OXYGEN FREE RADICALS AND CORNEAL ENDOTHELIUM* BY David S. Hull, MD INTRODUCTION

IT

IS SOMEWHAT PARADOXICAL THAT OXYGEN, WHICH IS VITAL TO THE

survival of all aerobic organisms, is potentially lethal and can kill at concentrations only five to ten times higher than that present in the air. The reduction of oxygen to water, proceeding by a series of four stepwise single electron transfers, leads to the production of toxic chemical species. The cellular metabolic processes of all aerobic organisms involve the production of these very reactive products. These products, namely the oxygen free radical (also called superoxide anion), hydroxyl free radical, and hydrogen peroxide, are the cause of oxygen toxicity. 12 These highly reactive products usually interact with unsaturated molecules such as deoxyribonucleic acid, proteins, and the polyunsaturated lipids found in cell membranes. 3 The survival of any aerobic organism is dependent upon its ability to regulate these toxic oxygen products. It is possible that oxygen at any concentration is toxic and that the aging process is the result of repeated cellular attack by toxic oxygen species over the life-time of an individual organism.4,5 Aging processes of the eye may also be related to toxic oxygen species and their products. These metabolic products, which are potentially damaging to the cell, are present in the aqueous humor and the surrounding tissues. Conceivably they may be related to aging and degenerative changes in the eye as well as cataract formation.5-7 Recent work has demonstrated low levels of hydrogen peroxide in the aqueous humor of humans and rabbits, and attempts have been made to associate this finding with cataract formation. 6-8 Elevated hydrogen peroxide concentration in the aqueous humor of certain cataract patients and the formation of cataracts in rabbits after catalase inhibition is suggestive of a possible link between free radical production and at least one ocular *From the Department of Ophthalmology, Medical College of Georgia, Augusta Georgia. This work was supported in part by Research grant EY04479 (Dr Hull) and EY04636 (Core Grant for Vision Research) from the National Eye Institute, Bethesda, Maryland and by an unrestricted Departmental Research Award from Research to Prevent Blindness, Inc., New

York.

464

Hull

disorder.6'8 Free radicals are also responsible for the appearance of fluorogens in the lens which may be related to the appearance of cataracts.9 10 Corneal endothelial damage has been related to hydrogen peroxide, and there is increasing evidence for a role of light-generated free radicals in causing the breakdown of retinal cell membranes. 11-14 Toxic oxygen species may also be produced during inflammatory disease processes. Oxygen free radicals, hydrogen peroxide, hydroxyl free radical, singlet oxygen, and hypohalite ions, produced during the "respiratory burst" of phagocytic cells are capable of oxidizing and damaging cells and tissues. 15-17 Oxygen free radicals have been shown to have an adverse effect on the microcirculation of brain, lung and heart, and they may play a role in shock, ischemia and organ preservation. 18-23 Ocular inflammatory disease processes may cause the alteration or destruction of previously normal eye tissues, and may damage corneal endothelial cells, trabecular meshwork, iris, and lens. 24-27 Toxic oxygen species which are known products of inflammatory cells may play a role in mediating this damage. Oxygen free radicals have been shown to cause an increase in iris vascular permeability.28 Corneal edema may be observed during intraocular inflammatory disease processes, and the pathophysiology may be related to an alteration of endothelial cell function caused by release of toxic products from polymorphonuclear leukocytes in the anterior chamber. 29-31 It is possible that superoxide anion, hydrogen peroxide, hydroxyl radical, hypohalite ion and singlet oxygen elaborated during the "respiratory burst" may play a role in compromising endothelial cell function during and after ocular inflammatory disease processes. MECHANISMS OF INJURY

A free radical is an atom or groups of atoms or molecules with an unpaired electron occupying an outer orbital. The oxygen free radical is also called the superoxide anion and is written 02. It has long been associated with the effects of ionizing radiation, but only relatively recently has it been associated with normal cell metabolism. The univalent reduction of oxygen in biologic systems results in the generation of 02. Ground state molecular oxygen is a biradical with two unpaired electrons with parallel spins in its outer orbital.32 Oxygen in its ground state is a weak oxidant because of this parallel spin arrangement and the dictates of quantum mechanics. The direct addition of a pair of elections (divalent pathway) with opposite spins would necessitate spin inversion, a slow and statistically improbable process. For this reason the univalent reduction of oxygen is the favored pathway since no spin inversion is required to

Corneal Endothelium and 20

465

accomplish this process. The univalent reduction of oxygen to H20 is accomplished by the stepwise addition of four single electrons and involves the intermediate production of the superoxide anion (also called oxygen free radical 02), hydrogen peroxide H202), and the hydroxvl free radical OH), all of which are potentially toxic (Fig 1).

UNIVALENT REDUCTION OF OXYGEN

02 eb O02

0 e - +2H +

Superoxide Anion

HO0

-

e

+H+

H 2 H-2

OH* OH e-

+

Hydrogen H20 Hydroxyl Free Peroxide Radical FIGURE 1

The stepwise addition of a single electron in the univalent reduction of oxvgen results in the production of the three toxic species - suiperoxide anion, hydrogen peroxide, hvdroxN l free radical.

These agents are capable of oxidizing tissue components thereby causing irreversible damage. Previous work has demonstrated that ervthrocytes sensitized with rose bengal and subsequently exposed to light hemolyze because of the production of toxic oxygen species. Those erythrocytes exposed to a similar concentration of rose bengal in the dark do not hemolyze.33-3 It is known that photosensitized reactions, which result in the production of toxic oxygen species, can cause alterations of membrane permeability and active transport. Rabbit red blood cells undergoing brief periods of photosensitization markedly accelerate the rate of Na+ uptake and K + loss.36 Studies on ocular tissues have indicated that free radical formation (either superoxide anion, hydroxyl radical, or hydrogen peroxide) probably occurs in both light-induced retinal damage involving the formation of lipid peroxides and in lens fluorogen formation.9'10'14 Although these changes are induced experimentally, they may represent an acceleration of normally occurring processes within the eve that result in aging changes. Comparisons of corneal swelling rates caused by hydrogen peroxide indicate that adverse effects on endothelial cell pump activity may be initiated by low concentrations of hydrogen peroxide with higher concentrations inducing endothelial cell structural and barrier defects. 13'37 The

