Hypoxia-Inducible Factor 1α in the Glaucomatous Retina and OpticNerve Head FREE

Tissue hypoxia in the optic nerve head and/or retina is thought to developsecondary to or independent from the elevated intraocular pressure in glaucomatouseyes and has been proposed to be associated with pathogenic mechanisms underlyingoptic nerve degeneration in glaucoma. Considerable evidence suggests thattissue hypoxia in the retina may adversely affect the survival of retinalganglion cells by inducing apoptosis.^{1 - 6} Analteration in the microcirculation of the optic nerve head blood supply, whichmay lead to an oligemic-hypoxic insult, has also been suggested to contributeto retinal ganglion cell death in glaucoma.⁷ Furthermore,chronic ischemia of the primate anterior optic nerve is accompanied by thediffuse loss of axons similar to that detected in glaucomatous human eyes.^{8 - 9} The pathophysiology of neuronal hypoxicdamage involves glutamate excitotoxicity,¹⁰ calciumoverload,¹¹ and oxidative stress, includingthat caused by nitric oxide.¹² In addition,tumor necrosis factor α has been implicated in hypoxia-induced neuronalapoptosis.^{6 ,13} Current knowledgerecognizes that most of these hypoxia-associated mediator mechanisms identifiedin the brain are likely involved in glaucomatous optic nerve degeneration.^{14 - 17}

Hypoxia has been postulated to occur in glaucomatous eyes on the basisof blood flow studies. Clinical evidence of vascular abnormalities in glaucomapatients, such as vasospasm, systemic hypotension, angiographic vascular perfusiondefects, and alterations in blood flow variables, has been suggested to resultin reduced vascular perfusion in the optic nerve head and/or retina.^{18 - 23} However,there is no direct evidence demonstrating that tissue hypoxia is present inglaucomatous eyes.

The identification of the molecular mechanisms responsible for the expressionof hypoxia-induced genes provides an opportunity to better understand thepresence of a hypoxic component of the neurodegeneration process in glaucoma.Hypoxia-inducible factor 1 (HIF-1) is an oxygen-regulated transcriptionalactivator that functions as a master regulator of oxygen homeostasis. Underhypoxic conditions, HIF-1 activates the transcription of a broad variety ofgenes, including those encoding erythropoietin, glucose transporters, glycolyticenzymes, vascular endothelial growth factor, inducible nitric oxide synthase,heme oxygenase 1, and other genes whose protein products increase oxygen deliveryor facilitate metabolic adaptation to hypoxia.^{24 - 30}

HIF-1 is a heterodimer composed of α (120 kDa) and β (approximately92 kDa) subunits that belongs to the basic helix-loop-helix Per/Arnt/Sim (PAS)protein family.^{27 ,31} AlthoughHIF-1α is constitutively expressed under normoxic conditions, it israpidly degraded by the ubiquitin-proteosome system.³² However,under hypoxic conditions, HIF-1α is stabilized and accumulated, allowingthe transcriptional activation of several genes.^{30 ,33} Theexpression and activity of HIF-1α have been shown to be tightly regulatedby the cellular oxygen concentration. For example, the expression of HIF-1αis exponentially increased as cells are exposed to oxygen concentrations lessthan 6%.³⁴

HIF-1α has been identified to be expressed in the mammalian centralnervous system,³⁵ including neurons.^{36 - 37} Although increased levels of HIF-1have been detected in the ischemic retina,³⁸ theexpression of HIF-1α has not been studied in glaucoma. To determinetissue hypoxia in glaucoma, we studied the cellular localization of the HIF-1αprotein in the retina and optic nerve head of human donor eyes with glaucomacompared with control eyes from healthy donors, using immunohistochemicalanalysis. Our findings demonstrated an increased immunostaining for HIF-1αin the retina and optic nerve head of glaucomatous eyes, which indicates thathypoxic tissue stress is present in these eyes. In addition, the retinal immunostainingfor HIF-1α in some of the glaucomatous eyes was found to exhibit a spatialrelationship with functional damage recorded in these eyes. These findingssupport the pathophysiological role of hypoxic signaling in glaucomatous neurodegeneration.

