IL-23–responsive innate lymphoid cells are increased in inflammatory bowel disease

In this study, we aimed to analyze the contribution of adaptive and innate sources of Th17 signature cytokines to chronic intestinal inflammation in patients with IBD. We confirmed the overexpression of Th17 signature cytokine genes in the inflamed intestinal mucosa in our cohort of patients with IBD, either ulcerative colitis (UC) or CD, compared with the nonaffected colon of patients undergoing colectomy for colorectal cancer as noninflammatory controls (Fig. 1 A). To evaluate the contribution of innate and adaptive sources of Th17 signature cytokines in the human systemic and intestinal immune response in the absence and presence of IBD, we compared T cell and non-T cell expression of Th17 genes in the peripheral blood (PB) versus the lamina propria (LP) of IBD patients and controls (Fig. 1, B and C). We found preferential expression of IL22, IL17A, IL17F, and IL26 in CD3⁺ cells isolated from the intestine compared with the blood of both control and IBD patients (Fig. 1 B). In line with these results, Kobayashi et al. (2008) described higher expression of IL-17 in LP CD4⁺ cells compared with the PB CD4⁺ population. Compartmentalization of Th17 gene expression was not restricted to T cells as we also found increased expression of IL22, IL17F, and IL26 in LP CD3⁻ cells with nondetectable or very low expression among PB CD3⁻ cells (Fig. 1 C) in both IBD and control individuals. IL17A expression was also increased in LP CD3⁻ cells compared with PB CD3⁻ in IBD patients. This increase was specific to intestinal inflammation because no significant difference was observed in noninflamed controls. Other Th17 genes such as IL21, IL23R, RORC, and aryl hydrocarbon receptor (AHR) were also expressed in both the intestinal T and non-T cell compartments, with very low or nondetectable expression in blood leukocytes (Fig. S1, A and B). These results confirm our hypothesis of a specific role for the IL-23 axis in the intestinal immune response and show that both T and non-T cells expressing Th17-related genes are present in the human intestine.

Interestingly, we found no significant differences between patients and controls in expression of Th17 genes in LP CD3⁺ cells (Fig. 1 B and Fig. S1 A), although absolute numbers of intestinal Th17 cells are increased in IBD, as suggested by Kleinschek et al. (2009). Strikingly, significantly higher expression of IL17A and IL17F was observed in LP CD3⁻ cells isolated from patients with IBD (Fig. 1 C) compared with controls. No significant difference was observed for IL22, IL26, RORC, AHR, and IL23R expression (Fig. 1 C and Fig. S1 B). IL21 expression was very low or undetectable in most PB and LP CD3⁻ cells from IBD patients (Fig. S1 B). These findings suggest that innate sources of IL-17A and IL-17F might contribute to intestinal inflammation in IBD.

To determine whether intestinal non-T cells are responsive to IL-23 and whether the innate response is altered in IBD, sorted LP CD3⁻ cells from patients and controls were cultured overnight with or without IL-23, and Th17 gene expression was evaluated (Fig. 2 A). IL22 was induced by IL-23 stimulation in non-T cells isolated from control colon, confirming the presence of IL-23–responsive innate cells in the human intestine. Induction of IL26 was not observed after IL-23 stimulation in LP CD3⁻ cells from either controls or IBD patients. However, IL26 transcripts were very low in some cultures, leading to a large variation in expression levels. Interestingly, IL17A in LP CD3⁻ cells after IL-23 stimulation was significantly higher in cells isolated from patients with IBD compared with controls. All together, these data suggest that an IL-23–dependent source of IL-17 is present in the CD3⁻ compartment in the inflamed colon of patients with IBD but not in controls.

In humans, CD3⁻CD127⁺ ILCs can be further subdivided based on expression of the NK marker CD56 (Crellin et al., 2010). Both populations can be isolated from human adult tonsils and share the expression of NKp44, NKp46, CD161, c-Kit, and RORC. In vitro expanded CD127⁺CD56⁻ and CD127⁺CD56⁺ cells showed a similar cytokine profile, including production of IL-17, IL-22, and IFN-γ. In vitro analysis suggests a precursor-product relationship with CD56 expression induced upon activation. However, the relative distribution and function of these ILC populations in the human intestine in health and disease have not been examined.

