Contribution of TNF-α to Leukocyte Adhesion, Vascular Leakage, and Apoptotic Cell Death in Endotoxin-Induced Uveitis In Vivo

Abstract

purpose. To investigate the effect of TNF-α on leukocyte adhesion, vascular leakage, and apoptotic cell death in endotoxin-induced uveitis (EIU) in the rat.

methods. EIU was induced in Long-Evans rats by a single footpad injection of lipopolysaccharide (LPS; 350 μg/kg) from Salmonella typhimurium. A single injection of recombinant TNF receptor P75 (etanercept) was given subcutaneously 24 hours before the administration of LPS. Twenty-four hours after administration of LPS, leukocyte adhesion was evaluated in vivo with SLO-acridine orange angiography and ex vivo with concanavalin A lectin staining of retinal flatmounts. Neutrophil activation was quantified by a myeloperoxidase activity assay. Vascular leakage was assessed by Evans blue extravasation. Retinal cell death was assessed with TUNEL staining and quantified with a modified ELISA protocol. Involvement of caspase-3 and -8 was determined by M30 antibody staining, Western blot analysis, and a test for enzymatic activity.

results. Twenty-four hours after the LPS injection, significant increases in leukocyte rolling, adhesion, and activation were observed. In addition, increased levels of apoptosis in the vascular endothelium and the ganglion cell and inner nuclear layers and activation of caspase-8 and -3 were observed. After administration of the TNF-α inhibitor, significant reduction in the leukocyte rolling, adhesion, activation, and apoptosis in all the affected layers was observed. The quantitative analysis of vascular leakage revealed a significant decrease after treatment with etanercept. Retinal cell death quantification showed a significant decrease after treatment with the TNF-α inhibitor.

conclusions. Anti-TNF-α treatment reduces the LPS-induced increases in leukocyte rolling, adhesion, and vascular leakage in this rat model of inflammatory uveitis. These results suggest the involvement of TNF-α in inflammatory uveitis and its potential use as a therapeutic agent in the reduction of ocular inflammation.

Uveitis is one of the most damaging ocular conditions and can lead to edema, high intraocular pressure, and, ultimately, destruction of the intraocular tissues and blindness. Uveitis is associated with a number of diseases, including Behçet’s, ankylosing spondylitis, juvenile rheumatoid arthritis, Reiter’s syndrome, and inflammatory bowel disease. 

Endotoxin-induced uveitis (EIU) is widely accepted as an animal model of clinical uveitis. ^{1 2} The acute, inflammatory response seen in EIU peaks 24 hours after the injection of a single sublethal dose of bacterial lipopolysaccharide (LPS) into the footpad of an experimental animal. ^{1 3 4} The course of EIU appears to be dependent on the dose of LPS and the route of administration. ^{4 5} Among the prominent features of EIU are leukocyte adhesion, retinal cell injury and death, and protein leakage into the eye. 

Use of the current animal model of EIU has revealed several cytokines to be involved in the initiation of uveitis, such as IL-1, -2, -4, and -6; tumor necrosis factor (TNF)-α; and transforming growth factor (TGF)-β. ^{6 7 8 9} TNF-α is involved in the pathogenesis of a variety of inflammatory processes. During induction of EIU, the TNF-α level also increases in both aqueous and serum. ^{10 11} Intravitreous injection of human TNF-α in rabbit eyes induces leukocyte infiltration and protein leakage. ¹²  

TNF-α exerts its actions through two distinct receptors: TNF RI (p55) and TNF RII (p75). The TNF-induced cytotoxicity has been attributed in the past to the p55 receptor, and TNF-induced proliferation to the p75 receptor. ¹³ However, it has been shown, that p75 can greatly enhance p55-induced cell death. ¹⁴ Recently, two drugs, etanercept (Enbrel; Amgen, Thousand Oaks, CA) and infliximab (Remicade; Centocor, Malvern, PA) became commercially available as TNF-α antagonists. Etanercept is a fusion protein of human TNF receptor p75 and infliximab is a monoclonal IgG1 antibody against TNF-α. 

In the present study, we investigated the role of TNF-α in LPS-induced leukocyte rolling, adhesion, and tissue infiltration, and the breakdown of the blood–retinal barrier. In addition, we determined the relevance of TNF-α in EIU-associated endothelial and retinal cell death in EIU. Our study demonstrated a potential role for etanercept in the treatment of all major pathologic manifestations of EIU in the retina. 