Twenty-eight donor eyes with a diagnosis of glaucoma (donor age, 56-94years) and 20 eyes from age-matched, normal donors (donor age, 55-96 years)were obtained from the Glaucoma Research Foundation (San Francisco, Calif),the Mid-America Eye Bank (St Louis, Mo), one of us (M.B.W.) (Alcon Laboratories,Ft Worth, Tex), and Douglas H. Johnson, MD (Mayo Clinic, Rochester, Minn).All of the human donor eyes were handled according to the tenets of the Declarationof Helsinki. Clinical findings of glaucomatous donors were well documented,which included intraocular pressure readings, optic disc assessments, andvisual field tests (Table 1).Normal donors had no history of eye disease. There was no diabetes, collagenvascular disease, infection, or sepsis in any of the donors. The cause ofdeath for all of the donors used in this study was acute myocardial infarctionor cardiopulmonary failure.

All of the donor eyes were enucleated within 2 to 4 hours after deathand fixed within 6 to 12 hours. To know retinal orientation of histologicsections and to be able to examine the relationship between the pattern ofimmunostaining and functional damage, 10 freshly obtained globes from glaucomatousdonors were marked for nasal, temporal, superior, and inferior sites beforetheir processing. The posterior poles were dissected from the surroundingtissues, washed extensively in 0.2% glycine in phosphate-buffered saline atpH 7.4, embedded in paraffin, and oriented sagittally to obtain 5-mm sections.Microscopic examination of histologic slides revealed that the retina andoptic nerve head tissues were relatively well preserved, with rare perimortemand/or processing-related alterations of photoreceptors.

Immunoperoxidase staining and double immunofluorescence labeling wereused to study the cellular localization of HIF-1α. All of the histologicslides subjected to immunohistochemical analysis were masked for the identityand diagnosis of donors and numbered by a technician unfamiliar with the retinaand optic nerve head condition before their immunostaining. To control variationsin the immunostaining, slides obtained from glaucomatous and control eyes,as well as the negative control slides, were simultaneously subjected to immunohistochemicalanalysis. The intensity of immunostaining was first qualitatively graded asnegative (−), faint (+), moderate (++), and strong (+++) using at least5 histologic sections from each donor eye. To obtain complementary information,we then performed quantitative image analysis as recently described.³⁹ For this purpose, chromogen quantity per pixel wasmeasured using the TIFFalyzer program,⁴⁰ andthe value obtained from the negative control slide was subtracted from theexperimental slide to determine the intensity of immunostaining.

To determine whether there is a relationship between the retinal regionsthat exhibit increased immunostaining for HIF-1α and the location ofvisual field defects, we correlated the gradings of HIF-1α immunostainingand visual field defects in glaucomatous eyes. Although immunoreactivity mayvary among different individuals, we considered that a masked evaluation couldbe informative in determining the correlation of glaucomatous damage withHIF-1α immunostaining in corresponding retinal quadrants of individualeyes.

Visual field defects in 4 quadrants were classified in these 10 eyesin a masked fashion. These classifications were based on the last availablevisual field test results (at most 2 years before death) in the patients'clinical records. The visual field test results evaluated had been obtainedusing the Humphrey Field Analyzer (30-2 program) and had met the reliabilitycriteria of a fixation loss less than 20% and false-positive and false-negativerates less than 30%. We calculated the mean of visual field indices withinquadrants of the total deviation plot, and the quadrants were defined as havingno, mild, moderate, or advanced visual field deficit if the mean defect wasmore than −2 dB, −2 to −6 dB, −7 to −15 dB,or less than −15 dB, respectively.^{41 - 42}

