Suppression of IL-22–Producing T Helper 22 Cells by RPE Cells via PD-L1/PD-1 Interactions

Abstract

Purpose.: To determine whether RPE cells can suppress a novel T helper subset, the Th22 cells, via costimulatory interactions.

Methods.: Primary RPE cells were established from normal C57BL/6 mice. The target CD4⁺ Th22 cells from spleen cells in wild-type control or knockout donors were used. T cell activation was assessed by examining BrdU incorporation (proliferation) and cytokine production. Expression of costimulatory molecules on RPE cells and expression of costimulatory receptors on target Th22 cells were evaluated by flow cytometry. Neutralizing antibodies were used to abolish the suppression function. In addition, human RPE cells and target Th22 cells induced from human CD4⁺ cells were also used in similar experiments.

Results.: Cultured RPE cells significantly suppressed activation of target Th22 cells (e.g., T cell proliferation and IL-22 production). Moreover, human RPE cells suppressed Th22 cell lines and T cell clones established from active uveitis patients. Although cultured RPE cells expressed various costimulatory molecules including programmed cell death 1 ligand 1 (PD-L1), only PD-L1 on the RPE cells was actually delivered to the target Th22 cells. Th22 cells greatly express programmed cell death 1 (PD-1), and RPE cells failed to suppress IL-22 expression by target Th22 cells from PD-1 knockout donors. In addition, if neutralizing antibodies for PD-L1 were cocultured with RPE cells, Th22 suppression was impaired.

Conclusions.: RPE cells express PD-L1 costimulatory molecules and suppress bystander Th22-type PD-1⁺ bystander T cells through negative costimulatory interactions.

Introduction

CD4⁺ T helper (Th) cells play an essential role in intraocular inflammation in experimental animals as well as in human inflammatory disorders. Activated CD4⁺ T cells can infiltrate the eye and cause an immune response and inflammation, resulting in damage of vision-related cells and tissues of the eye. However, the eye is an immune privileged site and has a unique immune system that protects important cells and tissues from inflammatory CD4⁺ T cells. Previous experimental evidence has demonstrated that some ocular parenchymal cells participate in the immune privilege of the eye. ^1,2  

Recent studies by our own and other various laboratories have investigated the immune regulation by ocular resident cells (i.e., endothelial cells of the cornea and pigment epithelial cells of the iris, ciliary body, and retina). ^3–15 Our research group focused on these ocular resident cells because they are located at the gate of the blood ocular barrier. Based on our current knowledge, it is obvious that immune regulation would work much more efficiently at the gate of the blood ocular barrier as opposed to after the activated CD4⁺ T cells enter the eye. Several investigators have previously demonstrated that pigment epithelial (PE) cells of the iris, ciliary body, and retina, as well as the corneal endothelial cells, possess immunoregulatory properties and contribute to the immune privileged site of the eye. ^3–6 For instance, murine RPE cells display immunomodulatory activity to various immune cells such as: downregulating Th1 cells, ⁷ Th17 cells, ⁸ CD8⁺ T cells, ⁹ B cells, ¹⁰ macrophages, ¹¹ and dendritic cells ¹² ; and upregulating regulatory T cells ^13,14 and suppressor myeloid cells. ¹⁵ The CD4⁺ T cells have identified at least five phenotypes such as Th1 cells, Th2 cells, Th9 cells, Th17 cells, and Th22 cells. 

In human studies performing gene expression profiling analysis, patients with noninfectious autoimmune uveitis exhibited significantly increased IL-22 genes compared with the nonuveitis normal controls. ¹⁶ Moreover, these studies showed that IL-22 can affect human RPE cells by decreasing the total tissue resistance, which includes apoptosis. In the presence of IL-6 and TNF-α, activated naïve CD4⁺ T cells can differentiate into Th22 cells, which produce IL-22 and TNF-α. ¹⁷ The expansion of Th22 cells appears to be regulated by the aryl hydrocarbon receptor (AHR) transcription factor, with the Th22 cells greatly expressing the CCR10 chemokine receptor. ¹⁸ Thus, the new subset of helper T cells may play an important role in the pathogenesis of ocular inflammation, which is a predominantly T cell-mediated autoimmune inflammatory disease. However, at the present time, there have been no reports showing that ocular resident cells or tissues can suppress Th22 cells. Moreover, if this does occur, the next thing to ascertain is which candidate immunoregulatory molecule is then produced by the ocular resident cells. For the assays in the current study, we prepared IL-22-producing Th22-type CD4⁺ T cells as the target cells, with primary cultured RPE cells established from normal mice as the effector cells. 