466

Hull

precise mechanism has not been determined as to how toxic oxygen species affect corneal endothelial cell function. It is known that the superoxide anion and its products are capable of perturbing cell membrane lipid bilayers sufficiently to cause leakage of ions.37-39 These reactions can also lead to sodium channel block, lipid peroxidation, oxidation of certain amino acid residues, depolymerization of acid polysaccharides, cross-linking of membrane proteins, and cell death.37-40 MECHANISMS OF DEFENSE

Since the univalent reduction of oxygen results in the production of toxic intermediates (superoxide anion, hydrogen peroxide, and hydroxyl radical), mechanisms have evolved to accomplish the majority of oxygen reduction by tetravalent means in the mitochondria using the enzyme cytochrome oxidase thereby avoiding the release of superoxide anion and hydrogen peroxide.4' However, about 2% of 02 reduction does occur by the univalent pathway and a variety of protective mechanisms have evolved to prevent cell damage. These protective agents include superoxide dismutase, catalase, glutathione peroxidase, ascorbic acid, and vitamin E

(Fig 2). ENZYMATIC DEFENSE MECHANISMS Cytochrome Oxidase

Tetravalent Pathway

02

* 2 -*

Oxygen

Superoxide

Univalent Pathway

Anion

H

H202

Hydrogen Peroxide

L~ J l L Superoxide Dismutase

-

OH

-

2 HO

Hydroxyl Free

Water

Radical Catalase,

I

Peroxidase

FIGURE 2

The enzyme cytochrome oxidase allows oxygen reduction to occur by an alternate pathwav resulting in the direct production of water. However, about 2% of oxygen reduction occurs by the univalent pathway and results in the production of the toxic intermediates, superoxide anion, hydrogen peroxide, and hydroxyl free radical. Protection from the toxic intermediates is provided by superoxide dismutase, catalase and the peroxidases.

Corneal Endothelium and 02

467

Superoxide Dismutase Superoxide dismutase provides an essential defense against oxygen toxicity. The superoxide anion is unstable with a lifetime of milliseconds at neutral pH. It is converted to hydrogen peroxide during a dismutation reaction that occurs spontaneously, but is catalyzed to occur more rapidly by the enzyme superoxide dismutase: 202 + 2H + -* H202 + 02. The enzyme is extremely stable, and has a molecular weight of about 35,000. Superoxide dismutase exists as a manganese containing protein in mitochrondria and as a copper-zinc containing enzyme in the cytoplasm of mammals.43-49 The rate constant of the catalyzed reaction is about 2 x 109 M-ls-I and appears to be largely independent of pH.4546 All aerobic organisms contain superoxide dismutase to protect against the toxic effects of univalent reduction of oxygen. In contrast, strict anaerobes do not contain any superoxide dismutase and therefore are rapidly destroyed in the presence of oxygen.47 Interactions between hydrogen peroxide and superoxide anion can also result in the production of the two toxic species the hydroxyl free-radical and single oxygen: H202 + 02. - OH- + OH + 02*.48 Therefore the rapid removal of superoxide anion from a system is necessary to prevent cell damage by the interactions between hydrogen peroxide and superoxide anion. Superoxide dismutase maintains superoxide anion levels below 10'1l M.44 Catalase and Peroxidase As previously mentioned, respiring cells produce hydrogen peroxide either directly by the divalent reduction of oxygen or more commonly by the dismutation of superoxide anion (oxygen free radical). The catalase and peroxidase enzymes catalyze the divalent reduction of hydrogen peroxide to water.47 2H202

02 + 2H20

catalase, peroxidase These enzymes are essential and serve the indispensable role of preventing hydrogen peroxide accumulation within the cell. Catalase is a heme protein in mammals and is contained in the subcellular peroxisomes.44 Catalase does not utilize a substrate other than H202 whereas the peroxidases such as glutathione peroxidase can reduce one of several substrates including hydrogen peroxide and lipid hydroperoxide.44 Glutathione peroxidase in conjunction with reduced glutathione may be a major system for the intracellular decomposition of hydrogen peroxide. Under normal conditions intracellular hydrogen peroxide levels are con-

468

Hull

trolled by glutathione peroxidase; however, when cells are under oxidative stress such as might be found during inflammation, glutathione peroxidase has less of a modulating effect.49 Catalase is better suited to respond to a sudden, high flux of hydrogen peroxide and to protect the cell from sudden oxidative stress. In addition to its role as a co-substrate for glutathione peroxidase, reduced glutathione may also serve as an antioxidant protecting the cell membrane from autoxidation. Glutathione peroxidase contains selenium and is widely distributed in mammalian cells. The interaction of reduced glutathione (GSH) and hydrogen peroxide in the presence of glutathione peroxidase results in the production of water and oxidized glutathione (GSSG). Glutathione reductase returns GSSG to GSH using NADPH as the hydrogen donor. The pyridoxine nucleotide NADPH is maintained by the hexose monophosphate shunt

(Fig 3). GLUTATHIONE OXIDATION - REDUCTION PATHWAY

02 + 2H20

2H202 Glutathione Peroxidase

GSSG Oxidized Glutathione

2GSH Reduced Glutathione Glutathions

Hexose Monophosphate Shunt

NADP

Glucose FIGURE 3

Reduced glutathione in the presence of the enzyme glutathione peroxidase helps to control intracellular levels of hydrogen peroxide. The GSSG produced in the reaction is reconverted to GSH by the enzyme glutathione reductase. NADPH produced by the hexose monophosphate shunt serves as the hydrogen donor for the production of GSH in the reaction.

Corneal Endothelium and °24

469

Riley and Giblin50 have shown that rabbit corneas swell rapidly when the endothelium is exposed to 50 ,uM H202 in the absence of glucose. However in the presence of glucose, the adverse effect is not noted until H202 concentrations reach 200 ,uM. The protection appears to be due to the ability of glucose to provide for the reduction of glutathione via the hexose monophosphate shunt. Catalase and glutathione peroxidase are present in the cornea, iris, lens, ciliary body, and retina. The highest levels of these enzymes in rabbit eyes have been found in iris and ciliary body.6 a-Tocopherol Alpha-tocopherol (vitamin E) may be of importance in protecting the cell membrane from lipid peroxidation. Cell membranes with their large content of polyunsaturated fatty acids can participate in self-propagating lipid peroxidation chain reactions that are damaging to their integrity. Damage to the cell membrane lipid bilayer is in part prevented by the presence of ox-tocopherol and its ability to terminate radical chain reactions.4'51 One manifestation of vitamin E deficiency is hemolysis of red cells in vitro in the presence of peroxide. In an experimental setting, vitamin E offered a 44% reduction of lipid peroxidation in cultured lens. In a rabbit model vitamin E was therapeutically effective in arresting 3-aminotriazole induced cataracts in 50% of the animals.52 Ascorbate Ascorbate can act as a free radical scavenger and it is present in both the blood and the aqueous humor. Its level in the aqueous humor is 40 times higher than its level in the blood because it is concentrated in the ciliary processes. In an aerobic medium ascorbic acid is an effective scavenger of the superoxide anion.13,53 Ascorbic acid may also quench the hydroxyl radical, forming the ascorbate radical.54 The ascorbate radicals that are formed during this detoxifying process are extremely stable and may exist for days before undergoing disproportionation thereby terminating the free radical reaction.55556 PREVIOUS WORK

