Retinal Neuroprotection against Ischemic Injury Mediated by Intraocular Gene Transfer of Pigment Epithelium-Derived Factor

Abstract

purpose. To determine whether intraocular gene transfer of pigment epithelium-derived factor (PEDF) protects the retina from ischemia-reperfusion injury.

methods. Four days before induction of pressure-induced ischemia, Lewis rats received intravitreous injection of 3 × 10⁹ particles of an adenovirus vector expressing PEDF (AdPEDF.11) in one eye and 3 × 10⁹ particles of an empty adenovirus vector (AdNull.11) in the contralateral eye. Seven days after reperfusion, eyes were enucleated and processed for morphometric analysis. Apoptotic cells stained by TdT-dUTP terminal nick-end labeling (TUNEL) in the retina were counted 12 hours after initiation of reperfusion. Retina levels of PEDF were measured by enzyme-linked immunosorbent assay.

results. PEDF levels in retinal homogenates from eyes receiving AdPEDF.11 injection were well above the background levels in the untreated baseline and control eyes (P = 0.04). Retinal thickness was preserved in AdPEDF.11-treated eyes. Retinal cell density was significantly greater in the ganglion cell layer (GCL; P = 0.014), inner nuclear layer (INL; P = 0.008), and outer nuclear layer (ONL; P = 0.008) of AdPEDF.11-treated eyes compared with the corresponding layers in AdNull.11-treated eyes. AdNull.11-treated eyes also had significantly more TUNEL-positive cells in these layers than AdPEDF.11-treated eyes (P < 0.05).

conclusions. Adenoviral vector-mediated intraocular expression of PEDF significantly increases cell survival after ischemia-reperfusion injury of the retina. The protective effect may result from inhibition of ischemia-induced apoptosis. This study provides proof of concept for a gene transfer approach directed at interrupting programmed cell death induced by retinal ischemic insult.

Oxidative stress is a mediator of apoptosis, ^{1 2} that has been implicated in the pathogenesis of a wide variety of retinal conditions including hereditary retinal degenerations, retinal light damage, hydrogen peroxide induced oxidative stress, glaucoma, retinal detachment, retinal ischemia, and others. ³ Ischemia-reperfusion induced apoptosis has been implicated in the neuronal cell death of diabetic retinopathy, retinal vein occlusion, sickle cell retinopathy, and other retinal vascular occlusive diseases. Each of these ischemic retinopathies is associated with retinal neovascularization that is commonly associated with severe visual loss. ⁴  

Pigment epithelium-derived factor (PEDF), a 50-kDa protein, is secreted by human fetal retinal pigment epithelial cells and has been shown to induce neuronal differentiation of human Y-79 retinoblastoma cells in vitro. ^{5 6} The human PEDF gene is located on the short arm of chromosome 17, region 13.3, where a locus for autosomal dominant retinitis pigmentosa is also found, ^{7 8} suggesting that PEDF could be a survival factor for neuronal cells of the retina. 

PEDF has been shown to be protective in models of inherited photoreceptor degeneration, ⁹ hydrogen peroxide-induced neuronal cell death, ³ ischemic retinal injury, ¹⁰ and retinal light damage. ¹¹ PEDF has recently been demonstrated to be a potent antiangiogenic agent that inhibits the migration of endothelial cells in vitro and has been a more potent antiangiogenic agent than angiostatin, thrombospondin-1, or endostatin in assays. ¹² Systemic injection of recombinant PEDF protein is reported to prevent the development of retinal neovascularization in mice with oxygen-induced ischemic retinopathy by promoting apoptosis of vascular endothelial cells. ¹³ PEDF is therefore potentially both a promising endogenous inhibitor of angiogenesis and a neuroprotective protein. 

