PD-L1 Expression on Human Ocular Cells and Its Possible Role in Regulating Immune-Mediated Ocular Inflammation

Abstract

purpose. To assess the expression of PD-L1 and PD-L2 on human ocular cells and their potential to regulate ocular inflammation.

methods. Five categories of human ocular cells were evaluated for PD-L1 and PD-L2 expression by RT-PCR and flow cytometry. Three normal eyes and an inflamed eye from a patient with sympathetic ophthalmia were examined by immunohistochemistry for in situ PD-L1 expression. The immunomodulatory functions of PD-L1 and PD-L2 were tested by coculturing untreated or IFN-γ–pretreated ocular cells with activated human peripheral blood T cells for 48 hours and assessing T-cell production of IFN-γ, TNF-α, IL-4, and IL-5 by ELISA and T-cell apoptosis by flow cytometry.

results. PD-L1 protein was expressed constitutively in 4 of 5 human ocular cell lines, and its expression was significantly upregulated after stimulation by IFN-γ. Moreover, in situ expression of PD-L1 in inflamed ocular tissues was remarkably upregulated compared with normal eyes. Although PD-L2 expression was detectable by flow cytometry on 3 of 5 ocular cell lines, immunohistochemical staining did not show expression of PD-L2 on either normal or inflamed ocular tissues. IFN-γ, TNF-α, and IL-5 production by activated T cells cocultured with ocular cells was significantly enhanced in the presence of anti–PD-L1 blocking antibody. However, ocular cell–expressed PD-L1 and PD-L2 did not induce T-cell apoptosis.

conclusions. PD-L1 expressed on human ocular cells has a presumptive role in controlling ocular inflammation by inhibiting the production of proinflammatory cytokines and a Th2 cytokine by activated T cells. This may represent an important mechanism for maintaining immune privilege in the eye.

The eye has limited capacity for regeneration, and, as a result, immune-mediated inflammation can have devastating consequences for normal ocular cells and can even lead to blindness. However, immune-mediated inflammation and allograft rejection are greatly reduced in the eye, a phenomenon called immune privilege. Ocular immune privilege is the product of multiple anatomic, physiologic, and immunoregulatory processes. ^{1 2 3} Antigens introduced into the eye elicit a unique form of systemic immune deviation, termed anterior chamber-associated immune deviation (ACAID). ^{3 4} ACAID has been shown to promote corneal allograft survival ^{5 6} and to mitigate experimental autoimmune uveitis. ⁷ The aqueous humor that fills the anterior chamber contains an enormous array of anti-inflammatory and immunosuppressive molecules, such as TGF-β and macrophage migration inhibitory factor, that act on innate and adaptive immune responses. ^{8 9 10 11 12} In addition, cell membrane–bound molecules such as Fas ligand (FasL) and tumor necrosis factor-related apoptosis-inducing ligand (TRAIL) expressed on ocular cells can induce apoptosis of inflammatory cells that pass the natural barriers and reach the eye. ^{13 14 15 16} Thus, multiple factors and mechanisms contribute to ocular immune privilege. 

Programmed cell death (PD)-1 is a receptor belonging to the CD28/CTLA-4 family that is expressed on a subset of thymocytes, ^{17 18} activated T and B cells, ^{19 20 21} and myeloid cells. ¹⁷ PD-1 has two known ligands: programmed death ligand-1 (PD-L1) and programmed death ligand-2 (PD-L2), both of which are cell membrane-bound molecules. PD-L1 and PD-L2 are type 1 transmembrane glycoproteins belonging to the B7 family. ^{22 23 24 25} PD-L1 is expressed in a wide variety of tissues and by a number of different cell types, and its expression is upregulated by the proinflammatory cytokine IFN-γ. ^{26 27} The expression of PD-L2 is more restricted and appears to be limited to a subset of bone marrow–derived cells, including dendritic cells and macrophages. ^{23 26} It has been reported that the interaction of PD-1 with PD-L1 or PD-L2 results in the downregulation of T-cell proliferation and cytokine production and the induction of T-cell apoptosis. ^{26 28 29 30} The expression of PD-L1 on murine cornea was also reported to play a role in downregulating local immune response and maintaining long-term acceptance of corneal allografts. ^{31 32}  

The aim of our study was to assess the expression of PD-L1 and PD-L2 on normal and inflamed human ocular cells and its potential to suppress immune-mediated inflammation by modulating proinflammatory cytokines and Th2 cytokine production by activated T cells. 

