A pathogenic role for T cell–derived IL-22BP in inflammatory bowel disease

See allHide authors and affiliations

Science  21 Oct 2016:
Vol. 354, Issue 6310, pp. 358-362
DOI: 10.1126/science.aah5903


Intestinal inflammation can impair mucosal healing, thereby establishing a vicious cycle leading to chronic inflammatory bowel disease (IBD). However, the signaling networks driving chronic inflammation remain unclear. Here we report that CD4+ T cells isolated from patients with IBD produce high levels of interleukin-22 binding protein (IL-22BP), the endogenous inhibitor of the tissue-protective cytokine IL-22. Using mouse models, we demonstrate that IBD development requires T cell–derived IL-22BP. Lastly, intestinal CD4+ T cells isolated from IBD patients responsive to treatment with antibodies against tumor necrosis factor–α (anti–TNF-α), the most effective known IBD therapy, exhibited reduced amounts of IL-22BP expression but still expressed IL-22. Our findings suggest that anti–TNF-α therapy may act at least in part by suppressing IL-22BP and point toward a more specific potential therapy for IBD.

Inflammatory bowel disease (IBD) is characterized by chronic intestinal inflammation and dysfunction of the epithelial barrier. The primary cause of the initiation of this disease is unclear. However, it is well accepted that inflammatory responses, most likely driven by the microbiome and defective barrier function, promote a vicious cycle that leads to chronic disease (1, 2). Because IBD is characterized by chronic inflammation, most therapies are directed against the inflammatory responses, but they do not directly promote mucosal healing. Relapsing flares and opportunistic infections that occur as a consequence of immune suppression are the resulting problems. Thus, one major goal is to identify the mechanism that links inflammation and barrier dysfunction. The proinflammatory cytokine tumor necrosis factor–α (TNF-α) is the primary target of IBD therapy at present. There are several potential mechanisms whereby therapy with antibodies against TNF-α (anti–TNF-α) might work in IBD. These include the expansion of regulatory T cells, induction of T cell apoptosis, and promotion of the epithelial cell barrier function [reviewed in (3)]. However, what determines the response to anti–TNF-α therapy in IBD remains unclear.

Interleukin-22 (IL-22) is up-regulated in the intestine in patients with IBD (4, 5). IL-22 is normally able to promote mucosal healing in the intestine (68); however, when uncontrolled, it can lead to intestinal pathogenesis (7, 912). Therefore, tight control of IL-22 activity is essential. IL-22 binding protein (IL-22BP) exerts this control by specifically binding IL-22 and preventing it from binding membrane-bound IL-22 receptor 1 (IL-22R1) (1316). The binding affinity of IL-22 and IL-22BP is 20- to 1000-fold that of the former and membrane-bound IL-22R1 (4, 17). IL-22 and IL-22BP exhibit an inverse expression pattern upon tissue damage in the intestine in mouse models: IL-22BP is most highly expressed in the colon during homeostasis and tissue repair, whereas IL-22 is most highly expressed at the peak of tissue damage (4, 7, 18). Thus, careful regulation of IL-22 and IL-22BP controls homeostasis in the intestine; however, the role of IL-22BP in humans in IBD is uncertain.