Materials and Methods

Animals

All protocols abided by the ARVO Statement for the Use of Animals in Ophthalmology and Vision Research and were approved by the Animal Care and Use Committee of the Regierungspräsidium, Köln, Germany. Male Long-Evans rats weighting 250 g were used in all experiments. The animals were fed standard laboratory chow and allowed free access to water in an air-conditioned room with a 12-hour light–dark cycle. Except as noted otherwise, the animals were anesthetized with ketamine hydrochloride (30 mg/kg; Ketalar, Parke-Davis, Morris Plains, NJ) and xylazine hydrochloride (5 mg/kg; Rompun, Harver-Lockhart, Morris Plains, NJ) before all experimental manipulations. 

Induction of EIU

Animals received a footpad injection of 350 μg/kg body weight lipopolysaccharide (LPS) from Salmonella typhimurium (Sigma, St. Louis, MO) in phosphate-buffered saline (PBS). Animals with systemic effects of the LPS injection such as hyperventilation or weakness, were excluded from the analysis. In addition, animals with an inflammatory reaction in the anterior chamber that prevented the visualization and analysis of the retina, were excluded from the analysis. 

Treatment with the Soluble TNF-Receptor Etanercept

Soluble p75 TNF-α receptor/Fc (etanercept) was reconstituted with sterile water according to the manufacturer’s instructions. The soluble TNF-α receptor was administered subcutaneously at a dose of 0.3 mg/kg 24 hours before the LPS injection. This is the recommended dose for treatment of rheumatoid arthritis in humans, and it has been previously shown to be effective in the rat. ^{15 16 17 18} A minimum of three animals (n = 3) per treatment group was used for each experiment. 

Acridine Orange Leukocyte Fluorography and Fluorescein Angiography

Leukocyte dynamics were evaluated using acridine orange leukocyte fluorography (AOLF). ^{19 20} The rats were anesthetized with a mixture of xylazine hydrochloride (4 mg/kg) and ketamine hydrochloride (25 mg/kg). Immediately before AOLF, the pupil of the right eye was dilated with 1% tropicamide (Alcon, Humancao, Puerto Rico) to observe the static leukocytes. A focused image of the peripapillary fundus was obtained with a scanning laser ophthalmoscope (SLO; Rodenstock Instrument, Munich, Germany). Acridine orange (AO; Sigma, St. Louis, MO) was dissolved in 0.9% sterile saline (1.0 mg/mL) and 3 mg/kg was slowly injected through a tail vein catheter (approximately 0.5 mL/min, maximum dose was 2 mL in each rat). AO binds noncovalently to double-stranded nucleic acids and causes leukocytes and endothelial cells to fluoresce. The fundus was observed with SLO using the argon blue laser as the illumination source and a standard fluorescein angiography filter in the 40° field setting for 10 minutes. The images were recorded on a videotape at a rate of 30 frames per second. We counted the number of rolling leukocytes in major veins according to the method of Yamashiro et al. ²¹ Rolling leukocytes were defined as bright intravascular spots that moved more slowly than free-flowing bright spots. We counted the spots that passed through the vein each minute, in each major vein of an area one papillary diameter away from the optic disc and calculated the average number of the spots per minute per major vein. 

Ex Vivo Quantitation of Retinal Leukostasis

After the induction of deep anesthesia in the rat, the chest cavity was opened and a 14-gauge perfusion canula was introduced to the left ventricle. The right atrium was opened with a 12-gauge needle to achieve outflow. With the heart providing the motive force, 250 mL/kg PBS was administered from the perfusion canula to remove erythrocytes and nonadherent leukocytes. Fixation was then achieved by perfusion with 1% paraformaldehyde and 0.5% glutaraldehyde at a pressure of 100 mm Hg. At this point, the heart stopped. A systemic blood pressure of 100 mm Hg was maintained by perfusing a total volume of 200 mL/kg over 3 minutes. The inhibition of nonspecific binding with 1% albumin in PBS (total volume 100 mL/kg) was followed by perfusion with FITC-coupled concanavalin A lectin (20 μg/mL in PBS [pH 7.4], total concentration, 5 mg/kg body weight; Vector Laboratories, Burlingame, CA). The latter stained adherent leukocytes and the vascular endothelium. Lectin staining was followed by 1% bovine serum albumin (BSA)/PBS perfusion for 1 minute, and PBS perfusion alone for 4 minutes, to remove excess concanavalin A. ²² The retinas were flatmounted in a water-based fluorescence anti-fading medium (Fluoromount; Southern Biotechnology, Birmingham, AL) and imaged by fluorescence microscopy (Axioplan, FITC filter, 40×; Carl Zeiss, Oberkochen, Germany). Only whole retinas in which the peripheral collecting vessels of the ora serrata were visible were used for analysis. Leukocyte location was scored as being either arteriolar, venular, or capillary. The total number of adherent leukocytes per retina was counted. All experiments were performed in a masked fashion. 

