Evaluating IL-21 as a Potential Therapeutic Target in Crohn's Disease

Background and Aim Interleukin-21 (IL-21) is primarily a T cell-derived cytokine; it is upregulated in patients with Crohn's Disease (CD) and could be a potential new therapeutic target in CD. Methods In human material, IL-21 and IL-21R expression was investigated by in situ hybridization (ISH) and immunohistochemistry (IHC) in noninflammatory bowel disease (non-IBD) controls and patients with CD. The pathologic role of IL-21 was examined in murine models of T cell-dependent and T cell-independent colitis, either with a neutralizing monoclonal antibody against IL-21 or with the transfer of CD4+CD45RBhighIL-21R−/− T cells. Colonic pathology was examined by endoscopy, histopathology, IHC, ELISA, and Luminex. Results In the human intestine, IL-21 and IL-21R mRNA and protein-expressing cells were observed in the mucosa, in lymphoid aggregates of submucosa in non-IBD controls, and in lymphoid aggregates of muscularis externa in patients with CD. IL-21 expression was most abundant in germinal centers (GCs) of the lymphoid aggregates, and IL-21R expression assessed semiquantitatively, was significantly higher in patients with CD compared to non-IBD controls. Following prophylactic and interventive anti-IL-21 mAb treatment in the adoptive transfer (AdTr) model, clinical and pathological parameters were significantly reduced. The most persistent finding was a reduction in colonic infiltrating neutrophils. As well, Rag2−/− mice receiving CD4+CD45RBhighIL-21R−/− T cells developed less severe colitis compared to Rag2−/− mice receiving CD4+CD45RBhighIL-21R+/+ T cells. No effect of reduced IL-21 signalling was observed in T cell-independent colitis. Conclusion Our study shows that patients with CD have significant expression of IL-21 and IL-21R in the gut. As well, we show that neutralization of IL-21 in experimental T cell-driven colitis is associated with a reduction in clinical and pathological findings. This amelioration seems to be associated with a reduction in colon-infiltrating neutrophils.