To address this, we first analyzed IL-22 and IL-22BP expression in intestinal biopsy specimens from patients with ulcerative colitis (UC) or Crohn’s disease (CD) (table S1) who either were in remission (UC, n = 13; CD, n = 12) or had active disease based on clinical, endoscopic, and histological findings (UC, n = 18; CD, n = 21). In patients with active IBD, we analyzed samples from both healthy (uninflamed) and diseased (inflamed) areas in the colon and terminal ileum, based on endoscopic and histological findings, which were further validated by analyzing expression of the genes for the inflammatory markers IL-22, IL-17A, and interferon-γ (IFN-γ) (Fig. 1A and fig. S1). In line with previous reports (4, 5), we found increased IL22 and IL17A mRNA expression in the colon and terminal ileum of patients with active CD and UC, compared with that in healthy controls. IL22BP mRNA was not decreased, but was rather increased in the colon of these patients (Fig. 1A), suggesting that the activity of IL-22 might be impaired. We also confirmed this finding at the protein level by immunohistochemistry of intestinal biopsy specimens (Fig. 1, B and C). As a control, we analyzed the IL-22 and IL-22BP expression pattern in a non–IBD-related intestinal disease, colonic diverticulitis (n = 5). Colonic diverticulosis is an acquired disease, developing as mucosal and submucosal herniation through the circular muscle layer at vulnerable weak points of the colonic wall. Subsequent inflammation of these diverticula is termed diverticulitis. We obtained tissue specimens from the inflamed area of the colon from patients with diverticulitis and, in line with the studies in mice (4, 7, 18), we observed increased IL22 and decreased IL22BP expression (Fig. 1A), suggesting that IL-22BP is not generally up-regulated during intestinal inflammation in humans. One possible explanation for the increased expression of IL-22BP in IBD relative to that in diverticulitis might be the chronic versus acute nature of the inflammatory response in these respective diseases.

Fig. 1 Increased IL-22 and IL-22BP expression in IBD.

(A) Relative IL22BP, IL22, and IL22R1 mRNA expression, as measured by reverse transcriptase polymerase chain reaction (RT-PCR) of intestinal specimens from patients with Crohn’s disease (CD), ulcerative colitis (UC), and diverticulitis, as well as from healthy controls (number of patients: UC in remission, n = 13; active UC, n = 18; CD in remission, n = 12; active CD, n = 21; diverticulitis, n = 5; controls, n = 12). Filled circles represent one sample, bars represent means, and error bars show SEM. (B) Representative IL-22BP immunohistochemical staining of colonic biopsy specimens (scale bars, 100 μm) and (C) quantitative assessment (each symbol represents one patient; horizontal lines indicate means ± SEM).

Next, we aimed to identify the cellular source of IL-22BP in the intestine. To that end, we sorted different cell populations from human intestine. In line with previous publications (7, 1921), we confirmed that dendritic cells (DCs) and eosinophils express high levels of IL-22BP (Fig. 2, A and B, and fig. S2). Unexpectedly, we also observed high IL-22BP expression in CD4+CD3+CD11c T cells in the small intestine (Fig. 2, A and B). We measured CD4 and CD11c (ITGAX) as controls (fig. S3). IL-22BP was also expressed in the colon by T cells, and it was further increased in T cells but not in DCs in the diseased area in patients with active IBD (Fig. 2, C and D, and tables S2 and S3). We found that murine CD4+ T cells in the lymph nodes also expressed Il22bp (Fig. 2E). The expression levels were highest in CD44+CD4+ T cells, reaching levels similar to those in DCs. In the mouse colon, however, we previously did not detect Il22bp in bulk populations of TCRβ+ cells in steady-state conditions (7). We therefore performed additional analysis of purified CD4+ T cells from the colon. We did detect Il22bp in colonic CD4+ T cells, but the expression level was low and close to the detection limit (Fig. 2E). In conclusion, both CD4+ T cells and DCs can produce IL-22BP in mice and humans.

Fig. 2 CD4+ T cells express IL-22BP in IBD.