For immunoperoxidase staining, retinal sections from normal and glaucomatouseyes were deparaffinized, rehydrated, and pretreated with 3% hydrogen peroxidein methanol to decrease endogenous peroxidase activity. Following washingwith phosphate-buffered saline solution containing 0.1% bovine serum albumin,the sections were incubated with 20% inactivated normal donkey serum (ChemiconInternational Inc, Temecula, Calif) for 30 minutes at room temperature toblock background staining. The sections were then incubated with a mouse monoclonalantibody to HIF-1α (1:1000; Novus Biologicals Inc, Littleton, Colo)for 16 hours at 4°C. After washing, the sections were incubated with thebiotinylated anti-mouse IgG (1:400; Chemicon International Inc) for 1 hourat room temperature and then with avidin-biotin complex (ABC reagent, VectastainABC Elite kit; Vector Laboratories, Burlingame, Calif) for 1 hour at roomtemperature. Following several washes, color was developed by incubation with3,3-diaminobenzidine tetrahydrochloride (Sigma-Aldrich, St Louis, Mo) as cosubstratefor 5 to 7 minutes. Sections were counterstained with Mayer hematoxylin andmounted with Permount (Fisher Scientific, Pittsburgh, Pa). The primary antibodywas eliminated from the incubation medium, or mouse serum (Sigma-Aldrich)was used to replace the primary antibody to serve as the negative control.Slides were examined in a microscope (Nikon, Tokyo, Japan), and images wererecorded by digital photomicrography (Optronics, Goleta, Calif).

To study cellular localization of HIF-1α to different cell typesin the retina, we performed double immunofluorescence labeling using antibodiesagainst specific markers. We used a rabbit antibody against glial fibrillaryacidic protein (GFAP) as a marker of macroglial cells. To identify retinalganglion cells, we used a rabbit antibody to Brn-3a (Chemicon InternationalInc) that is a member of the POU-domain genes, which are known to be expressedby most ganglion cells across a variety of mammalian species based on colocalizationstudies using retrograde tracers.⁴³

For double immunofluorescence labeling, sections were incubated witha mixture of anti–HIF-1α antibody and a rabbit antibody (1:400;anti-GFAP or anti–Brn-3a antibody) for 1 hour at room temperature. Thesections were then incubated with a mixture of rhodamine red–labeledanti-mouse IgG and Oregon green–labeled anti-rabbit IgG (1:400; MolecularProbes, Eugene, Ore) for another hour at room temperature. Negative controlswere performed by replacing the primary antibody with serum or by incubatingsections with each primary antibody followed by the inappropriate secondaryantibody to determine that each secondary antibody was specific to the speciesit was made against. Slides were examined in a fluorescence microscope (Nikon),and images were recorded by digital photomicrography (Optronics).

Immunoperoxidase staining using a monoclonal antibody against HIF-1αwas virtually negative in the retina of healthy donor eyes, except for veryfaint immunostaining in limited regions (Figure 1A). However, immunostaining for HIF-1α was detectablein retina sections obtained from the glaucomatous donor eyes (Figure 1B). HIF-1α immunostaining in the glaucomatous retinawas detectable in all of the slides examined, although the intensity of immunostainingand the number of cells immunostained for HIF-1α exhibited individualand regional differences. Immunostaining for HIF-1α in the glaucomatousretina was qualitatively graded as moderate or strong. Using digital imageanalysis, the intensity (mean ± SD) of HIF-1α immunostainingwas 76 ± 12 energy units per pixel in the glaucomatous retina but fewerthan 16 energy units per pixel in the control retina.