Introduction
Inflammatory bowel disease (IBD) is a refractory chronic inflammatory disease in the intestine. The disease is divided into ulcerative colitis (UC) and Crohn's disease (CD). IBD is defined by clinical symptoms and radiological, endoscopic, and histopathological findings [1,2]. Although its pathogenesis is still unclear, many studies have suggested it is a multifactorial disease involving genetic and environmental factors that cause an abnormal immune response to the gut microflora [3]. Differences in the histopathology of CD and UC clearly exist. Where UC is characterized by diffuse inflammation confined to the mucosa of the colon and rectum [2], the inflammation of CD is discontinuous and transmural and can affect the entire gastrointestinal tract [1]. Patients with CD often present complications like intestinal strictures, fistulas, and abscesses which make disease management challenging [1]. Within the last 15 years, tumour necrosis factor α (TNFα) antagonists have transformed the medical treatment of moderate to severe CD, even though a significant proportion of patients with CD do not respond adequately to treatment with these agents. Primary and secondary nonresponders to anti-TNFα therapy present a clinical challenge and require dose adjustment or switch to another medication. The efficacy of the treatment of anti-TNFα-resistant CD appears to be quite modest [4]. Thus, new drug candidates are clearly needed for the anti-TNFα nonresponders.
Disruption of the CD4 + T cell balance is an important part of CD pathogenesis [5,6], and cytokines are principal mediators orchestrating this disturbance [5]. Thus, various cytokines have been targeted or suggested as targets for the treatment of CD [7].
IL-21 is predominantly expressed by activated Th17 cells [8,9], T follicular helper cells [10] and NKT cells [11,12]. IL-21 signals through a receptor composed of a specific subunit, termed IL-21R, and the common γ-chain subunit, shared with IL-2, IL-4, IL-7, IL-9, and IL-15 receptors [13,14]. IL-21R is highly expressed on T and B lymphocytes [15,16], natural killer (NK) cells, and some nonhematopoietic cells, all of which functionally respond to IL-21 [13,17]. Previous studies have shown that IL-21 is important in the regulation of immunoglobulin synthesis by plasma cells [18], cytotoxic activity of NK cells, and CD8 + T cells [17]. As well, IL-21 has been reported to be a key modulator of TGF-β signalling, leading to reciprocal differentiation of Treg and Th17 cells [19,20]. During T cell differentiation, IL-21 is believed to sustain IL-23R expression, which then allows increased cellular response to IL-23 [11]. IL-21 may also act through the induction of the transcription factor retinoic acid receptorrelated orphan receptor (RORγt) [8,21]. In IBD patients, IL-21 has been reported to be expressed by IFN-γ-producing CD4 + T cells [22], as well as Th17 cells [6,23]. In the gut, Th17 cells can have tissue-protective effects and enhance epithelial barrier function, as well as increasing the synthesis of extracellular matrix-degrading proteases, inflammatory cytokines, chemokines, and adhesion molecules and recruitment of neutrophils [21,24]. In patients with CD, some of the IL-17A-producing cells coexpress IFN-γ and/or FoxP3 and may undergo transformation into typical Th1 or Treg cells dependent on the costimulatory environment [25,26].
In IBD, it is currently unknown whether IL-21 directly affects Treg proliferation, differentiation, and suppression or reduces the frequency of FoxP3-positive cells by promoting Th17 cells.
Enhanced expression of IL-21 and/or IL-21R has been documented in various diseases including CD, UC, and celiac disease [11,22,24,27,28]. The upregulation of T cellsecreted IL-21 has in particular been described in patients with CD and celiac disease [22,27]. Traditionally, CD can be categorized into ileum, ileocolonic, and colon subtypes, based on disease location [1]. Still, these subtypes behave similarly in terms of IL-21 upregulation [29], indicating that upregulation of IL-21 may be a general mechanism in CD. Moreover, there is evidence that the blockade of IL-21 signalling by either gene-targeting studies or IL-21R-Fc treatment limits the severity of inflammation in experimental models of immune-mediated diseases [11,[30][31][32][33][34]. In these studies, IL-21 has been described to work through a modulation of immune cell trafficking and downregulation of fibroblastsecreted matrix-degrading enzymes in the gut [35,36]. Mutations in IL-21 or IL-21R seem to work differently in humans, where severe immune dysregulation characterized by infectious susceptibility to opportunistic pathogens and, paradoxically, autoimmunity has been reported in a few cases [37][38][39]. The pleiotropic effect of IL-21 is also reflected in a range of clinical trials where IL-21 showed immune stimulatory effects, acceptable toxicity, and antitumour effects in a fraction of cancer patients [40]. Anti-IL-21 mAb treatment has been evaluated in rheumatoid arthritis, systemic lupus erythematosus, and CD [11]. More recently, the role of anti-TNFα mAb in regulating IL-21 expression and Th17 cell infiltration in the intestinal mucosa of CD patients was explored [41]. IL-21 and Th17 cells were found to be highly expressed in the inflamed mucosa of active CD patients compared with healthy controls. Ten weeks after anti-TNFα mAb infusion, intestinal mucosal healing was improved in CD patients and IL-21 expression and Th17 cell infiltration were found to be significantly decreased [41]. Similarly, IL-21 transcripts in the intestinal tissue from CD patients have been reported to be significantly downregulated in anti-TNFα responders, but not in nonresponders [42]. Thus, IL-21 seems to follow CD pathogenesis and could be a potential target in both anti-TNFα mAb responders and nonresponders.
We demonstrate that in human intestinal material IL-21 and IL-21R expression follows the hallmark of CD, the transmural inflammation, and that IL-21R is expressed by subsets of T cells, B cells, and plasma cells located in particular germinal centers (GCs) of lymphoid aggregates and in those infiltrating the intestinal wall in general.
We also demonstrate that IL-21 neutralization only is efficacious in models with CD4 T cell-driven colitis and primarily affects homing of calprotectin-positive cells. Approximately, 100 different anti-human IL-21 antibodies (commercially available and in-house produced) were evaluated on human IL-21 or mock transfected cells and positive control tissue (tonsil). One was found to be specific for recombinant human IL-21 NNC0114-0000-0032.