Measurement of Neutrophil Activation in Retinal Invasion

Our previous in vivo and ex vivo methods investigating the leukocyte dynamics (AOLF and concanavalin A staining) focus mainly on the study of intravascular leukocytes and they do not take into account the leukocytes that extravasate and resident in the retinal tissue. As the myeloperoxidase (MPO) enzyme is found in all activated leukocytes, we used a modified MPO assay to quantify the total amount in the retinal tissue. Each retina was homogenized in a 100 μL solution consisting of 20 mM phosphate buffer (pH 7.4), containing 0.1% Tween 20. After centrifugation at 12,000g for 15 minutes, the supernatant was diluted in phosphate buffer, containing 1% BSA, Tween 20, and sodium azide. Protein concentration was determined with the bicinchoninic acid (BCA) method (Pierce, Rockford, IL). MPO analysis was performed using a sandwich ELISA (Oxis Research, Portland, OR) according to the manufacturer’s instructions. Briefly, 100 μL standard or tissue sample were pipetted into 96 wells coated with an anti-MPO antibody and incubated for 2 hours at room temperature. After repeated washes, the wells were incubated with 100 μL of an anti-antibody conjugated with horseradish peroxidase for 2 hours at 25°C on a shaker. After that, the samples were washed five times, and incubated with 200 μL of substrate buffer for 1 hour at 37°C. The reaction was stopped and the absorption measured by an ELISA plate reader at 450 nm. The tissue sample concentration was calculated from a standard curve and corrected for protein concentration. 

Measurement of Retinal Vascular Leakage

Quantification of retinal vascular permeability was measured 24 hours after the LPS injection, according to our previously published method. ^{16 23 24} After the animals were deeply anesthetized, Evans blue dye (Sigma) dissolved in normal saline (30 mg/mL) was injected through the tail vein over 10 seconds at a dosage of 45 mg/kg. Beginning at 2 minutes after the injection, at 15-minute intervals, and just before perfusion at 2 hours, blood samples were obtained from the iliac artery or the left ventricle to obtain the time-average Evans blue plasma concentration. These blood samples were centrifuged at 12,000 rpm for 15 minutes and diluted to 1/10,000th of their initial concentration in formamide (Sigma). The absorbance was measured with a spectrophotometer at 620 nm. The concentration of dye in the plasma was calculated from a standard curve of Evans blue in formamide. After the dye had circulated for 2 hours, the chest cavity was opened, and the rats were perfused through the left ventricle with citrate buffer (0.05 M, pH 3.5) for 2 minutes at a physiological pressure of 120 mm Hg. The retinas were then carefully dissected under an operating microscope. After measurement of the retinal dry weight, Evans blue was extracted by incubating each retina in 0.3 mL of formamide for 18 hours at 70°C. The extract was ultracentrifuged at a speed of 70,000 rpm for 45 minutes at 4°C. Sixty microliters of the supernatant was used for spectrophotometric measurement at 620 nm. Each measurement occurred over a five-second interval, and all sets of measurements were preceded by evaluation of known standards. The background-subtracted absorbance was determined by measuring each sample at 620 nm (the absorbance maximum for Evans blue in formamide) and 740 nm (the absorbance minimum). The concentration of dye in the extracts was calculated from a standard curve of Evans blue in formamide. Blood–retinal barrier breakdown was calculated using the following equation, with results being expressed in microliters of plasma per gram of retina (dry weight) × hours.  

\[\frac{\mathrm{Evans\ blue\ ({\mu}g)/retina\ dry\ weight\ (g)}}{\mathrm{Time-averaged\ Evans\ blue\ concentration\ ({\mu}g)/plasma\ ({\mu}l)\ circulation\ time\ (hr)}}\]

 

Results were expressed as a percentage of the value in control animals. 