OXIDATIVE AGENTS AND THEIR EFFECT ON CORNEAL ENDOTHELIUM AND ANTERIOR SEGMENT STRUCTURES

Hydrogen peroxide has been found in the aqueous humor of humans and rabbits and attempts have been made to associate it with cataract formation.6-8'52 It is possible that a life-time of corneal endothelial exposure to

Hull

470

hydrogen peroxide may result in the decline of endothelial cell population and the altered morphology and function that has been noted with increasing age.57,58

It has been determined that the threshold of corneal endothelial toxicity is in the range of a nominal concentration of 0.3 to 0.5 mM hydrogen peroxide during in vitro perfusion with a Krebs-Ringer bicarbonate solution containing 27.8 mM glucose, with added adenosine and glutathione. 11 The toxic effect resulted in anatomic disruption of endothelial cells with swelling of endoplasmic reticulum, mitochondria, and nuclei. This was accompanied by a rapid increase in stromal thickness. The effect could be blocked with 5400 U/ml catalase thereby indicating the protective effect of catalase on a presumed direct hydrogen peroxide toxic effect. The toxic effect of hydrogen peroxide could be enhanced by the addition of EDTA-Fe(III), probably through production of the highly reactive hydroxyl radical via a Fenton-type reaction (superoxide-driven, ironpromoted Haber-Weiss reaction) (Fig 4).5,59-63

FENTONTYPE REACTION a) Fe + 2 (chelate) + H202 + H +_ Fe+ 3 (chelate) + OH* + H20

b) Fe + 3 (chelate) + 02 *

c) H22 + 02

+

H+-

Fe + 2 (chelate) + 02

-

OH

+ H2O + O2

FIGURE 4

Hydrogen peroxide and the oxygen free radical combine to form the extremely reactive hydroxyl free radical. The reaction is promoted by a cycling of iron from the ferrous (FeII) to the ferric (FeIII) state and back to the ferrous state.

Recent work has demonstrated rabbit and human aqueous humor levels of hydrogen peroxide ranging from 25 to 60 puM, or about one-tenth the concentration found in our study necessary to induce acute endothelial toxicity.7'8 It is possible that iron in hemoglobin found in the anterior chamber of hyphema patients may cause a similar Fenton-type reaction leading to the production of the extremely reactive hydroxyl free radical with resulting anterior segment and endothelial cell damage. Chelated iron may also enhance the peroxidation of polyunsaturated fatty acids. The ferric ion (Fe(III)) is required for the first step in the oxidative

Corneal Endothelium and °2

471

cleavage of polyunsaturated fatty acids and the resultant formation of phospholipid peroxides.64 The catalytic activity of Fe(III) can be stimulated in some model systems with the addition of a reducing agent such as ascorbic acid (which is present in aqueous humor) which regenerates Fe(II). 65-68 Neither the presence of ascorbic acid nor the absence of GSH modified the direct hydrogen peroxide toxic effect in one experimental system bringing into question the role that these agents may have in protecting the cornea from oxidative damage." It is known that aging processes may be related to the generation of free radicals as illustrated by the lens and retina. This work" demonstrated that corneal endothelial cells are susceptible to acute damage by hydrogen peroxide in concentrations only ten times that normally found in human aqueous humor. It is possible that a life-time of exposure to lower levels of hydrogen peroxide may result in damage and loss of endothelial cells. Other work has shown that the threshold of toxicity of an organic peroxide, t-butyl hydroperoxide, on human erythrocytes is 0.75 mM which is of the same general order of magnitude as has been found for hydrogen peroxide. 69 Prolonged exposure of erythrocytes to a lower concentration of t-butyl hydroperoxide, however, also resulted in an alteration of red cell membrane deformability, protein cross-linking, and osmotic fragility.69 In another experiment rabbit corneal endothelial cells were exposed to a flux of chemically generated superoxide anion (oxygen free radical) produced by the combination of 0.06 U/ml xanthine oxidase and 1 mM hypoxanthine. Exposure of endothelium to superoxide anion resulted in a corneal stromal swelling rate of 19 + 3 ,um/hr which was significantly more rapid than controls. Transmission electron microscopy (TEM) following exposure to superoxide anion showed the formation of large cytoplasmic vacuoles and disruption of the nucleus and cytoplasmic organelles.3 The adverse effect of the superoxide anion was not blocked by superoxide dismutase, which catalyzes the conversion of the superoxide anion to hydrogen peroxide: 202 + 2H + -- H202 + 02. Catalase prevented the adverse effect by catalyzing the divalent reduction of H202 to H20. It appears that the toxic effect on endothelial cells was, at least in part, secondary to hydrogen peroxide produced during the dismutation reaction of oxygen-free radical.47 Hydrogen peroxide toxicity also was confirmed by the fact that ascorbic acid, which in an aerobic medium scavenges 02- but not H202, did not offer protection. This experimental model demonstrated that hydrogen peroxide produced by the dismutation of superoxide anion is one of the species causing endothelial cell damage. Interactions between hydrogen peroxide and superoxide anion may result in the production of additional toxic species, ie, the hydroxyl-

