernatants were analyzed for production of IL-23. Under these conditions, monocytes and not their progeny immDC or lymphocytes produced IL-23, demonstrating that the IL-23 response to PTG results from direct interactions between monocytes and PTG (Fig. 3 A).
Monocytes are the cell source of IL-23 and related proinflammatory mediators triggered by gliadin. Highly purified lymphocytes, monocytes, or immDC were incubated with or without 100 μg/ml PTG for 24 h, and supernatants were analyzed for the production of IL-23 by ELISA (A) or production of IL-1β, IL-6, TNF-α (bioplex), and CCL20 (ELISA) (B). IL-23 data are the mean of five independent experiments. IL-1β, TNF-α, and IL-6 data are the means of three independent experiments. CCL20 data are from one of three independent experiments. Error bars indicate + SD.
Monocytes are the cell source of IL-23 and related proinflammatory mediators triggered by gliadin. Highly purified lymphocytes, monocytes, or immDC were incubated with or without 100 μg/ml PTG for 24 h, and supernatants were analyzed for the production of IL-23 by ELISA (A) or production of IL-1β, IL-6, TNF-α (bioplex), and CCL20 (ELISA) (B). IL-23 data are the mean of five independent experiments. IL-1β, TNF-α, and IL-6 data are the means of three independent experiments. CCL20 data are from one of three independent experiments. Error bars indicate + SD.
Because IL-1 signaling was required for IL-23 production in PBMC exposed to PTG, we presumed that IL-1β was also being produced by these monocytes. Again, we incubated purified monocytes or immDC with PTG overnight and analyzed supernatants for the production of IL-1β as well as that of IL-6, TNF-α, and CCL20, which have all been implicated in “Th17”-mediated autoimmune pathology. PTG only stimulated the production of IL-1β as well as that of IL-6, TNF-α, and CCL20 from monocytes (Fig. 3 B). Thus, our prediction was confirmed and shows for the first time that PTG stimulates robust production of “Th17”-polarizing mediators through direct interaction with human monocytes characterized as CD11c+CD14+HLA-DR+ and not with their CD14− progeny. These results may explain why investigators have been unsuccessful at detecting IL-23 in subsets of CD14− intestinal dendritic cells from CD patients (11).
To further examine the relationship between IL-1 cytokines and IL-23 production in human monocytes, purified cells were incubated with increasing doses of IL-1β alone or were treated with IL-1ra before stimulation with PTG or β-glucan. IL-1β alone triggered dose-dependent secretion of the IL-23 heterodimer, and IL-1ra significantly inhibited IL-23 responses to both PTG and β-glucan, illustrating the fundamental role of IL-1 signaling in the production of IL-23 from human monocytes (Fig. 4). IL-1ra also substantially reduced levels of IL-1β, TNF-α, and IL-6 in these culture fluids (supplemental Fig. 1).4 These data suggest that the initial engagement of PTG or β-glucan with their respective receptors stimulates the secretion of mature IL-1β, which, upon binding to its receptor, induces further production of IL-1β and initiates synthesis of IL-23 and other proinflammatory mediators in an autocrine feedback loop as described by Granowitz et al. (12). Activation of the IL-1R signaling pathway could explain the MyD88-dependent cytokine responses to PTG previously shown to occur independently of TLR2/4 (5).
The IL-1 system regulates IL-23 production in human monocytes. Purified monocytes were incubated with 0, 0.5, 5, or 50 ng/ml recombinant human IL-1β for 20 h or treated with 0.5 μg/ml IL-1ra before the addition of 100 μg/ml PTG or 500 μg/ml β-glucan(b-glucan) for 20 h. These data are mean values from five independent experiments. Error bars indicate + SD. p values compare IL-1β data sets to medium alone. ∗, p < 0.05; ∗∗, p < 0.005.
The IL-1 system regulates IL-23 production in human monocytes. Purified monocytes were incubated with 0, 0.5, 5, or 50 ng/ml recombinant human IL-1β for 20 h or treated with 0.5 μg/ml IL-1ra before the addition of 100 μg/ml PTG or 500 μg/ml β-glucan(b-glucan) for 20 h. These data are mean values from five independent experiments. Error bars indicate + SD. p values compare IL-1β data sets to medium alone. ∗, p < 0.05; ∗∗, p < 0.005.
In summary, our studies show that enzymatically digested wheat gliadin stimulates significantly greater production of IL-1β, IL-23, and TNF-α in PBMC from CD patients, advocating a role for IL-23 in the pathogenesis of CD. In addition to the robust proinflammatory cytokine response, PTG significantly reduced secretion of anti-inflammatory IL-1ra from CD patients but did not have the same effect from all healthy controls. The resulting increased ratio of IL-1β:IL-1ra may contribute to the higher amounts of IL-23, IL-1β, TNF-α, and IL-6 from CD patients, given that IL-1ra proved to be a powerful inhibitor of the responses of these cytokines to both PTG and β-glucan in our studies. Moreover, others have shown that IL-1ra inhibits IL-1β, IL-6, and TNF-α secretion in human monocytes stimulated with IL-1β alone (12).
IL-1 was one of the first cytokines to be described and has since proved to be an important mediator of inflammatory conditions in the gut (13). Elevated levels of IL-1β have been associated with inflammation in CD patients, and elimination of dietary glutens has been shown to reduce these levels and significantly increase the amount of IL-1ra in CD patients compared with healthy controls, indicating the involvement of these cytokines in the pathogenesis of CD (4). Our data confirm that PTG increases IL-1β production and reduces IL-1ra secretion from CD patients compared with controls and, in addition, provides new information that IL-1β alone activates secretion of IL-23 and that IL-1ra inhibits IL-23 responses in human monocytes.
Increased percentages of CD14+ cells thought to derive from circulating monocytes have been described in the inflamed intestines of patients with CD and irritable bowel disease (14, 15, 16). This subset has been identified as the major source of TNF-α, IL-1β, and IL-1ra in irritable bowel disease and was recently reported to produce increased levels of IL-23 in Crohn’s disease (15, 16). Given the similar phenotype and function of these intestinal CD14+ cells compared with the results of this study, we postulate that IL-1 controls IL-23 production from this proinflammatory subset in the gut and from equivalent subsets in other inflamed tissues. Indeed, IL-1β has been reported to induce IL-23p19 mRNA expression in human fibroblast-like synoviocytes and colonic subepithelial myofibroblasts (17, 18), which supports our hypothesis as to the identity of cell populations that are activated in response to various agents that initiate tissue-specific inflammatory conditions.
To our knowledge, this is the first report to demonstrate that peptic fragments of wheat gliadin stimulate the production of IL-23 from human monocytes and the first to detect increased production of this cytokine in individuals with CD. Moreover, we provide primary evidence that IL-1 cytokines control the secretion of IL-23. These novel findings may provide new therapeutic targets for this disease and other inflammatory conditions involving IL-23.
Acknowledgments
We thank the University of Maryland Center for Celiac Research for providing patient samples, PTG, and synthetic α-gliadin peptides.
Disclosures
A.F. has financial interest in Alba Therapeutics (stock holder). K.M.H. and D.L.M. have no financial conflicts of interest.
Footnotes
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Funding for this project was provided by an Internal Research Grant from Department of Pathology, University of Maryland School of Medicine, Baltimore, MD. These studies were partially supported by National Institutes of Health Grant DK078699 to A.F.
Abbreviations used in this paper: CD, celiac disease; IL-1ra, IL-1R antagonist; immDC, immature monocyte-derived dendritic cell; PTG, pepsin-trypsin digest of gliadin.
The online version of this article contains supplemental material.