Propidium Iodide Labeling In Vivo

Dead and injured endothelial cells were labeled in vivo using propidium iodide (PI; Molecular Probes, Eugene, OR). After the induction of deep anesthesia with 50 mg/kg intraperitoneal pentobarbital sodium, PI (1 mg/mL in PBS) was injected intravenously through the tail vein at a concentration of 5 μmol/kg (0.668 mL/200 mg BW). After 20 minutes, fixation by intracardiac perfusion and labeling with concanavalin A-lectin was performed as described earlier. Retinal flatmounts were examined by fluorescence microscopy. Labeled endothelial cells were distinguished from surrounding cells, especially pericytes, by distinct cellular outline and nuclear shape of the endothelial cells. All experiments were performed in a masked fashion. 

DNA Fragmentation ELISA

A modified ELISA method was used to quantify the amount of dead cells. This ELISA detects 5-bromo-2-deoxyuridine (BrdU)-labeled DNA fragments that occur during apoptosis, in tissue homogenates. Simultaneously with the LPS injection the animals received a single intraperitoneal injection of 20 mg/kg BrdU. Twenty-four hours after the injection, retinas were excised, lysed in buffer (4 M urea, 100 mM Tris, 20 mM NaCl, and 200 mM EDTA [pH 7.4]) and incubated with 4 mg/mL proteinase K at 55°C. Fragments of genomic DNA were isolated with a kit (Apoptotic DNA ladder kit; Roche Diagnostics, Mannheim, Germany) according to the manufacturer’s instructions and finally eluted in 100 μL of elution buffer per retina. The amount of fragmented DNA that corresponded to each eluate, and therefore to each retina, was subsequently quantified using the DNA fragmentation ELISA (Roche Diagnostics), according to the instructions of the manufacturer. Photometric readings were obtained at 450 nm and results expressed as a percentage of the level in LPS-treated eyes. 

TUNEL Histology

TUNEL was performed with horseradish peroxidase detection in 7-μm sections from formalin-fixed, paraffin-embedded retinas. Whole eyes from rats injected with LPS alone, LPS with the soluble TNF-α inhibitor, or the inhibitor alone were fixed in 4% paraformaldehyde overnight at 4°C. The TUNEL staining was performed as previously described. ²⁵  

Detection of Apoptosis-Related Enzymatic Activity

Cytokeratin 18 is altered in early apoptotic events. The M30 antibody (CytoDeath; Roche Diagnostics) is a monoclonal mouse IgG2b antibody that binds to a caspase-cleaved, formalin-resistant epitope of the cytokeratine 18 (CK18) cytoskeletal protein. Immunoreactivity of the M30 antibody is confined to the cytoplasm of apoptotic cells. Consequently, the M30 antibody allows reliable determination of caspase-related enzymatic activity early in the course of apoptosis. M30 immunoreactivity was evaluated in formalin-fixed, paraffin-embedded tissue sections of rats that received injections of LPS alone, LPS with the soluble TNF-α inhibitor, or the inhibitor alone. Briefly, 5-μm paraffin-embedded sections were deparaffinized, rehydrated, microwaved for 5 minutes in antigen retrieval reagent (Dako, Hamburg, Germany), treated for 30 minutes in methanol containing 0.5% H₂O₂, and incubated for 1 hour in incubation buffer (PBS containing 1% BSA and 0.1% Tween 20). The M30 antibody was applied for 1 hour at room temperature at a 1:50 dilution in incubation buffer. The sections were subsequently washed in PBS containing 0.1% Tween 20 and incubated with an anti-fluorescein-POD antibody (Roche Diagnostics) according to the manufacturer’s instructions. The peroxidase reaction was developed with 3,3′-diaminobenzidine (DAB), and the slides were counterstained with hematoxylin.  

Western Blot Analysis for Caspase-3 and -8

The levels of protein expression of caspase-3 and -8 were evaluated by Western blot analysis. Briefly, whole retinas were lysed for 30 minutes on ice in lysis buffer (50 mM Tris-HCl [pH 8], with 120 mM NaCl and 1% NP-40), supplemented with a mixture of proteinase inhibitors (Complete Mini; Roche Diagnostics). The samples were cleared by microcentrifugation (14,000 rpm, 30 minutes, 4°C) and assessed for protein concentration. Thirty micrograms of protein per sample was electrophoresed in a 12% sodium dodecyl sulfate (SDS)-polyacrylamide gel (SDS-PAGE), and electroblotted onto nitrocellulose membranes. After a 1-hour incubation in blocking solution (20% IgG-free normal horse serum, in PBS), the membranes were exposed overnight at 4°C to an anti-caspase-3 antibody (Santa Cruz Biotechnology, Heidelberg, Germany) or an anti-caspase-8 antibody (Upstate Biotechnology, Lake Placid, NY). After a wash in PBS, the respective secondary peroxidase-labeled antibody was applied at 1:10,000 dilution for 1 hour at room temperature. The proteins were visualized with the enhanced chemiluminescence technique (Amersham Pharmacia Biotech, Piscataway, NJ). 