Viral vectoring of genes into the ocular tissues provides for sustained local delivery of therapeutic agents. The eye, being both small and a relatively isolated compartment, requires a comparatively small amount of vector to transfect a large number of ocular cells. We recently demonstrated that intraocular injection of an expression construct for PEDF packaged in an adenoviral vector with E1, E3, and E4 deletions (AdPEDF.11), ¹⁴ as well as an adeno-associated viral construct, ¹⁵ inhibits retinal and choroidal neovascularization. These studies provided proof of concept of a gene transfer approach to treating ocular neovascularization. Currently, a phase 1 clinical trial evaluating AdPEDF.11 in eyes with choroidal neovascularization is enrolling patients. ¹⁶  

This study is intended to evaluate whether gene transfer approaches may be extended to include retinal neuroprotection, in the setting of ischemia-reperfusion injury. To this end, the potentially protective effects of AdPEDF.11 were evaluated in a rat model of retinal ischemia-reperfusion injury. An explanation of beneficial effect was sought by correlating retinal tissue morphometric analysis with apoptotic change as determined by TUNEL staining of retinal specimens from treated and control eyes. 

Methods

Animals

Female Lewis rats weighing 180 to 230 g were used at 4 to 8 weeks of age. The animals were anesthetized by intramuscular injection of 80 mg/kg of ketamine hydrochloride. The pupils were dilated with 0.5% tropicamide and 0.5% phenylephrine hydrochloride. All the animals were treated under deep sedation in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

Retinal ischemia-reperfusion alone was induced in 20 control rats by increasing intraocular pressure. Thirty-one rats received intravitreous injection of adenoviral vectors followed by induction of pressure-induced retinal ischemia-reperfusion injury. Six baseline untreated rats were used for an enzyme-linked immunosorbent assay (ELISA) of PEDF. 

Adenoviral Vectors of PEDF and Intraocular Injection Procedures

Serotype 5 adenoviral vectors expressing PEDF from a cytomegalovirus (CMV) immediate early promoter expression cassette have been described and characterized. ^{14 17 18} The vectors are deleted for E1A, E1B, E3, and E4 (AdPEDF.11). The same vector without transgene expression was used as a null virus control (AdNull.11). 

We have reported on the time course of intraocular expression of similar adenoviral vectors containing reporter genes and have shown that peak expression occurs at 3 to 5 days with elevated expression persisting well beyond the 7-day experimental period used in this investigation. ¹⁸ Rats in this experiment were injected 4 days before ischemic insult, receiving intravitreous injection of 3 × 10⁹ particles of AdPEDF.11 in one eye and 3 × 10⁹ particles of AdNull.11 in the contralateral eye. Intravitreous injection was performed with a Hamilton syringe fitted with a 33-gauge beveled needle. The needle was passed through the sclera at the equator into the vitreous cavity. Injection occurred with direct observation of the needle in the center of the vitreous cavity. 

ELISA of PEDF

Rats were assigned to four groups: no treatment, ischemia-reperfusion alone, and ischemia-reperfusion with intravitreous injection of 3 × 10⁹ viral particles of AdNull.11 or AdPEDF.11. Intravitreous vector injection was performed 4 days before ischemia-reperfusion insult. Twelve hours after the initiation of reperfusion, retinas were removed and immediately frozen. Whole rat retinas were extracted with 0.1% Triton X-100 in PBS with a protease inhibitor cocktail (Roche Molecular Biochemicals, Mannheim, Germany). The mixture was sonicated for 15 minutes at 4°C (model 1510; Branson, Shelton, CT). The retinas were then fully homogenized by mechanical disruption. After microfuging, total protein was measured in the supernatant with a protein assay (Bio-Rad, Hercules, CA). 

PEDF concentrations were determined using a sandwich ELISA in immunoplates (Easywash 96-well plate; Corning, Corning, NY). Wells were coated with 5 μg/mL of a rabbit polyclonal anti-human PEDF antibody in 100 μL of PBS for 16 hours at 4°C. Wells were then blocked for 2 hours at room temperature with 300 μL of nonprotein blocking reagent (Synblock; Immunochemistry Technologies, Bloomington, MN). Either 100 μL of retina extracts or recombinant human PEDF standards ¹⁹ were added for 2 hours at 37°C. Wells were then washed and incubated for 1 hour at 37°C with 100 μL of rabbit anti-human PEDF polyclonal antibody conjugated to horseradish peroxidase. After the wells were washed, an ELISA substrate-peroxide mixture (Turbo TMB; Pierce Biotechnology, Rockford, IL) was added for 20 minutes. The reaction was terminated with 100 μL of 2 M sulfuric acid, and the plate was read with a microplate reader. 