In (A) and (B), cells were isolated from the ileum of patients undergoing laparoscopic gastric bypass surgery because of morbid obesity. (A) Relative IL22BP mRNA expression, as measured by RT-PCR of sorted CD45+CD4+CD3+CD11c T cells, CD45+CD11c+MHC-II+CD3 DCs, and CD45+CD4CD3CD11c cells (rest). Bars represent means, error bars show SEM (n = 3 patients). (B) A representative example of Western blot analysis of IL-22BP from sorted CD4+CD3+CD11c T cells and CD45+CD4CD3CD11c cells (rest). In vitro differentiated monocyte-derived DCs and recombinant (rec.) IL-22BP were used as positive controls. Results are representative of at least three independent experiments. (C) Relative IL22BP mRNA expression, as measured by RT-PCR of sorted CD4+CD3+CD11c T cells and CD11c+MHC-II+CD3 DCs isolated from intestinal biopsy specimens from active CD, active UC, or healthy control patients. When possible, biopsies were taken from inflamed (red circles) and uninflamed (black circles) regions of the intestine. Bars represent means, error bars show SEM (number of patients: control, n = 4; CD, n = 7; UC, n =11). (D) Analysis of IL-22BP expression by flow cytometry. Bars represent means, error bars show SEM, and each circle represents one sample (control, n = 6; IBD uninflamed regions, n = 5; IBD inflamed regions, n = 7) (number of patients: control, n = 6; UC, n = 4; CD, n = 3). *One outlier not depicted. (E) Relative Il22bp mRNA expression, as measured by RT-PCR of sorted CD4+ T cells and DCs from murine colon and lymph node (LN). Horizontal lines indicate means ± SEM; each symbol represents one experiment using pooled samples from three to six mice per group.

T cell–specific up-regulation of IL-22BP in active IBD suggests a pathogenic role in intestinal inflammation. On the basis of these findings, we aimed to further decipher the relevance of DC- versus T cell–derived IL-22BP in colitis. To that end, we used the murine CD45RBhigh transfer colitis model, in which IL-22 mediates a protective function (10, 22). To discriminate between CD4+ T cell–derived and innate immune cell–derived IL-22BP, we transferred Il22bp−/− and Il22bp+/+ CD4+CD25CD45RBhigh T cells into Rag1−/− and Rag1−/−Il22bp−/− mice. Transfer of wild-type T cells into Rag1−/− and Rag1−/−Il22bp−/− mice caused equally severe disease, characterized by weight loss and endoscopic and histological scores of colitis (Fig. 3, A to E). However, both Rag1−/− and Rag1−/−Il22bp−/− mice that received Il22bp-deficient T cells were largely protected from disease development (Fig. 3, A to E). Similar to the results observed in humans (Fig. 2C), the expression of IL-22BP by CD4+ T cells was increased in the murine colitis model (fig. S4). T cells from Il22bp–/+ mice showed an intermediate level of Il22bp expression and caused an intermediate phenotype of colitis development (fig. S5). We also confirmed the pathogenic effect of CD4+ T cell–derived IL-22BP in a bacterial-driven colitis model (fig. S6). T cell–derived IL-22BP thus plays an important pathogenic role in multiple mouse IBD models.

Fig. 3 A pathogenic role of CD4+ T cell–derived IL-22BP in a murine colitis model.

CD4+CD25CD45RBhigh cells were isolated from the spleen and lymph nodes of Il22bp+/+ and Il22bp−/− mice and transferred into Rag1−/− and Rag1−/−Il22bp−/− recipients. Disease development was assessed by (A) weight loss, (B) endoscopic, and (C) histological findings 5 weeks after transfer. Each symbol represents one mouse. Horizontal lines indicate means ± SEM. Results are representative of four independent experiments. (D) Representative histological (scale bar, 1 mm) and (E) endoscopic findings (asterisks, stool inconsistency; Xs, granularity; pound sign, abundant fibrin; arrow, ulceration).

