Inflammatory Cytokines Induce Apoptosis of Corneal Endothelium through Nitric Oxide

Abstract

purpose. Proinflammatory cytokines are integral components of the allogeneic response to a corneal transplant and contribute to the pathogenesis of graft failure that results from irreversible damage to donor corneal endothelium. As yet, the mechanism and effectors of tissue damage during graft rejection remain unidentified. In the current study, the synergistic apoptotic effect of sustained proinflammatory cytokine insult was investigated in excised cornea and in transformed and primary corneal endothelial cells.

methods. Apoptosis was assessed by tissue- and flow cytometry–based TUNEL staining. Downstream signaling events of cytokine stimulation and subsequent activation status of endothelium were studied by RT-PCR and Western blot analysis. Cellular production of NO was examined by the Griess reaction.

results. Prolonged exposure (48 hours) of corneal endothelium to IL-1, IFNγ, and TNF (100 ng/mL each) resulted in induction of apoptosis. Synergy in induction of apoptosis was found after exposure to cytokine combinations. Cytokine-mediated cytotoxicity was correlated with high and sustained (up to 36 hours) endothelial activation (specifically through NF-κB, p38, and STAT-1), upregulation of inducible nitric oxide synthase (iNOS), and elevated de novo production of NO. Pharmacologic inhibition of iNOS elicited complete cytoprotection from inflammatory cytokine insult.

conclusions. The specific release of proinflammatory cytokines from alloreactive infiltrating cells, in combination with the inflamed environment of a corneal allograft, results in apoptosis in the corneal endothelium. This effect is mediated by the de novo generation of NO and sustained activation of NF-κB, p38, and STAT-1. Inflammatory cytokine-induced apoptosis presents a new target for the development of interventions to prevent or attenuate endothelial injury in graft rejection.

The cornea is the most commonly transplanted tissue. The nondividing corneal endothelial cells (CECs) are critical in maintenance of corneal transparency. The integrity of the endothelial monolayer of the donor cornea is thus essential to transplant function and survival. Allogeneic rejection is the commonest cause of corneal graft rejection in all reported series. ^{1 2} Understanding the final effector phase of endothelial cell loss during corneal allograft rejection is key to the development of therapeutic strategies to prolong graft survival. In the setting of inflammation after corneal transplantation, several mechanisms of cellular damage are implicated, for example Fas/Fas-ligand interaction, perforin and granzyme release, reactive oxygen species, and inflammatory cytokines. ^{3 4 5}  

Proinflammatory cytokines TNF, IFNγ, and IL-1 are integral components of an inflammatory allogeneic response and there are several potential sources of cytokine release within the cornea and anterior chamber. Cellular infiltrates of professional antigen-presenting cells (APCs), activated lymphocytes, ^{6 7} neutrophils, and NK cells ⁸ have all been detected in rejected corneal grafts in human and experimental transplantation. Infiltrating macrophages and CD4⁺ lymphocytes are the primary sources of TNF during a rejection episode as well as releasing IFNγ and IL-1. Contributing to the inflammatory milieu, activated cytotoxic T lymphocyte (CTLs), NK cells and neutrophils predominantly generate IFN-γ. The cellular constituents of the cornea themselves have a profound capacity to produce inflammatory cytokines. Zhu et al. ⁹ detected elevated levels of IL-1α and TNF in a murine model of orthotopic allograft transplantation, which were attributed to resident corneal cells, as cytokine upregulation was observed before peak infiltration of host inflammatory cells. Under physiological stress, the cornea itself and more specifically the corneal endothelium have been found to synthesize IL-1α, IL-1β, and TNF. ^{10 11} Strategies to interfere with the activities of IL-1 and TNF have been shown to extend corneal graft survival, indicating a possible role for these inflammatory cytokines in the pathogenesis of corneal allograft rejection. ^{12 13} However, the incomplete protection of grafts in these studies suggests there are alternative mechanisms of cellular injury involved. Thus, although the effector cell phenotypes in corneal allograft rejection have become better defined, there is much less information on the effector molecules in graft cellular injury. 

Apoptosis is a distinctive mechanism of active cellular death and plays a key role in the attrition of organ transplants in vivo, primarily as a mediator of immune-mediated transplant rejection. ^{14 15 16} Of the cytokines mentioned, TNF is the only one capable of directly inducing cellular apoptosis through a well-characterized signaling pathway. More recently, several studies have shown that with particular combinations of inflammatory cytokines, there is a potentiation of cytotoxicity. Indeed, elevated proinflammatory cytokine expression and their synergistic activities are implicated as a mechanism of pathogenesis in several clinical and inflammatory disorders, such as autoimmune diabetes IFNγ/TNF, ¹⁷ IFNγ/IL-1β, ¹⁸ and chronic autoimmune thyroiditis (IL-1β/TNF/IFNγ). ¹⁹ For example, the combined effects of IFNγ, IL-1β, and TNF have been found to induce apoptosis of smooth muscle cells in vitro and are implicated as a mechanism of plaque development in the chronic inflammatory condition atherosclerosis. ²⁰ Furthermore, this effect has previously been reported in murine vascular endothelial cells, where combined stimulation of endothelium with TNF and IFNγ again results in a substantial increase in levels of apoptosis observed relative to cells stimulated with each cytokine alone. ²¹  

We present evidence suggesting cytokine-induced death as a mechanism for injury to the corneal endothelium, as a consequence of the synergistic proapoptotic effect of sustained exposure to TNF, IFNγ, and IL-1. We have identified physiologically relevant conditions in which inflammatory cytokines can mediate cell death of murine CECs. We have gone on to investigate the apoptotic signaling cascades resulting from cytokine exposure, examining the roles of nitric oxide synthase (NOS) signal transduction and nitric oxide (NO). 

