Pathogenic T helper type 17 cells contribute to type 1 diabetes independently of interleukin‐22

Summary

We have shown that pathogenic T helper type 17 (Th17) cells differentiated from naive CD4⁺ T cells of BDC2·5 T cell receptor transgenic non‐obese diabetic (NOD) mice by interleukin (IL)‐23 plus IL‐6 produce IL‐17, IL‐22 and induce type 1 diabetes (T1D). Neutralizing interferon (IFN)‐γ during the polarization process leads to a significant increase in IL‐22 production by these Th17 cells. We also isolated IL‐22‐producing Th17 cells from the pancreas of wild‐type diabetic NOD mice. IL‐27 also blocked IL‐22 production from diabetogenic Th17 cells. To determine the functional role of IL‐22 produced by pathogenic Th17 cells in T1D we neutralized IL‐22 in vivo by using anti‐IL‐22 monoclonal antibody. We found that blocking IL‐22 did not alter significantly adoptive transfer of disease by pathogenic Th17 cells. Therefore, IL‐22 is not required for T1D pathogenesis. The IL‐22Rα receptor for IL‐22 however, increased in the pancreas of NOD mice during disease progression and based upon our and other studies we suggest that IL‐22 may have a regenerative and protective role in the pancreatic islets.

Introduction

In type 1 diabetes (T1D) both T helper type 1 (Th1) and Th17 cells contribute to the disease development 1, 2. We have shown recently that pathogenic Th17 cells derived from BDC2·5 T cell receptor transgenic non‐obese diabetic (NOD) mice polarized with interleukin (IL)‐23 and IL‐6 produce IL‐17 and IL‐22 and transfer disease adoptively in NOD mice 3. It is not yet clear if IL‐22 has any role in the pathogenic function of Th17 cells in T1D. IL‐22 is involved in the process of cell regeneration, maintains the integrity of the epithelial cell layer, contributes to tissue repair and induces anti‐bacterial peptides in epithelial cells 4, 5, 6. Targets of IL‐22 are mainly non‐haematopoietic cells, including intestinal epithelial cells. Of note, the receptor for IL‐22 is expressed highly in the pancreas in both beta and alpha cells 7, 8, 9. We hypothesized that IL‐22 might contribute to the pathogenic potential of Th17 cells in diabetes development. We therefore explored protocols to maximize IL‐22 production by the pathogenic Th17 cells. We also isolated IL‐22‐producing Th17 cells and characterized the expression of IL‐22 receptor in the NOD mouse pancreatic islets. The functional effect of IL‐22 in diabetes progression was explored with anti‐IL‐22 monoclonal antibody combined with a disease‐inducing adoptive transfer model. Neutralization of IL‐22 per se in the course of adoptive transfer of Th17 cells did not reduce the pathogenic potential of these Th17 cells. Therefore, IL‐22 produced by pathogenic Th17 cells plays a redundant role in T1D pathogenesis. Conversely, we and others have found that the receptor for IL‐22 increased in the pancreas of NOD mice during disease progression and IL‐22 may have a regenerative and protective role in the pancreatic islets 10, 11.

Materials and methods

Mice

NOD/Ltj and BDC2.5 TCR transgenic (Rag^+/–) NOD mice were obtained from the Jackson Laboratory (Bar Harbor, ME, USA). Mice were bred and housed in a pathogen‐free environment at the animal care facility of the University of Western Ontario (London, Canada) and both BDC2·5 T cell receptor (TCR) transgenic (Rag^+/+ or Rag^+/–) NOD mice were used for these studies. C57BL/6 (B6) mice were generously provided by Dr Mansour Haeryfar from our Department. All experiments were performed according to institutional guidelines and those of the Canadian Council for Animal Care. Mice were monitored for disease development by measuring urine glucose output with Diastix strips (Bayer, Elkhart, IN, USA). Mice were considered diabetic after two consecutive positive (>11·5 mmol/l) urine glucose tests, and where needed diabetic NOD mice were used within 2 weeks of the diagnosis of disease for tissue or lymphocyte isolation.