Pressure-Induced Ischemia-Reperfusion Model

The anterior chamber was cannulated with a 27-gauge needle connected to a bag containing normal saline. Raising the bag of saline to a predetermined height raised the intraocular pressure of the cannulated eye to 110 mm Hg. This was maintained for 60 minutes in all animals. Sham surgery was performed without increasing the intraocular pressure. The corneal wound was covered by cyanoacrylate adhesive to avoid fluid leakage. After 60 minutes of retinal ischemia, the intraocular pressure was lowered to normal. Both retinal ischemia and reperfusion were confirmed by ophthalmoscopic evaluation. The body temperature was maintained at 37°C with a heating blanket throughout the period of ischemia. 

Morphometric Analysis

Control eyes receiving only an ischemia-reperfusion insult were enucleated at 0, 1, 7, 14, or 28 days after reperfusion. Eyes with viral vector administration before ischemia-reperfusion treatment were enucleated at 7 days. All eyes were immediately fixed in 4% paraformaldehyde in phosphate-buffered saline (PBS) for 60 minutes. After they were rinsed with PBS, the eyes were frozen in optimum cutting temperature embedding compound (OCT; Miles Diagnostics, Elkhart, IN) and snap frozen in liquid nitrogen, after which they were stored at −80°C until sectioning. At cryosectioning, five serial sections (10 μm) were obtained at 100-μm intervals on each side of the optic nerve. Sections through the optic nerve were also taken, but optic nerve tissue was not included in cell counts. All specimens were processed for hematoxylin and eosin (Sigma-Aldrich, St. Louis, MO) staining. 

The numbers of nuclear cells in the GCL, INL, and ONL were counted per 200-μm length at more than 10 points selected randomly. The mean cell count of these points was then used to determine a representative cell number for each layer. 

Identification of Apoptotic Cells by TUNEL

Apoptotic cells were detected by TdT-dUTP terminal nick end-labeling (TUNEL). Based on our previous baseline studies ²⁰ and those of others, ²¹ control eyes, without viral vector injection, were enucleated at 0, 6, 12, 24, or 72 hours after reperfusion, and immediately fixed in 2% paraformaldehyde in phosphate-buffered saline (PBS) for 60 minutes. After rinsing with PBS, eyes were frozen in OCT compound. Frozen serial sections (10 μm) were obtained and sections that included the optic disc were processed for morphometric analysis of TUNEL-stained tissues as described later. Sections were stained with the in situ cell death detection kit (Roche Molecular Biochemicals) in accordance with the manufacturer’s protocol, with minor modifications. The specimens were also double stained with propidium iodide, after which they were examined with a scanning laser confocal microscope (Radiance 2000; Bio-Rad, Hercules, CA). In this way, the timing of peak TUNEL staining for full-thickness retina was determined to occur from 12 to 24 hours. Eyes pretreated with viral vector and then subjected to ischemia-reperfusion conditions were therefore enucleated at 12 hours and processed in the same manner. 

Because TUNEL-positive cells appears sporadically, the number of TUNEL-positive cells in the GCL, INL, and ONL per retina was counted in three or more sections that included the optic disc. The mean cell count of these sections was then used to determine a representative cell number for each layer of each eye. Statistical analysis was performed with a paired t-test for morphologic analysis and an unpaired t-test assuming unequal variances for ELISA study. P < 0.05 was considered to be statistically significant. 