Methods

Corneal Endothelial Cells

An SV40-immortalized endothelial cell line derived from the BALB/c mouse corneal endothelial cells (MCECs) was the kind gift of Jerry Niederkorn (University of Texas Southwestern Medical Center, Dallas, TX). ²² Cells were passaged in Eagle’s MEM with Earles balanced salt solution (EBSS; BioWhittaker, Walkersville, MD), enriched with 10% fetal calf serum, 2 mM l-glutamine (Invitrogen, Paisley, UK), 1% antibiotic mixture (100 U/mL penicillin, 100 μg/mL streptomycin, 25 μg/mL amphotericin B; Cambrex Biosciences, Wokingham, UK), 1% sodium pyruvate (Sigma-Aldrich, Dorset, UK) and 1% MEM vitamin mixture (Cambrex Biosciences). Routine subculture and harvesting of cell monolayers was performed with 0.02% EDTA and 0.05% Trypsin (Invitrogen). 

Primary murine corneal endothelial cells (PMCECs) were generated by placing dissected corneas from BALB/c mice, endothelium-side down on gelatin-coated (1% vol/vol in PBS) culture dishes and allowing endothelial cells to migrate onto the dish over 48 to 72 hours in complete Eagle’s MEM medium. Corneas were subsequently removed, and adherent cells maintained for a further 2 to 4 days in medium without amphotericin B. PMCECs were trypsin/EDTA harvested and placed in fresh gelatin-coated culture dishes. Second-passage cells were used for experimentation. Cells were daily replenished with fresh medium until grown to confluence. Purity of PMCEC cultures was validated by RT-PCR analysis of corneal cell-specific expression based on a method previously described by Chen et al. ²³ All cell cultures were maintained in a humidified incubator at 37°C and 5% CO₂. 

Cytokine Treatment

Recombinant murine cytokines TNF, IL-1β, IL-1α, and IFNγ (all from PeproTech EC, London, UK) were added directly to cell cultures at final concentrations of between 0.01 and 100 ng/mL (TNF and IFNγ, 200 U/ng). Cell cultures were always cultured to 70% confluence during routine cell culture and before experimental stimulation to avoid a differential response to TNF. 

Manipulation of NO

Cell samples were pretreated with the NOS inhibitor, N ^G-nitro-l-arginine-methylester (l-NAME) hydrochloride (Alexis Corp., Nottingham, UK), for 2 hours at concentrations of 1 to 5 mM and then cocultured with the appropriate cytokines for the remaining incubation period. ²⁴ The specific iNOS enzyme inhibitor, 1400W (N-(3-[aminomethyl]benzyl)acetamidine; Sigma-Aldrich), was cocultured with experimental samples at various concentrations (5–30 μM) for the complete incubation period. Cells were exposed to exogenous NO by addition of the NO donor compounds, DD1 (3-bromo-3,4,4-trimethyl-3,4-dihydrodiazete 1,2-dioxide) and DetaNONOate (NOC-18l both from Alexis Corp.). Both compounds were added to culture medium alone and assayed between 6 and 48 hours by using the Griess reaction (detailed later) to measure the relative amount of NO released. 

Apoptosis Analysis

Apoptosis was quantified by detecting DNA fragmentation using a TUNEL and propidium-iodide–based assay involving a DNA fragmentation kit (ApoDIRECT; Cambridge Biosciences, Cambridge, UK), according to the manufacturer’s instructions. Samples were analyzed by flow cytometry (Facscalibur; BD Biosciences, Oxford, UK), with FITC-dUTP detected on the FL-1 channel and PI on the FL-3 channel. 

Murine corneas were maintained in complete EMEM for up to 48 hours with 100 ng/mL of each TNF, IL-1α, and IFNγ and also in combination with l-NAME (5 mg/mL). Corneas were also stimulated with 1 μg/mL staurosporine (Sigma-Aldrich) for 24 hours as a positive control for apoptosis. Whole-cornea TUNEL staining was performed with another kit (ApoAlert kit; BD Biosciences). Briefly, tissues were washed in PBS by immersion for 30 minutes and then incubated in 100 μL of 20 μg/mL proteinase K solution for a further 10 minutes at room temperature. Tissues were washed in PBS and fixed in 4% formaldehyde for 10 minutes. After two further washes in PBS, tissues were stained in 100 μL staining solution provided with the kit and incubated at 37°C for 60 minutes. Corneas were washed with PBS for 45 minutes, mounted in medium containing PI (DakoCytomation, Cambridge, UK), and analyzed by confocal microscopy (Radiance 2000; Bio-Rad, Hercules, CA) of the corneal endothelium. 

Reverse Transcription–Polymerase Chain Reaction

Preparation of cDNA and RT-PCR analysis of experimental cell samples and corneal tissues was undertaken using methods previously described. ²⁵ PCR reactions using 0.1 μg of cDNA template were performed on a thermal cycler (Omnigene; Hybaid, London, UK) using the following conditions: 2 minutes at 95°C (1 cycle), 30 seconds at 95°C, 45 seconds at the appropriate annealing temperature, and 30 seconds at 72°C (28 cycles). PCR reactions were completed by incubation for a further 10 minutes at 72°C (1 cycle). Amplified DNA bands were separated by 1.5% agarose gel electrophoresis, stained with ethidium bromide, and visualized using a UV transilluminator. ²⁶ Relative changes in levels of amplified PCR products were determined by visual comparison with the corresponding expression of a stable housekeeping gene. 