Cytokines and antibodies

Murine cytokines IL‐6 and IL‐23 were purchased from BioLegend (San Diego, CA, USA). All cytokines were reconstituted and used according to the manufacturer's instructions. The following anti‐mouse antibodies were purchased from BioLegend: anti‐CD3ε (clone 145‐2C11) was used to coat 24‐well plates overnight in 1 ml sterile 1× phosphate‐buffered saline (PBS) at 4°C; anti‐CD28 (clone 37·51) was added to cultures on anti‐CD3 coated plates; anti‐interferon (IFN)‐γ (clone XMG1·2) was added to splenic or T cell cultures as required. The following anti‐mouse, fluorophore‐conjugated antibodies were purchased from eBioscience: anti‐CD4‐fluorescein isothiocyanate (FITC) and anti‐allophycocyanin (APC), anti‐CD8‐FITC, anti‐phycoerythrin/cyanin7 (PE‐Cy7) or ‐APC, anti‐IFN‐γ‐FITC, anti‐IL‐22‐PE, anti‐IL‐17A‐APC, anti‐CD8‐PE, PE‐conjugated rat IgG1 isotype control and peridinin chlorophyll (PerCP)‐conjugated streptavidin were purchased from Becton‐Dickinson (BD, Franklin Lakes, NJ, USA). Anti‐CD4‐PE/Cy7 was purchased from BioLegend. For Western blotting, the primary antibody monoclonal rat anti‐mouse IL‐22Rα1 was purchased from R&D systems (Minneapolis, MN, USA) and polyclonal goat anti‐mouse actin was purchased from Santa Cruz Biotechnology (Dallas, TX, USA). Secondary antibodies used were horseradish peroxidase (HRP)‐conjugated goat anti‐rat immunoglobulin (Ig)G and HRP‐conjugated donkey anti‐goat IgG both purchased from R&D Systems.

Naive T cell isolation

Splenocytes from BDC2·5 mice were extracted and naive T cells isolated using kits from Miltenyi Biotec (Auburn, CA, USA) to isolate CD4⁺CD62L⁺ cells according to the manufacturer's guidelines. Briefly, magnetic labelling of CD4⁺ T cells and separation using an LS column led to the depletion of non‐CD4⁺ cells. Then, positive selection of CD62L⁺ cells from this fraction was performed using an MS column to achieve a highly enriched (>90%) sample of CD4⁺CD62L⁺ cells. These cells were then washed, counted and plated at 3 × 10⁶ cells per well in a 24‐well plate that had been coated overnight with anti‐CD3 and anti‐CD28. Cells were cultured for 4 or 5 days as stated in complete RPMI [RPMI‐1640 medium supplemented with 2 mM L‐glutamine, 0.5% HEPES, 5 μg/ml penicillin, 100 U/ml streptomycin and 10% (v/v) fetal calf serum (HyClone Laboratories, Logan, UT, USA]. In our experiments the non‐diabetic control NOD mice were the same age (18–25 weeks) as the diabetic NOD mice. The lymphocytes are derived mainly from the peri‐insulitic lesions, which are known to persist during the prediabetic and early diabetic states 1, 2.

In‐vitro stimulation of splenocytes

Splenocytes from BDC2·5 mice were extracted and seeded into a 96‐well plate at 2 × 10⁵ cells per well with 1 μM PS3 mimotope peptide, SRLGLWVRME, that induces proliferation in BDC2·5 T cells 12. This mimotope was synthesized, purified and characterized by mass spectrometry in our laboratory as described previously 13. Cytokines were added at the following concentrations: IL‐6 (20 ng/ml) and IL‐23 (20 ng/ml) similar to the Th17 induction concentrations used by Sugita et al. 14. Where needed, 5 μg/ml anti‐IFN‐γ antibody was added to cultures. Cells were cultured for 4 or 5 days as stated.

Flow cytometry

Fluorescence‐activated cell sorting (FACS) was used to determine cytokine profiles and surface marker expression of cells after in‐vitro culture. Cells in culture were stimulated with 50 ng/ml [phorbol myristate acetate (PMA)] and 500 ng/ml ionomycin for 4–6 h. Brefeldin A (5 μg/ml) was also added for the last 4 h. Cells were then harvested and supernatants collected and stored at −20°C for future applications. Cells were washed with 1× PBS followed by surface staining for 30 min at 4°C in 2% bovine serum albumin (BSA) in PBS with the FITC‐, PE‐, PE/Cy7 and APC‐labelled antibodies stated above. Cells were then washed twice with 1× PBS and fixed in 2% formaldehyde with 1% BSA in PBS for 20 min at room temperature. Cells were again washed and permeabilized with 0·5% saponin (Sigma, St Louis, MO, USA) in 2% BSA in PBS with antibodies such as those against IFN‐γ, IL‐22, IL‐17 or appropriate isotype controls either overnight or for 1 h. Cells were then washed with 1× PBS and data for 100 000 events (unless stated otherwise) were collected using a FACSCalibur (BD Biosciences) and analysis of data was performed using FlowJo software (Tree Star Inc., Ashland, OR, USA).