Methods

Mice

Adult C57BL/6 mice (CLEA Japan, Inc., Tokyo, Japan) were used as donors of the lymphoid cells and ocular pigment epithelium. ^5,19 The PD-1 knockout (KO) donor (PD-1^−/−) mice, as well as the wild type, were used as the target T cell donors. ^20–22 All experiments were approved by the Institutional Animal Research Committee of Tokyo Medical and Dental University, and conformed to the Association for Research in Vision and Ophthalmology Statement for the Use of Animals in Ophthalmic and Vision Research. 

Cultured RPE Cells

Eyes were enucleated, and cut into two halves along a circumferential line posterior to the ciliary process, creating a ciliary body-free posterior eyecup. The eyecup was incubated in 0.2% trypsin (Biowhitaker, Walkersville, MD) for 1 hour. The RPE tissues were triturated to make a single cell suspension, and then resuspended in Dulbecco's modified Eagle's medium (DMEM), placed in 24-well plates and incubated for 2 weeks. DMEM containing 10% fetal bovine serum (FBS) was used for the primary cultures of RPE. ^5,19 Other murine ocular PE cells—such as iris pigment epithelium (IPE) and ciliary body pigment epithelium (CBPE), ⁵ and murine cornea endothelial cells (CE) ²³ —were also prepared. As determined by flow cytometry, the primary RPE cultures were found to be greater than 98% cytokeratin positive. Human RPE cell lines (ARPE-19) were also used for human experiments. 

Preparation of Target T Cells and Induction of Th22 Cells

T cells were enriched with mouse CD4⁺ cells using beads (MACS cell isolation kits; Miltenyi Biotec, Auburn, CA; >95% of cells expressed the relevant surface marker). Purified CD4⁺ T cells were added (2 × 10⁵ cells/well) to culture wells with RPE cells (1 × 10⁵ cells/well). The cultures were maintained for 4 days (for evaluation of cytokine production) or 5 days (for evaluation of cell proliferation). The cultures were then assayed for BrdU uptake (the last 8 hours of culture, BrdU cell proliferation ELISA; Roche Diagnostic, Mannheim, Germany), as a measure of cell proliferation. Serum-free medium was used in cultures and assays involving T cells stimulated by anti-CD3 antibodies to mimic, as close as possible, the intraocular microenvironment outside the blood ocular barrier. Serum-free medium was composed of RPMI 1640 medium without the addition of FBS, and supplemented with 0.1% bovine serum albumin (BSA; Sigma-Aldrich, St. Louis, MO) and 0.2% insulin, transferrin, selenium culture supplement (ITS⁺; Collaborative Biochemical Products, Bedford, MA). 

Mouse Th22 cells were induced with a similar method that has been described in a previous report. ¹⁷ Splenic CD4⁺ T cells were cocultured with anti-mouse CD3 antibody (2 μg/mL; BD Pharmingen, San Diego, CA); anti-mouse CD28 antibody (2 μg/mL; BD Pharmingen); anti-mouse IFN-γ antibody (5 μg/mL, R&D Systems, Minneapolis, MN); anti-mouse IL-4 antibody (5 μg/mL, R&D Systems); recombinant mouse TNF-α (50 ng/mL, R&D Systems); and recombinant mouse IL-6 (20 ng/mL, R&D Systems). The concentration of IL-22 in the supernatants of the T cell cultures was measured by ELISA (R&D Systems). 

Establishment of Human T Cell Clones and T Cell Lines

All subjects were uveitis patients with Behçet's disease who were examined at Tokyo Medical and Dental University Hospital between 2011 and 2012. The research followed the tenets of the Declaration of Helsinki, and the study was approved by the Institutional Ethics Committee of Tokyo Medical and Dental University. 

Samples of aqueous humor were collected from patients after informed consent was obtained. At the time of sampling, patients had active intraocular inflammation but were not being treated with any systemic therapy such as corticosteroids and infliximab. Peripheral blood mononuclear cells (PBMCs) were also obtained from both the Behçet's disease patients and healthy donors. Freshly purified T cells were enriched for CD4⁺ cells using MACS cell isolation kits (Miltenyi Biotec, >94% CD4⁺) and applied to in vitro assays or flow cytometry. 

T cell clones were established by the previously described limiting dilution method. ^24,25 CD4⁺ T cells were obtained from patients with uveitis who had Behçet's disease, as it has been shown that the Th22-type T cell clones can produce large amounts of IL-22 and TNF-α, but not IFN-γ and IL-17. ²⁶ Thus, in line with this previous report, we used PBMCs from Behçet's disease or healthy donors to establish the CD4⁺ Th22 cell lines. 