Results

Increased Retina Levels in Eyes with AdPEDF.11 Administration

Rats given an intravitreous injection of AdPEDF.11 showed levels of human PEDF of 44.7 ± 15.1 (mean ± SE) pg/μg in total retinal protein. All rats in the other three groups had undetectable levels of PEDF. PEDF levels in the retina with AdPEDF.11 injection were well above the background levels in the untreated baseline and other control eyes (P = 0.04). There was no detectable expression of PEDF induced by ischemia-reperfusion (Fig. 1) . 

Time Course of Ischemia-Reperfusion Damage

Lewis rats exposed to retinal ischemia-reperfusion insult showed progressive retinal damage over the studied time course. Immediately after reperfusion, the retina was normal to mildly edematous (Fig. 2A) . Examination 1 day after reperfusion showed a retina that was both reduced in thickness and had fewer nuclei in the GCL, INL, and ONL (Fig. 2B) . Reduction in the number of cells and retinal thinning plateaued after day 7 (Figs. 2C 2D 2E 2F 2G 2H) . 

Effect of Adenoviral Vector-Mediated Expression of PEDF on Ischemic Rat Retina

Four days before induction of pressure-induced ischemia, rats received intravitreous injection of 3 × 10⁹ particles of AdPEDF.11 in one eye and 3 × 10⁹ particles of AdNull.11 in the contralateral eye. Because visible ischemic damage plateaued after day 7, experimental eyes with viral vector administration followed by ischemia-reperfusion insult were enucleated at 7 days from onset of reperfusion and processed for tissue preparation. Typical histologic changes are shown in Fig. 3 . Retinal thickness and cell density in the GCL, INL, and ONL were reduced in eyes treated with AdNull.11 compared with those in eyes treated with AdPEDF.11. Morphometric analysis was performed to analyze quantitatively the cell nuclei in the GCL, INL, and ONL. Nuclear cell counts of the GCL, INL, and ONL in eyes treated with AdNull.11 were 14.7 ± 2.8, 74.0 ± 15.3, and 275.6 ± 45.7 (mean ± SD), respectively. In eyes treated with AdPEDF.11 the cell counts in the GCL, INL, and ONL were 20.6 ± 6.0, 109.5 ± 20.0, and 366.7 ± 46.3, respectively. The preservation of cell counts in eyes treated with AdPEDF.11 was statistically significant in all layers: GCL (P = 0.014), INL (P = 0.008) and ONL (P = 0.008; Fig. 4 ). 

Expression of PEDF Inhibits Apoptosis in Ischemic Retinal Cells

Based on our prior studies ²⁰ and those of others, ²¹ we know that the detection of TUNEL-positive cells is a transient phenomenon. Therefore, we examined the time course of development of TUNEL-positive staining in this setting and determined that the appearance of TUNEL-positive cells peaked from 12 to 24 hours in whole retina (Fig. 5) . Twelve hours after reperfusion, AdPEDF.11- and AdNull.11-injected eyes were enucleated and double stained with TUNEL and propidium iodide. There were many TUNEL-positive cells in the GCL, INL, and ONL which corresponded to pyknotic cells stained by propidium iodide. AdPEDF.11-treated eyes had fewer of these cells than did AdNull.11-treated eyes (Fig. 6) . 

Mean TUNEL-positive cell counts in the GCL, INL, and ONL in retina after AdNull.11 treatment were 22.8 ± 8.8, 123.4 ± 72.3, and 153.4 ± 68.6 (mean ± SD), respectively. In eyes treated with AdPEDF.11, the mean cell counts in the GCL, INL, and ONL were 8.6 ± 2.4, 70.1 ± 35.6, and 47.6 ± 35.0, respectively. The TUNEL-positive cell counts in eyes treated with AdPEDF.11 was significantly lower in all layers: GCL (P = 0.007), INL (P = 0.026), and ONL (P = 0.002; Fig. 7 ). 