Enzyme‐linked immunosorbent assay (ELISA)

Duoset ELISA kits from R&D Systems were used to analyse supernatants from in‐vitro cultures for IL‐22 and IL‐17. The manufacturer's protocols were followed directly. Standard curves were generated for each plate to determine sample concentration. Absorbance was determined using a Benchmark Microplate reader (BioRad, Hercules, CA, USA) and data were analysed using Microplate Manager version 4·0 software (BioRad).

RNA extraction

For RNA extraction from whole pancreatic tissue, mice were killed and approximately 50 mg of pancreatic tissue was placed in buffer RLT (Qiagen, Mississauga, ON, USA) containing β‐mercaptoethanol (Sigma‐Aldrich, St Louis, MO, USA). Tissues were homogenized using a PowerGen 700 homogenizer (Fisher Scientific, Pittsburgh, PA, USA). Total RNA was then extracted using an RNeasy Midi Kit (Qiagen, Mississauga, ON, USA). For RNA extraction from splenocytes, peripheral lymph nodes (PLNs) or cultured lymphocytes, cells were disrupted in buffer RLT and β‐mercaptoethanol and homogenized by adding lysate to a QIAshredder spin column (Qiagen). Total RNA was then extracted using an RNeasy Mini Kit (Qiagen). Contaminating DNA was removed from all RNA samples using the DNase treatment and removal kit purchased from Ambion (Austin, TX, USA). The concentration of RNA in each sample was then determined by measuring absorbance at 260 nm using a NanoDrop 1000 spectrophotometer (NanoDrop Products, Wilmington, DE, USA).

Quantitative real‐time reverse transcription–polymerase chain reaction (qRT–PCR)

For quantification of specific genes using qRT–PCR, 1–2 μg RNA from each sample was reverse‐transcribed into first‐strand cDNA using oligo dT12–18 primers from Superscript II reverse transcriptase kit (Invitrogen, Carlsbad, CA, USA) and using a GeneAmp PCR System 2400 from Applied Biosystems (Foster City, CA, USA). The resultant cDNA was diluted in diethylpyrocarbonate (DEPC) water to a consistent concentration for each experiment, usually 225 ng/μl for RNA extracted from in‐vitro cultures, and 500 ng/μl for whole tissues. cDNA was then amplified using Quantifast SYBR Green PCR Kit (Qiagen) according to the manufacturer's protocols. Gene‐specific primers were purchased from Sigma‐Aldrich and used at a concentration of 1·25 μM. Amplification was performed using a Corbett Rotor‐Gene 6000 thermocycler (Corbett Life Sciences, San Francisco, CA, USA) and analysed using the manufacturer‐provided Rotor‐Gene software. The assay uses a two‐step melting/annealing programme over 40 cycles of amplification. The Pfaffl method was used to quantify threshold cycle (CT) data values, and all primers were determined to be 95–100% efficient 15.

Tissue preparation

For lymphocyte isolation from the pancreas, the tissue was extracted, cut into 1‐mm pieces and incubated in 1 mg/ml collagenase V from Clostridium histolyticum (Sigma‐Aldrich) in complete RPMI at 37°C for 30 min while shaking. Digested pieces were then forced through a cell strainer and erythrocytes lysed.

Restimulation of tissue‐extracted lymphocytes

Lymphocytes were extracted from tissues and restimulated for FACS and quantitative real‐time polymerase chain reaction (RT–qPCR) analysis. Cells were plated at 3 × 10⁶ cells per well on a 24‐well plate that was coated overnight with 1 μg/ml anti‐CD3 and cultured with 1 μg/ml anti‐CD28 in complete RPMI. After 48–72 h, supernatants were collected and either PMA/ionomycin was added for FACS analysis (see above) or RNA extracted for RT–qPCR analysis.

Western blot

Mice were killed and approximately 10 mg of whole pancreatic tissue was snap‐frozen immediately in liquid nitrogen. Tissue was then homogenized in lysis buffer containing 1% Triton X‐100 and protease inhibitors using a PowerGen 700 homogenizer (Fisher Scientific, Pittsburgh, PA, USA). The homogenate was centrifuged for 1 min at 2000 g and supernatants were then centrifuged at 10000 g for 20 min at 4°C. Supernatants were extracted and protein concentration was determined using the Bradford assay using Bio‐Rad protein assay dye reagent concentrate and a Benchmark Microplate reader (BioRad). Data were analysed using Microplate Manager version 4·0 software (BioRad) using bovine serum albumin (BSA) as a protein standard (Roche, Laval, QC, Canada).