472

Hull

free radical and/or singlet oxygen: H202 + 02 - OH- + OH- + 02*.48,70,71 However, in one model, the hydroxyl radical scavengers Dmannitol (3 mM and 15 mM) and ethanol (0.07 mM) did not alter the corneal swelling rate. 13 Therefore, participation of the hydroxyl-free radical in the mediation of endothelial cell damage in this experimental model appears to be unlikely. Additional, indirect evidence against hydroxylfree radical and singlet oxygen participation is provided by the fact that superoxide dismutase did not protect the endothelial cells. Because hydroxyl-free radical and singlet oxygen are produced by the interaction of H202 and 02., superoxide dismutase would be expected to offer some protection, which it did not. Hydrogen peroxide production always accompanies superoxide production since it is a product of the dismutation reaction of oxygen free radical. Superoxide anion, hydrogen peroxide, hydroxyl radical, singlet oxygen, hypohalite ions and possibly other oxidant species are produced during activation of phagocytic cells with the resultant "respiratory burst." All of these species have been implicated as being capable of oxidizing tissue components and causing irreversible cellular damage. 15-17 Since hydrogen peroxide, and other oxidant species are produced by phagocytic cells, it is possible that the development of therapeutic agents having a specific modifying effect on hydrogen peroxide and other oxidant species may be beneficial in reducing the corneal endothelial cell damage that results during ocular inflammatory disease processes. In a separate experiment superoxide anion was generated by adding potassium superoxide to Krebs-Ringer bicarbonate (KRB) solution. 12 The reduction of nitroblue tetrazolium measured in a spectrophotometer, proved that superoxide anion was produced. In the experiment rabbit corneal endothelial cells which were perfused with a KRB solution to which 0.7 mM potassium superoxide (KO2) had been added resulted in a stromal swelling rate of 39 + 7 ,um/hr. This value was significantly higher than the 5 + 1 ,im/hr swelling rate of paired control corneas that were not perfused with potassium superoxide. The corneal endothelial cells following potassium superoxide perfusion showed severe anatomic disruption with irregularity of the plasma membrane and swelling of the cytoplasm and cytoplasmic organelles. The addition of 1 ,um/ml catalase to the solution reduced the corneal swelling rate to 4 + 5 ,um/hr and prevented the anatomic damage to the endothelium. Catalase, by breaking down hydrogen peroxide, effectively removed it from the system. Superoxide dismutase, which accelerates the dismutation of superoxide anion to produce hydrogen peroxide was, as expected, without a protective effect. Neither of the hydroxyl radical scavengers D-mannitol nor DMSO (di-

Corneal Endothelium and °2

473

methyl sulfoxide) were protective demonstrating that hydroxyl free radical was not involved in producing the toxic effect. Previous studies have indicated that an iron chelate may be necessary for catalyzing a HaberWeiss type reaction resulting in production of hydroxyl free radical when hydrogen peroxide and superoxide anion interact. 72-74 Neither DETAPAC (diethylenetriamine-pentaacetic acid) nor EDTA (ethylenediaminetetraacetic acid) offered protection by complexing with trace amounts of iron in the solution. In addition, the combination of EDTA-FeCI2 did not enhance the anatomic and physiologic alteration of endothelial cells. This provided additional evidence against hydroxyl free radical participation in the system. Endothelial intracellular glutathione levels following a 1 hour perfusion with 0.3 mM KG2 and a subsequent 1/2 hour perfusion with KRB were 249 + 56 ng glutathione/corneal endothelium with 1.3% oxidized (n = 5). This was similar (P > 0.01) to the 333 ± 61 ng glutathione/corneal endothelium with 0.6% oxidized which was found following perfusion of paired controls with KRB alone. It was of interest that the total intracellular glutathione levels and redox state of endothelial cells were unaltered. It has been shown that the presence of glucose in a perfusion medium prevents the depletion and oxidation of glutathione by hydrogen peroxide thereby reducing the overall toxic effect of hydrogen peroxide on corneal endothelial cells. 50 As in previous studies, the data again suggests that hvdrogen peroxide is the toxic species and that superoxide anion, hydroxyl free radical, and singlet oxygen do not participate in causing the endothelial cell damage. 12 Since oxygen free radicals, hydrogen peroxide, hydroxyl free radical, singlet oxygen, and hypohalite ions are elaborated by inflammatory cells during the "respiratory burst," an in vivo model was utilized to determine the effect of chemically generated oxygen free radicals on anterior segment structures.28 Xanthine oxidase combined with hypoxanthine was injected into the anterior chamber of rabbit eyes giving a final concentration of 0.06 U/ml xanthine oxidase and 1 mM hypoxanthine. This combination is known to result in the production of oxygen free radicals and was confirmed by the precipitation of nitroblue tetrazolium. Iris fluorescein angiography performed 2 hours following injection of xanthine oxidase and hypoxanthine into the anterior chamber resulted in increased iris vascular permeability with Tmarked leakage of fluorescein dye from the iris vessels. Injection of bovine serum albumin into the anterior chamber did not result in increased permeability of the iris to fluorescein. This ruled out the possibility of an immune response in the time frame of this experiment. Twenty-four hours following intracameral injection the increased iris vascular permeability persisted in 20% of eyes whereas 80%

474

Hull

of the eyes showed no persistent leakage. The increased permeability was not modified by either of the prostaglandin inhibitors aspirin or naproxen nor by the free radical scavenger D-penicillamine. The study demonstrated that the increased iris vascular permeability that occurs during ocular inflammatory processes may in part be mediated by oxygen free radical products. It demonstrated that iris vascular and possibly blood aqueous barrier permeability is increased in the presence of a chemical system known to result in the production of oxygen free radicals. These toxic oxygen species, which are also elaborated by inflammatory cells may in part be responsible for tissue damage in the anterior segment following ocular inflammatory disease processes where they may overwhelm anterior chamber ascorbate and other antioxidants in the aqueous humor. Therapeutic modalities directed at preventing the damaging effect of oxygen free radical products may be of benefit in reducing ocular tissue damage following inflammatory disorders.28 LIGHT TOXICITY AND CORNEAL ENDOTHELIUM: DYE MEDIATED OXIDATIVE EFFECTS

Rose bengal is a photosensitizing dye related to fluorescein. The fluorescein dyes, including rose bengal, are taken up on the cell surface and, when the cell is exposed to light, free radicals are formed leading to cell membrane damage and lysis. In one experiment, corneal endothelial cells were perfused in the specular microscope with rose bengal at 5 x 10-6 M and were exposed to a 25 watt (W) incandescent light at 5 cm (1050 RiW/cm2).75 A Petri dish filled with water was placed between the cornea and the light source to absorb the heat generated by the lamp. Corneas exposed to light for 5 minutes swelled at 39 ,um/hr whereas those which were exposed to light for only 0.5 minutes swelled at 6 ,um/hr. Corneas perfused in the dark with a similar concentration of rose bengal did not swell. Combining rose bengal with 100 ,ug/ml superoxide dismutase or 10 mM D-penicillamine (a free radical scavenger) did not reduce corneal swelling following exposure to light. The addition of 200 jig/ml catalase to the rose bengal perfusing solution eliminated the corneal swelling following exposure of corneas to light, indicating that the photodynamic effect on corneal endothelium is secondary to cell alteration from the hydrogen peroxide produced during the dismutation reaction of oxygen free radical. 75 Another study was performed to determine the effect of rose bengal photosensitization on anionic fluxes across the endothelium.38 Cornea endothelial cells perfused with 5 x 10-6 M rose bengal and exposed to a 25 W incandescent light at 5 cm demonstrated a 16% increase and a 19% increase in passive bicarbonate flux (J endothelium to stroma) following 1