Materials and Methods
2.6. In Situ Hybridization on Human Tissue. ISH was performed on frozen sections as previously described with minor modifications [43,44]. The protocol is described in S1.
2.7. Immunohistochemistry on Human Tissue. The sections were deparaffinised in xylene and rehydrated in decreasing concentrations of alcohols. Antigen retrieval was performed in Tris-EGTA buffer (10 mM; 0.5 mM), pH 9.0 in a microwave oven for 15 min. Endogenous peroxidase activity was blocked with 3% H 2 O 2 and dual block (Dako, S2003). Endogenous biotin was blocked by incubation with avidin and biotin blocking solutions for 10 min, according to the manufacturer. Nonspecific binding was blocked by incubation with TBS containing 3% skimmed milk, 7% donkey serum, 3% human serum, and 3.2 mg/ml poly-L-lysine (PLL) for 30 min. The primary and secondary antibodies were diluted in a Tris buffer containing 0.5% skimmed milk, 7% donkey serum, 3% human serum, and 3.2 mg/ml PLL, and incubation was performed overnight at 4°C and 60 min at room temperature, respectively. The first amplification step was performed by incubation with Vectastain ABC peroxidase kit, diluted in 0.1 M Tris-HCl buffer (pH 7.5) containing 0.5% Du Pont blocking reagent (TNB) for 30 min, followed by a second amplification step with incubation in biotinylated tyramide for 6 min. The final amplification was performed by an additional incubation with the Vectastain ABC peroxidase kit, diluted as previously described for 30 min. The chromogenic reaction was achieved with diaminobenzidine. Nuclei were counterstained with haematoxylin, and the sections were rehydrated, cleared in xylene, and mounted with Pertex. The evaluation of the slides was performed by the blinded observer. The sections were evaluated by virtual microscopy using a Hamamatsu NanoZoomer (Hamamatsu Denmark, Ballerup, Denmark) and conventional microscopy using an Olympus BX51 microscope equipped with a DP70 digital camera (Olympus Denmark A/S, Ballerup, Denmark).

Double
Immunofluorescence on Human Tissue. The immunohistochemical demonstration of the IL-21R was in general performed as described above. Upon the second amplification with biotinylated tyramide for 3 min, sections were incubated with the Vectastain ABC peroxidase kit, diluted as previously described for 30 min. The final amplification was performed by incubation with Alexa-594 conjugated tyramide diluted according to the manufacturer for 6 min. The sections were then subjected to second rounds of microwave oven treatment in Tris-EGTA buffer (10 mM; 0.5 mM), pH 9.0 for 10 min, blocking of endogenous peroxidase activity with 3% H 2 O 2 , and endogenous biotin by incubation with avidin and biotin blocking solutions for 10 min, respectively, according to the manufacturer. Additional blocking of nonspecific binding was performed by incubation with TBS containing 3% skimmed milk, 7% donkey serum, and 3% human serum for 30 min. The second primary and secondary antibodies were diluted in a Tris buffer containing 0.5% skimmed milk, 7% donkey serum, and 3% human serum, and incubation was performed overnight at 4°C and 60 min at room temperature, respectively. The final amplifications were performed by an additional incubation with the Vectastain ABC peroxidase kit, diluted as previously described for 30 min, and incubation with Alexa-488 conjugated tyramide. The nuclei were counterstained with Hoechst for 15 min, and the sections were finally mounted in fluorescence mounting media (Dako). The sections were evaluated by virtual microscopy using a Hamamatsu NanoZoomer (Hamamatsu  [45] as follows: 10-12 mice per cage and a 12-hour light/dark cycle. 90 percent of the cage bedding was changed weekly, and 10 percent of the cage bedding was transferred between cages in order to ensure a homogenous microbial environment. The clinical status of the mice was evaluated three times weekly by visual inspection, percentage weight change, and faecal consistency.
2.11. Induction of Adoptive Transfer Colitis Using CD4 + CD25 − T Cells. The transfer colitis with CD4 + CD25 − T cells was conducted as previously described [46,47]. The protocol is described in S1 and S6.
2.12. Induction of Adoptive Transfer Colitis Using CD4 + CD25 − CD45RB high T Cells. The protocol is described in S1 and S6.
2.14. Induction of DSS Colitis. Induction of dextran sulphate sodium (DSS) colitis was performed as previously described [50,51]. The protocol is described in S1 and S6.
2.15. Monitoring of Disease in Mice. Body weight was monitored three times weekly, and the mice were sacrificed if they lost more than 20% of their initial body weight. Faecal consistency was evaluated before the start of the treatment and subsequently 3 times a week using a semiquantitative score (normal stool = 0; slightly soft = 1; soft but formed = 2; not formed = 3; and liquid stools or no faeces in colon at sacrifice = 4). Disease activity index (DAI) score was calculated as previously described [48,49,51].
2.16. Collection of Serum and WBC Counts from Mice. During the experiment, the mice were anesthetized with isoflurane, and blood from the periorbital venous plexus was collected in EDTA-containing tubes for white blood cell (WBC) counts, FACS analysis, and exposure. The number of WBC per liter was analysed in samples (20 μl) of EDTAstabilized peripheral whole blood, using a Medonic CA 620 (Boule Nordic, Denmark) blood analysis apparatus according to the manufacturer's instructions. Serum samples were collected directly in microtainer SST tubes (BD, US). The isolated serum samples were then transferred to micronic tubes and stored at −80°C.
2.17. Flow Cytometry of Blood Samples from Mice. The flow cytometry analysis of mouse blood samples was performed as described in S1 and S6.