Immunostaining for HIF-1α in the glaucomatous retina was predominantin the inner retinal layers, mostly in the retinal ganglion cell layer. Inthis layer, immunostaining was detectable in cells with large cell bodies,which likely correspond to retinal ganglion cells. In addition, based on themorphologic characteristics and known localization of retinal cell types,some glial cells, including astrocytes and Müller cells, exhibited immunostainingfor HIF-1α in the glaucomatous retina. For example, retinal astrocytesare localized in the inner retina and can be differentiated from the retinalganglion cells by their characteristic darker, smaller, and irregular nucleusrelative to that of ganglion cells and by their proximity to the blood vesselsof the inner retina.^{44 - 45} Anothermacroglial cell type in the retina, Müller cells, are characterized bytheir radial orientation and processes that extend all through the retina,⁴⁶ although their cell bodies are located in the innernuclear layer.^{44 - 45} Glial immunostainingfor HIF-1α was mostly associated with the nuclei, although glial cellprocesses that extend all through the retina exhibited some faint immunostaining.However, immunostaining of retinal ganglion cells for HIF-1α was mostly,not exclusively, localized to the cytoplasm (Figure 1C).

Some retinal immunostaining for HIF-1α was also detectable inscattered cells located in the inner nuclear or inner plexiform layers, whichwere likely amacrine cells, based on their morphologic characteristics. Inaddition, faint immunostaining was detectable at the inner wall of the bloodvessels in the glaucomatous retina, as well as in the perivascular glial cells(Figure 1D). Control slides in whichthe primary antibody was omitted or replaced with nonimmune sera were allnegative for specific immunostaining for HIF-1α.

Although increased immunostaining for HIF-1α in the glaucomatousretina was mostly detectable in the inner retinal layers, HIF-1α immunostainingwas also detectable in the photoreceptor cells in the eyes of a patient withnormal-pressure glaucoma (eyes 20 and 21), whose clinical and histopathologicfindings were previously documented⁴⁷ (Figure 2). This observation suggests thatindividual factors play a role in determining tissue hypoxia and/or cellularresponses to tissue hypoxia and/or immunoreactivity in different retinal layers.

Similar to the control retina, immunostaining of the control optic nervehead for HIF-1α was virtually negative (Figure 1E). However, optic nerve head sections from glaucomatouseyes exhibited immunostaining of glial cells for HIF-1α (Figure 1F) in all of the slides examined, despite interindividualdifferences. Immunostaining for HIF-1α in the glaucomatous optic nervehead was qualitatively graded as moderate or strong. Using digital image analysis,the intensity of the HIF-1α immunostaining was 84 ± 14 energyunits per pixel in the glaucomatous optic nerve head but 16 ± 5 energyunits per pixel in the control optic nerve head. Glial immunostaining forHIF-1α was mostly detectable at the prelaminar and laminar regions ofthe glaucomatous optic nerve head. Both the processes and the nuclei of someglial cells around the nerve bundles and blood vessels were immunostainedfor HIF-1α. HIF-1α immunostaining was also detectable in nervebundles that passed through the prelaminar region of the optic nerve head.

To identify retinal cell types that exhibit HIF-1α immunostaining,double immunofluorescence labeling was performed. We observed that the increasedimmunostaining for HIF-1α in glaucomatous eyes was localized to theretinal ganglion cells and glial cells. As shown in Figure 3, HIF-1α immunostaining was colocalized with the immunostainingfor Brn-3a, which is a marker for retinal ganglion cells⁴³ (Figure 3A-C), or with the immunostainingfor GFAP (Figure 3D-F), which isa marker for astrocytes and Müller cells in glaucomatous eyes. The cellularpattern of HIF-1α immunostaining using double immunofluorescence labelingwas consistent with the observations using immunoperoxidase staining, in whichHIF-1α immunostaining was detectable mostly in the cytoplasm of Brn-3a–positiveganglion cells, although the immunostaining of GFAP-positive glial cells wasmostly localized to the nuclei.