Statistical Analysis

All results are expressed as the mean ± SD. Student’s t-test for populations with normal distribution and equal variance or two-sample t-test with the Welch correction was used. Differences were considered statistically significant at P < 0.05. 

Results

Leukocyte Rolling and Adhesion in the Retina after LPS Injection

According to the AFLO method, the amount of rolling leukocytes per major retinal vein per minute increased significantly in the rats after LPS injection, whereas it was significantly suppressed with the administration of the soluble TNF-α receptor (Fig. 1) . Twenty-four hours after LPS injection, the number of rolling leukocytes was markedly increased compared with the number in control animals (0.00 ± 0.00 vs. 51.75 ± 6.02; n = 3; P < 0.001). The soluble TNF-α receptor etanercept (0.3 mg/kg) reduced leukocyte rolling in retinal veins by 20% (41.50 ± 3.12; n = 3; P < 0.05), compared with nontreated LPS-injected animals. To elucidate the static leukocyte adhesion in more detail in the larger retinal vessels, we applied a perfusion technique using concanavalin A lectin (Fig. 2) . ²² Typically, leukocyte adhesion was found in venules after LPS injection (Fig. 2B) . Leukocyte adhesion was elevated 4.5-fold compared with the levels in control animals (48.8 ± 6.11 vs. 218.7 ± 92.5; n = 6; P < 0.01). Etanercept (0.3 mg/kg) reduced leukocyte adhesion in the retina by 41% (128.7 ± 48.0; n = 6; P < 0.05) compared with nontreated LPS injected animals. 

LPS-Induced Upregulation of the MPO Levels in the Retina

MPO is a hemoprotein that is stored in primary granules of neutrophils and released on activation of the cell. Because our homogenization procedure is mild and does not break the primary granules, the amount of MPO levels correlate in our model with the total number of activated neutrophils. Compared with the retinas of normal controls, the retinas of LPS injected animals demonstrated 1.7-fold higher MPO levels (57.4% ± 8.8% vs. 100% ± 17.3%; n = 6). Treatment with etanercept reduced the retinal MPO levels by 28.8% (71.2% ± 17.8%; n = 6; P < 0.05; Fig. 3 ). 

LPS-Induced Blood–Retinal Barrier Breakdown

Blood–retinal barrier breakdown is one of the major pathologic manifestations of uveitis. The effect of the TNF-α inhibitor on blood–retinal barrier breakdown was measured with the Evans blue permeability assay. In agreement with previously published data, blood–retinal barrier breakdown increased approximately 4.2-fold, 24 hours after the LPS injection (0.44 ± 0.29 vs. 1.89 ± 0.42 arbitrary units; n = 8; P < 0.001; Fig. 4 ). Treatment with etanercept resulted in a reduction of LPS-induced blood–retinal barrier breakdown by 60.3% (0.76 ± 0.34; n = 8; P < 0.001). 

LPS-Induced Apoptosis in the Retina

To quantify apoptotic cell death in the retina, a modified ELISA for fragmented DNA was used (Fig. 5) . Twenty-four hours after the injection of LPS, the amount of fragmented DNA had increased by almost 11.6-fold (8.61% ± 0.12% vs. 100% ± 4.81%; n = 6; P < 0.00001) compared with the control. Systemic treatment with etanercept reduced fragmented retinal DNA by 46.6% (53.42% ± 3.60%; n = 6; P < 0.00001), which was still 6.16-fold higher than in the control animals. 

Endothelial Cell Injury

To investigate the in vivo role of TNF-α in mediating retinal endothelial cell death in LPS induced uveitis, we assessed the effect of etanercept on retinal endothelial injury using a PI labeling assay (Fig. 6) . LPS injection increased significantly the amount of PI-labeled cells, which were mainly located in clusters, often adjacent to adherent leukocytes (12.83 ± 6.21 vs. 280 ± 101.6 PI-positive cells per retina; n = 6; P < 0.005). In contrast, the total number of PI-labeled endothelial cells per retina in eyes of animals treated with etanercept was reduced by 67.3% (101.7 ± 44.46; n = 6; P < 0.01). 