Exposure and Pharmacokinetics in Mice.
Exposure and pharmacokinetic analysis was performed as described in S1 and S6.

Endoscopic Analysis on Mice.
Mice were anaesthetized with isoflurane and placed on the back. A rigid telescope (HOPKINS straight forward, 0°) was connected to a light source/air pump (Xenon 175) and a camera (Telecam SL) all purchased from Karl Storz (Tuttlingen, Germany), as well as a monitor (Sony Triniton) and a video recorder (SCB aida DVD, Karl Storz) as described by Becker et al. [52]. The endoscope was coated with a lubricant containing lidocaine hydrochloride (Farco-pharma, Köln, Germany) and introduced via the anus into the distal 4 cm of the colon. The evaluation of the colonoscopy findings was done by two blinded observers using the Murine endoscopic index of colitis severity (MEICS) score as previously described in [52].

Post Mortem Analyses on Mouse
Tissue. Mice were sacrificed by cervical dislocation, and the caecum, colon, and rectum were excised. The length of the colon was measured from the caeco-colonic junction to the anus. The colon was rinsed with PBS and weighed. The colonic weight : length (W : L) ratio (cm/g) was used as a macroscopic, objective parameter to verify the presence of established colitis, since it is known to correlate with the histology score [48]. The proximal 1/3 of the colon was removed, and the remaining 2/3 of the colon was bisected or trisected longitudinally. The dissected colon biopsies were processed for histological analysis, qPCR, and cytokine/chemokine profiling.

Histological Analysis on Mouse
Tissue. Histological analyses on mouse tissues were performed as described in details in S1 and S6. The severity of the histopathological lesions of colon segments was examined in a blinded manner, using the criteria previously described in [46, 48, 51, 53].

IL-21 Intracellular Cytokine Staining of Mesenteric
Lymph Node (mesLN) CD4 + T Cells. On day 7, 16, 29, and 38 post cell transfer, mesenteric lymph node (mesLNs) and colons were aseptically removed from 5 mice per time point. Three (3) naïve BALB/c mice were included as controls. Restimulation mesLN lymphocytes or LPLs were stimulated for 4 h at 37°C/5% CO 2 with PMA/ionomycin in vitro in the presence of GolgiSTOP (w/monensin) and GolgiPLUG (w/Brefeldin A). Following stimulation, cells were surface stained with murine Fc Block (α-CD16/32) followed by surface staining with the antibodies indicated in the following: anti-mouse CD45 Qdot605, anti-mouse TCRb FITC, antimouse CD4 PE, and anti-mouse CD8 Pacific Blue, NEAR-IR (APC-Cy7). Subsequently, the cells were washed and fixed in 4% PFA. The fixed cells were permeabilized by Perm/ Wash buffer. Hereafter, either anti-mouse IL-21 or mouse IgG1 conjugated to AlexaFlour647 was added.

Statistical Analyses
2.24.1. Human Study. The semiquantitative scoring of the immunohistochemical data for IL-21R protein expression was analysed by Mann-Whitney's two-tailed nonparametric t-test in GraphPad Prism 5, and p < 0 05 was considered significant.

Experimental Mouse Studies.
In the experimental colitis studies, GraphPad Prism version 6.0 was used for all statistical analyses. The statistical significance of differences between normally distributed, parametric data from the two groups was evaluated using unpaired Student's t-test. The comparison of nonparametric, non-Gaussian distributed data from the two groups was performed using a nonparametric Mann-Whitney U test. The statistical significance between more than two groups was evaluated by a one-way ANOVA or Kruskal-Wallis test with Holm-Sidak's or Dunn's multiple comparison test, respectively. Prophylactic treatment studies were evaluated by two-sided tests, whereas therapeutic studies were evaluated by one-sided tests.