An examination of the relationship between the retinal regions thatexhibited increased immunostaining for HIF-1α and the location of visualfield defects in 10 glaucomatous donor eyes, in which visual field data couldbe referenced to histologic sections, revealed a close correspondence betweenthe increased immunostaining for HIF-1α and the functional damage. Asshown in Figure 4, retinal immunostainingfor HIF-1α was more prominent in retinal regions corresponding to visualfield defects. Although the intensity of immunostaining varied among differentindividuals, the grading of immunostaining was generally faint or moderatein retinal regions, corresponding to relatively normal visual field sensitivityin individual eyes. Table 2 documentsthe corresponding gradings of visual field defects and HIF-1α immunostainingin 10 glaucomatous eyes. Because the expression of HIF-1α is an indicatorof hypoxic tissue stress, the retinal areas corresponding to visual fielddefects in these glaucomatous eyes are the retinal areas in which tissue hypoxiais predominant.

Our findings demonstrate that immunostaining for HIF-1α in theretina and optic nerve head of glaucomatous eyes is greater than in the controleyes matched for several characteristics. Additionally, we examined the relationshipbetween the retinal regions exhibiting increased immunostaining for HIF-1αand the location of visual field defects recorded in some of these glaucomatouseyes. This demonstrated that the increased immunostaining for HIF-1αin glaucomatous eyes was most prominent in retinal regions that correspondedto areas of decreased light sensitivity as determined by achromatic thresholdvisual field testing. Because the regions of HIF-1α induction representthe areas of decreased oxygen delivery and hypoxic stress, these observationssuggest that tissue hypoxia is present in glaucomatous eyes, and hypoxic signalingis a likely component underlying the pathogenic mechanisms of glaucomatousneurodegeneration.

Immunostaining for HIF-1α was virtually negative in the controltissues. This is consistent with the fact that HIF-1α is constitutivelyexpressed yet hardly detectable in normoxic cells due to its rapid degradation.³² Tissue oxygen tension is not homogeneously distributedin the retina as a direct consequence of oxygen delivery to the tissue bydiffusion from the capillary network, which varies throughout the retina.^{48 - 49} Thus, there are regions of normoxicretinal tissue with low oxygen levels, which could explain the very faintregional immunostaining in the control sections.

There was a prominent increase in the immunostaining for HIF-1αin glaucomatous eyes compared with control eyes, although the intensity ofimmunostaining exhibited regional differences, which were positively correlatedwith the locations of visual field defects. In addition to regional differences,the intensity of immunostaining for HIF-1α exhibited differences amongglaucomatous donor eyes. This may indicate differences in tissue hypoxia amongdifferent individuals. In addition, the individual differences in HIF-1αimmunostaining may partly be associated with individual factors that determinethe immunoreactivity and/or cellular responses to tissue hypoxia. Alternatively,differences in the intensity of immunostaining between different glaucomatousdonor eyes might have been associated, in part, with the fixation time ofglobes before their processing for immunohistochemical analysis, which variedamong donors.

Retinal immunostaining for HIF-1α in glaucomatous eyes was mostlyassociated with cells located in the inner retinal layers. This observationis consistent with previous observations in the ischemic retina.³⁸ Immunostainingof the glaucomatous optic nerve head for HIF-1α was similarly most prominentin the inner layers. Ischemia of the axons at the optic nerve head can initiatea retrograde signal to their cell bodies to up-regulate HIF-1α. However,in addition to the immunolabeling of retinal ganglion cells, immunolabelingof retinal glial cells for HIF-1α suggests that tissue hypoxia is presentin the retina as well.