Equal amounts of 15–20 μg total protein from each sample were separated by sodium dodecyl sulphate–polyacrylamide gel electrophoresis (SDS‐PAGE) using a 12% separating gel and then transferred to a nitrocellulose membrane using a TransBlot Western blotting apparatus (Bio‐Rad). The nitrocellulose membranes were then blocked with 5% skimmed milk powder (Carnation, Don Mills, ON, USA) overnight on a shaker at room temperature. Primary antibody was applied at a concentration of 1 μg/ml overnight on a shaker at 4°C. Secondary antibody was applied at a dilution of 1 : 2000 for the IL‐22R antibody and 1 : 8000 for the beta‐actin antibody. Blots were developed in 3,3′,5,5′‐tetramethylbenzidine (TMB) solution. The expression of IL‐22Rα was quantified by densitometry using SigmaGel Analysis software (Jandel Scientific, San Rafael, VA, USA).

Adoptive transfer

In the transgenic adoptive transfer model, splenocytes from BDC2·5 mice were cultured with the indicated cytokines and antibodies as well as PS3 mimotope 12, 13 as antigen for 4 days. Cells were then washed with sterile PBS twice to remove excess cytokines and resuspended in PBS. Five‐week‐old female NOD mice were then injected with 5 × 10⁶ cells each in 200 μl intravenously (i.v.) through the tail vein. Mice were monitored for diabetes development by urine glucose tests. The Kaplan–Meier survival estimate was used to determine differences in the treatment and control groups.

Anti‐IL‐22 neutralizing antibody treatment

Mice were injected with anti‐IL‐22 antibody in an adoptive transfer model. Pathogenic cells were injected i.v. in 5‐week‐old mice as described above on day 0. These mice were also either injected intraperitoneally (i.p.) with 300 μg in 100 μl of rat anti‐mouse IL‐22·01 monoclonal neutralizing antibody (Pfizer, New York, NY, USA) 15 or sterile PBS on days −1, 2, 6 and 14 for the transgenic adoptive transfer model. The amount of neutralizing antibody was chosen based upon previous studies by Ma et al. with the same clone 16. We have tested the in‐vivo efficacy of this antibody in blocking the induction of regenerating Reg2 gene by complete Freund's adjuvant 10.

Statistical analyses

Statistical analysis was performed using GraphPad Prism 5·01 (La Jolla, CA, USA). Error bars represent standard error of the mean (s.e.m.) among samples. Significant differences between samples were determined using the analysis of variance (anova) and Student's t‐test, with P < 0·05 considered significant for all experiments. For these experiments, P‐values were *P < 0·05, **P < 0·01, ***P < 0·001 and n.s. = not significant.

Results

IL‐22‐producing Th17 cells in diabetic NOD mice

To determine the presence of IL‐22‐producing Th17 cells in type 1 diabetes (T1D), lymphocytes from diabetic and non‐diabetic NOD mice were isolated from pancreas. Infiltrating lymphocytes were extracted from the pancreas by using collagenase V. The presence of infiltrating IL‐22‐producing cells in the pancreas is diabetes‐specific; diabetic and non‐diabetic NOD mice of the same age were compared. Importantly, when examining CD4⁺ pancreatic‐infiltrating T cells, the diabetic NOD mice demonstrated a greater than fivefold increase in IFN‐γ⁺ cells compared to the non‐diabetic mice and an almost twofold increase in IL‐22‐producing cells (Fig. 1a). The cytokines in CD8⁺ cells from the same mice were also explored to discover any differences that may be present in diseased mice. As expected, the diabetic NOD mice demonstrated a very high percentage of IFN‐γ‐producing CD8⁺ cells, and an increase compared to the non‐diabetic mice (Fig. 1b). A much smaller percentage of these cells also expressed IL‐22; however, the diabetic NOD mice showed a greater than sevenfold increase in IL‐22‐producing CD8⁺ cells compared to the same age, non‐diabetic control. The cells producing both IL‐22 and IFN‐γ in Fig. 1a are probably of Th17 and not Th22 lineage. Th22 cells generally produce IL‐22 but not IFN‐γ 17, 18.

Figure 1.