Retinal Endothelial Cell Apoptosis

To identify whether, and localize where, apoptosis is involved in LPS-induced retinal damage, formalin-fixed, paraffin-embedded retinal sections were stained using TUNEL and the M30 antibody (Fig. 7) . TUNEL reacts to fragmented DNA that is localized in the nucleus, therefore the staining was nuclear, whereas M30 stains caspase-cleaved cytokeratins that are located in the cytoplasm; therefore, the staining was cytoplasmic. In LPS-injected animals, TUNEL- and M30-positive cells were located predominantly in the ganglion cell layer, in the vascular endothelium as well as in the inner nuclear layer. Retinal vessels were not frequently encountered in our sections, but, where evident, they contained TUNEL- and M30-positive cells. In contrast, control animals showed little staining anywhere in the retina. Treatment of LPS-injected animals with the TNF-α inhibitor etanercept reduced the amount of TUNEL- and M30-positive cells. During apoptosis, caspases are activated in a hierarchical manner and serve to cleave structural and functional proteins that are needed for the cell survival. Cleaved products for both caspase-8 and -3 were found in the retinal lysates of rats that received LPS injection, proving the activation of these caspases in LPS-induced uveitis in the rat (Fig. 8) . Administration of etanercept significantly reduced the cleaved forms of both caspase-8 and -3 in the LPS-injected rats, which correlated with the reduced levels of apoptotic cells in these rats. 

Discussion

In the present study we investigated the potential involvement of the apoptotic factor TNF-α in the inflammatory processes of endotoxin-induced uveitis in a rat model. Our data demonstrate that LPS-induced uveitis resulted in increased leukocyte rolling; leukocyte sticking, predominantly in the retinal veins; and leukocyte activation, as assessed by the retinal MPO activity levels. The increase leukocyte activation and adhesion causes increased endothelial cell death, as indicated by positive PI staining. Increased apoptosis can be demonstrated by TUNEL staining in retinal sections in endothelial cells and throughout the ganglion cell and inner nuclear layers. Increased apoptosis in the retinal tissue was also demonstrated by cell-fragmentation ELISA. Active caspase-3 and -8 was found after LPS injection, and activity was localized to endothelial cells and neuronal cells by the M30 antibody, in a pattern comparable to that of TUNEL staining, indicating that these caspases are involved in LPS-induced apoptosis. The increased apoptosis resulted in increased levels of vascular leakage. Treatment with the TNF receptor p75 fusion protein etanercept reduced retinal leukostasis, apoptotic cell death, and vascular leakage. Our data demonstrate a role for TNF-α in all major aspects of EIU. 

The beneficial effect of the inhibition of endogenous TNF-α in our rat model of endotoxin-induced uveitis points to a major role of TNF-α in the pathogenesis of inflammatory uveitis. The role of TNF-α in EIU has been controversial. Although Smith et al. ²⁶ has found reduced inflammation in TNFR p55/p75–knockout mice after administration of LPS, previous studies have failed to replicate the findings. ²⁷ This variant effect of TNF-α inhibition on EIU is an indirect indication that other factors are operative in the pathogenesis of EIU. This is consistent with our data that show reduction but not complete regression of the EIU phenotype. In agreement, studies by different investigators have shown a milder EIU phenotype in knockout mice for various inflammatory cytokines and their receptors such as IL-1 and -6 (Kievit P, Park JM, O’Rouke LM, Planck SR, Rosenbaum JT, ARVO Abstract 2489, 1996), ^{27 28} pointing out that these cytokines also play a role in the pathogenesis of EIU. The degree of the cross talk between the pathways that are activated by all these factors has not been fully investigated. Another attractive hypothesis for the failure of the genetic ablation of TNF-α to reduce the symptoms of EIU in these prior studies is that compensatory pathways are operative in the TNF p55/75 mice that allow for a marked inflammatory reaction. ³  

In addition, differences in the therapeutic regimen and the time of evaluation in relation to the treatment can explain the variable result of the anti-TNF therapy with agents similar to the one used in our study. The beneficial effect of the inhibition of TNF-α in our study can be attributed to the early administration of the soluble TNF-α inhibitor. According to our hypothesis, which emanates from earlier findings, ^{10 29} TNF-α plays a crucial role early in the course of EIU. The increase of TNF-α levels early in the course of EIU mediates the early inflammatory phase and contributes to the initial accumulation of neutrophils and monocytes that invade the retina. The administration of the soluble TNF-α receptor before the LPS injection in our study allowed the inhibitor to reach therapeutic levels at the plasma early enough to block this initial phase ^{4 5 10 30} and thus resulted in a positive effect. Delayed treatment after the initiation of this early inflammatory phase by TNF-α may be less effective in blocking the deleterious sequelae that characterize EIU. 