Similar to the inner retina, superficial layers of the optic nerve headare known to be supplied mostly by the retinal circulation with cilioretinalcapillary anastomoses.^{50 - 52} However,the predominant increase of HIF-1α immunostaining in the inner layersof the retina and optic nerve head in glaucomatous eyes does not necessarilymean that tissue hypoxia is manifest mainly in the regions supplied by theretinal circulation. First, blood flow studies demonstrate vascular abnormalitiesin both the retinal and the choroidal systems in glaucomatous eyes,^{20 ,53} although choroidal and retinal vasculaturemay be differently affected in primary open-angle glaucoma and normal-pressureglaucoma.^{54 - 55} Second, it is feasiblethat all layers of the retina and/or optic nerve head sustain an equal amountof hypoxic stress in glaucomatous eyes, whereas hypoxia-induced regulationof HIF-1α expression may vary among their different layers, resultingin the prominent up-regulation in the inner layers. For example, recent observationsin glaucomatous eyes provide evidence that the signaling molecules, also involvedin the regulation of HIF-1α, are differentially activated through theretina.³⁹ In addition, in one of the patientswith normal-pressure glaucoma, whose clinical and histopathologic findingswere previously documented,⁴⁷ HIF-1αimmunostaining was also detectable in the outer retinal layer, including thephotoreceptor cells. This observation signifies the role of individual factorsin determining tissue hypoxia and/or cellular responses to tissue hypoxiain different retinal layers. Individual factors that determine immunoreactivitymay also be associated with the differential pattern of immunostaining amongdifferent eyes.

We also noticed cellular differences in the immunostaining for HIF-1αin glaucomatous eyes. For example, HIF-1α immunostaining was more prominentwithin the cytoplasm of retinal ganglion cells, whereas glial immunostainingfor HIF-1α occurred predominantly in the nucleus. Nuclear immunostainingfor HIF-1α is consistent with the fact that HIF-1α is a DNA bindingprotein and is localized to the nucleus for its transcriptional activity.²⁵ However, cytoplasmic immunostaining may be associatedwith the insufficient transcriptional activity and thereby stabilized cytoplasmicexpression of HIF-1α and/or its insufficient degradation in the ubiquitin-proteosomesystem.³² It has been demonstrated that thestabilization, transcriptional activation, and proteosomal degradation ofHIF-1α are independently regulated.⁵⁶ Thesignaling components controlling the regulation of these intracellular events,which require phosphorylation,^{57 - 58} mayvary among different cell types, depending on kinase activity. Therefore,the variability we observed in the intracellular localization of HIF-1αimmunostaining in retinal cell types may be associated, in part, with thedifferential activity of the regulatory signaling cascades between retinalganglion cells and glial cells, which is evident in glaucomatous eyes.³⁹ Additional studies are warranted to identify theregulation of HIF-1α and the associated signaling cascades in glaucoma.Such information may also explain why the sensitivity of these cell typesto glaucomatous damage, as well as to hypoxic stress, is not uniform, despitethe fact that both neuronal and glial cells are exposed to similar hypoxicstress and exhibit increased immunostaining for HIF-1α in the glaucomatouseyes.

In addition to retinal ganglion cells and glial cells, some immunostainingfor HIF-1α was also detectable in the blood vessels. This may supportthe proposed role of HIF-1α in metabolic adaptation to hypoxia, suchas an increase in oxygen delivery and glucose transport through the up-regulationof vascular endothelial growth factor and glucose transporters.

During prolonged exposure to hypoxia, HIF-1α is expressed as longas the balance between oxygen supply and tissue utilization has not been reached.For example, it has been shown that the expression of HIF-1α rapidlyincreases during the onset of hypoxia in rat brain and remains increased forat least 14 days. Despite the continuously low arterial oxygen tension, however,it returns to normal within 21 days following compensatory adaptations.⁵⁹ Therefore, our results demonstrating immunostainingfor HIF-1α in a diverse sample of glaucomatous donor eyes may signifya sustained hypoxic insult or possibly recurrent episodes of tissue hypoxiain these eyes.

Glaucomatous donors whose eyes were used in immunohistochemical analysiswere under antiglaucomatous treatment, and their last available intraocularpressure readings were within normal limits. Because of this and the retrospectivenature of our data collection, we considered that the determination of a relationshipbetween the intraocular pressure and HIF-1α immunostaining in theseeyes would not be precisely informative. No evidence is currently presentthat any medication could induce HIF-1α expression and explain HIF-1αimmuostaining in glaucomatous eyes, which exhibits a specific pattern closelyconcordant with the functional damage. Whether it is secondary to or independentfrom the elevated intraocular pressure,^{20 ,60} ourfindings provide evidence for the presence of hypoxic stress in the glaucomatousretina and optic nerve head and suggest that hypoxic signaling is associatedwith the initiation and/or progression of neuronal damage. HIF-1α hasbeen shown to coordinate the expression of not only adaptive but also pathogenicgenes, such as p53, and has therefore been implied to promote delayed neuronalcell death.³⁷ HIF-1α may similarly beassociated with the activation of a cell death program in glaucomatous eyesthrough p53, which has been suggested to be a transcriptional activator ofneuronal apoptosis in glaucoma.⁶¹