Corneal Endothelium and °24

475

minute and 5 minutes, respectively, of light exposure when compared to control corneas which were perfused with rose bengal but which were not exposed to light. Active bicarbonate flux (J stroma to endothelium) was reduced 11% after 5 minutes of light exposure, but was not reduced after 1 minute of light exposure. This indicates that the rose bengal-induced photoreaction had an effect on bicarbonate permeability before it had an effect on endothelial bicarbonate transport. Corneas perfused with 5 x 10-6 M rose bengal in a solution with PO2 of 124 mm Hg and exposed to light swelled at 26 ,um/hr whereas corneas perfused with a similar rose bengal solution and exposed to light, but with a P02 of 20 mm Hg swelled at 13 ,um/hr (P < 0.05). This study demonstrated that corneal endothelial cell bicarbonate flux is altered by photosensitization and that the photosensitization reaction is oxygen dependent.38 The study also confirmed prior studies in other cell systems where the rose bengal photoreaction has been shown to alter membrane permeability and transport.35'36 Since glutathione can be effective in the removal of hydrogen peroxide via glutathione peroxidase (Fig 3) it was of importance to learn the effect of the rose bengal photoreaction on corneal endothelial glutathione. Rabbit corneal endothelial cells perfused with 5 x 10-6 M rose bengal and exposed to incandescent light (25 W, 5 cm, 5 min, 1050 pLW/cm2) demonstrated endothelial intracellular glutathione levels in the range of 350 ng/ corneal endothelium with 20% GSSG.76 This was similar to levels found in nonphotosensitized control corneas. Addition of 5400 U/ml catalase to the perfusing solution had no effect on total glutathione levels, but lowered the oxidized fraction of glutathione to 1% to 2% of the total glutathione in corneas exposed to light as well as in those not exposed to light. It appears that catalase removes a source of oxidative stress from the system and shifts the glutathione redox cycle to a more reduced state. It was concluded that the photodynamically induced swelling of corneas, whether due to a failure of a fluid pump or permeability barrier, or to oxidation of membrane lipids and proteins, is not the result of failure of the endothelial cell glutathione redox system.76 It had been previously established that the rate of corneal swelling that follows the rose bengal photoreaction is dependent on both rose bengal concentration and the duration of light exposure. It was of further importance to determine the relationship between corneal swelling and the wavelength of the incident light. Corneal endothelial cells sensitized with rose bengal and exposed to 25 W incandescent light demonstrated a wavelength dependent alteration in physiologic response. Rose bengal perfused corneas exposed to 451, 500, 612, and 651 nm swelled at rates less rapid than those exposed to 550 nm. The photodynamically induced

476

Hull

physiologic alteration of corneal endothelial cells approximately paralleled the spectral curve of rose bengal with its absorptive peak at 550 nm. T'he study demonstrated that rose bengal-sensitized corneas exposed to similar energy levels at narrow wavelengths other than at the peak of the spectral curve swell less rapidly than those exposed to light at the peak of the spectral curve for rose bengal.77 With establishment of the fact that corneal perfusion with rose bengal and exposure to incandescent light resulted in corneal swelling, it was of importance to determine the effect of the photoreaction on endothelial membrane permeability and ionic and nonionic fluxes. In a series of experiments, rabbit corneal endothelial cells demonstrated an increase of endothelial membrane permeability to inulin (effective molecular radius = 14 A) following perfusion with 5 x 10-6 M rose bengal and exposure to 1050 ,uW/cm2 incandescent light for either 1 minute or 5 minutes when compared to corneas similarly perfused with rose bengal but not exposed to light.78 Endothelial membrane permeability to dextran (effective molecular radius = 38 A) was unaltered following the photosensitization reaction. This indicated that the size of the photodynamically-induced lesion in the endothelial membrane was large enough to allow penetration of a 14 A particle but not large enough to allow penetration of a 38 A particle. Exposure of rose bengal perfused corneal endothelial cells to incandescent light resulted in an increase in the unidirectional passive sodium flux in both directions across the endothelium, and a reduction in the active component of the sodium flux from stroma to endothelium. This resulted in a reduction of the net active transport of sodium across the endothelium from the stroma to the aqueous facing surface following the photosensitization reaction. TEM showed no alteration of endothelial cell ultrastructural integrity following rose bengal perfusion in the absence of light. However, rose bengal perfusion accompanied by exposure to 1050 11W/cm2 incandescent light caused swelling of endothelial cells, endoplasmic reticulum, and mitochondria. This study showed that rose bengal photosensitization, which from previous studies was known to result in the production of superoxide anion and hydrogen peroxide, has an adverse effect on endothelial cell energy production as well as ionic transport and membrane barrier function. 78 It was postulated that rose bengal photosensitization of corneal endothelium results initially in increased membrane permeability and only later does it have an effect on ionic transport function. It appeared from this study in conjunction with previous work that corneal endothelial cells are more susceptible to the photodynamic effects on permeability than they are to the photodynamic effects on ionic transport.