Materials and Methods

Cells

Human retinal pigment epithelial ARPE-19 cells and chronic myelogenous leukemia K562 cells were obtained from the American Type Culture Collection (Rockville, MD). Human telomerase-immortalized corneal epithelial cells (HCE) were a generous gift from James Jester (University of California, Irvine). Human HPV E6/E7-transformed corneal endothelial cells (HCNs) were established in our laboratory as previously described. ³³ Human corneal stromal cells and iris/ciliary body (ICB) cells were primary cultures derived from scleral rims obtained from the Transplantation Center in the University of Texas Southwestern Medical Center. Briefly, to establish a primary culture of corneal stromal cells, the sclera and iris remnants were removed, and epithelial and endothelial surfaces of each rim were swabbed with 70% EtOH to kill any residual epithelial or endothelial cells. The tissues were then placed in complete KGM medium, minced to approximately 1-mm pieces with a sterile scalpel blade, and incubated in 5% CO₂ at 37°C. The method of primary culture of iris/ciliary body cells was similar to that of the corneal stromal cells except that iris remnants from the scleral rims were isolated and cultured in Dulbecco modified Eagle medium (DMEM). 

HCE cells and corneal stromal cells were maintained in keratinocyte medium (KGM Bullet Kit; Clonetics, Walkersville, MD). All the other cells were maintained in complete DMEM (BioWhittaker, Walkersville, MD) containing 10% fetal bovine serum (Hyclone, Logan, UT). The establishment of the human ocular cells and all research performed in this study adhered to the tenets of the Declaration of Helsinki. All research in this study involving human subjects was approved by the Institutional Review Board at the University of Texas Southwestern Medical Center (no. 032008-025). 

Cytokines, Reagents, and Antibodies

Human recombinant interferon (IFN)-γ was purchased from Sigma-Aldrich (St. Louis, MO). Anti-human CD3 antibody (clone HIT3a) and PE–anti-human CD3 antibody were purchased from BD PharMingen (San Jose, CA). Functional grade purified mouse anti-human PD-L1 antibody (clone MIH1), anti-human PD-L2 antibody (clone MIH18), mouse IgG1 isotype, human IFN-γ ELISA, IL-4 ELISA, IL-5 ELISA, and TNF-α ELISA were purchased from eBioscience (Ready-SET-Go kit for all; San Diego, CA). Anti-human/mouse caspase 3 (active) antibody was purchased from R&D Systems (Minneapolis, MN). 

Cell Stimulation

Cells were grown to 80% confluence and stimulated with 500 U/mL recombinant human IFN-γ in complete DMEM for 48 hours. Cells were then tested for PD-L1, PD-L2 mRNA, and protein expression by RT-PCR and flow cytometry, respectively. 

Reverse Transcription–PCR

Total cellular RNA was prepared from lysed tumor cells using an RNA isolation kit (RNAqueous; Ambion, Austin, TX). The first strand of cDNA was synthesized with a synthesis kit (iScript cDNA; Bio-Rad, Hercules, CA), and 0.5 μL resultant cDNA was used in a 50-μL reaction containing 0.1 μM each primer, 200 μM dNTP, 1.5 mM MgCl₂, 1× reaction buffer, and 1 U Taq polymerase (Invitrogen, Carlsbad, CA). Primer sequences for human PD-L1 were as follows: forward, 5′-TTG GGA AAT GGA GGA TAA GA-3′; reverse, 5′-GGA TGT GCC AGA GGT AGT TCT-3′ (IDT, Coralville, IA). Primer sequences for human PD-L2 were as follows: forward, 5′-AAA GAG CCA CTT TGC TGG AG-3′; reverse, 5′-TGA AAG CAA TGA TGC AGG AG-3′. Human GAPDH was used as an internal control, and the primer sequences were as follows: forward, 5′-ACC ACA GTC CAT GCC ATC AC-3′; reverse, 5′-TCC ACC ACC CTG TTG CTG TA-3′. 