Primers and annealing temperatures used for the generation of a 983-bp PCR product for glyceraldehyde-3-phosphate-dehydrogenase (GAPDH) were 5′-atgcccccatgtttgtgatg-3′ and 5′-atggcatggactgtggtcat-3′ (55°C); for a 329-bp product for collagen VIII were 5′-cctggacccaaaggagaaggtgg-3′ and 5′-cctttggggcccggaatcccag-3′ (66°C); for a 499-bp product for decorin 5′-caagaacctgaaggacttgc-3′ and 5′-ggagttccctcagatgagg-3′ (60°C); for a 765-bp product for keratin 14 5′-cactgaactggaggtgaag-3′ and 5′-ttctgctgctccatctcg-3′ (58°C); and for a 505-bp product for endothelial (e)NOS 5′-gcagaagagtccagcgaaca-3′ and 5′-ggcagccaaacaccaaagtc-3′ (59.4°C). Primers used for the detection of iNOS have been described. ²⁷  

Quantification of NO Production

Levels of NO generated by experimental cell cultures (10⁶ cells) were detected with a modified Griess reaction protocol, based on detection of total concentrations of NO stable end products, nitrate (NO₃ ⁻ ) and nitrite (NO₂ ⁻ ). ²⁸ Sampled cell culture supernatants were first treated with nitrate reductase to convert all NO₃ ⁻ into NO₂ ⁻ . Concentrations of nitrite were measured using a spectrophotometric-based colorimetric assay kit (Cambridge Biosciences), according to the manufacturer’s instructions, and analyzed on a plate reader at 540 nm. Nitrite levels are expressed as micromolar concentrations determined from 10⁶ cells in 1-mL culture volumes, and accordingly correspond to total nitrite quantities in picomoles per cell. 

Western Blot Analysis

Cell lysates (40 μg protein) were prepared by standard techniques and separated on 12% SDS-polyacrylamide gels for Western blot analysis. ²⁹ Protein was transferred and immunoprobed using previously described electrophoretic transfer and blotting methods. ²⁹ Membranes were probed using the following primary antibodies and dilutions: mouse IgG₁ anti-phosphorylated-IκBα (Ser32/36; 1:200; New England Biolabs, Hitchin, UK); goat anti-phosphorylated-Stat-1 (Ser727; 1:250; Autogen Bioclear, Calne, UK); mouse IgG₁ anti-β-actin (1:7500; Sigma-Aldrich); mouse anti-phosphorylated-p38 (Thr180/Try182; 1:200; New England Biolabs); rabbit anti-Bcl-x_L (1:200; Autogen Bioclear). Horseradish peroxidase (HRP)–conjugated secondary antibodies (all from DakoCytomation) were used for detection. Blots were developed using the electrochemiluminescence system (Amersham Pharmacia Biotech, Buckinghamshire, UK) and x-ray photographic film (Eastman Kodak, New York, NY). Expression of the housekeeping gene β-actin was used as an internal control for equal sample loading. In addition, developed films were used for semiquantitative densitometry analysis (GelDoc-It Bioimaging; UVP Ltd., Cambridge, UK). Protein expression was measured in optical density units (ODU) and normalized to the corresponding sample expression of β-actin. Relative x-fold differences between untreated and treated samples were calculated using these normalized values (normalized expression of treated cells divided by normalized expression of untreated cells). 

Statistical Analysis

For calculation of statistical significance, each of the treated cell samples were compared to unstimulated control cells with an unpaired one-tailed t-test. P < 0.05 was considered to be statistically significant. Data are reported as the mean ± SD. 

Results

Inflammatory Cytokine-Induced Apoptosis in CECs

The cytotoxic effects of TNF, IL-1, and IFNγ were evaluated by exposing the CEC line MCECs to various combinations of cytokines (1–100 ng/mL each cytokine). At 24 hours, no significant effect on cell survival resulted from exposure to any cytokine (P > 0.05; Fig. 1 ). After 48 hours, treatment with each cytokine alone had a minimal effect on cell survival, inducing at most a 2.9 ± 1.1-fold increase (9% ± 6.1% apoptotic cells) in apoptosis relative to untreated cells. However, at 48 hours, considerable apoptosis was observed after stimulation with particular combinations of cytokines: TNF with IFNγ (9.5 ± 0.6-fold increase; P < 0.05), TNF/IFNγ with IL-1α (15.6 ± 2.5-fold increase; P < 0.01), and TNF/IFNγ with IL-1β (17.2 ± 2.0-fold increase; P < 0.01), with the latter two cytokine combinations resulting in up to 59% and 62% apoptosis (Fig. 1) . This pattern of cytokine-induced cell death was also seen after exposure to proinflammatory cytokines at lower concentrations of 1 to 10 ng/mL each (data not shown). 

Similar patterns and kinetics of cytokine-induced apoptosis were also seen in the cultured PMCECs. Maximum apoptosis of PMCECs was induced by stimulation with TNF, IL-1, and IFNγ compared with unstimulated cells after 48 hours of treatment (Fig. 2) . Similarly, exposure to each cytokine alone induced minimal apoptosis (maximum 1.3-fold increase, TNF) and combined treatment with TNF and IFNγ resulted in substantial induction of apoptosis (5.3-fold increase, 10.2% apoptosis; data not shown) relative to control samples. 