To further characterize the intestinal IL-23–responsive non-T cell source of Th17 cytokines in the inflamed intestine of patients with IBD, we sorted CD3⁻CD19⁻CD14⁻CD16⁻(Lin⁻) CD45⁺CD56⁺ cells and Lin⁻CD45⁺CD56⁻ cells from the ileum and colon of patients with CD and cultured them with IL-23. IL22 and IL26 were induced by IL-23 in the CD56⁺ population. In contrast, IL17A and IL17F were preferentially expressed in the Lin⁻CD45⁺CD56⁻ population analyzed directly ex vivo, and levels were not increased upon addition of IL-23 (Fig. 2 B). Expression of IFNγ was detected among ileal as well as colonic Lin⁻CD45⁺CD56⁺ and Lin⁻CD45⁺CD56⁻ cells from patients with CD (Fig. S2). Further analysis of the phenotype of IL-17A– and IFN-γ–producing cells in the Lin⁻ compartment by intracellular FACS staining showed that these cells were primarily CD127⁺CD56⁻ (77% and 90%, respectively; Fig. 2 C). Analysis of the frequency of intestinal Lin⁻CD45⁺CD127⁺CD56⁻ cells and Lin⁻CD45⁺CD127⁺CD56⁺ (termed CD56⁻ and CD56⁺ ILCs) in patients with CD, UC, and controls showed that although both populations were present at similar frequencies in the uninflamed colon and ileum of control individuals, there was a marked increase specifically in CD56⁻ ILCs in the inflamed ileum and colon of CD patients. Interestingly, no difference in the frequency of CD56⁻ and CD56⁺ ILCs was observed in the colon of UC patients versus controls, indicating that accumulation of CD56⁻ ILCs might be a specific feature of CD (Fig. 3, A–C). Consistent with low IL22 expression, only a small percentage of CD56⁻ ILCs that accumulated in CD expressed NKp44, suggesting that they represent a distinct population from IL-22–producing NK22 cells, which expressed NKp44 and represented a quarter of CD56⁺ ILCs (Fig. S3 A). Both intestinal CD56⁻ and CD56⁺ ILCs also expressed the chemokine receptor CCR6 (Fig. S3 B). It has been reported that CCL20 and β-defensins, which are known CCR6 ligands, are both increased in the inflamed intestine of patients with IBD (Wehkamp et al., 2002; Kwon et al., 2003; Kaser et al., 2004). These results raise the possibility that a CCR6-mediated mechanism might be responsible for the recruitment of ILCs to the intestine in IBD.

In this study, an IL-23–dependent innate lymphoid source of IL-17A and IL-17F, which shares features of human LTi cells, has been identified in the intestine of patients with CD. The differential accumulation of IL-17A– and IL-17F–producing ILCs in the inflamed CD intestine is very similar to the accumulation of IL-17A–secreting ILCs that mediate innate colitis in mice (Buonocore et al., 2010). Further understanding of the factors that selectively promote IL-17–producing ILCs at the expense of tissue-protective IL-22–producing populations during intestinal inflammation may provide novel insights into immune pathology in the intestine. It is notable that increases in NKp44⁻NKp46⁺CD56⁺CD3⁻ cells capable of secreting IFN-g in response to IL-23 have been observed in CD patients, emphasizing a more pathogenic phenotype amongst ILC populations (Takayama et al., 2010). These results also raise important questions about the developmental relationship between CD56⁻ and CD56⁺ ILC populations in vivo.

CD56⁻ ILCs may contribute to chronic intestinal inflammation not only through the production of inflammatory cytokines such as IL-17A and IL-17F but potentially through induction of adhesion molecules and recruitment of other lymphocytes. Indeed, increased numbers of isolated lymphoid follicles are typically found in the colon of patients with IBD. This study opens the way to further work on the functional role of distinct ILC populations in intestinal inflammation and identifies a potential tissue-specific target for the treatment of patients with IBD.

All patients and controls were recruited from the gastroenterology unit and the colorectal surgery department at the John Radcliffe Hospital in Oxford. The diagnosis of IBD was confirmed by established clinical, radiological, endoscopic, and histological criteria. Blood samples and gut specimens were obtained from patients with IBD undergoing surgery for severe disease, chronically active disease, or complications of disease. Blood samples and gut specimens from macroscopically healthy areas were collected from colorectal cancer patients as noninflammatory controls. Biopsies were collected from inflamed areas of the colon and small bowel of patients with IBD, undergoing endoscopy for assessment of disease activity, extension or surveillance, and the noninflamed intestine of healthy subjects. Ethical approval was obtained from the Oxfordshire Research Ethics Committee (reference number 07/Q1605/35), and informed written consent was given by all study participants.