An initial PCR denaturation step was performed at 94°C for 4 minutes. General cycling parameters for PCR were as follows: denaturation at 94°C for 45 seconds, annealing at 56°C (PD-L1), 53°C (PD-L2), or 56°C (GAPDH) for 45 seconds, and extension at 72°C for 45 seconds for 35 cycles with a final extension step at 72°C for 10 minutes. PCR amplification products were run on 1.5% agarose gels (Bio-Rad), prestained with 1× nucleic acid stain (GelStar; Cambrex Bioscience Rockland Inc., Rockland, ME), and visualized with an imager (Typhoon 9410; GE Healthcare, Piscataway, NJ). 

Flow Cytometric Analysis

Expression of human PD-L1 and PD-L2 protein on ocular cells was assessed by flow cytometry. In brief, single-cell suspensions were prepared and washed in fluorescence-activated cell sorter (FACS) buffer consisting of phosphate-buffered saline (PBS; pH 7.2) containing 2% fetal bovine serum. Cells were incubated with anti–PD-L1 antibody (2 μg/mL), anti–PD-L2 antibody (2 μg/mL), or mouse IgG1 isotype control (2 μg/mL) for 60 minutes at 4°C, washed three times, and incubated with FITC-labeled secondary antibody (BD PharMingen) for 30 minutes at 4°C. The cells were then washed three additional times in PBS, fixed in 2% formalin, and assessed for fluorescence using a flow cytometer (FACScalibur; BD Biosciences, San Diego, CA). 

Expression of PD-1 on T cells was assessed using anti-human PD-1 primary antibody followed by FITC-conjugated secondary antibody and PE–anti-human CD3 antibody by flow cytometry. The percentage of PD-1⁺ T cells was determined as PD-1⁺ cells among gated PE-CD3⁺ cells. 

T-cell apoptosis after coculturing with ocular cells was assessed using anti-human caspase-3 active antibody, followed by FITC-conjugated secondary antibody and PE–anti-human CD3 antibody by flow cytometry. Apoptotic T cells were identified as caspase-3–active positive cells among gated PE-CD3⁺ cells. 

Immunohistochemistry on Normal and Inflamed Eyes

Cell membrane expression of PD-L1 and PD-L2 in three normal eyes and in one inflamed eye (sympathetic ophthalmia) was determined by immunohistochemistry using anti-human PD-L1 or anti-human PD-L2 monoclonal antibody, respectively. Normal eye and inflamed eye samples were embedded in paraffin and cut into 4-μm sections. Tissue sections were incubated with 1 μg/mL anti–PD-L1, anti–PD-L2, or mouse IgG1 isotype control, followed by use of an ABC kit (Vectastain Elite; Vector Laboratories, Burlingame, CA) and counterstaining with methyl green, as described. ³⁴  

Coculture of Human Ocular Cells with Peripheral T Cells

Heparinized blood samples were obtained from six healthy donors. Peripheral blood mononuclear cells (PBMNs) were isolated by density gradient centrifugation. Human ocular cells were pretreated with or without recombinant IFN-γ (500 U/mL) and were grown to 80% confluence, harvested, washed to remove IFN-γ, and cocultured with PBMNs in the presence of 1 μg/mL anti-human CD3 at the effector/target (E:T) ratio of 1:10 for 48 hours. In parallel experiments, 10 μg/mL anti–PD-L1 antibody, anti–PD-L2 antibody, or mouse IgG1 isotype control was added to each coculture system for 48 hours. Supernatants were harvested 48 hours later and assessed for IFN-γ, TNF-α, IL-4, and IL-5 secretion by ELISA. Cells in the cocultures were assessed for T-cell apoptosis by flow cytometry, as described. 

Statistical Analysis

Results were expressed as mean ± SD of at least triplicate samples. Data were analyzed using the Student’s t-test. P < 0.05 was considered statistically significant. Each assay was performed at least three times. 