Cytokine-Induced Apoptosis and Sustained Endothelial Activation

To demarcate cell signal transduction of each inflammatory cytokine, Western blot analysis of key molecules in the signaling cascade was performed at 6 to 36 hours after cytokine treatment. NF-κB activity was evaluated based on levels of phosphorylated I-κBα (phosphorylation of I-κBα releases the active NF-κB complex). Resting CECs have a basal level of constitutive NF-κB activation, with low levels of phosphorylated I-κBα detected at all time points studied (Fig. 3A) . After 6 hours, NF-κB activity increased by 1.9-fold on stimulation with TNF alone and by 1.3-fold by IL-1 alone, whereas IFNγ had no detectable effect compared with untreated samples. Double cytokine combinations of TNF with IL-1 or IFNγ resulted in stronger (2.4-fold increase) NF-κB activation. In contrast, cell stimulation with IL-1 and IFNγ resulted in a smaller increase in phosphorylated I-κBα (1.8-fold). Greater induction of NF-κB activity was observed on stimulation with all three proinflammatory cytokines simultaneously (3.4-fold). A similar pattern of NF-κB activation was induced by all cytokine stimulations after 24 hours of treatment, where combined proinflammatory cytokines induced a maximum 2.8-fold increase. After 36 hours, cytokine-induced NF-κB activation remained elevated (maximum, 2.6-fold) although detection of the housekeeping gene β-actin showed protein levels were diminished due to substantial apoptosis in the cell population. This sustained high activation of NF-κB indicates that endothelial susceptibility to proinflammatory cytokines was not a consequence of downregulated antiapoptotic genes such as Bcl-x_L, which are positively regulated by this transcription factor (Fig. 3B) . 

Protein kinase p38 participates in signaling cascades controlling cellular responses to cytokines and environmental stresses. P38, activated by MAP kinase (MKK3)–mediated phosphorylation, can activate several transcription factors (MEF2, Elk-1, ATF-2, CREB, AP-1, and STAT-1), as well as inducing a proapoptotic cellular response. No p38 activation was detected in resting cells; however, elevated levels were observed after cytokine stimulation with TNF, IL-1, and, to a lesser extent, IFNγ. Phosphorylated p38 increased after dual stimulation with IL-1 and TNF and increased most substantially after stimulation with all three cytokines at all time points (Fig. 3B) . 

The STAT proteins operate another important pathway of cellular activation by cytokines. Activation of STATs by phosphorylation allows their nuclear translocation where they bind to DNA regulatory elements to affect gene transcription. Low levels of STAT-1 activation were observed after MCEC stimulation with IL-1α and -1β alone (Fig. 3B) , with increased levels of activation after stimulation with IFNγ alone. Although TNF itself did not induce phosphorylation, combined treatment with IFNγ augmented STAT-1 activation. Stimulation of MCECs with all three cytokines simultaneously resulted in similar high levels of STAT-1 activation. This pattern of STAT-1 phosphorylation persisted and was detectable at 24 hours and 36 hours after stimulation. 

Our study of key transcriptional activators induced by cytokine stimulation shows that CEC susceptibility to cytokine-induced apoptosis is associated with elevated and sustained (up to 36 hours) activation of NF-κB, p38, and STAT-1. 

Cytokine-Induced De Novo NO Generation through iNOS Upregulation

To understand the mechanism of proinflammatory cytokine synergism resulting in potentiation of their cytotoxicity, MCEC and PMCEC culture supernatants were sampled at 24 and 48 hours after cytokine treatment and assayed for the presence of nitrite (NO₂ ⁻ ) as a measure of NO production. Resting MCEC cultures were found to generate nitrite at relatively low levels, with up to 14.6 μM of NO₂ ⁻ detected per 10⁶ cells over 48 hours (Fig. 4A) , corresponding to 14.6 pmol NO₂ ⁻ produced per cell. Similarly, an equivalent number of PMCECs generated up to 10.3 μM NO₂ ⁻ over 48 hours (Fig. 4B) . Highest levels of nitrite were induced by simultaneous stimulation with TNF, IL-1, and IFNγ, resulting in 61.8 ± 10.9 μM NO₂ ⁻ generated by MCECs (P < 0.05; Fig. 4A ) and up to 101.3 ± 17.4 μM NO₂ ⁻ by PMCECs (P < 0.01; Fig. 4B ). Elevated levels of nitrite were also evident after stimulation of endothelial cells with dual cytokine combinations, particularly with TNF and IFNγ, which stimulated PMCECs to generate up to 51.8 ± 9.0 μM NO₂ ⁻ (P < 0.01). No significant change in nitrite generation relative to unstimulated control cell samples was detected in MCECs (P > 0.05) or PMCECs treated with each cytokine alone. Similarly, at lower concentrations (0.5–50 ng/mL each), combined cytokine treatment of MCECs induced significant and equivalent production of NO₂ ⁻ with ≤0.1 ng/mL, resulting in no changes compared with untreated cells (Fig. 4C) . 

An increase in cellular nitrite production was evident after 12 hours of double and triple cytokine treatment (Fig. 5) . Generated nitrite accumulated at a constant rate; however, after 24 hours of nitrite, production by triple-cytokine–stimulated cultures occurred at a higher rate (MCEC 1.1 ± 0.1 μM/h) than in cultures stimulated with double-cytokine combinations (maximum 0.69 ± 0.09 μM/h, TNF with IL-1β). The delayed kinetics of nitrite accumulation suggest that cytokine-induced production of NO necessitates protein synthesis. 