Molecular mechanisms responsible for oxygen sensing are still poorlyunderstood. Recent experimental studies⁶² haveprovided evidence in support of the hypothesis that mitochondrial generationof reactive oxygen species (which are implicated in glaucomatous neurodegeneration¹⁵ ) is required for the induction of HIF-1α expressionand activity. Although the signal transduction pathways remain enigmatic,recent advances in this field imply that HIF-1 activity is also regulatedby different signals, including nitric oxide and cytokines, such as tumornecrosis factor α,^{63 - 66} whichare also implicated in glaucoma.^{16 - 17} Thus,it is clear that both the expression and the transcriptional activity of HIF-1αare regulated by the cellular oxygen concentration and the redox modificationsof the protein,^{24 ,63 ,67} althoughmany details remain elusive.

Despite the unique informative value of immunohistochemical studiesof postmortem human tissues, their findings can be difficult to interpretmainly due to perimortem tissue alterations. Obviously, such tissue alterationscannot fully be ruled out for the tissues used in our study, because the photoreceptorlayer shown in the figures may be suspected of having such abnormalities.However, all of the tissues used in this study were fixed within 12 hoursafter death. Within this period, retinal neurons, including human photoreceptorcells, have been reported to maintain high viability.⁶⁸ Inaddition to the death-to-tissue fixation period, glaucomatous and controleyes were also carefully matched for donor age and cause of death, and anyeyes with ocular vascular diseases were excluded. These features should minimizethe vulnerabilities intrinsic to such a study and validate the interpretationof immunostaining differences between glaucomatous eyes and control eyes fromhealthy donors.

In conclusion, immunostaining for HIF-1α is increased in the retinaand optic nerve head of glaucomatous eyes and exhibits a spatial relationshipwith functional damage recorded in these eyes. Because the expression of HIF-1αis a direct indicator of hypoxic stress, our findings demonstrate that hypoxicstress is present in the glaucomatous retina and optic nerve head and HIF-1αsignaling may have a pathophysiologic role in the development and/or progressionof neurodegeneration in these eyes. HIF-1α is known to play a centralrole in the cellular regulation of a broad variety of hypoxia-inducible proteinsand is critical for the determination of ultimate cell fate in response tohypoxic stress. Therefore, better understanding of the regulation of HIF-1αand the associated signaling cascades can provide insights into the pathogenicmechanisms of glaucomatous neurodegeneration, thereby providing new strategiesfor neuroprotection in glaucoma.

Correspondence: Gülgün Tezel, MD, Department of Ophthalmologyand Visual Sciences, University of Louisville School of Medicine, Louisville,KY 40202 (gulgun.tezel@louisville.edu).

Submitted for publication February 12, 2003; final revision receivedJanuary 16, 2004; accepted March 26, 2004.

This study was supported in part by the Glaucoma Foundation, New York,NY (Dr Tezel); American Health Assistance Foundation, Clarksburg, Md (Dr Tezel);National Eye Institute, Bethesda, Md (1R01 EY13813; Dr Tezel); Universityof Louisville, Louisville, Ky; and an unrestricted grant to Department ofOphthalmology and Visual Sciences, University of Louisville, from Researchto Prevent Blindness Inc, New York, NY.

We thank Douglas H. Johnson, MD, Mayo Clinic, Rochester, Minn, for providingglaucomatous donor eyes.