Results

Expression of PD-L1 and PD-L2 by Human Ocular Cells with or without IFN-γ Stimulation

PD-L1 and PD-L2 mRNA expression in human ocular cells was evaluated by RT-PCR. Total cellular RNA from human ocular cells was prepared, and PD-L1 (534 bp) or PD-L2 (492 bp) mRNA was detected with specific primers by RT-PCR. As shown in Figure 1 , all five human ocular cell types tested constitutively expressed PD-L1 and PD-L2 mRNA. Proinflammatory cytokine IFN-γ stimulation for 48 hours induced a modest upregulation of PD-L1 and PD-L2 mRNA expression in all five categories of human ocular cells. Human chronic myelogenous leukemia K562 cells that did not express PD-L1 or PD-L2 mRNA served as the negative control. ²⁸  

PD-L1 protein expression on human ocular cells was assessed by flow cytometry. Examples of PD-L1 protein expression on HCE and retinal pigment epithelial (ARPE-19) cells are depicted in Figures 2A and 2B , respectively. PD-L1 protein was constitutively expressed at various levels on four of the five ocular cells (HCE, HCN, ICB, ARPE-19; Fig. 2C ). Although corneal stromal cells expressed PD-L1 mRNA, they did not constitutively express PD-L1 protein on the cell membrane, which suggests a posttranscriptional control of PD-L1 surface expression. Moreover, as shown in Figure 2C , the percentage of PD-L1–expressing cells in all the ocular cells except ARPE-19 increased significantly after IFN-γ stimulation (P < 0.01). Although 92% of the ARPE-19 cells constitutively expressed PD-L1 protein, IFN-γ stimulation further increased PD-L1 expression, as shown by the elevated mean fluorescence intensity of PD-L1 protein on these cells (Fig. 2B) . 

We also examined PD-L2 protein expression on human ocular cells by flow cytometry. Examples of PD-L2 protein expression on HCE and ARPE-19 cells are depicted in Figures 3A and 3B . Three of five ocular cells (HCN, ICB, ARPE-19) constitutively expressed PD-L2 protein. A posttranscriptional control of PD-L2 surface expression was evident in HCE and corneal stromal cells because they constitutively produced PD-L2 mRNA but failed to express its protein. Proinflammatory cytokine IFN-γ stimulation for 48 hours induced significant upregulation of PD-L2 expression on all five of the ocular cell lines (Fig. 3C ; P < 0.01). Mean fluorescence intensity of PD-L2 surface expression on ARPE-19 cells increased twofold after IFN-γ stimulation. 

Expression of PD-L1 and PD-L2 in Normal and Inflamed Eyes

Given that in vitro cultures of ocular cells might not reflect PD-L1 and PD-L2 expression in situ, we examined eye sections from three normal eyes and one sympathetic ophthalmia eye for PD-L1 and PD-L2 expression by immunohistochemical staining. Examples of PD-L1 staining in the normal eye and the inflamed eye are shown in Figure 4 . In normal eye samples, PD-L1 staining was detected in corneal epithelial cells, corneal endothelial cells, iris/ciliary body cells, and retinal pigment epithelial cells, which was consistent with the flow cytometry data. The inflamed eye tissue showed more intensive PD-L1 staining in corneal epithelial cells, corneal endothelial cells, iris/ciliary body cells, and retinal pigment epithelial cells. Although corneal stromal cells of normal eyes did not express PD-L1 in situ, the stromal cells of the inflamed eye showed a remarkable upregulation of PD-L1 expression. However, contrary to flow cytometry data, in situ PD-L2 expression was not detected in either the normal eyes or the inflamed eye (data not shown). 

Effect of Ocular Cell–Expressed PD-L1 and PD-L2 on Proinflammatory Cytokine Production by T Cells

IFN-γ and TNF-α are important proinflammatory cytokines involved in immune-mediated ocular inflammation. To test whether ocular cell–expressed PD-L1 and PD-L2 affect immune-mediated ocular inflammation, we cocultured HCE or ARPE-19 cells with human peripheral blood mononuclear cells at an E:T ratio of 1:10 for 48 hours in the absence or presence of blocking antibodies to PD-L1, PD-L2, and mouse IgG1 isotype control, and we assessed IFN-γ and TNF-α production of T cells by ELISA. Anti-CD3 was added to the cocultures to activate T cells to produce IFN-γ and TNF-α. After 48 hours of stimulation with 1 μg/mL anti-CD3, 63% to 85% of the PBMNs were CD3⁺, and 81% to 96% of the CD3⁺ cells expressed PD-1 on their surfaces (Table 1) . 