Flow Cytometry

Flow cytometric analysis of the Th22 cell lines and Th22 cells exposed to RPE was performed using phycoerythrin (PE)-labeled anti-human IL-22 monoclonal antibody (R&D Systems). T cells were precultured with a protein transport inhibitor (GolgiPlug; BD Biosciences, San Jose, CA), ionomycin (0.5 mg/mL; Sigma-Aldrich, St. Louis, MO), and phorbol-12-myristate-13-acetate (PMA, 40 ng/mL; Merck Chemical, Darmstadt, Germany) for 5 hours before intracellular staining. Prior to the staining, T cells were incubated with a mouse Fc block (Fcγ III/II Receptor; BD Pharmingen) at 4°C for 15 minutes. After permeabilization, Th22 cells were stained with PE-labeled anti-mouse IL-22 and FITC-labeled anti-mouse CD4 abs. PE-conjugated rat IgG (R&D Systems) was used as the isotype control. Cells (1 × 10⁶) were stained for 30 minutes at room temperature in the dark. 

Flow cytometry was also used to analyze the expression of the AHR on the Th22 cells or RPE-exposed Th22 cells (Abcam, Tokyo, Japan). After permeabilization, the cells were stained with anti-mouse AHR antibody or rabbit IgG (isotype control) at 4°C for 30 minutes. The cells were washed, with the bound primary antibody detected by incubation with biotin-conjugated anti-rabbit IgG (BD Pharmingen) at 4°C for 30 minutes, followed by FITC-conjugated streptavidin (BD Pharmingen) at 4°C for 15 minutes. 

In another experiment, various T cells such as Th22 cells from wild-type mice, Th22 cells from PD-1 KO mice, anti-mouse programmed cell death 1 ligand 1 (PD-L1, B7-H1) antibody-pretreated Th22 cells, intraocular T cells from the experimental autoimmune uveitis (EAU) model, and EAU T cells exposed to RPE were also stained with PE-labeled anti-mouse IL-22 abs and FITC-labeled anti-mouse CD4 abs. 

Human T cells, Th22-type T cell clones and Th22 cell lines derived from Behçet's disease patients were stained with PE-anti-human IL-22 (R&D Systems) and FITC-anti-human CD4 abs. Cells (1 × 10⁶) were stained for 30 minutes at room temperature in the dark. PE-conjugated mouse IgG (R&D Systems) was used as the isotype control. These T cells were also stained with PE-labeled anti-human CCR10 (BioLegend, San Diego, CA) and FITC-anti-human CD4 abs. PE-conjugated hamster IgG was used as the isotype control. Cells (1 × 10⁶) were stained for 30 minutes at 4°C in the dark. 

Flow cytometric analysis of three RPE cells such as Th22 supernatant-exposed RPE cells, control RPE cells without supernatant, or RPE cells in the presence of recombinant mouse IL-22 (R&D Systems) was performed using PE-labeled anti-mouse PD-L1 antibody (B7-H1; eBioscience, San Diego, CA). RPE cells were harvested and stained with anti-PD-L1 abs. RPE cells were also stained with anti-PD-L2 (B7-DC, Clone 122), CD80 (B7-1), CD86 (B7-2), and ICOS ligand (ICOSL/B7-H2: all eBioscience). Before staining, RPE cells were incubated with anti-CD16/CD32 antibodies (mouse Fc block) for 15 minutes at 4°C. PE-conjugated rat IgG isotype (eBioscience) was used as the control. Flow cytometry was also used to analyze the expression of the costimulatory receptors on the target T cells. PE-conjugated anti-mouse PD-1 mAb (eBioscience) was used to stain the T cells, naïve T cells, Th22 cells from wild-type mice, and Th22 cells from PD-1 KO mice. These T cells were harvested and stained with anti-PD-1 (PE) and anti-CD4 abs (FITC). As an isotype control for the molecules, we used PE-conjugated rat IgG isotype. 

Blocking Antibodies

In some in vitro experiments, purified anti-mouse PD-L1/B7-H1 mAb (10 μg/mL), anti-mouse PD-L2/B7-DC mAb (10 μg/mL) or rat IgG (isotype control, 10 μg/mL) was added to cultures with RPE cells plus target Th22 cells. 