Discussion

Currently, there is no effective treatment for retinal neuron cell death resulting from ischemic disease. The eye is a relatively isolated anatomic compartment that can present difficulties for drug delivery. This in large part is because the blood-retinal barrier (BRB) limits access of certain systemically administered agents. Bypassing the BRB with repetitive intraocular injections of a therapeutic agent carries multiple risks, including intraocular infection, retinal detachment, vitreous hemorrhage, and cataract formation. Intraocular gene transfer of neurotrophic agents, mediated by viral vectors, provides a novel way to achieve sustained local delivery to the eye. The eye’s small size and relative anatomic isolation are advantageous in the setting of gene transfer. A relatively small amount of viral vector can be used, efficient transfection of predominantly ocular cells can occur, relatively little vector is exposed to the systemic circulation, and a single intraocular injection replaces multiple injections. 

We have described the transduction that follows intraocular injection of AdPEDF.11 ¹⁴ and have presented reporter gene expression data using a vector with the same viral backbone. ¹⁸ After intravitreous injection, reporter gene expression was predominantly in the iris, cornea, and ciliary body, with sporadic transduction of retinal cells. ¹⁴ Intravitreous injection of AdPEDF.11 resulted in increased production of pedf mRNA, measured by RT-PCR, and increased immunohistochemical staining for PEDF protein in both the anterior segment and the retina. ¹⁸ In the current study, after ischemia-reperfusion and intravitreous administration of AdPEDF.11, we measured significant PEDF levels in the retina that were not present in untreated or ischemic control retinas. 

The marked injury response in AdNull.11-treated eyes compared with otherwise identically handled AdPEDF.11-treated eyes in this rat model of ischemia-reperfusion insult, strongly supports the hypothesis that adenoviral vector-mediated expression of PEDF significantly increases retinal cell survival after ischemia-reperfusion insult. We have provided evidence that the protective effect in the neural retina is at least in part dependent on inhibition of ischemia-induced apoptotic processes. These findings serve to provide proof of concept of a gene transfer approach to modifying the course of programmed cell death in the setting of retinal injury mediated by oxidative stress. The extent to which gene transfer and PEDF will be useful in altering the course of apoptosis induced by injury caused by other than oxidative stress will be the subject of future study. 

The mechanism of neuroprotection by PEDF in cerebellar granule neurons is believed to be mediated in part by the activation of NFκB. ²² The role of NFκB in the regulation of neuronal survival and neuronal death is currently under intense investigation. NFκB plays a critical role in neuronal cell rescue in several models, including glutamate toxicity, low K⁺-induced apoptosis, ischemia-reperfusion-induced apoptosis, β-amyloid peptide-induced toxicity, optic nerve transection, IκB kinase-deficient mice, oxidative stress, and death of developing peripheral neurons. ²² That PEDF may affect pathways involving NFκB and that NFκB is broadly implicated in neuronal survival pathways implies that there is much left to discover about potential therapeutic roles of PEDF in both the central and peripheral nervous systems. Expression of PEDF through adenoviral gene transfer techniques is an exciting way to investigate proof-of-concept questions relating to this protein. 

The major disadvantages of adenoviral-mediated gene transfer include vector-related cytotoxicity and a decline in transgene expression to low levels over the course of weeks. It is not yet known whether repeated intraocular injection of adenoviral vectors can be achieved. Prolonged transgene expression and little evidence of cytotoxicity have been demonstrated with intraocular delivery of adeno-associated virus vectors. ^{23 24} A recent report suggests that adenoviral vector-related toxicity may not be as much of a problem as was once thought (Rasmussen HS, et al. IOVS 2002;43:ARVO E-Abstract 1289). Currently, a phase I clinical trial of the same AdPEDF.11 construct used in this study is under way in patients with choroidal neovascularization. ¹⁶ This study should give preliminary insight into the relative ocular toxicity of this vector in the human eye. The goal of sustained nontoxic transgene expression using adenoviral vectors is one that will generate a great deal of interest in the future. This study presents proof of concept of yet another therapeutic role for PEDF, that of retinal neuroprotection from oxidative stress, and provides evidence of a potential affect on the broader area of apoptosis-mediated cell death. 

 Supported in part by a Grant-in-Aid for Scientific Research 11771070 from the Ministry of Education, Culture and Science in Japan (KM).