Results
IL-21 and IL-21R mRNA and protein expression analyses were assessed on frozen OCT-embedded and fixed paraffin-embedded intestinal samples from non-IBD controls histopathologically within normal limits and from patients with CD, respectively. The intestinal types included in each study are listed in Table 1. 3.1. IL-21 and IL-21R Expression in the Intestine from Patients with CD. As summarized in Table 1, IL-21 and IL-21R mRNA and protein-expressing cells were observed in IL-21R mRNA expression in the intraepithelial lymphocyte (IEL) compartment (e.g., surface epithelium including intraepithelial lymphocytes) could not be clearly demonstrated.
As summarized in Table 2, IL-21 and IL-21R immunopositive immune cells were present in intestinal non-IBD samples within normal limits in the mucosa and submucosa. In patients with CD, additional expression was observed within lymphoid aggregates of the submucosa and in the muscularis externa, with abundant expression in GCs. Thus, the general compartmentalization of IL-21 and IL-21R immunopositive cells resembled that described for IL-21 and IL-21R mRNA expressing cells, that is, within the IEL compartment, in follicle-associated epithelium and as solitary cells in the lamina propria of the mucosa (Figures 3(a)  Finally, IL-21R immunopositive cells were generally more abundantly present than IL-21 immunopositive cells. These IL-21 immunopositive cells were present within the same areas as IL-21R immunopositive cells throughout the different immune compartments of the intestine from patients with CD. No immunoreactivity was observed with the isotypespecific control antibody (Figures 3(g) and 3(h)).

Dark field
Bright field  The semiquantitative analysis was performed on mucosaassociated lymphoid tissues identified according to the revised nomenclature assigned by Brandtzaeg et al. [54] and semiquantitatively scored as described in Materials and Methods.
Based on the semiquantitative score the IL-21R was significantly highly expressed (p < 0 05) in the intestine from patients with CD compared to non-IBD control intestinal samples within normal limits (Figure 4(a)). The IL-21R expression did not reach significance in the mucosa (p = 0 47) (Figure 4(b)), but did show a significant difference in the submucosa (p < 0 01) (Figure 4(c)) and muscularis externa (p < 0 01) (Figure 4(d)).
There was no significant difference in the IL-21R expression in the IEL compartment in samples from patients with CD compared to non-IBD controls (data not shown). The study on IL-21 protein expression was not powered to perform a semiquantitative analysis of IL-21 expression. IL-21R + CD3 + T cells were present in the mucosa, submucosa ( Figures 5(a)-5(c)), and muscularis externa, interspersed in general in the different intestinal layers and in particular within lymphoid aggregates. Likewise, IL-21R + CD20 + B cells were found in the mucosa, lymphoid aggregates, and particularly GCs of the submucosa and muscularis externa from patients with CD ( Figures 5(d)-5(f)). In addition, IL-21R + CD138 + plasma cells were observed only in the mucosa in the normal intestine (data not shown), whereas CD IL-21R + plasma cells were also present in lymphoid aggregates of the submucosa (Figures 5(g)-5(i)) and muscularis externa. Finally, IL-21R + CD68 + macrophages were present in the mucosa and lymphoid aggregates of the submucosa (Figures 5(j)-5(l)) and muscularis externa from patients with CD as well as in non-IBD controls (data not shown). The study was not powered to perform a quantitative analysis of the individual subsets of IL-21R + cells.
3.3. IL-21-Producing CD4 + T Cells Are Upregulated in the CD4 + CD25 − T Cell Adoptive Transfer (AdTr) Colitis Model. Because IL-21 is reported to be produced by both human and mouse Th17 cells [6,24] and since Th17 cells have been shown to be involved in adoptive transfer colitis, we initially focused on the CD4 + CD25 − T cell transfer model of colitis. Moreover, before treatment, studies were initiated; target validation was conducted to describe the dynamics of IL-21 protein expression by CD4 + T cells during the development of colitis.
The percentage of IL-21 + CD4 + T cells was shown to increase over time, in both the mesLN and the lamina propria (Figures 6(a) and 6(b)). As well, the percentage of mesLN IL-21 + CD4 + cells correlated with the degree of body weight change in individual mice with colitis, indicating that IL-21 may be involved in colonic disease development in this model (Figure 6(c)).