In accordance with previous studies we have demonstrated increased leukocyte rolling in vivo in EIU. ^{19 20 30} According to our ex vivo studies, EIU is characterized by an increased leukocyte adhesion, predominantly in the veins. This pattern is in agreement with previous studies ^{1 30} and in sharp contrast to the pattern observed in other conditions, such as diabetic retinopathy, which demonstrates an equal distribution of adherent leukocytes between arterioles, venules, and capillaries. The factors contributing to this preferred pattern is currently unknown, but research is directed toward surface molecules that are known to play a role in adhesion and are differentially regulated in the different parts of the circulation, such as subcategories of ephrins. 

According to our data, administration of the TNF-α inhibitor reduced leukocyte rolling and adhesion significantly in our rat model of endotoxin-induced uveitis. This is in accordance to our previous findings in a rat model of streptozotocin induced diabetes, in which administration of the same inhibitor reduced leukocyte adhesion. This reduction correlated with a reduction in intercellular adhesion molecule (ICAM)-1 levels and NF-κB levels. TNF-α has been shown to regulate the activation of NF-κB in various models, whereas ICAM-1 is an adhesion molecule that has multiple NF-κB binding sites in its promoter. It is highly likely that early in the course of EIU, the increased levels of TNF-α, results in the activation of NF-κB and the upregulation of the ICAM-1 levels that lead to increased leukocyte adhesion in the vasculature. The role of ICAM-1 in EIU has been demonstrated in previous studies in which inhibition of ICAM-1 or leukocyte function-associated antigen (LFA)-1 resulted in reduced leukocyte adhesion, fewer infiltrating cells, and milder initial symptoms. ^{28 30} In addition, the reduced leukocyte rolling that we observed in our rat model of EIU after inhibition of TNF-α could be explained with the effect of the inhibitor to selectins, a class of leukocyte surface molecule that is influenced by TNF-α levels in a variety of models. 

We have demonstrated previously, that leukocyte adhesion is causally linked to endothelial cell injury and death in diabetes. ²² We hypothesize that similar mechanisms are operative in EIU. In our rat model of EIU, apoptotic death, as determined by TUNEL positivity, was localized in endothelial cells, ganglion cells, and the inner nuclear cell layer. After administration of the TNF-α inhibitor and the concomitant reduction of leukocyte adhesion, apoptosis was significantly reduced in the retinas of the EIU rats, in all layers. This result leads to the conclusion that leukocytes that adhere to the vasculature are directly responsible for the observed endothelial apoptosis, whereas extravasation of leukocytes into the retinal tissue is responsible for the apoptotic death observed in all the other cell layers. This is in accordance with our previous data in diabetes, in which inhibition of leukocyte-mediated death with administration of a neutralizing antibody against FasL reduces apoptotic death, not only in the endothelial cell layer but also in the ganglion and inner cell layers. 

Last, according to our data, blood–retinal barrier breakdown increased significantly early in the course of EIU, but inhibition of TNF-α reduced it. The increase in blood–retinal barrier breakdown correlates with the increase in leukocyte adhesion and apoptotic death, demonstrating the potential role of leukocyte-mediated endothelial injury to the integrity of the vasculature. The reduction of blood–retinal barrier breakdown after inhibition of TNF-α can be explained either by a direct effect of TNF-α on the leukocyte-mediating injury by the Fas/FasL mechanism, as just discussed, or by an indirect effect on other factors regulating the vascular leakage. Although our previous work failed to show downregulation of vascular endothelial growth factor (VEGF) after inhibition of TNF-α in a rat model of diabetic retinopathy, additional studies are necessary to investigate the role of this factor in TNF-α–induced vascular leakage in EIU. 