Induction of EAU and Harvested Intraocular T Cells

Adult C57BL/6 mice were immunized subcutaneously in the neck region with 200 μg of interphotoreceptor retinoid-binding protein peptide (IRBP_1-20) as per a previous report. ²⁶ Inflammation was evaluated by a funduscopic examination at 21 days after the immunization. Intraocular T cells were collected from EAU mice and evaluated by the IRBP retinal antigen-specific assay. EAU eyes (n = 20) were cut into two halves along a circumferential line posterior to the ciliary process, with the neural retina then detached and removed. The ocular fluids, triturated retina, and eyecup were incubated in 0.2% trypsin for 1 hour. After the removal of the visual tissues, culture cells were collected. The collected intraocular T cells were enriched with mouse pan-T cells using beads (MACS). The EAU T cells (1 × 10⁵/well) were cocultured with antigen-presenting cells (20 Gy x-irradiated spleen cells: 1 × 10⁴/well) plus mouse IRBP peptide (10 μg/mL) in the presence of RPE cells. Supernatants of the T cell-RPE cultures were collected for 48 hours in order to measure the IL-22 cytokine concentration. Intraocular cells were also harvested from the eyes of EAU mice at day 21, and cocultured with RPE cells for flow cytometry (CD4/IL-22 staining). 

Statistical Analysis

Each experiment was repeated at least twice with similar results. All statistical analyses were conducted using the Student's t-test. Values were considered statistically significant if P < 0.05. 

Results

Capacity of RPE Cells to Suppress Th22 Cells That Produce IL-22 Inflammatory Cytokine

Several groups have previously reported that the cultured RPE cells suppress bystander target inflammatory cells. ^5,7–11,19 Recently, a subset of IL-22-producing T (Th22) cells distinct from Th1, Th2, or Th17 cells have been reported to play a crucial role in the induction of the autoimmune inflammatory response. ^17,18,27,28 Therefore, we first examined whether the RPE cells could suppress the production of IL-22 by activated CD4⁺ T cells and then determined if the RPE cells also suppress polarized Th22 cells. 

Using ELISA, we confirmed that the RPE cells did suppress the IL-22 production from anti-CD3-stimulated CD4⁺ T cells in vitro (Fig. 1A). To examine the induction of polarized Th22 cells, we used rIL-6 and rTNF-α, as per a previously described report. ¹⁷ As shown in Fig. 1A, RPE cells profoundly suppressed the IL-22 produced by the Th22 cells. On the other hand, cultured RPE alone without T cells did not produce IL-22 (data not shown). To confirm the suppression by RPE, different concentrations of rIL-6 and rTNF-α were used in the subsequent assay. Cultured RPE cells significantly suppressed the IL-22 produced by the Th22 cells in a dose-dependent manner (Fig. 1B). In particular, the RPE cells were greatly suppressed when 20 ng/mL rIL-6 and 50-100 ng/mL rTNF-α were used for induction of the Th22 cells (Fig. 1B). Based on these results, all subsequent experiments used induced Th22 cells with 20 ng/mL rIL-6 and 50 ng/mL rTNF-α. 

Figure 1

 

Capacity of RPE cells to suppress activation of Th22 cells. (A) Purified naïve CD4⁺ T cells in the presence of anti-CD3 antibodies were cocultured with RPE cells. Th22 cells induced by rTNF-α and rIL-6 were also cocultured with RPE cells. After 96 hours, the supernatants of the cells were harvested for mouse IL-22 ELISA analysis. Data are the mean ± SEM of three ELISA determinations. **P < 0.005, ***P < 0.0005, compared with the positive control (open bar). (B) CD4⁺ T cells in the presence of rTNF-α and rIL-6 were cocultured with RPE cells using several different doses: (1) Th22 cells: rTNF-α, 20 ng/mL and rIL-6, 10 ng/mL; (2) Th22 cells: rTNFα, 20 ng/mL and rIL-6, 5 ng/ml; (3) Th22 cells: rTNF-α, 20 ng/mL and rIL-6, 10 ng/mL; (4) Th22 cells: rTNF-α, 20 ng/mL and rIL-6, 20 ng/mL; (5) Th22 cells: rTNF-α, 50 ng/mL and rIL-6, 20 ng/mL; (6) Th22 cells: rTNF-α, 100 ng/mL and rIL-6, 20 ng/mL. *P < 0.05, **P < 0.005, ***P < 0.0005, compared with the positive control (open bar). (C) Th22 cells were stimulated in the absence (upper panel) or presence (lower panel) of RPE cells. After 96 hours, the T cells were harvested for flow cytometric analysis. The numbers in the histograms indicate the percentage of cells that were double-positive for IL-22/CD4. (D) Th22 cells were cocultured with RPE cells for proliferation. Cells were then harvested after 5 days and assayed for BrdU incorporation by ELISA. Data are the mean ± SEM of three ELISA determinations. **P < 0.005, compared with the positive control (open bar). (E) T cells cocultured with RPE were harvested for detection of AHR by flow cytometry. The numbers in the histograms indicate the percentage of cells for AHR. (F) Th22 cells were cocultured with IPE, CBPE, and CEs for 96 hours. The supernatants of cells were harvested for IL-22 ELISA analysis. Data are the mean ± SEM of three ELISA determinations. **P < 0.005, ***P < 0.0005, compared with the positive control (open bar).