In the resting state, MCECs constitutively express endothelial nitric oxide synthase (eNOS), as shown by RT-PCR (Fig. 6) . The inducible isoform of NOS enzyme, iNOS, which was absent in resting cells, was markedly upregulated by stimulation with inflammatory cytokine combinations (Fig. 6) . An increase in iNOS mRNA levels also resulted from endothelial activation by TNF alone and by IFNγ alone. Double-cytokine combinations of TNF with IL-1 and IFNγ with IL-1α caused more striking induction of iNOS transcription. In comparison, expression of eNOS was downregulated after stimulation with all three inflammatory cytokines. Observed levels of nitrite accumulation (Fig. 4) thus corresponded well to iNOS expression levels, as maximal expression and activity of iNOS was induced by combined cytokine stimulation of MCECs with TNF, IFNγ, and IL-1. 

NO in Cytokine-Induced Apoptosis

To determine whether NO might be responsible for cytokine-induced apoptosis, MCECs were exposed to exogenous sources of NO by coculturing cells with the NO donor compounds DetaNONOate and DD1 for up to 48 hours. As both compounds have distinct kinetics of NO release in physiological conditions, the levels of NO liberated were quantified by detecting the relative amounts of nitrite produced by a range of donor compound concentrations. ³⁰ DetaNONOate (250–1000 μM) release of NO was rapid on addition to cell cultures, generating 69 to 126 μM of NO₂ ⁻ within 30 minutes (Fig. 7A) . NO release continued more slowly, generating 3.1 μM of NO₂ ⁻ per hour (1000 μM DetaNONOate). Comparatively, DD1 (500 μM) liberated much lower levels of NO, producing 0.22 μM NO₂ ⁻ per hour (Fig. 7B) . Endothelial susceptibility to exogenous NO was studied at 24 and 48 hours after exposure to NO donor compounds. After 24 hours, significant MCEC apoptosis (23.8% ± 6.2% apoptotic cells) was induced only by exposure to 1000 μM DetaNONOate, which liberated 126 to 183 μM NO₂ ⁻ over 24 hours (Fig. 7C) . A dose-dependent apoptotic response to DetaNONOate was observed after 48 hours, with induction of apoptosis on exposure to 250 to 1000 μM DetaNONOate (8.8%–38.5% apoptotic cells) corresponding to 69 to 111.5 μM of NO₂ ⁻ released over 48 hours. DD1 liberation of NO, which, at its highest applied dose, generated a maximum of 15.5 μM over 48 hours, failed to induce any cytotoxic effects (Fig. 7D) . These results correlate well with our observations of cytotoxicity induced by cytokine-mediated endogenous NO production, where cytokine stimulation induced MCEC production of up to 61.8 μM NO₂ ⁻ and resulted in significant apoptosis. In comparison, resting MCECs generated up to 14.6 μM NO₂ ⁻ with no cytotoxic effects. Thus, MCECs were able to withstand sustained exposure to relatively low levels of NO (15–40 μM NO₂ ⁻) and apoptosis of MCECs was induced only on exposure to higher (>60 μM NO₂ ⁻) concentrations of NO. 

Effect of Inhibition of iNOS on Cytokine-Induced Apoptosis

To determine whether NO is indeed the toxic mediator of cytokine-induced apoptosis, MCECs were pretreated with increasing doses of 1400W or l-NAME, after which inflammatory cytokine combinations were added and cultured for a further 48 hours. Both l-NAME and 1400W inhibit all isoforms of NOS; however, 1400W acts as a more potent and selective inhibitor of iNOS. ^{24 31} Cell cultures were assayed for production of nitrite and apoptotic cell death. Control cultures were found to accumulate 15 ± 3.3 μM of NO₂ ⁻ after 48 hours and up to 65 ± 13 μM of NO₂ ⁻ on stimulation with TNF, IFNγ, and IL-1. Coculture of cytokine-stimulated cells with 1400W and l-NAME reduced nitrite output to 19.4 ± 2.3 and 30.5 ± 10.8 μM, respectively (Figs. 8A 8B) . Although both NOS inhibitors suppressed cytokine induced NO production, 1400W was more effective than l-NAME, presumably due to the efficacy with which 1400W inhibits iNOS activity. Both NOS inhibitors significantly abrogated the cytotoxic effects of inflammatory cytokines and elicited almost complete cytoprotection. Inflammatory cytokine combinations induced a 10.6 ± 0.9-fold increase in apoptosis relative to control samples (Figs. 8C 8D) , which was reduced to a 1.5 ± 0.02- and 1.3 ± 0.9-fold increase in apoptosis in cell cultures containing 1400W and l-NAME, respectively. The correlation between knockdown of nitrite production and increased survival on treatment with NOS-inhibiting compounds clearly indicates that cytokine-induced damage of corneal endothelium is a consequence of iNOS-mediated production of NO. 

Prevention of Cytokine-Induced Corneal Injury by NOS Inhibitors

To study whether cytokines exert a cytotoxic effect on the corneal endothelium in situ, whole murine corneas were excised and incubated in medium containing 100 ng/mL each of TNF, IFNγ, and IL-1 for 24 and 48 hours. Corneas were studied for apoptosis by TUNEL staining and the endothelium examined en face by confocal imaging. Unstimulated corneal specimens showed areas of intact endothelial cells with their well-characterized hexagonal monolayer morphology (Fig. 9A) . Exposure to inflammatory cytokines for 24 hours had no effect on the integrity of the tissue (data not shown); however, after 48 hours, there was extensive damage to the cornea, with the disappearance of most endothelial cells and significant TUNEL staining of the tissue (Figs. 9B) . Preincubation of the cornea with l-NAME fully prevented cytokine-mediated tissue damage, demonstrated by maintenance of endothelial cell integrity and absence of any detectable TUNEL-positive staining (Fig. 9C) . 