As shown in Figure 5 , adding anti–PD-L1 blocking antibody to the cocultures significantly enhanced IFN-γ and TNF-α production by activated T cells (P < 0.01), which indicated that PD-L1 expressed on ocular cells suppressed the production of these two proinflammatory cytokines. Adding mouse IgG1 isotype controls to the cocultures did not have any effect on IFN-γ or TNF-α production, which reconfirmed the specificity of ocular cell–expressed PD-L1 in the suppression of IFN-γ and TNF-α production. Although PD-L2 can also affect T-cell cytokine production, ³⁰ adding anti–PD-L2 blocking antibody to the cocultures did not have any effect on IFN-γ or TNF-α production, which indicated that PD-L2 expressed on ocular cells did not inhibit proinflammatory cytokine production by human T cells. 

IFN-γ stimulation of HCE cells or ARPE-19 cells resulted in a profound suppression of IFN-γ and TNF-α production by human T cells compared with the ocular cells not treated with IFN-γ, probably because of upregulation of PD-L1 on ocular cells. 

Effect of Ocular Cell–Expressed PD-L1 and PD-L2 on Th2 Cytokine Production by T Cells

To examine whether ocular cell–expressed PD-L1 and PD-L2 affect Th2 cytokine production by T cells, we used ELISA to examine IL-4 and IL-5 production by T cells in the cocultures. As shown in Figure 6 , IL-4 could not be detected in supernatants from activated T-cell cultures. However, IL-5 was detected but was significantly reduced when T cells were cocultured with human ocular cells (P < 0.05). Adding anti–PD-L1 to the cocultures significantly restored the production of IL-5 (P < 0.05), which indicated that PD-L1 expressed on ocular cells specifically suppressed the production of IL-5 cytokine by activated T cells. Adding mouse IgG1 isotype controls to the cocultures did not have any effect on IL-5 production, which reconfirmed that specificity of ocular cell–expressed PD-L1 in the suppression of IL-5 production. By contrast, adding anti–PD-L2 blocking antibody to the cocultures did not have any effect on IL-5 production. 

Effect of Ocular Cell–Expressed PD-L1 and PD-L2 on Apoptosis of T Cells

To examine whether ocular cell–expressed PD-L1 and PD-L2 induced T-cell apoptosis, we set up cocultures, as described, and assessed T-cell apoptosis by flow cytometry using anti-human caspase-3 active antibody and anti-human CD3 antibody. As shown in Figure 7 , T cells in cocultures with human ocular cells underwent significantly enhanced apoptosis compared with activated T cells alone (P < 0.05). However, adding anti–PD-L1 or anti–PD-L2 did not affect the apoptosis of T cells, indicating that ocular cell–expressed PD-L1 or PD-L2 did not induce T-cell apoptosis and that other immunosuppressive molecules expressed on ocular cells, such as FasL ¹⁴ and TRAIL, ^{15 16} might account for the T-cell apoptosis. 

Discussion

These results demonstrate that PD-L1 is expressed constitutively in 4 of 5 categories of human ocular cells, and its expression is significantly upregulated after stimulation by the proinflammatory cytokine IFN-γ. Moreover, PD-L1 expression in inflamed eye tissues is remarkably upregulated in situ compared with the normal eyes. The results also indicate that PD-L1 in the eye is capable of suppressing T-cell production of the proinflammatory cytokines IFN-γ and TNF-α and the Th2 cytokine IL-5. 

Ocular immune privilege is maintained by multiple anatomic, physiologic, and immunoregulatory processes. No single mechanism truly defines immune privilege of the eye, and none of the mechanisms of immune privilege is unique to the eye. However, the molecules and mechanisms associated with immune privilege are best developed and are expressed more extensively in the eye than in any other organ, with the possible exceptions of the brain and the pregnant uterus. 