In the next step, flow cytometric analysis (Fig. 1C) and the BrdU proliferation assay (Fig. 1D) both confirmed that the RPE cells greatly suppressed the IL-22 production by the Th22 cells. The RPE cells suppressed the AHR transcript factor in the Th22 cells, whereas the Th22 cells without RPE cells highly expressed the transcript factor (Fig. 1E). 

We further examined whether other types of ocular resident cells could suppress activation of Th22 cells in vitro. As expected, these cells along with IPE cells, CBPE cells, and CE cells all significantly suppressed the IL-22 production by the Th22 cells (Fig. 1F). Thus, based on this large body of in vitro data, this demonstrates that RPE-exposed CD4⁺ T cells down-regulate Th22-mediated cytokine production. 

Capacity of RPE Cells to Suppress IL-22-Secreting CD4⁺ T Cells in an Experimental Uveitis Model

Th1 and Th17 inflammatory cells are both associated with the ocular immune responses that occur in the EAU model. To evaluate the purified T cells, we used the IRBP retinal-antigen specific assay. We collected purified splenic CD4⁺ T cells from an EAU model in addition to harvesting ocular infiltrating cells from the inflamed eyes of EAU donors. A significant response was observed (i.e., increased levels of IL-22 were produced in purified eye or splenic CD4⁺ T cells from EAU donors immunized with IRBP antigens; Fig. 2A). In contrast, the response was significantly reduced when these T cells were cocultured with RPE cells. Compared with control EAU T cells without RPE, there was a decrease in the Th22 population of EAU eye-derived CD4⁺ T cells when the T cells were exposed to RPE (Fig. 2B). These results suggest that RPE cells directly suppress retinal antigen-specific Th22 cells in the inflamed eye. 

Figure 2

 

Detection of IL-22-producing cells in EAU donors. (A) For the assay, normal mice (n = 20) were immunized with IRBP_1-20. The intraocular T cells were collected from EAU donors on day 21, with spleen CD4⁺ T cells also prepared in order to evaluate the IRBP retinal antigen-specific assay. The EAU T cells (1 × 10⁵/well) were cocultured with antigen-presenting cells (20 Gy x-irradiated spleen cells: 1 × 10⁴/well) plus mouse IRBP peptide (10 μg/mL) for 48 hours. As controls, spleen CD4⁺ T cells from the EAU mouse (1 × 10⁵/well) were prepared in the presence of IRBP peptide. These T cells were cocultured with (black bars) or without (open bars) RPE cells. ELISA was used to measure the IL-22 cytokine concentration in the supernatants of the T cell cultures. **P < 0.005, between two groups. (B) T cells collected from EAU eyes (upper histograms) and T cells from EAU eyes cocultured with RPE cells (lower histogram) were stained with anti-mouse IL-22 and anti-mouse CD4 abs after permeabilization and then analyzed by flow cytometric analysis. The numbers in the histograms indicate the percentage of cells that were double-positive for IL-22/CD4.

Capacity of Human RPE Cells to Suppress Th22 Cell Lines and Th22-Type T Cell Clones in Uveitis Patients

In a subsequent experiment, we examined whether human RPE cells could suppress Th22 cells in vitro. Th22 cell lines from uveitis patients and Th22-type T cell clones established from the ocular fluids of uveitis patients with Behçet's disease were assayed as per a previously described method. ²⁶ T cells were stained with anti-IL-22 and anti-CD4 antibodies. As seen in Figure 3A, polarized Th22 cell lines from uveitis patients or healthy donors expressed IL-22. When examined by flow cytometric analysis, patients with Behçet's disease particularly expressed Th22 cells. In contrast, T cells in the presence of RPE poorly expressed IL-22. T cell clones established from eyes with Behçet's disease greatly expressed IL-22, although the expression of IL-22 was poor if the T cells were exposed to RPE (Fig. 3A). ELISA similarly showed that human RPE cells significantly suppressed production of IL-22 by these human Th22-type T cells (Fig. 3B). In addition, cultured RPE cells significantly suppressed both the proliferation of Th22 cells from Behçet's disease patients (Fig. 3C) and the expression of CCR10 molecules, which are highly expressed by Th22 cells (Fig. 3D). Taken together, these results indicate that both mouse and human RPE cells are able to suppress activation of Th22 cells in vitro. 