Corneal Endothelium and 024

477

Hematoporphyrin derivative (HpD) is a systemically administered photosensitizing agent that may be of value in the treatment of solid tumors. When corneal endothelial cells were perfused in the specular microscope with HpD and exposed to a 25 W incandescent light at 5 cm, there was anatomic disruption of corneal endothelial cells and swelling of the corneal stroma.79 Perfusion with 0.2 RI/ml HpD and 5-minute exposure to light resulted in a corneal swelling of 71 + 4 ,um after 3 hours, whereas perfusion with 0.2 uLw/ml and a 1-minute exposure to light resulted in a corneal swelling of 36 + 4 pum after 3 hours. Perfusion with 0.2 ,ul/ml HpD with no light exposure resulted in a corneal swelling of 22 ± 4 ,um after 3 hours. Inclusion of 100 ,ug/ml catalase in the perfusion solution resulted in a 38% reduction of the corneal swelling that was induced by perfusion with 0.2 ,ul/ml HpD and 5-minute exposure to light. The inclusion of 100 ,ug/ml superoxide dismutase, 10 mM L-histidine, or 1 mM sodium azide did not modify the corneal swelling induced by the photosensitization reaction. TEM of corneal endothelial cells following the photosensitization reaction with HpD showed swelling of cellular cytoplasm and organelles with disruption of nuclear chromatin. The photoreaction had no effect on either endothelial cell total glutathione content or percent GSSG. This indicates that the endothelial cell damage was not the result of failure of the redox system. The data suggested that corneal endothelial cells are damaged by hydrogen peroxide produced by the dismutation of superoxide anion formed during the photoreaction. Superoxide anion itself and hydroxyl free radical did not appear to participate in causing endothelial cell damage. The role of singlet oxygen remained somewhat unclear.79 This study demonstrated that corneal endothelial cells can be adversely affected by HpD in the presence of incandescent white light. Because the normal thickness cornea is known to transmit 80% of incandescent light at 400 nm and about 100% of light at 700 nm, it may be of value to determine the possible susceptibility of endothelial cells in a clinical setting where patients are receiving photoradiation therapy for tumors. The possibility of adverse effects of HpD photoradiation therapy on the cornea should be kept in mind during treatment of ocular disease processes where intense levels of light passing through the cornea may produce endothelial damage mediated by HpD in the aqueous humor. Additional studies were performed with chlorpromazine and trifluoperazine, two commonly used psychotropic medications that undergo chemical alteration in the presence of light. Corneal endothelial cells perfused in the specular microscope in the dark with 0.5 mM chlorpromazine resulted in a stromal swelling rate of 18 + 2 ,um/hr whereas corneas

478

Hull

exposed to long wavelength ultraviolet (UV) for 3 minutes in the presence of 0.5 mM chlorpromazine swelled at the significantly more rapid rate of 37 ± 9 pLm/hr.80 Preirradiation of 0.5 mM chlorpromazine solution with UV light for 30 minutes and subsequent corneal perfusion with the solution resulted in a corneal swelling rate of 45 + 19 ,um/hr. Cornea endothelial cells perfused with 0.5 mM chlorpromazine that was preirradiated with UV light showed marked swelling on scanning electron microscopic examination, whereas those perfused with nonirradiated chlorpromazine were flat and showed a normal mosaic pattern. The adverse effect on endothelium was found to result from toxic photoproducts produced during the UV irradiation of chlorpromazine solution and was not related to superoxide anion or to hydrogen peroxide since neither 5400 U/ml catalase nor 290 U/ml superoxide dismutase modified the adverse response on corneal endothelial cells.80 Additional work on preirradiated 0.5 mM chlorpromazine demonstrated that endothelial permeability to inulin and dextran were markedly increased during perfusion with the UV irradiated compound. This correlated well with the previously noted anatomic disruption of endothelial cells following perfusion with photoactivated chlorpromazine. Chlorpromazine appeared to also have an influence on active and passive ionic fluxes in addition to affecting endothelial membrane permeability to large molecules. Perfusion of corneal endothelial cells with a second psychotropic agent trifluoperazine 2.5 x 10-4 M concurrent with 3 minutes of exposure to long wavelength UV light resulted in a corneal swelling rate of 22 ± 1 ,um/hr which was significantly greater than the swelling rate of 10 ± 2 ,um/hr found in cornea where endothelial cells were perfused with a similar concentration of trifluoperazine and not exposed to UV light.8' The toxic effect could also be produced by pre-exposure of the trifluoperazine perfusion solution to UV light suggesting the production of toxic photoproducts during exposure of trifluoperazine to UV light. TEM of endothelial cells showed that cell membranes and nuclei as well as tight and gap junctions were maintained; however, there was an alteration of mitochondria and a loss of cytoplasmic hemogeneity of cells perfused with trifluoperazine that had been preirradiated with UV light. Perfusion with trifluoperazine and exposure to incandescent light did not result in a photoinduced toxic effect. The relationship of these photoreactions may be of importance to individuals who use chlorpromazine and trifluoperazine for the treatment of psychiatric disorders. The work demonstrated that, within the confines of the experimental model, photosensitized chlorpromazine and trifluoperazine can cause endothelial cell phototoxicity.

Corneal Endothelium and °2

479

SUMMARY OF INFORMATION OBTAINED PRIOR TO THIS THESIS

Most laboratory work concerning the biologic effects of free radicals has been done with cell homogenates. The endothelial specular microscope, allowing observations of morphology and function of a cell monolayer, has permitted a unique model for the observation of free radical effects in vitro. The response of corneal endothelium to photosensitization with rose bengal and the subsequent formation of free radicals following exposure to incandescent light, indicate a sensitivity of the corneal endothelium to these agents. Free radical formation in vivo over many years could be associated with aging changes of the corneal endothelium and the cell loss that has been observed with increasing age. Catalase rather than superoxide dismutase appears to play a greater role in protecting the corneal endothelium in vitro. This indicates a more pronounced effect of hydrogen peroxide rather than of the oxygen free radical itself Potential roles for glutathione and ascorbic acid in the corneal endothelial membranes are in areas where they may serve as free radical scavengers and protect the cornea from light-induced damage. Glutathione apparently can also do this in the lens. Iris vascular permeability is increased in the presence of oxygen free radicals. The toxic concentration of hydrogen peroxide on the corneal endothelium in the presence of a glucose-containing perfusion solution is at the nominal concentration of between 0.3 and 0.5 mM H202. This is approximately ten times the level normally found in the aqueous humor. Hydroxide peroxide, the dismutation product of superoxide anion, is the toxic species on the corneal endothelium. The superoxide anion itself and other free radical products such as singlet oxygen and hydroxyl free radical probably do not participate in causing the damage. Photodynamically induced corneal endothelial cell alteration results in increased passive bicarbonate flux and a time dependent decrease in active bicarbonate flux. The phototoxic effect is oxygen dependent. The rose bengal photosensitization effect on corneal endothelial cells results in an increase in the unidirectional passive sodium flux in both directions across the corneal endothelium. In addition, there is a reduction in the active component of sodium flux from the stromal side of the endothelium to the posterior side of the endothelium. This causes a net reduction in the active transport of sodium across the endothelium following the photosensitization reaction.