Discussion

The results of the study demonstrate that (1) synergistic action of TNF, IL-1, and IFNγ trigger NF-κB, p38, and STAT-1 activities in corneal endothelium; (2) cytokine induction of high and sustained endothelial activation induces expression of iNOS and subsequent generation of NO; (3) de novo generation of NO by iNOS mediates cytokine-induced apoptosis; (4) cytokine-induced apoptosis of corneal endothelium can be prevented by inhibition of iNOS activity. We deduce from these observations that cytokines released from anterior chamber and graft-infiltrating cells in the course of the allogeneic response to cornea could, in combination, induce death of CECs, leading to graft failure. 

The synergistic activity of proinflammatory cytokines, increasing the cytotoxic potential of each cytokine, must be a consequence of the coupling of signal transduction pathways. TNF, along with its numerous other effector functions, is a potent inducer of apoptosis. As demonstrated in this study, corneal endothelium, akin to vascular endothelium, is largely resistant to TNF-induced apoptosis, even though corneal endothelium expresses TNFR1 (p55) and TNFR2 (p75) receptors. ³² This property is attributed to dual pro- and antiapoptosis signals initiated by TNF receptor stimulation and the subsequent activation of NF-κB, which promotes cell survival (Fig. 10A) . ³³ Our data suggest that for TNF to induce apoptosis in CECs, the concurrent signaling of one or more combining stress signals is necessary to overcome the inherent cytoprotective TNF signaling. TNF also mediates activation of JNK kinases, specifically p38. Activation of JNK/p38 has been associated with several forms of endothelial stress including the inflammatory response. ^{34 35} Furthermore, JNK/p38 signaling is believed to act cooperatively with NF-κB to modulate its activity. ³⁶ The exact mechanism by which p38 activates apoptosis is not clearly understood, as it is also associated with regulating cell survival in certain systems; however, during NO-induced cellular stress, p38 MAPK activation of p53 has been incriminated. ³⁷  

IL-1 signaling, which is mediated through MAPK activation of JNK and p38, can also lead to NF-κB signaling (Fig. 10B) . This convergence between cytokine signaling pathways may account for the capacity of IL-1 to induce cellular apoptosis when combined with other stimuli. 

In the cornea, IFNγ can activate infiltrating monocytes and other resident cells, resulting in upregulation of major histocompatibility complex (MHC) class I and II molecules, as well as regulating the expression of other proinflammatory cytokines, such as IL-2 and TNF. ³⁸ IFNγ signaling involves activation of receptor-associated JAK and the subsequent phosphorylation of STAT-1 (Fig. 10C) . Although this cytokine has not been demonstrated to exert any direct cellular cytotoxicity, it is known to as act as a sensitizing agent to other stimuli through its ability to activate vascular endothelium. ³⁹  

Our data suggest that the synergistic effect of cytokine combinations on apoptosis is mediated through the induction of iNOS. In an effect similar to that in vascular endothelium, ²¹ TNF stimulation results in NF-κB activation of CECs which is known to upregulate iNOS. ⁴⁰ However, in contrast to vascular endothelium, ⁴¹ no induction of iNOS was detected in corneal endothelium after IL-1 stimulation. IFNγ also induces iNOS expression through activation of JAK, STAT-1, and IFNγ response factor (IRF)-1 proteins (Fig. 10C) . IFNγ-activated STAT-1 interacts with the GAS element in the enhancer sequence of the iNOS promoter and acts to augment iNOS transcription. STAT-1 also induces expression of the transcription factor IRF-1, which then itself binds to specific DNA elements of the iNOS promoter to further promote iNOS expression. ⁴² Stimulation of corneal endothelium with cytokine combinations that induced high levels of activated NF-κB and STAT-1 also resulted in parallel levels of iNOS expression and activity. The synergistic induction of iNOS by TNF and IFNγ is known to occur at several levels of cell signaling. For example, as we have shown, NF-κB activity is synergistically increased by IFNγ and TNF. The mechanism may be the enhanced degradation of I-κB. ⁴³ Alternatively NF-κB, activated by TNF or IL-1, interacts with IFNγ-induced IRF-1. ⁴⁴ The resultant protein–protein interaction alters the iNOS promoter DNA structure, resulting in a conformational change to form an enhanceosome nucleoprotein complex with improved promoter activity. ⁴⁵ Manna et al. ⁴⁶ suggest that synergy between TNF and IFNα is mediated through suppression of TNF-induced cell survival signaling of NF-κB and p38-mediated AP-1 expression. This mechanism does not apply in the corneal endothelium, as we see no reduction in p38 or NF-κB activity after TNF, IL-1, and IFNγ stimulation of CECs. Thus, cooperation between major signal transducers of TNF, IFNγ, and IL-1 accounts for the synergistic induction of iNOS expression and substantial increase in NO generated by proinflammatory cytokine stimulation of CECs. 