Pharmacokinetics of Anti-IL-21 mAb in Mice.
In a pilot study, we determined the exposure and the pharmacokinetics of our mouse anti-IL-21 mAb and performed modelling using a 1-compartment model. Exposure from AdTr colitis mice using two selected doses of mouse anti-mouse IL-21 (3.3 mg/kg and 25 mg/kg) was compared with the simulated exposure levels based on pharmacokinetics in healthy NMRI mice. At both doses (3.3 and 25 mg/kg), circulating unbound mouse anti-IL-21 mAb was detected ( Figure S1) and matched the simulated exposure. The 25 mg/kg dose interval was selected in subsequent studies in order to have the highest possible peripheral exposure.
3.6. Prophylactic Treatment with Anti-IL-21 mAb Ameliorates CD4 + CD25 − T Cell AdTr Colitis. Next, we wanted to assess whether treatment with 25 mg/kg anti-IL-21 mAb could prevent adoptive transfer colitis, compared to mice treated with mIgG1. The change in body weight over the course of the experiment is shown in Figure 8(a). As previously described [46,47], mice treated with mIgG1 started to lose body weight in the 3rd-4th week following transfer. In comparison, mice treated with anti-IL-21 mAb showed some weight loss in week 3, but stabilized at a level above the isotype control. The difference did not reach statistical significance (P < 0 13) at study termination.
Blood samples were obtained on day 28 in order to evaluate whether anti-IL-21 merely prevents T cell proliferation. Mice treated with anti-IL-21 mAb did not show any reduction in WBC counts compared to control treated mice (Figure 8(b)). Neither analysis of the peripheral blood did reveal any changes in CD4 + T cell frequencies (Figure 8(c)).    In the AdTr colitis model, the disease can be associated with thickening of the colon wall. This is related to the infiltration of donor T cells and the resulting host inflammatory response. The host response includes infiltration of additional inflammatory cells (e.g., neutrophils and macrophages), vascular leakage resulting in intestinal oedema, and varying degrees of mucosal epithelial hyperplasia, all of which contribute to the thickness of the wall of the colon. In turn, there is a compensatory shortening of the colon. The ratio of colon weight (W) to length (L) provides a colon thickening index, which correlates with colitis severity and histopathology. Mice treated with mouse anti-IL-21 mAb had a significant reduction in the colon W : L ratio, compared to mIgG1-treated mice (P < 0 02) (Figure 8(d)).
To acquire a translational clinical endpoint, a 2 mm endoscope was introduced via the mouse anus. Endoscopic pictures were obtained to allow the monitoring and grading of inflammation. The grading uses the murine endoscopic index of colitis severity (MEICS) score encompassing (thickening of the colon, changes of vascular pattern, visible fibrin, granularity of mucosal surface, and stool consistency). Mice treated with the anti-IL-21 mAb had a significant reduction in their endoscopic score on day 28, compared to mice treated with mIgG1 (P < 0 01) (Figures 8(e) and 8(f)).
A significant reduction in the histopathology score was also observed in mice treated with anti-IL-21 mAb compared to mIgG1 (Figures 9(a)-9(c)). The analysis of the inflammatory subsets by immunohistochemistry revealed that only calprotectin-positive cells (neutrophils and macrophages) were significantly reduced (P < 0 006), whereas no significant effect could be detected in the CD3 density (Figures 9(g)-9(i) and Figures 9(d)-9(f), resp.). We also evaluated changes in mRNA transcripts (Th1, Th2, Th17, and Treg) in colonic biopsies. The only transcript that was significantly downregulated following the anti-IL-21 mAb treatment was GATA3 (Supplementary Figure 2). Anti-CD3 10 ng/ml −

Interventive Treatment with Anti-IL-21 mAb Has a
Moderate Effect on Disease Parameters in the CD4 + CD25 − T Cell AdTr Model. We have previously shown that mild to moderate colonic inflammation is present at day 21 [47]. Thus, we randomized our treatment groups according to weight on day 21. Anti-IL-21 mAb treatment had a significant effect on body weight loss over the treatment period (Figure 10(a)). However, even though there was a tendency to lower colonic disease score in mice treated with anti-IL-21 mAb, no significant effect could be detected on WBC counts, CD4 + TCR + T cell frequency, colon W : L ratio, and endoscopic score (Figures 10(b)-10(e)). Colonic inflammation was further assessed by the evaluation of the histopathology score, density of CD3, and calprotectin staining. Similar to the prophylactic study, both histopathology score (Figures 11(a)-11(c)) and calprotectin density (Figures 11(g)-11(i)) were significantly reduced following anti-IL-21 mAb treatment, whereas no significant effect on CD3 density could be detected (Figures 11(d)-11(f)). Cytokine and chemokine profiling of colon biopsies from mice confirmed a significant interventive treatment effect of anti-IL-21 mAb on chemokine and Th17-related parameters (KC, CCL3, CCL5, IL-17, TNFα, IFN-γ, CCL2, and IP-10) ( Figure 12). However, TNFα, IFN-γ, CCL2, and IP-10 were only borderline significant and determined by few mice with high colonic levels. Four SCID control mice were included in the analysis to show the cytokine and chemokine background level. In the isotype-treated mice all cytokines and chemokines besides CCL3 were elevated compared to SCID control mice.