LP mononuclear cells (LPMCs) were isolated using a modified version of the protocol described by Bull and Bookman (1977). In brief, the mucosa was dissected, cut in pieces <25 mm², and washed in 1 mM DTT solution at room temperature for 15 min to remove adherent mucus. Specimens were washed three times in 0.75 mM EDTA solution at 37°C for 45 min to detach the epithelial crypts and then digested overnight in 0.1 mg/ml collagenase D solution (Roche). Cells were then centrifuged for 30 min in a Percoll gradient and collected at the 40–60% interface. All solutions used were supplemented with antibiotics (penicillin/streptomycin, 40 µg/ml gentamicin, and 0.025 µg/ml Amphotericin B).

PB was diluted in an equal volume of PBS and centrifuged over a Ficoll-Hypaque layer at 2,000 rpm. Cells were collected at the Ficoll–dilute plasma interface. LPMCs were isolated from biopsies (up to 10 per patient) using a combined mechanical (GentleMACS; Miltenyi Biotech) and enzymatic digestion process.

CD3⁺ and CD3⁻ cells were sorted either by magnetic cell sorting with positive selection of CD3⁺ (CD3 Micro Beads; Miltenyi Biotech) or by FACS using a MoFlow (Dako). CD3⁻, CD3⁻CD19⁻, Lin⁻CD45⁺CD56⁺, and Lin⁻CD45⁺CD56⁻ cells were FACS sorted. The following antibodies were used: anti-CD3, anti-CD19, anti-CD56 (BD), anti-CD14, anti-CD16 (eBioscience), and anti-CD45 (BioLegend).

Cells were cultured in RPMI with 10% FCS, antibiotics, and l-glutamine with or without recombinant human IL-23 (R&D Systems) at 10 ng/ml concentration.

RNA was isolated from cells using the RNeasy Mini kit (QIAGEN) and cDNA synthesized using Superscript III and oligo (dT) primers (Invitrogen). Quantitative PCR was performed with ACTB-, IL17A-, IL17F-, IL22-, IL21-, IFNγ–, IL23R-, RORC-, and AHR-specific primers (QuantiTect Primer Assays; QIAGEN) and Platinum SYBR green qPCR super mix (Invitrogen). TaqMan Gene Expression Assays for ACTB and IL26 were also used in some experiments (Applied Biosystems). cDNA samples were assayed in triplicate using the Chromo4 detection system (GMI), and gene expression levels for each individual sample were normalized to β-actin. Mean relative gene expression was determined and expressed as 2^−ΔCT (ΔCT = CT_gene − CT_β-actin) × 10,000.

Cells were preincubated in 2% normal rat serum. The following antibodies were used for flow cytometry: anti-CD3, anti-CD19, anti-CD56 (BD), anti-CD14, anti-CD16, anti-CD127, anti–IL-17, anti–IFN-γ, anti-NKp44, anti-CCR6 (eBioscience), and anti-CD45 (BioLegend). For the intracellular staining, cells were stimulated with PMA and ionomycin in the presence of brefeldin A for 4 h and fixed/permeabilized (eBioscience). Analysis was performed using FlowJo software (Tree Star), and gates were set using relevant IgG isotype controls.

The nonparametric, two-tailed Mann-Whitney test was performed in Prism software (GraphPad Software) in all cases. Mean ± standard error of the mean is represented on bar charts.

Fig. S1 shows that Th17 signature genes are expressed in intestinal CD3⁻ cells and overexpressed in IBD. Fig. S2 shows that IFN-γ is expressed in both intestinal Lin⁻CD45⁺CD56⁺ and Lin⁻CD45⁺CD56⁻ cells. Fig. S3 shows phenotyping of intestinal CD56⁻ and CD56⁺ ILCs.

We thank Professor Derek P. Jewell for his support and advice, Dr. Holm H. Uhlig, Dr. Hergen Spits, and Dr. Tom Cupedo for critically reading the manuscript, and the patients and their families for agreeing to take part in this study.

This work was supported by the Wellcome Trust (F. Powrie and A. Geremia), the Lee-Placito Medical Fund (A. Geremia), and the Broad Medical Research Program of the Broad Foundation (C.V. Arancibia-Cárcamo).

The authors have no conflicting financial interests.