We next sought to address whether the effect of IL-22BP is due to the free activity of IL-22 or to indirect mechanisms. Therefore, we performed the transfer colitis experiment in an IL-22–deficient environment. We transferred Il22−/− and Il22−/−Il22bp−/− CD4+CD25CD45RBhigh T cells into Rag1−/−Il22−/− mice. In this setting, both Il22−/− and Il22−/−Il22bp−/− CD4+CD25CD45RBhigh T cells caused equally severe colitis (fig. S7). As a control, we transferred Il22bp−/− CD4+CD25CD45RBhigh T cells into Rag1−/− mice, which again were largely protected from colitis development (fig. S7), indicating that IL-22BP aggravates colitis by blocking IL-22. We furthermore wanted to test whether Il22bp-deficient T cells have a cell-intrinsic defect. To this end, we cotransferred congenic wild-type and Il22bp-deficient CD4+CD25CD45RBhigh T cells into Rag1−/− mice and analyzed the transferred T cells upon colitis development. We could not find a significant difference in T cell numbers or cytokine production between wild-type and Il22bp-deficient T cells in this setting, arguing against a T cell–intrinsic defect of Il22bp-deficient T cells (fig. S8). Taken together, our data suggest that IL-22 is sufficient to protect mice from effector T cell–mediated colitis in the absence of T cell–derived IL-22BP.

Anti–TNF-α therapy is the most effective treatment of IBD at present. We therefore tested whether the efficacy of this therapy is linked to IL-22BP. We found a positive correlation between IL22BP and TNFa expression in the intestine of IBD patients with active disease (P = 0.004, correlation coefficient r = 0.36; Fig. 4A). Genes encoding other cytokines, such as IL18, IL6, and IL23, did not correlate with IL22BP expression (fig. S9). Moreover, IL22 also positively correlated with IL22BP (fig. S9). However, Il22bp expression in the colon and lymph nodes of Il22−/− mice was not significantly reduced (fig. S10), suggesting that IL-22 does not regulate IL-22BP. We therefore further investigated the link between IL-22BP and anti–TNF-α therapy.

Fig. 4 Anti–TNF-α therapy correlates with reduced IL-22BP expression by CD4+ T cells in IBD (UC and CD) patients.

(A) Correlation between TNFα and IL22BP mRNA expression, as measured by RT-PCR (P = 0.004; r = 0.36) of specimens from the intestines of patients with active IBD. In (B) to (D), CD4+CD3+CD11c T cells and CD11c+MHC-II+CD3 DCs were isolated from intestinal biopsy specimens from patients with IBD (CD and UC). When possible, biopsies were taken from inflamed (red circles) and uninflamed (black circles) regions. (B and C) Relative IL22BP mRNA expression, as measured by RT-PCR of (B) CD4+ T cells and (C) DCs isolated from intestinal biopsy specimens from IBD patients being treated with anti–TNF-α therapy or other therapies. (D) Relative IL22, IL17A, IFNg, and IL5 mRNA expression, as measured by RT-PCR of sorted CD4+ T cells from IBD patients being treated with anti–TNF-α therapy or other therapies. Horizontal lines indicate means ± SEM; dotted lines indicate the detection limit (number of patients: other therapies, n = 16; anti–TNF-α responder, n = 5; anti–TNF-α nonresponder, n = 5).

To that end, we isolated CD4+ T cells and DCs from intestinal biopsy specimens obtained from IBD patients who were receiving anti–TNF-α therapy (adalimumab or infliximab) or other immune-modulating treatment in order to test whether anti–TNF-α treatment influences the expression of IL22BP. IL22BP expression was markedly reduced in CD4+ T cells of IBD patients who were responsive to anti–TNF-α treatment, compared with those of patients on other medications (Fig. 4B and table S4). This effect did not seem to be due to differences in disease activity [disease activity index (mean ± SEM) for anti–TNF-α, 0.6 ± 0.25; for other treatment, 0.3 ± 0.13; P = 0.29] and appeared to be specific to CD4+ T cells, because DCs did not show a significant down-regulation of IL22BP in IBD patients treated with anti–TNF-α (Fig. 4C). Furthermore, the expression of genes for other T cell signature cytokines and transcriptional regulators, such as IL17A, IL22, IFNg, IL5, FOXP3, TBX21, GATA3, and RORC, were not different between these groups (Fig. 4D and fig. S11). These data argue against a general and broad effect of anti–TNF-α treatment on CD4+ T cells. However, it remained unclear whether TNF-α would regulate IL-22BP in a direct manner. We found that IL22bp expression is not significantly reduced in Tnfr1- and Tnfr2-deficient T cells in the transfer colitis model, compared with that in wild-type controls (fig. S12). Moreover, TNF-α did not induce IL22BP in T cells in vitro (fig. S12). Taken together, these data suggest that TNF-α might regulate IL-22BP in an indirect manner that is as yet unknown.