Ablation of IL-21
Signalling Ameliorated Colitis in the CD4 + CD45RB high Model. Since absolute neutralization of cytokine signalling on CD4 + T cells may be difficult to achieve with an antibody, we tested whether naïve CD4 + T cells would require the IL-21R to develop chronic mucosal inflammation in the CD4 + CD45RB high model using B6.129S6-RAG2 −/− mice as recipients. In short, no effect was detected on body weight change, but mice transferred with CD4 + CD45RB high IL-21R −/− T cells had a significantly lower colon W : L ratio as well as reduced numbers of granulocytes in the blood, compared to mice transferred with CD4 + CD45RB high IL-21R +/+ T cells (Figures 13(a)-13(c)). Moreover, no significant difference between Tregs expressing the IL-21R and Tregs lacking the IL-21R could be detected in cotransfer experiments. However, CD4 + IL-21R +/+ Tregs did reduce the colon W : L ratio significantly when compared to mice only transferred with CD4 + CD45RB high IL-21R +/+ T cells (Figure 13(c)).  First, the frequency of CD4 + T cells expressing IL-21 was determined in mesLN. As seen in Figure 14(a), the percentage of IL-21 + CD4 + T cells was significantly elevated in PAC IL-10 k.o. compared to IL-10 k.o. littermates. Still, the percentage of IL-21 + CD4 + T cells was lower compared to the AdTr colitis model ( Figure 6). Following prophylactic treatment with anti-IL-21 mAb, no significant treatment effect could be observed on systemic parameters. At day 18, the mean body weight change of anti-IL-21 mAb and mIgG1treated mice was minus 5%, and no significant difference in weight change could be observed (Figure 14(c)). The systemic serum level of the acute phase protein (haptoglobin) level at day 11 was unchanged ( Figure 14(d)). As well, anti-IL-21 mAb treatment had no effect on colonic parameters such as the colon W : L ratio at termination or endoscopic score day 12 (Figures 14(e) and 14(f)). The only indication of a positive treatment effect was a significant reduction in colonic MPO levels ( Figure 14(b)). Indicating a potential effect on neutrophils and macrophages.