Lymphocytes isolated from the pancreas and draining lymph nodes from diabetic (NOD) mice show more interleukin (IL)‐22 production than non‐diabetic mice of the same age. Pancreata were extracted from diabetic, non‐diabetic age‐matched control (18–25 weeks) or young (4 weeks) NOD mice and infiltrating lymphocytes were isolated by incubating minced pancreatic fragments in 10% RPMI media containing collagenase type V for 30 min to release pancreatic islets. Cells were stimulated in 24‐well plates that had been precoated overnight with 1 μg/ml anti‐CD3 and 1 μg/ml anti‐CD28 and cultured for 72 h, and then phorbol myristate acetate (PMA) and ionomycin were added on the last day. Cells were then collected and stained for (a) CD4⁺ or (b) CD8⁺ cells. Plots are representative of multiple experiments. (c) IL‐17, IL‐22, interferon (IFN)‐γ, IL‐10 and aryl hydrocarbon receptor (AhR) expression in restimulated lymphocytes extracted from PLNs of diabetic NOD mice and control young NOD mice. Also cells from PLNs were restimulated with 1 μg/ml anti‐CD3 and 1 μg/ml anti‐CD28 and collected after 72 h and RNA was extracted and analysed by reverse transcription–quantitative polymerase chain reaction (RT–qPCR). Relative results using beta‐actin as a housekeeping gene are shown as mean expression ± standard error of the mean of multiple experiments, n = 6 in both groups. Student's t‐test was used to determine significance (P < 0·05). There was significant increase in IL‐22 (*P < 0·05), IL‐17 (**P < 0·01), IFN‐γ (*P < 0·05) and a decrease in IL‐10 (*P < 0·05) and no change in AhR (P = ns, non significant) in the cells from diabetic mice compared to 4‐week‐old non‐diabetic mice.

Pancreatic lymph nodes (PLN) from these mice were also extracted and cells were restimulated with anti‐CD3 and anti‐CD28 antibodies for 48–72 h. These cells were collected and RNA extracted for quantitative analysis using RT–qPCR. Th17 cytokines IL‐22 and IL‐17 were significantly higher in the lymph nodes of diabetic mice compared to young controls (Fig. 1c, P < 0·05 and < 0·01, respectively). The Th1 cytokine IFN‐γ also increased significantly in the diabetic mice, supporting the presence of Th1‐like cells in the islet infiltrating cells (P < 0.05). We found that the expression of aryl hydrocarbon receptor (AhR) did not change significantly in these cells. We have shown previously that pathogenic Th17 cells induced by IL‐23 + IL‐6 express little AhR 3. Conversely, the IL‐10 level was significantly lower in the diabetic group versus the non‐diabetic NOD mice supporting the down‐regulation of regulatory T cells (T_reg) upon disease progression (P < 0·05).

Diabetogenic potential of IL‐22‐producing pathogenic Th17 cells and the effect of neutralizing IFN‐γ on pathogenicity

We have recently shown the diabetogenic potential of pathogenic Th17 cells derived from BDC2·5 NOD mice by IL‐23 plus IL‐6 cytokine cocktail polarization 3. To maximize IL‐22 production in Th17‐polarized BDC2·5 cells we added anti‐IFN‐γ antibody to the polarization cell culture. Briefly, splenocytes from BDC2.5 NOD mice were stimulated with PS3 mimotope as antigen in the presence of IL‐23 + IL‐6 cytokine cocktail with or without anti‐IFN‐γ antibody. After 5 days, analysis of the supernatant revealed that the cells polarized with α‐IFN‐γ had significantly more IL‐22 (P < 0·01) and IL‐17 (P < 0·05) production (Fig. 2a).

Figure 2.

Diabetogenic potential of interleukin (IL)‐22‐producing T helper type 17 (Th17) cells derived from BDC2·5 T cells polarized with IL‐23 plus IL‐6 with or without anti‐interferon (IFN)‐γ antibody. Spleens were harvested from 6–8‐week‐old BDC2·5 NOD mice, and single‐cell suspensions were prepared. The cells (2 × 10⁵ cells/well) were cultured in 96‐well plates in the presence of PS3 mimotope (1 µM) in complete RPMI‐1640 medium at 37°C, 5% CO₂ for 5 days. Cells were either polarized with IL‐23 (10 ng/ml) plus IL‐6 (20 ng/ml) cytokine cocktail with or without 5 μg/ml anti‐IFN‐γ antibody. (a) Supernatants were collected on day 5 from both groups and IL‐17 and IL‐22 cytokine levels were measured by enzyme‐linked immunosorbent assay (ELISA). A significant increase in IL‐22 (***P < 0·001) and IL‐17 (*P < 0·05) was observed. (b) The polarized cells from each of the groups in (a) were collected, washed and 5 × 10⁶ cells were adoptively transferred via the tail vein into 5‐week‐old non‐obese diabetic (NOD) mice and the diabetes incidence was monitored. The Kaplan–Meier survival estimate was used to determine differences in the two groups. Student's t‐test was used to determine significant differences between the two groups and no difference was observed (P = ns, non significant).