480

Hull

Photosensitizing agents used in the treatment of solid tumors can also cause photosensitization of corneal endothelial cells in vitro. HpD is such an agent. Corneal endothelial cells perfused with HpD and exposed to incandescent light are damaged. The endothelial cell damage is caused by hydrogen peroxide generated by the dismutation of superoxide anion produced during the reaction. Trifluoperazine, a commonly used agent for treatment of psychiatric disorders, is also a photosensitizer. Perfusion of corneal endothelial cells with trifluoperazine hydrochloride (HCl) with concurrent exposure to UV light results in the production of a toxic photoproduct which causes corneal swelling. Chlorpromazine, a phenothiazine derivative, is used for the treatment of psychiatric disorders. Exposure of chlorpromazine to UV light results in the production of a toxic photoproduct that causes corneal endothelial cell damage. The phototoxic product results from photodynamically induced decomposition of chlorpromazine and is not caused by hydrogen peroxide or superoxide anion generated during the phototoxic reaction. The rose bengal-induced photosensitization of corneal endothelial cells has no effect on either the total glutathione concentration or the proportion of GSSG found in corneal endothelial cells. The toxic effect is, therefore, not due to failure of the glutathione redox system. Corneal endothelial cells perfused with rose bengal and exposed to light swelled maximally when the inciting wavelength was 549 nm. The photodynamically induced physiologic alteration parallels the spectral curve of rose bengal which has its absorptive peak at 550 nm. PRESENT STUDIES: EFFECT OF HYDROGEN PEROXIDE ON CORNEAL ENDOTHELLAL IONIC AND MOLECULAR FLUXES

Previous studies have demonstrated that hydrogen peroxide is the agent causing endothelial damage in oxygen free radical mediated toxic reactions. It has also been shown that hydrogen peroxide perfusion in the nominal concentration of 0.3 to 0.5 mM results in stromal swelling." Since net water movement across the corneal endothelium (which maintains its state of deturgescence) is presumably linked to ion movement, it is of importance to determine the effect of hydrogen peroxide on endothelial transport and permeability functions. In this study we evaluated some of the specific cellular transport and barrier functions that are altered when corneal endothelial cells are exposed to hydrogen peroxide. We examined the effect of hydrogen peroxide on corneal endothelial sodium and bicarbonate fluxes and on endothelial permeability to inulin

Corneal Endothelium and

°2

481

(molecular weight 5200, equivalent molecular radius = 14 A) and dextran (molecular weight 70,000, equivalent molecular radius = 38 A). GENERAL EXPLANATION OF CORNEAL ENDOTHELLAL FLUXES AND PERMEABILITY

Sodium and Bicarbonate Flux Corneal clarity is maintained by the movement of water out of the corneal stroma into the aqueous humor. Water movement is presumably linked to the movement of sodium and bicarbonate ions across the endothelial membrane into the anterior chamber. The net movement of each of these two ions (J net) is the algebraic sum of the stroma to endothelium flux (J stroma to endothelium) and the endothelium to stroma flux (J endothelium to stroma). The stroma to endothelium flux (J stroma to endothelium) is made of both an active "pump" component and a passive "leak" component. The endothelium to stroma flux ( endothelium to stroma) is made up of only a passive of "leak" component which is of the same magnitude as the passive or "leak" component of the J stroma to endothelium flux. The net flux (J net) is the algebraic sum of the fluxes in both directions which in effect cancels out the bi-directional passive or "leak" component, leaving only the active or "pump" component (Fig 5). In this experiment we determined the effect of hydrogen peroxide on endothelial sodium and bicarbonate fluxes.

J stroma to endothelium plus J endothelium to stroma

equals J net

Ionic flux across the corneal endothelium FIGURE

5

The net movement of sodium or bicarbonate (J net) across the endothelium is the algebraic sum of the stroma to endothelium flux a stroma to endothelium) and the endothelium to stroma flux U endothelium to stroma).

482

Hull

Inulin and Dextran Permeability The cornea endothelial membrane can act as a barrier to certain molecules attempting to cross it. A molecule colliding with the corneal endothelial membrane can either be reflected back from that surface or it can pass through a physiologic "hole" in the surface. Since inulin has an equivalent molecular radius of 14 A, an increase in inulin permeability leads to the conclusion that more physiologic "holes" equal to or greater than 14 A in diameter exist in the membrane. Similarly increased permeability to the dextran molecule with an equivalent molecular radius of 38 A implies the existence of cornea endothelial physiologic holes equal to or greater than 38 A in diameter. The presence of more physiologic "holes" between 14 A and 38 A in size would allow for increased endothelial permeability to inulin but not to dextran (Fig 6). In this experiment we determined the effect of hydrogen peroxide on corneal endothelial permeability to inulin and dextran.

0

-/

25A "hole"

( ~~~0

\ lnulin= 14 A easily passes 0

\\ Dextran=38 A is reflected

Behavior of inulin andodextran molecules in the presence of 25 A endothelial physiologic "holes " FIGURE 6

Inulin (equivalent molecular radius = 14 A) and dextran (equivalent molecular radius = 38 A) may either pass through an endothelial physiologic "hole" or be reflected back depending on the size of the physiologic "hole."