Discussion
IL-21 has been shown to be upregulated in the intestine from patients with IBD in several studies by different methods such as Western blotting of total protein extracts from tissue specimens, culture of lamina propria mononuclear cells (LPMCs) followed by ELISA on supernatants, intracellular flow cytometry on LPMCs [22,55,56], and immunohistochemistry of mucosal biopsies [57]. In the present study, using resected intestinal human samples, the presence of IL-21 mRNA and protein-expressing cells was described in different immune cell compartments in the intestine from patients with CD. IL-21 mRNA was primarily expressed by solitary cells in the IEL compartment, lamina propria, and follicle-associated epithelium of the mucosa, in isolated lymphoid follicles in the submucosa as well as in infiltrates in the muscularis externa. Upregulation of the IL-21R in the intestine from patients with IBD has likewise been demonstrated by different groups using immunohistochemistry [35,56]. In the present study, these data were confirmed, as the IL-21R was significantly upregulated in the intestine from patients with CD compared to non-IBD control samples. Moreover, the IL-21R mRNA expression was similar to the IL-21R protein expression pattern in the immune cell compartment characteristic of the intestinal wall. Specifically, IL-21R mRNA and protein were expressed by solitary immune cells in the lamina propria and follicle-associated epithelium of the mucosa, in isolated lymphoid follicles in submucosa, as well as in cellular infiltrates of the muscularis externa. Moreover, increased IL-21R protein expression was restricted to the submucosa and muscularis externa. Thus, IL-21 and IL-21R expression followed the histopathological hallmark of CD, that is, the transmural inflammation that was observed in the resected material investigated herein as compared to studies performed on mucosal biopsies [35,56,57]. Furthermore, we  identified the IL-21R + cells as the subsets of T cells, B cells, plasma cells, and macrophages; however, the study was not powered to detailed quantitative analyses.
In this context, one may speculate that the expression of the IL-21R on CD20 + B cells, CD138 + plasma cells, and the CD68 + macrophages directly and/or indirectly has an effect on T cell activation, B cell maturation, and immunoglobulin production from plasma cells. Whether a similar situation applies to patients with UC remains to be examined in full size biopsies. However, it may be hypothesized that the effect of IL-21 in patients with UC is related to the cytokine pleiotropic effect on neutrophils and/or macrophages, a hallmark for the inflammation in these patients.
Previous studies have shown that mice lacking IL-21 are unable to upregulate Th17-associated molecules during experimental colitis and are largely protected from DSSinduced colitis [31,34,58,59]. Similarly, immunodeficient mice adoptively transferred with CD4 + IL-21 −/− T cells develop less severe colitis [60,61], and mice treated in a prophylactic setup with a neutralizing IL-21R/Fc fusion protein or anti-IL-21 mAb are protected from both DSS and TNBS-induced colitis [31,58,59]. In our AdTr study with CD4 + CD25 − T cells, we observed a clear production of IL-21 from the lamina propria and mesenteric lymph node CD4 + T cells. Moreover, we observed a reduction of colonic inflammation following both prophylactic and interventive treatments with 25 mg/kg anti-IL-21 mAb. This treatment effect was not associated with a noticeable reduction in WBC, CD4 + T cell frequency, or colonic CD3-infiltrating T cells, suggesting that the effects of neutralizing IL-21 in this model may not depend on the prevention of T cell proliferation or migration. However, we observed a reduction in colonic calprotectin (marker of neutrophils), several chemokines, and proinflammatory cytokines, supporting the idea that IL-21 neutralization in vivo not only affects CD4 + T cell differentiation but also regulates neutrophil infiltration through direct or indirect mechanisms. To further dissect the importance of IL-21 signalling in innate lymphoid cells and nonimmune cells, future studies should investigate the in vivo effects of transferring CD4 + T cells to IL-21R −/− Rag2 −/− recipient mice. In our piroxicam-accelerated colitis study, where colitis symptoms are more acute and not directly driven by CD4 T cells [48,49], we did not observe any protective effect of IL-21 neutralization. However, if CD4 + T cells are the primary producer of IL-21, this may not be surprising.
The underlying mechanism of IL-21 signalling in experimental colitis is still unclear. Most groups describe a downregulation of Th17 cells and associated molecules [31,34,58,59], while other groups report that IL-21 is not essential for Th17 cell development [60][61][62]. We explored the direct effect of IL-21 signalling in CD4 + T cells by transferring either CD45RB high IL-21R −/− CD4 + T cells or CD45RB high IL-21R +/+ CD4 T cells to Rag2 k.o. mice. This concept is clearly different from transferring CD4 + T cells lacking IL-21 or IL-21 neutralization by treatment, since other cell types, for example, innate lymphoid cells or fibroblasts may respond to IL-21 secreted from the CD4 + IL-21R −/− T cells. In our study, we observed less severe colitis in Rag2 k.o. mice transferred with CD45RB high IL-21R −/− CD4 + T cells compared with mice transferred with CD45RB high IL-21R +/+ CD4 + T cells. In addition, mesLN CD4 + T cells isolated from mice with colitis responded with excessive production of proinflammatory cytokines and chemokines following stimulation with exogenous IL-21. The observation supports that IL-21 signalling may play a part in the initial development and differentiation of pathogenic CD4 + T cells. However, we could not detect a marked reduction in proinflammatory cytokine or chemokine production following in vitro neutralization of endogenous IL-21 production by mesLN CD4 + T cells. Thus, only high IL-21 levels seem to mediate the secretory effect. In addition, we could not detect any significant changes in Th17, Th1, or Treg transcription factors in colonic biopsies; only Th2 transcripts were downregulated by anti-IL-21 mAb treatment.
Moreover, several studies suggest that IL-21 mediates inhibitory effects of peripheral differentiation of Tregs and makes CD4 + T cells resistant to Tregs-mediated immune suppression [34,63]. However, in our CD45RB high CD25 +-CD4 + T cell transfer study, only minor effects of IL-21R ablation could be detected on Treg frequency or in vivo suppressive function. We speculate that IL-21 in our model primarily affects the differentiation of Th17 in mesLN rather than affecting Tregs directly.
Recently, Wang et al. [64] reported that IL-21/IL-21R signalling may actually suppress intestinal inflammation induced by DSS through regulation of Th responses. Still, in most studies, IL-21 −/− mice have been reported to be protected from DSS-induced colitis [31,34,58,59]. In our facility, IL-21R −/− mice are not protected from DSS-induced colitis neither in an acute nor in a chronic setup ( Supplementary Figures 3  and 4). The discrepancy between colitis studies with IL-21 or IL-21R −/− mice may be due to variability in microflora between laboratories, T cell dependency in the model setup, or a consequence of transcriptional changes in the areas around the targeted gene in the k.o. mice. However, it also highlights the complex biology associated with IL-21 signalling.
IL-21 is overproduced in many chronic inflammatory disorders, and studies in experimental models indicate that IL-21 plays an important role in sustaining tissue damage. Moreover, since IL-21 is a pleiotropic cytokine and since CD most likely is caused by multiple immunological subsets, the stimulatory role of IL-21 on non-CD4 + T cells, for example, neutrophils, IgA-producing B cells, fibroblasts, and macrophages, should be further investigated.

Conflicts of Interest
All authors are present or past employees at Novo Nordisk A/S.