 Submitted for publication January 16, 2003; revised April 5, April 22, and April 27, 2003; accepted May 11, 2003.

 Disclosure: H. Takita, None; S. Yoneya, None; P.L. Gehlbach, None; E.J. Duh, None; L.L. Wei, GenVec (E), K. Mori, GenVec (F)

 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.

 

Intraocular levels of human PEDF were significantly elevated after intravitreous injection of AdPEDF.11. Rats received no treatment, ischemia-reperfusion alone or ischemia-reperfusion with intravitreous injection of 3 × 10⁹ viral particles of either AdNull.11 or AdPEDF.11. PEDF levels in each group were 0.015 ± 0.002, 0.020 ± 0.002, 0.037 ± 0.001, 44.7 ± 15.1 pg/μg of total retinal protein (mean ± SE), respectively. PEDF levels in retina after AdPEDF.11 injection were significantly elevated over levels measured in untreated and ischemic control eyes (P = 0.04).

Figure 2.

 

Hematoxylin and eosin staining of the ischemic rat retina without viral vector administration. Retinal ischemia was induced by increasing intraocular pressure above systolic pressure. Immediately after reperfusion (A); 1 day (B); 7 days (C); 14 days (D); 28 days (E). Time-course of change in the nuclear cell count in the GCL (F), in INL (G), ONL (H). Note that retinal thinning and reduction of nuclei counts in the GCL, INL, and ONL plateaued after day 7. Bar, 20 μm.

Figure 3.

 

Hematoxylin and eosin staining of the ischemic retina treated by intravitreous injection of AdNull.11 (A) and AdPEDF.11 (B). Note the retina with intravitreous injection of AdPEDF.11 is well preserved. Inset, arrowhead: pyknotic photoreceptor nucleus. Bar, 20 μm.

Figure 4.

 

Quantitative analysis of nuclear cell count in the GCL (A), INL (B), and ONL (C) of ischemic rat retina stained with hematoxylin and eosin. Four days before induction of pressure-induced ischemia, rats received intravitreous injection of AdPEDF.11 in one eye and AdNull.11 in the contralateral eye. Each data point indicates an individual eye and is connected to the data point for the contralateral the eye by a line. Eyes treated with AdPEDF.11 showed significantly higher nuclear counts in the GCL (P = 0.014), INL (P = 0.008), and ONL (P = 0.008) when compared with eyes treated with AdNull.11.

Figure 5.

 

In situ labeling of ischemic rat retina without viral vector administration by the TUNEL methods. Immediately after reperfusion (A), 6 hours (B), 12 hours (C), 24 hours (D), and 72 hours (E). Double staining with propidium iodide (PI, red) and TUNEL (green). Higher magnification of inset in (C) with PI staining (F), TUNEL staining (G), and PI+TUNEL staining (H). Note that TUNEL staining reached a peak from 12 to 24 hours. Bar, (A–E) 20 μm; and (F–H) 5 μm.

Figure 6.

 

TUNEL staining of ischemic rat retina treated with intravitreous injection of AdNull.11 (A–C) and AdPEDF.11 (D–F). PI staining (A, D), TUNEL staining (B, E), and double staining with PI+TUNEL (C, F). The number of pyknotic (A, arrowheads) TUNEL-positive cells were reduced in eyes treated with AdPEDF.11. Bar, 20 μm.

Figure 7.

 

Quantitative analysis of TUNEL-positive cell count in the GCL (A), INL (B), and ONL (C) per ischemic rat retina. Four days before induction of pressure-induced ischemia, rats received intravitreous injection of AdPEDF.11 in one eye and AdNull.11 in the contralateral eye. Each data point represents an individual eye and is connected to the data point for the contralateral eye by a line. Eyes treated with AdPEDF.11 showed significantly lower TUNEL-positive cell counts in the GCL (P = 0.007), INL (P = 0.026), and ONL (P = 0.002) when compared with eyes treated with AdNull.11.

The authors are thankful to Manesh Dagli for technical assistance.