Figure 3

 

Capacity of human RPE cells to inhibit Th22 cells from uveitis patients with Behçet's disease. (A) Harvested Behçet's Th22 cells were stained with anti-human IL-22 and anti-human CD4 abs after permeabilization and then analyzed by flow cytometric analysis. The left histogram is for the polarized Th22 cell lines from Behçet's disease (BD), while the middle histogram is for the Th22 cell lines (a healthy donor, HD), and the right histogram is for the Th22-type T cell clones (TCC) established from a BD patient. Lower histograms show the T cells plus human RPE cells. The numbers in the histograms indicate the percentage of cells that were double-positive for IL-22/CD4. (B) For the ELISA analysis, polarized Th22 cell lines (BD or HD) or Th22-type T cell clones (BD) were cocultured with (black bars) or without (open bars) RPE cells. The graph indicates the amount of IL-22 determined by ELISA. *P < 0.05, ***P < 0.0005 between the two groups. (C) For the proliferation assay, Th22 cell lines (BD) or Th22-type T cell clones (BD) were cocultured with (black bars) or without (open bars) RPE cells. The graph indicates the absorbance of BrdU determined by ELISA. **P < 0.005, ***P < 0.0005 between the two groups. (D) For the flow cytometry, Th22 cell lines from Behçet's disease (upper histogram) or Th22 cells exposed to RPE (lower histogram) were stained with anti-human CCR10 and anti-human CD4 abs. The numbers in the histograms indicate the percentage of cells that were double-positive for CCR10/CD4.

Expression of PD-L1 Costimulatory Molecules by Th22 Exposed RPE Cells

The immunomodulation by RPE cells is mediated by various soluble or membrane-bound molecules. ^13,29–31 From all of these, we attempted to determine the candidate molecules for Th22 inhibition by RPE cells. 

In the first step, we used blocking antibodies for TGFβ, TSP-1, and CTLA-2α in line with previous RPE studies. ^13,14,19,29 Anti-mouse TGFβ, TSP-1 or CTLA-2α blocking antibodies were unable to neutralize the RPE cell suppressive activity on the Th22 cells (data not shown). In the next step, after adding recombinant TGFβ, TSP-1 or CTLA-2α proteins to the Th22 cell culture medium, we found a greatly enhanced secretion of IL-22 (data not shown), which indicated that these molecules were not relevant to Th22 suppression by RPE cells. 

Subsequently, we focused on costimulatory molecules in order to examine the expression of CD80 (B7-1), CD86 (B7-2), PD-L1 (B7-H1), PD-L2 (B7-DC), and ICOSL (B7-H2) on RPE cells exposed to Th22 supernatants. For this step, we collected supernatants from the polarized mouse Th22 cell lines. Th22 supernatant-exposed RPE expressed CD80, CD86, and especially the PD-L1 molecules at much higher levels than was seen for the control RPE without the supernatants (Fig. 4A). For confirmation, we used recombinant mouse IL-22 proteins to treat the RPE cells. As shown in Figure 4B, these recombinant-treated RPE cells highly expressed PD-L1 compared with the nontreated cells. These results suggest that the PD-L1-PD-1 interaction might be associated with the Th22 cell suppression by RPE cells. Based on these results, we then examined whether Th22 cells could express the PD-1 receptor. As expected, the Th22 cells greatly expressed the PD-1 receptor on their surface, whereas the naïve T cells did not (Fig. 4C). 

Figure 4

 

Detection of PD-L1 costimulatory molecules by RPE cells under Th22 conditions. (A) After RPE cells exposed to Th22 supernatants and RPE cells without supernatants (normal condition) were stained with specific mouse antibodies (CD80 [B7-1], CD86 [B7-2], PD-L1 [B7-H1], PD-L2 [B7-DC], ICOSL [B7-H2], and isotype control), they were then examined by flow cytometric analysis. Percentages in the upper-right corners indicate the positive cells. Primary cultured RPE cells were cocultured with the Th22 supernatants for 48 hours. (B) RPE cells in the presence (open histogram) or absence (dotted histogram) of recombinant mouse IL-22 were harvested and stained with anti-mouse PD-L1 abs. Numbers in the upper-right corners indicate the mean fluorescence intensity. (C) Detection of PD-1 on mouse Th22 cells. Naïve CD4⁺ T cells and Th22 cells were harvested and stained with anti-mouse PD-1 and anti-mouse CD4 abs, followed by a flow cytometry examination. Percentages in the upper right corners indicate PD-1/CD4 double positive cells.