Recently, the newly recognized cell membrane–bound molecules PD-L1 (B7-H1), PD-L2 (B7-DC), and B7-H4 have been found to contribute to immune regulation. Among them, PD-L1 is thought to deliver inhibitory signals to T cells. There is evidence that engagement of PD-L1 with its receptor, PD-1, inhibits T-cell proliferation, induces apoptosis of immune cells, and suppresses cytokine secretion by T cells. ^{21 26 28} Our results indicate that PD-L1 is constitutively expressed in the eye and is upregulated in at least one ocular inflammatory disease, sympathetic ophthalmia. In vitro results support this observation and indicate that the proinflammatory cytokine IFN-γ upregulates PD-L1 expression on multiple types of ocular cells. Although PD-L1 mRNA was detected in all ocular cells tested, including HCE, corneal stromal, HCN, ICB, and ARPE-19 cells, the primary cultured corneal stromal cells did not constitutively express PD-L1 protein on their surfaces, which indicates posttranscriptional control of this immune regulatory factor. The mechanisms regulating posttranscriptional PD-L1 expression are still unknown. The proinflammatory cytokine IFN-γ significantly upregulated PD-L1 expression on all ocular cells, even on corneal stromal cells which did not express PD-L1 constitutively. In situ immunohistochemical staining of PD-L1 in normal or inflamed eye tissues was consistent with the findings with cultured cells. PD-L1 was detected in normal eye tissues, including corneal epithelium, endothelium, iris/ciliary body, and retinal pigment epithelium, which indicates that it is readily available to function in the normal eye. Importantly, PD-L1 expression was remarkably higher in inflamed ocular tissues than in normal ocular tissues, suggesting that in situ upregulation of PD-L1 might be a compensatory response to immune-mediated ocular inflammation and might serve to reestablish ocular immune homeostasis. PD-L1 is strategically expressed in the eye near areas that constitute the blood-ocular barrier and in locations with the opportunity for interaction between ocular tissue and inflammatory cells. In the cornea, PD-L1 is expressed on the epithelium and endothelium, suggesting its importance in controlling inflammatory cells that would enter from the anterior chamber or the conjunctiva. PD-L1 is expressed on the iris and ciliary body, where it can contact and inactivate inflammatory cells that emigrate from vessels during the rejection of corneal transplantation. ³⁵ In the retina, PD-L1 is expressed on the retinal pigment epithelium, the outermost retinal layer that interfaces with the highly vascularized choroid. 

Expression of PD-L1 on ocular cells could suppress the production of IFN-γ and TNF-α by activated T cells in a number of ways. IFN-γ is a Th1 proinflammatory cytokine that has pleiotropic effects. It can increase major histocompatibility complex class II expression on many cells, including APCs. ³⁶ It upregulates the expression of cell adhesion molecules on vascular endothelial cells, thereby enhancing the migration of immune cells into the eye, ³⁷ and it induces the production of chemotactic molecules leading to the recruitment of inflammatory cells. ^{38 39 40 41} Thus, IFN-γ not only influences Th1 inflammation, it enhances the capability of immune cells to gain entry to extravascular tissue in Th1 and Th2 inflammatory responses. 

TNF-α is another proinflammatory cytokine that plays an important role in ocular inflammatory diseases. It plays a key role in the pathogenesis of experimental noninfectious posterior segment intraocular inflammation and experimental autoimmune uveoretinitis, both of which can be mediated by Th1 cells. ^{42 43 44} Neutralization of TNF-α inhibits macrophage activity and NO production, which contributes to the reduction of photoreceptor damage in these models. ⁴⁵ It was also reported that inhibition of TNF-α activity prolongs corneal allograft survival. ⁴⁶ TNF-α may also have a role in experimental retinal neovascularization in rabbit ⁴⁷ and murine models. ⁴⁸ Thus, suppression of IFN-γ and TNF-α production by PD-L1 expressed on ocular cells may be an important mechanism for controlling immune-mediated inflammation in the eye. 

IFN-γ stimulation of HCE cells or ARPE-19 cells resulted in a profound suppression of IFN-γ and TNF-α production by human T cells when compared with the ocular cells not treated with IFN-γ, probably because of the upregulation of PD-L1 on ocular cells. Another explanation for this phenomenon is the induction of other immune-suppressive molecules by IFN-γ, such as IDO, ^{49 50 51} FasL, ^{52 53} and TRAIL, ¹⁶ because adding anti–PD-L1 blocking antibody to those cocultures did not restore IFN-γ or TNF-α production to the same level as the cocultures in which ocular cells were not prestimulated by IFN-γ. 