Next we determined if neutralizing IFN‐γ has an effect on the pathogenicity of the IL‐22‐producing Th17 cells. We transferred adoptively 5 × 10⁶ cells from the above two cultures (Fig. 2a) into each of 5‐week‐old NOD mice. Diabetes incidence was monitored in the recipient mice and they began to develop diabetes within a week. According to the Kaplan–Meier survival estimate, cells from both groups developed disease and no significant difference was seen between the two groups, although all animals that received cells from the α‐IFN‐γ antibody treatment group developed disease compared to 75% in the non‐antibody treatment group (Fig. 2b, P = n.s.).

IL‐27 blocks IL‐22 production from Th17 cells

It is known that IL‐27 blocks IL‐17 production from Th17 cells 19. We sought to determine if IL‐27 affects the levels of IL‐22 produced by Th17 cells. To this end, BDC2·5 splenocytes stimulated with PS3 mimotope and polarized with IL‐23 and IL‐6 were also treated with titrating amounts of IL‐27 (10, 20 and 50 ng/ml) on day 0 to analyse the influence of IL‐27 on this cell population (Fig. 3). As expected, IL‐27 neutralized IL‐17 production and but also drastically reduced IL‐22 production (P < 0·05). IL‐27 also caused a significant reduction in expression of mRNA for the IL‐23 receptor (P < 0·05) that is essential for IL‐22 production. Analysis of gene expression for AhR and IL‐21 showed no significant difference by IL‐27 treatment.

Figure 3.

Interleukin (IL)‐27 neutralizes IL‐22 production from T helper type 17 (Th17) cells. Spleens were harvested from 6–8‐week‐old BDC2·5 non‐obese diabetic (NOD) mice, and single‐cell suspensions were prepared. The cells (2 × 10⁵ cells/well) were cultured in 96‐well plates in the presence of PS3 mimotope (1 µM) in complete RPMI‐1640 medium at 37°C, 5% CO₂ for 5 days. Cells were polarized with IL‐23 (10 ng/ml) + IL‐6 (20 ng/ml) cytokine cocktail. Titrating concentration of IL‐27 (10, 20 and 50 ng/ml) were added to the cells. Cells stimulated with PS3 mimotope without cytokine treatment were used as a control. IL‐27 neutralized IL‐17 and IL‐22 production (P < 0·05) and caused a significant reduction in expression of mRNA for the IL‐23 receptor (P < 0·05). It did not change aryl hydrocarbon receptor (AhR) and IL‐21 expression (P = ns, non significant). Figure shows representative data from two reproducible experiments.

Neutralizing IL‐22 in pathogenic Th17 adoptive transfer of T1D into NOD mice

To examine the effect IL‐22 has on diabetes pathogenesis, an IL‐22 neutralizing antibody was used in adoptive transfer models of NOD mice. BDC2·5 NOD splenocytes were stimulated with PS3 mimotope in the presence of IL‐23 + IL‐6 for 4 days. On day 0, 5‐week‐old NOD mice were injected i.v. each with 5 million polarized IL‐22‐producing Th17 cells. These mice were also injected with 300 μg neutralizing IL‐22 antibody 16 or sterile PBS on days −1, 2, 6 and 14. Diabetes incidence was monitored using urine glucose test strips. The Kaplan–Meier survival estimate was used to determine differences in incidence; however, no statistically significant difference was found (Fig. 4). These experiments were repeated twice with similar results.

Figure 4.

Neutralizing anti‐interleukin (IL)‐22 monoclonal antibody does not change the pathogenic capacity of the adoptively transferred T helper type 17 (Th17) cells in non‐obese diabetic (NOD) mice. Splenocytes from BDC2·5 mice were cultured with 1 μM PS3, 20 ng/ml IL‐6 and 10 ng/ml IL‐23 for 4 days. Cells were collected and 5 × 10⁶ cells were injected intravenously into 4‐week‐old female NOD mice in each group. On days −1, 2 and 6, mice were also injected intraperitoneally with either 300 μg of IL‐22 neutralizing antibody or phosphate‐buffered saline (PBS) control. Mice were monitored for diabetes onset. The Kaplan–Meier survival estimate was used to determine differences in the two groups. No significant difference in disease incidence was observed (P = ns, non significant).

In preliminary experiments using anti‐IL‐22 antibody we also did not find a role for IL‐22 in the conventional adoptive transfer of diabetes in NOD.SCID mice by splenocytes from diabetic NOD mice (data not shown).