Ability of RPE Cells to Suppress Th22 Cells From PD-1 KO Mice

We next examined whether RPE cells could suppress Th22 cells induced from a PD-1 KO donor. Results showed that cultured RPE cells readily suppressed PD-1⁺ Th22 cells from wild-type mice (48% → 9%; Fig. 5A). On the other hand, flow cytometric analysis indicated that RPE cells failed to suppress the bystander T cells from the PD-1 KO mice (4% → 10%). Interestingly, induced Th22 cells from PD-1 KO donors poorly expressed IL-22 (Fig. 5A: upper right panel), which suggests that the PD-1 signal in CD4⁺ T cells might be required in order to differentiate the Th22 cells. Importantly, anti-mouse PD-L1 antibody neutralized the suppression of Th22 cells by RPE, but not the isotype control abs (Fig. 5B). In contrast, anti-mouse PD-L2 (B7-DC) mAb failed to block the suppression (data not shown). These results indicate that the interaction between PD-L1-expressing RPE cells and PD-1-expressing Th22 cells provides negative costimulatory signals, thereby suppressing the bystander T cells. 

Figure 5

 

Capacity of RPE cells to suppress Th22 cells from PD-1 KO donors. (A) Target Th22 cells were obtained from C57BL/6 wild-type (WT) controls or from PD-1 KO mice. Polarized Th22 cells (WT or KO T cells) were cocultured with RPE cells, harvested, and then stained with anti-mouse IL-22 and anti-mouse CD4 abs after permeabilization. (B) Polarized Th22 cells were cocultured with RPE cells for 96 hours. Anti-mouse PD-L1 neutralizing antibodies (lower-right histogram) or isotype rat IgG (upper-right histogram) were added in some wells. The numbers in the histograms indicate the percentage of cells that were double positive for IL-22/CD4.

Discussion

The eye has been shown to be protected from invasion of inflammatory cells by two systems, an anatomical barrier and an immunological barrier. ^1,2 The blood ocular barrier anatomically blocks harmful cells in the peripheral bloodstream from invading the eye and protects sight-related important cells and tissues. PE cells of the iris, ciliary body and retina, vascular endothelial cells in the retina, and corneal endothelial cells are all important components of the blood ocular barrier. Once the blood ocular barrier is disrupted, another defense system (regional immunity of the eye) works to suppress the activation of the pathogenic T cells, thereby protecting the eye. Recent studies have shown that ocular resident cells at the inner layer of the blood ocular barrier, such as the iris PE cells, ciliary body PE cells, retinal PE cells, and corneal endothelial cells, have the capacity to suppress the activation of bystander inflammatory immune cells, including the CD4⁺ T helper cells. ^3–12  

RPE cells are the ocular resident cells that participate in the immune regulation of the posterior segment in the eye. Furthermore, RPE cells have the capacity to convert activated T cells into regulatory T (Treg) cells. ^13,14 Thus, the balance between the activated CD4⁺ T cells in the eye and the regional immune system of the ocular resident cells and the Treg cells they induce is able to maintain homeostasis in the eye. 

Many investigators have previously demonstrated that cultured RPE cells constitutively express immunosuppressive factors. ^{8–14,19,29–31} To identify the soluble factors that participate in the immune regulation by the RPE cells, we previously performed a GeneChip microarray assay using RPE cells and IFN-γ pretreated RPE cells. ³⁰ Several eye-derived immunoregulatory genes were found to be expressed at high levels on both of these RPE cells. Among them, the one molecule that has been identified as a key factor for the immunoregulation of the RPE cells is the PD-L1 (B7-H1) costimulatory molecule. ³⁰ Our current study demonstrated that while the RPE cells exposed to Th22 supernatants and recombinant IL-22 proteins greatly expressed this molecule, no other costimulatory factor was expressed. 

PD-L1 is involved in the suppression of T cells by retinal PE cells, other ocular tissues, and eye-specific Treg cells. ^30–34 The PD-L1 molecules are widely expressed by the thymus, spleen, heart, pancreas, endothelium, epithelium, tumors, and immunocytes such as dendritic cells and monocytes. In the eye, PD-L1 molecules are constitutively expressed by the corneal endothelial cells, ^32,34 and provide a negative costimulation for the effector T cell, thereby helping to inhibit corneal allograft rejection. ³² In addition to these cells, human RPE cell lines constitutively express PD-L1 costimulatory molecules, while retinal PE cells can suppress the PD-1 expressing T cells. ³¹ We further examined whether mouse RPE cells can suppress bystander T cells during inflammatory conditions. To achieve this, we used Th1 cytokine IFNγ-pretreated retinal PE cells. ³⁰ Although primary murine RPE cells poorly expressed PD-L1, expression of PD-L1 was greatly induced on the surface of IFNγ-pretreated RPE cells. Thus, PD-L1 costimulatory molecules are expressed on ocular resident tissues/cells in the presence of Th1-type inflammatory cytokine IFNγ, as has been previously reported in other studies. ^30,34  