NO is a multifunctional molecule with roles in synaptic signaling, regulation of vascular tone, and innate immunity. At low concentrations, NO is relatively unreactive, and most of its physiological functions are mediated through cellular cGMP. Free radical NO is generated by conversion of l-arginine to citrulline by NOS, which is expressed in several ocular structures, including the cornea. ^{47 48} Basal levels of NO released by corneal epithelium, stroma, and endothelium can regulate endothelial Na⁺/K⁺ ion pump function, maintaining corneal dehydration. ⁴⁹ Low level NO can also exert cytoprotective functions in other endothelial systems. ^{50 51}  

As demonstrated in this study, at higher concentrations, NO is also a potent cytotoxic effector molecule. As well as being implicated in the pathophysiology of clinical conditions such as myocardial infarction, ⁵² elevated NO is associated with inflammatory conditions within the eye, in which inflammatory cells infiltrating the anterior chamber contribute NO to the local environment. ^{53 54} The continuous release of NO by human corneas while in storage ex vivo, is also implicated in tissue injury before transplantation. ⁵⁵ Furthermore, increased incidence of NO and its byproducts in corneal disorders such as Fuchs’ endothelial disease, keratoconus, experimental allergic conjunctivitis, and LPS-induced inflammation, suggests NO contributes to the onset and pathologic course of these diseases. ^{56 57} O’Brien et al. ⁵⁸ reported an upregulation of iNOS in rabbit corneal fibroblasts and endothelial cultures exposed to TNF, IL-1β, and IFNγ and suggested that NO plays a regulatory role in corneal hydration during anterior ocular inflammation. Our study suggests that cytokine-induced iNOS and NO directly mediates corneal endothelial injury. 

We have demonstrated the complete inhibition of cytokine-induced apoptosis through application of iNOS inhibitor compounds. The low toxicity and high efficacy of 1400W and l-NAME allows their potential therapeutic use in ocular inflammatory diseases. Inhibition of iNOS activity alone would be of considerable interest because of the important homeostatic role and regulatory functions of low-level endogenous NO generated by eNOS. Recently, Strestikova et al. ⁵⁹ have demonstrated the role of iNOS in murine corneal allograft rejection, in which systemic administration of the iNOS inhibitor aminoguanidine significantly extended corneal allograft survival. Modulation of other cytoprotective proteins such as Bcl-2 have also been shown to elicit protection from NO-induced apoptosis and provides an alternative target for inhibition of cytokine-induced cellular damage. ⁶⁰  

Understanding the molecular processes of allograft rejection is paramount in the development of strategies to prolong corneal graft survival. The data presented by our study support the hypothesis that NO and iNOS play an important role in damaging the endothelium in inflammation. It also suggests that effective intervention to minimize damage to donor corneal endothelium in allograft rejection will either require an upstream multifaceted approach, in which, for example, either bioactive TNF or IFNγ must to be inhibited, or, alternatively, a more targeted downstream approach, in which key molecular mediators of endothelial function, such as NF-κB, JNK, and JAKs are targeted. 

 Supported by the Medical Research Council UK. AJTG is a Biotechnology and Biological Research Council research leave fellow.

 Submitted for publication April 19, 2004; revised July 26, 2004; accepted August 9, 2004.

 Disclosure: P. Sagoo, None; G. Chan, None; D.F.P. Larkin, None; A.J.T. George, 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.

 

Synergistic induction of apoptosis by combined inflammatory cytokine stimulation of MCECs. Cells were treated with various combinations of proinflammatory cytokines (100 ng/mL each) for 24 or 48 hours and analyzed with a flow-cytometry–based TUNEL assay. Apoptosis was quantified as an x-fold increase in the levels of apoptosis detected relative to the unstimulated control samples (% apoptosis in experimental samples ÷ % apoptosis of control sample). Data are the mean change (x-fold) in apoptosis ± SD (n = 4). Statistical significance is shown at *P < 0.05 and **P < 0.01, according to Student’s t-test, comparing cytokine-treated with untreated cell samples.

Figure 2.

 

Cytokine-synergistic induction of apoptosis of PMCECs, after simultaneous stimulation with all three proinflammatory cytokines for 48 hours. (A) Images of control PMCEC cultures at 0 and 48 hours with background apoptosis of 1.9% after 48 hours of culturing, as detected by the TUNEL assay. PMCEC cultures stimulated with 100 ng/mL each of TNF, IFNγ, and IL-1β, showed significant cell death (8.05-fold increase, 15.3%; P < 0.05) at 48 hours relative to control samples. Apoptosis was evident in cultured cells that became rounded and detached (arrows). Similarly, PMCECs stimulated with 100 ng/mL each of TNF, IFNγ, and IL-1α, resulted in the maximum levels of apoptosis being detected at 48 hours (11.1-fold increase, 21.1%; P < 0.05). (B) Representative flow cytometry dot plots of control (0 and 48 hours) and triple-combination, cytokine-stimulated PMCECs after 48 hours of cytokine treatment, where apoptotic cells are detected with the selected region (R1) defined by propidium iodide staining and incorporation of dUTP-FITC by TUNEL staining.

Figure 3.

 

MCECs were treated with combinations of proinflammatory cytokines (100 ng/mL each) for 6, 24, and 36 hours. Equal amounts of protein fractions from prepared cell lysates (40 μg protein) were separated by 12% SDS-PAGE and then transferred onto nitrocellulose membrane. The differential expression of (A) phosphorylated I-κBα (Phos-I-κBα) and (B) Bcl-x_L, phosphorylated p38 (Phos-p38) and phosphorylated STAT-1α/β (Phos-STAT-1α/β) in response to cytokine stimulation was analyzed by using standard immunodetection techniques. Elevated activation of NF-κB, p38, and STAT-1 was induced for up to 36 hours by simultaneous stimulation by TNF, IFNγ, and IL-1. Relative changes in protein expression were verified by comparison with the stable expression of the cellular cytoskeletal protein β-actin, and are shown only for Phos-I-κBα at all three time points (A). Semiquantitative densitometry was also performed and the results recorded as x-fold increases in protein expression compared with untreated cells, after correction for corresponding β-actin expression.