Expression of IL‐22 receptor increases in the pancreatic tissue of diabetic NOD mice

As described above, we isolated IL‐22 expressing Th17 cells from the pancreas of diabetic NOD mice. As pancreatic islets have the highest expression of IL‐22 receptor 7, 9, we explored the effect of diabetes development on its expression in the islets. Small pieces of the pancreas from diabetic and young NOD mice as well as B6 control mice were homogenized and RNA extracted for the analysis using RT–qPCR. Interestingly, the unique IL‐22Rα subunit was found to be expressed in significantly higher amounts in the diabetic NOD pancreas compared to young and healthy controls (Fig. 5a). As this expression is shown at the transcription level, we sought to determine the presence of IL‐ 22Rα in the pancreas at the protein level. Therefore, a Western blot of protein extracted from pancreatic tissue was performed and confirmed that the protein is expressed in the pancreas of NOD mice at all ages (Fig. 5b). The expression of IL‐22Rα increased by 1·5 to twofold in diabetic islets relative to 4‐ and 12‐week‐old prediabetic mice as measured by densitometry.

Figure 5.

Increased expression of interleukin (IL)‐22 receptor (IL‐22R) in pancreatic tissue of diabetic non‐obese diabetic (NOD) mice. (a) RNA was extracted from the pancreas of diabetic NOD mice (n = 6), 4–5‐week‐old NOD mice (n = 5), and 10‐week‐old B6 mice (n = 3) and analysed by reverse transcription–quantitative polymerase chain reaction (RT–qPCR) using gene‐specific primers. Relative results using beta actin as a housekeeping gene are shown as mean expression ± standard error of the mean. The analysis of variance (anova) test was used to analyse all values, with P < 0·05 termed as significant. (b) Pancreata from diabetic NOD mice, pre‐diabetic mice (12‐week‐old) and 4‐week‐old NOD mice were excised and snap‐frozen in liquid nitrogen before homogenization in lysis buffer to collect total protein. Western blots were performed using a monoclonal antibody against IL‐22Rα with beta‐actin as a loading control. The increase in IL‐22Rα expression was quantified by densitometry and in diabetic NOD mice 1·5 to 2 fold increase was observed for 4‐ and 12‐week‐old pre‐diabetic NOD mice.

Discussion

In this study we investigated the role of IL‐22 in T1D pathogenesis by first examining the presence of IL‐22‐producing cells in the pancreas and related tissues. To observe directly the phenotypes of these cells we isolated infiltrating lymphocytes from the pancreas of diabetic NOD mice. These cells produce IL‐17, IL‐22 and IFN‐γ. Flavell et al. also demonstrated the presence of both IFN‐γ‐, IL‐22‐ and IL‐17‐producing T cells in infiltrating lymphocytes isolated from the pancreas of BDC2·5 NOD transgenic mice 20. IL‐22 is produced mainly by IFN‐γ‐producing T cells 17), and as Th22 cells generally do not produce IFN‐γ 18, we suggest that cells producing both IL‐22 and IFN‐γ are of Th17 cells. Moreover, Th22 cells are present mainly in skin or anatomical barriers 18. In addition, AhR is the transcription factor for the polarization of CD4⁺ T cells into Th22 cells 18, and we showed previously that polarization of pathogenic Th17 cells by IL‐23 plus IL‐6 down‐regulated AhR expression 3. Therefore, the cells producing both IL‐22 and IFN‐γ are of Th17 lineage. In our study we compared IL‐22 production by restimulated infiltrating lymphocytes between diabetic and non‐diabetic mice of the same age to determine whether the presence of IL‐22 was disease‐specific. By restimulating lymphocytes we aimed to achieve near‐physiological cell phenotypes at the time of disease induction, and also to make it more likely that we were looking at T cells, as stimulation may distinguish them from non‐stimulated innate or naive T cells 21. A higher amount of both IFN‐γ‐ and IL‐22‐producing CD4⁺ cells was demonstrated in the diabetic mice compared to the controls. A similar result was seen with CD8⁺ cells. The diabetic mice certainly have more infiltration and therefore more CD4⁺ and CD8⁺ T cells present in the pancreas; however, when looking at the characteristics of these cells as a whole, more cells produce inflammatory cytokines IL‐22 and IFN‐γ than in the non‐diabetic mice. The lymphocytes are derived mainly from the peri‐insulitic lesion and persist during the prediabetic and early diabetic states 1, 2.

The finding that IL‐22 levels are elevated in diabetogenic Th17 cells 3, 21 and receptors for IL‐22 are expressed highly in pancreas 7, 9 prompted us to functionally investigate the role of IL‐22 in Th17 cells in the pathogenesis of T1D. For this purpose we used an adoptive transfer model system of IL‐22‐producing activated Th17 cells to elucidate the effect of anti‐IL‐22 monoclonal antibody in disease pathogenesis in NOD mice. We discovered that the IL‐22‐producing Th17 cells induced T1D both in anti‐IL‐22 antibody‐treated and untreated control mice. Therefore IL‐22 is not required for the pathogenesis of T1D. This is supported by the recent study by Flavell et al., which demonstrated that knocking out IL‐22 in NOD mice does not protect against diabetes 20. To rule out the role of IL‐22 produced by other cell types we used the NOD.SCID mouse model for adoptive transfer of diabetogenic spleen cells. Here again anti‐IL‐22 antibody did not prevent T1D development. Therefore, IL‐22 does not contribute to the pathogenesis of T1D in NOD mice.