Ocular cell–expressed PD-L1 not only suppresses the production of proinflammatory cytokines by T cells, it inhibits the production of IL-5, a Th2 cytokine. Th2 cytokines are crucial in the pathogenesis of some ocular inflammatory diseases, such as allergic conjunctivitis. Thus, inhibition of IL-5 production by PD-L1 expressed on ocular cells may be a mechanism for controlling ocular allergic diseases. 

Although PD-L1 can influence T-cell apoptosis, we did not detect specific PD-L1–induced T-cell apoptosis. The higher apoptotic rate of T cells in the cocultures might be attributed to other immunosuppressive molecules expressed on ocular cells, such as FasL and TRAIL. Moreover, this confirms that the suppression of cytokine production by PD-L1 was not caused by the apoptosis of activated T cells. 

Usui et al. ⁵⁴ reported that cultured human retinal pigment epithelial cells constitutively expressed PD-L2. Although we could also detect PD-L2 expression on 3 of 5 ocular cell types by flow cytometry, in situ immunohistochemical staining did not show the expression of PD-L2 on normal or inflamed eye tissues, probably because the in vitro–cultured ocular cells could change phenotypes by immortalization with telomerase or transformation with HPV E6/E7. Moreover, PD-L2 expressed on the cultured ocular cells did not affect the proinflammatory cytokine production by activated T cells in vitro. It was reported that PD-L2 may interact with receptors other than PD-1. ^{55 56} Although the anti–PD-L2 antibody we used is a functional blocking antibody to PD-1, it might not block the binding of PD-L2 to other receptors, which may be why we did not see any effect on cytokine production when using anti–PD-L2 antibody. 

Inflammatory ocular diseases are characterized by marked accumulation of activated T cells and monocytes/macrophages in the inflamed eye lesions. Our findings demonstrate that the expression of PD-L1 on ocular cells may play an immunosuppressive role in ocular inflammation by inhibiting proinflammatory cytokine production by activated T cells and, as a result, may contribute to the immune privilege of the eye. This mechanism should be considered when designing novel immunotherapy for immune-mediated inflammatory diseases in the eye. 

 Supported by National Institutes of Health Grant CA30276 and by an unrestricted grant from Research to Prevent Blindness, Inc.

 Submitted for publication June 6, 2008; revised August 13, 2008; accepted October 27, 2008.

 Disclosure: W. Yang, None; H. Li, None; P.W. Chen, None; H. Alizadeh, None; Y. He, None; R.N. Ho gan, None; J . Y. Niederkorn, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

Expression of PD-L1 and PD-L2 mRNA in ocular cells. Total cellular RNA was prepared from untreated or IFN-γ–treated ocular cells. PD-L1 or PD-L2 mRNA was detected with specific primers by RT-PCR. (A) HCE. (B) Primary stromal cells. (C) HCN. (D) Iris/ciliary body cells. (E) ARPE-19. (F) K562 served as negative control. −, untreated cells; +, IFN-γ–pretreated cells.

Figure 2.

 

Expression of PD-L1 protein on ocular cells. Untreated or IFN-γ–pretreated human ocular cells were incubated with anti-human PD-L1 antibody or isotype control antibody, washed, incubated with FITC-labeled secondary antibody, and examined by flow cytometry. (A, B) FACS profiles of HCE cells (A) and ARPE-19 cells (B). Filled profile: isotype control. Dashed line: without IFN-γ stimulation. Bold line: with IFN-γ stimulation. (C) Percentage of ocular cells expressing PD-L1 protein. Data represent the average of three independent experiments performed in triplicate. **P < 0.01 compared with IFN-γ–untreated cells.

Figure 3.

 

Expression of PD-L2 protein on ocular cells. Untreated or IFN-γ–pretreated human ocular cells were incubated with anti-human PD-L2 antibody or isotype control antibody, washed, incubated with FITC-labeled secondary antibody, and examined by flow cytometry. (A, B) FACS profiles of HCE cells (A) and ARPE-19 cells (B). Filled profile: isotype control. Dashed line: without IFN-γ stimulation. Bold line: with IFN-γ stimulation. (C) Percentage of ocular cells expressing PD-L2 protein. Data represent the average of three independent experiments performed in triplicate. **P < 0.01 compared with untreated cells.