We next tested whether anti–TNF-α therapy is simply inversely correlated with IL-22BP expression or whether the effect of this treatment is dependent on IL-22BP regulation, and thus on the protective effect of IL-22. In case of the latter, one would expect that this therapy would not work in an Il22-deficient environment. To test this point, we transferred wild-type and Il22−/− CD4+CD25CD45RBhigh T cells into Rag1−/− and Rag1−/−Il22−/− mice, respectively, and treated the mice with anti–TNF-α upon colitis development. Anti–TNF-α treatment was not effective in the Il22–deficient environment (fig. S13), but anti–TNF-α therapy significantly reduced colitis severity in the Il22–sufficient environment. In addition, we transferred Il22bp−/− CD4+CD25CD45RBhigh T cells into Rag1−/−Il22bp−/− mice and treated the mice with anti–TNF-α upon colitis development. As expected, these mice developed a mild colitis, which, however, was not further improved by anti–TNF-α therapy. To further support these data that we obtained in the murine system, we analyzed CD4+ T cells isolated from the intestine of patients who did not respond to anti–TNF-α therapy. T cell–derived IL-22BP was not down-regulated in these patients (Fig. 4B). Thus, our data indicate that one mechanism whereby anti–TNF-α therapy may reduce disease activity is by down-regulating expression of IL-22BP.

TNF-α antibodies are one of the most effective therapies for IBD, but what determines the response to anti–TNF-α therapy in IBD has remained elusive (23, 24). Our data suggest that anti–TNF-α therapy may block IL-22BP expression by intestinal T cells, thus allowing IL-22–induced mucosal healing. Targeting IL-22BP directly might allow for a more effective and specific therapy for IBD without invoking the undesirable and potentially dangerous side effects of anti–TNF-α, such as susceptibility to infections. In addition, anti–TNF-α therapy may increase patients’ risk of developing cancer. Therefore, long-term anti–IL-22BP treatment might have similar effects, for which these potential patients should be closely screened.

Supplementary Materials

Materials and Methods

Figs. S1 to S13

Tables S1 to S4

References (2527)

References and Notes

  1. Acknowledgments: The authors thank C. Lieber, P. Musco, J. Alderman, and F. Huber for expert administrative assistance and C. Möller-Koop for performing the immune histochemistry. The authors also thank all members of the Departments of Endoscopy and Surgery at the University Medical Center Hamburg-Eppendorf for their key help in obtaining the human samples involved in this study. This work was supported by the Deutsche Forschungsgemeinschaft (grant SFB841 to S.H., A.W.L, and D.B.), the European Research Council (starting grant 337251 to S.H.), Ernst Jung-Stiftung Hamburg (to S.H.), Stiftung Experimentelle Biomedizin (to S.H.), Werner Otto Stiftung Hamburg (to S.H. and M.W.), and the Howard Hughes Medical Institute (to R.A.F.). All data necessary to understand and assess the conclusions of the manuscript are available in the body of the paper and in the supplementary materials. Il22- and Il22bp-deficient mice are available from Regeneron under a material transfer agreement with Yale University. S.H., R.A.F., and P.P. are inventors on a patent application (LU 92982) submitted by the University Medical Center Hamburg-Eppendorf together with Yale University that covers the use of IL-22BP as biomarker in anti–TNF-α treatments.
View Abstract

Navigate This Article