In conclusion, in the present study, increased leukocyte rolling and adhesion were associated with increased levels of apoptotic death and vascular leakage early in the course of EIU, whereas TNF-α inhibition reduced all phenomena. Several investigators have reported anti-TNF-α therapy for uveitis in experimental animals and also in humans; however, the effect of anti-TNF-α therapy on uveitis is still controversial in both experimental and clinical studies. ^{3 10 26 27 31 32} The presented data warrant further studies on the effect of the soluble TNF-α receptor (etanercept) on inflammatory eye diseases. 

²  Contributed equally to the work and therefore should be considered equivalent authors.

 Supported by German Research Foundation Grant DFG Jo 324-4/1, the Ernst and Berta-Grimmke Fund, the Meyer-Schwarting Stiftung, the Kämpgen, the Ernst and Gertrud Müller Fund, and Grant TV76 from the Center for Molecular Medicine.

 Submitted for publication June 17, 2002; revised October 6 and November 25, 2002; accepted December 6, 2002.

 Disclosure: K. Koizumi, None; V. Poulaki, None; S. Doehmen, None; G. Welsandt, None; S. Radetzky, None; A. Lappas, None; N. Kociok, None; B. Kirchhof, None; A.M. Joussen, 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.

 

(a) In vivo AOLF. Etanercept reduced leukocyte rolling in the retina after injection of LPS. Representative images of a normal eye (aA), 24 hours after LPS injection (aB), a normal eye 48 hours after treatment with etanercept (aC), and after treatment with etanercept and LPS injection (aD). Rolling leukocytes were visible in the large veins in the animal with EIU (arrows). Vascular endothelial cells stained brightly after AO circulation (⋆). Quantification of rolling leukocytes in retinal vessels per vascularized area (aE). (b) To demonstrate rolling leukocytes, time frame observations are shown. Representative image of an eye 24 hours after LPS injection (bA). Arrows: rolling leukocytes. According to their movement along the vessel wall, the leukocytes were distant from their original locus, when observed 5 seconds later (arrows) (bB).

Figure 2.

 

Typically, leukocyte adhesion was found in venules after LPS injection (A). Leukocyte adhesion after LPS injection was elevated 4.5-fold compared with the levels in control animals. Leukostasis in the retinal vasculature was determined by concanavalin A staining. A retinal vein of a control animal demonstrated only a few adherent leukocytes (B). Etanercept (0.3 mg/kg) reduced leukocyte adhesion after LPS injection in the retina by 41% (C, D).

Figure 3.

 

Etanercept reduced LPS-induced levels of MPO in the retina. Compared with the retinas of normal control animals, the retinas of LPS-injected animals demonstrated 1.7-fold higher MPO levels. Treatment with etanercept reduced the retinal MPO levels by 28.8%.

Figure 4.

 

Etanercept reduced LPS-induced blood–retinal barrier breakdown. Blood–retinal barrier breakdown was increased by approximately 4.2-fold 24 hours after LPS injection. Treatment with etanercept resulted in a reduction of LPS induced blood–retinal barrier breakdown by 60.3%.

Figure 5.

 

Eternacept reduced LPS-induced apoptosis in the retina. Twenty-four hours after injection of LPS, fragmented DNA had increased by 11.6-fold compared with the control level. Systemic treatment with etanercept reduced fragmented retinal DNA by 46.6%.

Figure 6.

 

Etanercept-reduced retinal endothelial cell injury as determined by propidium iodide. After LPS injection the retina exhibited a marked increase in PI-labeled cells that were mainly located in clusters (A). In contrast, untreated controls demonstrate almost no PI-positive endothelial cells (B). The total number of PI-labeled endothelial cells per retina in eyes of animals treated with etanercept was reduced by 67.3% (C, D).

Figure 7.

 

Eternacept reduces retinal endothelial cell apoptosis. Formalin-fixed paraffin-embedded retinal sections were stained by TUNEL assay (A) and the M30 antibody (B). In LPS-injected animals, TUNEL and M30 staining was readily observed, and positive cells were predominantly located in the ganglion cell layer, the vascular endothelium, and inner nuclear layer. Treatment with etanercept attenuated the number of TUNEL-positive cells as well as M30 positivity.

Figure 8.

 

Western blot analysis for caspase-3 and -8. After LPS injection both caspase-3 (A) and caspase-8 (B) protein levels were increased. The active form of both caspases was reduced after treatment with etanercept.

The authors thank Christina Esser and Claudia Gavranic for expert technical assistance.