Corneal Endothelium and °2

483

MATERIALS AND METHODS

In Vitro Corneal Perfusion Experiments

Eyes of freshly killed rabbits were proptosed and the epithelium was removed by scraping. A procedure for preparing and atraumatically mounting corneas as is done in preparing specimens for the specular microscope was followed, except that the corneal endothelium and remaining stroma were placed between two water-jacked chambers and perfused at 37°C with Ringer-bicarbonate solution. The composition of the solution was as follows: Na+, 143 mM; Cl-, 127.7 mM; K+, 5.9 mM; Mg+ +, 1.18 mM; S04- -, 1.2 mM; H2PO4-, 1.2 mM; HCO3-, 25 mM; glucose, 27.8 mM; adenosine, 0.5 mM; GSH, 0.3 mM. In experiments where perfusion was performed with glucose-free Ringer, either sucrose or 0-methyl glucose (3-0-methyl-d-glucopyranose) was added to maintain the correct osmolarity. Neither sucrose nor 0-methyl glucose is metabolized by the endothelial cell. Hydrogen peroxide was additionally included in experimental perfusion solutions. The osmolarity was 307 + 2 mOsM as determined by a Fiske OM osmometer, and the pH was adjusted to 7.3. The volume of the endothelial (posterior) chamber was 0.45 ml and the stromal (anterior) chamber was 1.2 ml. After mounting the cornea between the chambers the anteriorly placed vacuum was released after a 10 mm Hg hydrostatic pressure was applied to the endothelial surface to prevent wrinkling. After 5 or 10 minutes this posteriorly applied "mounting" pressure was released leaving only about 1.5 mm Hg pressure on the upward-facing (concave) endothelial surface. The solution on the downward-facing, or stromal surface, of the preparation was stirred at 400 rpm using a small teflon-coated magnetic stirrer which was driven by an external magnet. For molecular permeability studies 3H-Inulin or 14C dextran were utilized, and for ionic fluxes H14CO3- or 22Na+ was used. After a 60minute equilibration period with the radioactive Ringer which was on one side of the cornea immediately after mounting, the "cold" chamber was flushed with 5 ml of nonradioactive Ringer which was discarded and then a further flush of 5 ml was made and collected in a tared vial and counted. Paired corneas were used, one experimental and one control. Sampling of the cold chambers, at 30-minute intervals was performed by total replacement of the chamber solutions. Experiments in our laboratory have shown that a flush volume of 5 ml is sufficient to reduce the count rate to background levels. The flushed volume was collected in tared vials and the volume determined gravimetrically. A 100 RI sample of the flushed volume was taken and counted, after admixing with 10 ml of Aquasol (New England Nuclear Corp, Boston, MA) in an Isocap 300 Liquid Scintillation Counter (Searle, Des Moines, IL) at 92% efficiency

484

Hulll

for 14C and 56% efficiency for 3H. At the termination of the experiment, the "hot" solution was flushed out of the appropriate chamber with a 5 ml volume and subsequently sampled, as was the solution prior to placement in the chambers. Sodium and bicarbonate fluxes were determined in each direction across the endothelium using paired cornea. The unidirectional fluxes were calculated as follows.82

Unidirectional flux (J) =

increase in counts on chamber solute unlabeled side x volume x concentration counts on labeled side x time

In experiments where net ionic fluxes were determined the "hot" and "cold" chambers were reversed in paired corneas so as to determine ionic movement in the opposite direction. The algebraic sum of the ion movements in each direction is the net flux rate.83 J Net flux = J stroma to endothelium - J endothelium to stroma

In experiments where corneal permeability to inulin and dextran was studied the following equation was used: K trans =

increase in counts on unlabeled side concentration of counts x area of membrane x time on labeled side

This formula has the dimensions centimeter/second.84-86 Experimental and control data at 30-minute intervals was computed and the standard error of the mean determined. A comparison of experimentals and controls was made with the t-test. A P level of 0.01 was used to determined statistical significance. In Vivo Corneal Endothelial Response to Intracameral Injection of Hydrogen Peroxide A separate set of experiments was performed to evaluate the in vivo corneal endothelial response to the intracameral injection of hydrogen peroxide. In this experiment the in vivo response in adult rabbits was compared to that of young rabbits. Intracameral hydrogen peroxide was injected into the anterior chamber of young and adult rabbits and endothelial permeability to inulin and dextran was determined. In this series

Corneal Endothelium and °2

485

of experiments young rabbits ranged from 5 to 7 weeks of age and weighed approximately 1 kg. The adult rabbits in this series of experiments were from 4 to 6 months of age and weighed approximately 2 to 3 kg. Rabbits were anesthetized with a 1:1 mixture of ketamine HCI (100 mg/ml) and xylazine (20 mg/ml). Two cubic centimeters of the mixture was used for adult rabbits. One cubic centimeter was used for young rabbits. An intracameral injection of 10 pAl for adults and 7 ,u for young rabbits of 62.5 mM hydrogen peroxide in distilled water was performed. This resulted in a concentration of 2.06 mM H202 in both the adult and young animals assuming an aqueous volume of 160 ,ul in young animals and 200 R1 in adult animals. Previous experiments in our laboratory demonstrated that hydrogen peroxide injected into the anterior chamber was not restricted to the anterior chamber, but diffused into the iris and the posterior chamber, and resulted in an aqueous humor concentration of 2.06 mM. The paired control eye of each animal received a 10 ,ul (adult animal) or 6 ulI (young animal) intracameral injection of distilled water. Animals were then sacrificed at 48, 72, or 144 hours. Corneas were removed, mounted in flux chambers and inulin and dextram permeabilities determined as described above. Data was determined on six animals at each time period studied. RESULTS

Endothelial Bicarbonate Fluxes During Perfusion with 0.3 mM Hydrogen Peroxide and Ringer's-Bicarbonate Solution Containing Glucose 1 Hour - The unidirectional flux of bicarbonate across the corneal endothelium (from the stroma to the endothelium or J stroma to endothelium) at 1 hour for corneas perfused with 0.3 mM hydrogen peroxide and for controls was statistically similar, P > 0.01 (Fig 7). Endothelial barrier function to bicarbonate (J endothelium to stroma) was also well maintained after 1 hour of hydrogen peroxide perfusion. This resulted in a statistically similar net flux for experimentals and controls at 1 hour (Fig 7). 2 Hours - After 2 hours of perfusion the flux of bicarbonate from the stroma to the endothelium (J stroma to endothelium) was significantly lower (P < 0.01) in experimental corneas perfused with 0.3 mM hydrogen peroxide than it was in control corneas (Fig 7). The endothelial barrier to bicarbonate (J endothelium to stroma) was statistically similar in experiments and controls. This led to a net flux of bicarbonate of 0.15 + 0.28 p.eq/cm2/hr for corneas perfused with 0.3 mM hydrogen peroxide which was significantly less (P < 0.01) when the 1.82 ± 0.20 p.eq/cm2/hr for controls (Fig 7).

486

Hull

3 Hours - Following 3 hours of perfusion the movement of bicarbonate from the stroma to the endothelium (J stroma to endothelium) is similar in corneas perfused with 0.3 mM hydrogen peroxide and controls (Fig 7). However the endothelial barrier to bicarbonate (J endothelium to stroma) following hydrogen peroxide perfusion is compromised compared to control corneas (P < 0.01). The net flux of bicarbonate at 3 hours is similar in both hydrogen peroxide perfused and control corneas (Fig 7). Bicarbonate Flux: mean ±standard error 6 * *Experimental (0.3 mM H202) _-Control (no H202) p