To understand more clearly the effect of IL‐22 we investigated the increase of IL‐22 production by Th17 cells. It was found that neutralizing IFN‐γ demonstrated a profound increase in IL‐22 production, as the Th1‐differentiation pathway was unfavourable for the polarization of Th17 cells. Conversely, neutralization of transforming growth factor (TGF)‐β, an IL‐22 inhibitor, did not increase IL‐22 production, as Th17 induction was also impeded (data not shown). We therefore conclude that for maximal IL‐22‐production from Th17 cells, naive T cells should be polarized with IL‐23 + IL‐6 and anti‐IFN‐γ antibody. In addition, we found that IL‐27 inhibited the production of IL‐22 from Th17 cells. It can act potentially as a negative regulator of Th17 cells and their ability to produce IL‐22 and IL‐17 19. IL‐27 represents a novel agent for the treatment of autoimmune diseases 22. Our studies suggest that IL‐27 could also block pathogenic Th17 cells that are involved in T1D.

We found that IL‐27 did not change the expression of AhR in these IL‐23 + IL‐6 polarized Th17 cells. We have shown that AhR is expressed in Th17 cells that have been differentiated with TGF‐β plus IL‐6, but not in the pathogenic Th17 cells that have been polarized with IL‐23 + IL‐6. These pathogenic Th17 cells expressed much‐reduced levels of AhR 3. The induction of Th22 cells is regulated by AhR transcription factor 18. Because IL‐27 did not reduce AhR expression, but reduced IL‐22 expression significantly, we believe that IL‐22 was produced by pathogenic Th17 cells and not by Th22 cells. As discussed above, these results further support that Th22 cells are not involved in our studies.

The role of IL‐22 in autoimmune diseases affecting various organs is not fully understood 6, 23. Becher et al. showed that IL‐22 is not required for the development of experimental autoimmune encephalomyelitis (EAE) 24, and in rheumatoid arthritis (RA) a less severe course of disease is observed in the absence of IL‐22 25. In general, IL‐22 targets non‐inflammatory cells such as fibroblasts, smooth muscle and endothelial cells. In addition, IL‐10 and IL‐22 share a common IL‐10Rβ2 subunit in their receptors. Therefore, high amounts of IL‐22 may compete for binding to this IL‐22 receptor subunit, and may not allow binding of IL‐10 to exert its regulatory functions.

The pancreas has the highest levels of IL‐22R1 expression among tissues, which suggests that IL‐22 has an important role in the pancreas 7, 9. Furthermore, in these cells IL‐22 up‐regulates directly anti‐apoptotic factors such as B cell lymphoma (BCL)‐2 and BCL‐X_L 26, 27, 28. IL‐22R1 is expressed by pancreatic islet cells and in the majority of insulin‐expressing β cells and/or glucagon‐expressing α cells 9. Increased IL‐22 expression can also protect mice from acute pancreatitis 29. It was shown recently that IL‐22 regulates oxidative stress caused by cytokines or glucolipotoxicity in mouse and human beta cells. IL‐22 administration modulated endoplasmic reticulum (ER) stress and inflammation in islets and promoted secretion of insulin and restored glucose homeostasis 11. We suggest that IL‐22 may have a protective role by fortifying pancreatic islets against immune destruction via suppression of ER stress. It may be involved in modulating glucose metabolism the beta cells and could play a role in the pathophysiology of diabetes 11.

We have shown recently that IL‐22 increases the expression of islet regenerative (Reg) genes REG1 and REG2 in the islets of NOD mice and induces proliferation in these cells 10. The expression of REG1 and REG2 genes also occurs in vivo and can be abolished by neutralizing anti‐IL‐22 antibody 10. Whether this effect is due to an inflammatory signal, or because IL‐22 plays an active role via IL‐22R in islet cells, thereby turning on the Reg to regenerate beta cells, is not understood fully 30. Further work in this area may uncover an important mechanism of IL‐22 action in the pancreas that is not yet known.

In summary, IL‐22 is not required for the diabetogenic effect of Th17 cells and this is supported by the results obtained by Flavell et al. in IL‐22 knock‐out NOD mice 20. In general, IL‐22 may have a regenerative and protective role in pancreatic islets 10, 11, 30.

Disclosure

Authors have no financial disclosures to declare.