Figure 4.

 

Cytokine-induced (100 ng/mL each) production of nitrite by (A) MCECs and (B) PMCECs was detected after 24 and 48 hours of stimulation. (C) A titration of cytokine-induced production of nitrite was performed by treatment of MCECs with 0.01 to 100 ng/mL each of TNF, IL-1α, and IFNγ for 48 hours. Cell culture supernatants were assayed for relative levels of endothelial NO production, based on the Griess reaction, which detects the stable oxidation products of NO, nitrate (NO₃ ⁻) and nitrite (NO₂ ⁻). Data shown summarize mean nitrite levels ± SD (n = 3). Statistical significance is shown at *P < 0.05 and at **P < 0.01, according to Student’s t-test, comparing experimental samples with untreated cell samples.

Figure 5.

 

Kinetics of nitrite production by MCECs after cytokine stimulation. MCEC cultures (10⁶ cells) were stimulated with various combinations of cytokines (100 ng/mL each) and levels of NO₂ ⁻ measured from cell culture supernatants at 6, 12, 24, 36, and 48 hours. Cytokine-induced production and accumulation of NO₂ ⁻ began at 12 hours, with combination, but not single, cytokine-stimulated cells showing an elevated rate of production.

Figure 6.

 

RT-PCR expression analysis of eNOS and iNOS after 36 hours of cytokine stimulation. To ensure accurate detection of differential gene expression, equal amounts of cDNA from experimental samples were used for PCR amplification and validated by PCR amplification of the housekeeping gene GAPDH.

Figure 7.

 

An exogenous source of NO was applied to MCECs by NO donor compounds DetaNONOate (A) or DD1 (B). Detection of NO₂ ⁻ production using the Griess reaction assessed the relative liberation of NO by these two compounds at 0.5, 6, 24, and 48 hours when applied to culture medium (37°C). Exogenous NO induced apoptosis of MCECs was detected by TUNEL staining and flow cytometric quantification, on exposure to 1 to 1000 μM DetaNONOate (C) and 1 to 500 μM DD1 (D) after 24 and 48 hours exposure.

Figure 8.

 

Endothelial production of NO on exposure to TNF, IFNγ, and IL-1 (100 ng/mL each) was determined with the Griess reaction, measuring NO₂ ⁻ accumulation. Cytokine-stimulated cells were cocultured with increasing doses of 1400W (15–30 μM) (A) and l-NAME (2–5 μM) (B) to assess the ability of each compound to return cytokine-induced nitrite generation to the basal levels produced by control cell cultures (15 ± 3.3 μM NO₂ ⁻, 48 hours). The ability of 1400W (C) and l-NAME (D) to inhibit cytokine-induced apoptosis was determined by coculturing cytokine-stimulated cells with increasing doses of each compound and detecting apoptosis with the TUNEL assay.

Figure 9.

 

Whole murine corneas were maintained in complete culture medium alone (A), or stimulated with 100 ng/mL each TNF, IFNγ, and IL-1 for 48 hours (B). Corneas were also cocultured with all three cytokines and l-NAME (5 mM) for up to 48 hours (C). Apoptotic controls were generated by exposing corneas to 1 μg/mL staurosporine for 24 hours (D). Apoptotic damage was assessed by whole-tissue TUNEL staining (FITC-green) and PI counterstaining (red) and visualized by confocal microscopy. Apoptotic cells appeared as distinct, rounded, bright yellow nuclei (arrows), due to colocalization of condensed nuclear staining by PI (red) and the TUNEL reaction (green). Areas of intact CECs (elongated bars) were visible in control (A) and cytoprotected (C) samples, but were lost with cytokine treatment alone (B).

Figure 10.

 

Outline of intracellular cytokine signaling mechanisms that are active within endothelial cells. (A) TNF initiates signal transduction on TNFR1 receptor trimerization, which results in the recruitment of several adaptor molecules (TRADD, TRAF-2, RIP) and then in the dual activation of NF-κB and JNK/p38, and the subsequent expression of iNOS. Initiation of NF-κB signaling is mediated through activation of NF-κB–inducing kinase (NIK). TNF stimulation also activates the apoptotic caspase cascade, which, in noninflammatory conditions, is counteracted by the cytoprotective effects of NF-κB. (B) IL-1R1 recruits a series of cytoplasmic proteins (MyD88, IL-1RacP, IRAKS, and TRAF-6) forming a proximal IL-1R–signaling complex that links IL-1 stimulation to some of the same signaling pathways as TNF, such as NF-κB signaling. IL-1 signaling, similar to TNF, involves the concordant activation of both NF-κB and JNK/p38 pathways, also resulting in iNOS production. (C) IFNγ mediated activation of endothelium results in the expression of iNOS, through STAT-1 activation and also IRF-1, a STAT-1–inducible gene. IFNγ-activated STAT-1 forms hetero- and homodimeric complexes and translocates to the nucleus, where it binds to GAS sites of IFNγ-inducible promoters. Cytokine synergistic induction of apoptosis may be a consequence of combined signal-mediated induction of NO production and JNK/p38 proapoptotic stress-response effects.

The authors thank Stephen Moss, Institute of Ophthalmology, for the generous use of imaging facilities.