Interestingly, other groups have previously demonstrated that the inflammatory cytokine IL-17, as well as IFNγ, can up-regulate the expression of PD-L1 on RPE. ³⁵ Our results also confirmed this new evidence. Although the novel cytokine IL-22 can promote expression by RPE cells, and polarized Th22 cells can highly express the receptor PD-1, Th22 cells from PD-1 null donors do not promote this expression. Moreover, while the RPE cells readily suppressed the PD-1⁺ bystander Th22 cells, the RPE cells failed to suppress the target T cells from the PD-1-deficient mice. Additionally, anti-PD-L1 antibody neutralized the suppression of the Th22 cells by RPE, which indicates that negative costimulatory signals are able to suppress bystander T cell types. It is assumed that ocular resident cells including RPE cells are able to suppress infiltrating inflammatory cells (i.e., Th1/Th17/Th22 cells, under inflamed conditions). In fact, these activated CD4⁺ T cells have been shown to play a critical role in the immune response of eye inflammations such as uveitis. ^36–39  

A recent gene analysis study showed that increased levels of the IL-22 gene were exhibited in patients with autoimmune noninfectious uveitis. ¹⁶ More recently, we have demonstrated that the Th22 cytokines play a key role in the ocular immune response in Behçet's uveitis. ²⁶  

Th22 cells are the novel T helper cell lineage of the CD4⁺ T cells, and these cells are relevant in both autoimmune and dermal inflammatory diseases. Induction of Th22 differentiation is achieved through the activation of naïve CD4⁺ T cells that then differentiate into Th22 cells in the presence of IL-6 and TNF-α. ¹⁷ These T cells express IL-22, TNF-α, CCR4, CCR6, and CCR10. ¹⁸ It should be noted that these Th22 cells do not express the Th17 markers IL-17, Th2 marker IL-4, or the Th1 marker IFN-γ. Thus, these characteristics are distinct from the Th17, Th2, and Th1 subtypes. The expansion of the Th22 cells appears to be regulated by the AHR transcription factor, ¹⁸ with this transcript factor significantly reduced when Th22 cells are cocultured with RPE (Fig. 1E). The inflammatory cytokines IL-22, as well as TNF-α, may play a key role in the immune response in the eye. However, ocular resident cells can inhibit both the inflammatory cytokines and the IL-22–producing cells. Our current study demonstrated that ocular resident cells can suppress the Th22-type cells, and that one of the candidate immune regulatory molecules produced by the ocular cells is the PD-L1 negative costimulator. 

However, both our mouse and human in vitro experiments demonstrated that RPE from autoimmune uveitis eyes inhibited Th22 activity and that IL-22 production was much more efficient than normal RPE. The reason this occurs is that when RPE is exposed to inflammatory cytokines, it is able to express the negative costimulatory molecules, PD-L1. In addition, RPE is able to convert CD4⁺ T cells into Treg cells, while the RPE-induced Treg cells may suppress these inflammatory T cells. Nevertheless, this ability does not prevent the in vivo intraocular T cell infiltration and inflammatory reaction. Thus, the balance between inflammatory intraocular T cells and the regional immune system of the RPE is very important for maintaining homeostasis in the eye. 

In conclusion, cultured RPE cells greatly suppress IL-22-producing CD4⁺ T cells via a negative costimulatory signal. The immunomodulation by the ocular resident RPE cells is mediated by various immunoregulatory molecules, including TGFβ, ^8–10 thrombospondin-1, ²⁹ CTLA-2α, ^13,14 retinoic acid, ⁴⁰ and PD-L1. ^30,31 Among these, PD-L1 molecules expressed by the RPE cells are able to directly suppress the inflammatory T cells (e.g., Th1 cells ³⁰ and Th17 cells). ³⁵ Results of this study also further clarified the actions of the Th22 cells. 

Acknowledgments

We thank Ikuyo Yamamoto for her expert technical assistance. The PD-1 knockout donor (PD-1^−/−) mice were provided by Taku Okazaki (University of Tokushima, Tokushima, Japan) and Tasuku Honjo (Kyoto University, Kyoto, Japan). Murine corneal endothelial cells were provided by Satoru Yamagami (University of Tokyo Graduate School of Medicine, Tokyo, Japan). 

Supported by Grants-in-Aid for Scientific Research (No. 20592073) from the Ministry of Education, Culture, Sports, Science and Technology, Japan. 

Disclosure: S. Sugita, None; Y. Kawazoe, None; A. Imai, None; Y. Usui, None; M. Takahashi, None; M. Mochizuki, None