Figure 4.

 

PD-L1 expression on normal and inflamed eyes. Paraffin-embedded sections of the normal or inflamed eyes were incubated with anti–PD-L1 or mouse IgG1 isotype control antibody and counterstained with methyl green. (A–C) Immunohistochemical staining of anti–PD-L1 on epithelium and stroma (A), endothelium (B), and retina (C) of the normal eye. (D) Mouse IgG1 isotype control staining of normal cornea. (E–G) Anti–PD-L1 staining on epithelium and stroma (E), endothelium (F), and retina (G) of the inflamed eye. (H) Mouse IgG1 isotype control staining on corneal epithelium of the inflamed eye. Scale bar: (A, C–E, G, H) 100 μm; (B, F) 20 μm.

Table 1.

 

Expression of PD-1 on T Cells from Six Donors

Donor	T Cells in PBMNs (%)	PD-1+ T Cells (%)
1	65.67 ± 0.30	91.21 ± 1.91
2	64.16 ± 0.89	96.55 ± 0.15
3	63.66 ± 1.15	88.19 ± 0.12
4	79.00 ± 0.26	85.89 ± 1.11
5	85.30 ± 0.79	81.85 ± 0.83
6	79.20 ± 0.36	84.38 ± 0.38

 Isolated PBMNs from healthy donors were stimulated with 1 μg/mL anti-human CD3 for 48 hours. The percentages of T cells were assessed using PE–anti-human CD3 antibody by flow cytometry. The expression of PD-1 on T cells was examined using anti-human PD-1 primary antibody followed by FITC-conjugated secondary antibody and PE–anti-human CD3 antibody by flow cytometry. The percentage of PD-1⁺ T cells was determined as PD-1⁺ cells among gated PE-CD3⁺ cells.

Figure 5.

 

Influence of ocular cell–expressed PD-L1 or PD-L2 on proinflammatory cytokine production by T cells. Human PBMNs were isolated from healthy donors and cocultured with untreated or IFN-γ–pretreated ocular cells at the E/T ratio of 1:10 in the presence of anti-CD3 for 48 hours. Ten μg/mL anti–PD-L1, anti–PD-L2, or mouse IgG isotype control was added to the cocultures, respectively. Supernatant of the cocultures was assessed for IFN-γ (A) or TNF-α (B) production of activated T cells by ELISA. Data are representative of three independent experiments performed in triplicate. *P < 0.05; **P < 0.01.

Figure 6.

 

Effect of ocular cell–expressed PD-L1 or PD-L2 on Th2 cytokine production by T cells. Human PBMNs were isolated from healthy donors and cocultured with untreated or IFN-γ–pretreated ocular cells at the E/T ratio of 1:10 in the presence of anti-CD3 for 48 hours. Ten μg/mL anti–PD-L1, anti–PD-L2, or mouse IgG isotype control was added to the cocultures, respectively. Supernatants were collected and assessed for IL-4 (A) and IL-5 (B) production by ELISA. Data are representative of three independent experiments performed in triplicate. *P < 0.05; **P < 0.01.

Figure 7.

 

Effect of ocular cell–expressed PD-L1 or PD-L2 on T cell apoptosis. Human PBMNs were isolated from healthy donors and cocultured with untreated or IFN-γ–pretreated ocular cells at the E/T ratio of 1:10 in the presence of anti-CD3 for 48 hours. Ten μg/mL anti–PD-L1, anti–PD-L2, or mouse IgG isotype control was added to the cocultures, respectively. T-cell apoptosis was assessed using anti-human caspase-3 active antibody followed by FITC-conjugated secondary antibody and PE–anti-human CD3 antibody by flow cytometry. The apoptotic T cells were identified as caspase-3 active positive cells among gated PE-CD3⁺ cells. Resting T cells and anti-CD3 stimulated T cells served as background controls. Staurosporine-treated T cells served as the positive control for the apoptosis assay. Data are representative of three independent experiments performed in triplicate. *P < 0.05; **P < 0.01.

The authors thank Elizabeth Mayhew, Jessamee Mellon, and Christina Stevens for technical assistance.