Neovascularization plays a major role in several disease

Transcription

1 Regression of Ocular Neovascularization in Response to Increased Expression of Pigment Epithelium Derived Factor Keisuke Mori, 1 Peter Gehlbach, 1 Akira Ando, 1 Duncan McVey, 2 Lisa Wei, 2 and Peter A. Campochiaro 1 PURPOSE. Several pharmacologic treatments have been shown to reduce ocular neovascularization when administered before the onset of angiogenic stimuli, but none have been shown to cause regression of already established ocular neovascularization. In this study, the authors tested the effect of adenoviral vectored pigment epithelium-derived factor (PEDF) gene transfer on established neovascularization in transgenic mice with expression of vascular endothelial growth factor (VEGF) in photoreceptors (rho/vegf mice) and in a model of choroidal neovascularization. METHODS. Two weeks after the onset of VEGF transgene expression in rho/vegf mice or 2 weeks after laser-induced rupture of Bruch s membrane in wild-type mice, subgroups of mice were killed, and the baseline amount of neovascularization was measured by image analysis. The remainder of the mice received an intravitreous or subretinal injection of adenoviral vector containing a PEDF expression construct (AdPEDF.11) or control vector (AdNull.11). RESULTS. Seven days after injection in rho/vegf mice or 10 days after injection in the choroidal neovascularization model, the amount of neovascularization in AdPEDF.11-injected eyes was significantly less than the baseline level, indicating that regression of neovascularization had occurred. There was TUNEL staining within choroidal neovascular lesions in eyes injected with AdPEDF.11. Eyes given a subretinal injection of AdNull.11 had TUNEL-positive cells in the retina, but none within areas of choroidal neovascularization. CONCLUSIONS. These data indicate that increased expression of PEDF causes regression of ocular neovascularization by promoting apoptosis of cells within neovascular lesions and possibly represents a new treatment paradigm for patients with established ocular neovascularization. (Invest Ophthalmol Vis Sci. 2002;43: ) From the 1 Departments of Ophthalmology and Neuroscience, The Johns Hopkins University School of Medicine, Baltimore, Maryland; and 2 GenVec Inc., Gaithersburg, Maryland. Supported by grants from the Michael Panitch Fund and Foundation Fighting Blindness; Grants EY05951, EY12609, K08 EY13420 (PG) and P30EY1765 from the National Eye Institute; the Steinbach Foundation, the Juvenile Diabetes Research Foundation (PG), Knights Templar (PG), a Lew R. Wasserman Merit Award (PAC) and unrestricted funds from Research to Prevent Blindness. PAC is the George S. and Dolores Dore Eccles Professor of Ophthalmology and Neuroscience, The Johns Hopkins University School of Medicine. Submitted for publication October 26, 2001; revised February 21, 2002; accepted March 12, Commercial relationships policy: E (DM, LW); C, R (PAC); N (all others). 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 solely to indicate this fact. Corresponding author: Peter A. Campochiaro, The Johns Hopkins University School of Medicine, Maumenee 719, 600 N. Wolfe Street, Baltimore, MD ; pcampo@jhmi.edu. Neovascularization plays a major role in several disease processes including tumors, arthritis, atherosclerosis, and ocular neovascularization. Because large numbers of patients with many different underlying disease processes could benefit from antiangiogenic therapies, there is great interest in their development. Neovascularization in different vascular beds throughout the body has some common features, but there are also some unique characteristics conferred by the microenvironment. Similarly, tumor angiogenesis probably shares some characteristics with the angiogenesis that occurs in other disease processes, but there are also likely to be some differences. Even neovascular processes in different vascular beds of the same organ show overlapping, but not identical, characteristics. 1 Understanding both similarities and differences can provide important insights into how neovascular disease processes can be modulated. The structure of the eye in which the avascular outer retina and subretinal space are sandwiched between two very different vascular beds, the choroidal vasculature and the retinal vasculature, provides an ideal setting for studying and quantifying vascular invasion. There are good animal models for both retinal and choroidal neovascularization (CNV), and they have been used to provide important baseline knowledge on which to build. The transgenic mouse, in which the rhodopsin promoter drives expression of VEGF in photoreceptors (rho/vegf mice), provides a well-characterized model in which neovascularization sprouts from the deep capillary bed of the retina and invades the photoreceptor layer and subretinal space. 2,3 CNV originates from the choroidal vessels and invades the subretinal space from the other direction by penetrating Bruch s membrane and the retinal pigmented epithelium. Laser-induced rupture of Bruch s membrane provides a reliable model of CNV that is easily quantified. 4 Thus, retinal and CNV are interesting to all vascular biologists, because they provide important benchmarks for comparison with other types of neovascularization. In addition, they are particularly interesting to ophthalmologists, because they account for most severe visual loss in developed countries. 5,6 One goal of antiangiogenic treatments is to prevent the development of pathologic neovascularization in patients who are at high risk. In animal models of retinal or CNV, treatments have been identified that decrease the amount of neovascularization when the treatments are given before the onset of the inciting stimuli Another goal, which is much more formidable and has not yet been achieved, is to identify treatments that cause regression of already established neovascularization. Recently, it has been demonstrated that pigment epitheliumderived factor (PEDF) is a potent antiangiogenic agent 16 that reduces the development of retinal or choroidal neovascularization when administered before the onset of angiogenic stimuli, either by systemic injection 13 or by intraocular gene transfer. 14 In this study, we sought to determine whether increased expression of PEDF causes regression of established neovascularization. Investigative Ophthalmology & Visual Science, July 2002, Vol. 43, No Copyright Association for Research in Vision and Ophthalmology

2 IOVS, July 2002, Vol. 43, No. 7 Effect of PEDF on Neovascularization 2429 MATERIALS AND METHODS Adenoviral Vectors Expressing PEDF Adenoviral vectors expressing PEDF from a cytomegalovirus (CMV) immediate early promoter expression cassette have been previously described and characterized. 14 They are deleted for the E1A, E1B, E3, and E4 regions of the viral genome (AdPEDF.11). Vectors that do not express transgene products, used as the vector-infected control in these studies, are designated as Null and were constructed and produced in the same manner (AdNull.11). Mouse Model of Laser-Induced CNV Procedures involving mice were conducted in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Adult C57BL/6 mice were anesthetized with ketamine hydrochloride (100 mg/kg body weight), pupils were dilated with 1% tropicamide, and diode laser photocoagulation was used to rupture Bruch s membrane at four locations in each eye of each mouse, as previously described. 4 Briefly, laser photocoagulation (532-nm wavelength, 100- m spot size, 0.1-second duration, and 120-mW intensity) was delivered with the slit lamp delivery system and a handheld cover slide as a contact lens. Burns were performed in the 9, 12, 3, and 6 o clock positions 2 to 3 disc diameters from the optic nerve. Production of a vaporization bubble at the time of laser, which indicates rupture of Bruch s membrane, is an important factor in obtaining CNV, 4 and therefore only burns in which a bubble was produced were included in the study. Intraocular Injection of Vectors Two weeks after rupture of Bruch s membrane, 10 mice were killed, and the baseline amount of CNV was measured at each rupture site, as described later. The remainder of the mice received an intravitreous injection of 10 9 particles or a subretinal injection of 10 8 particles of AdNull.11 or AdPEDF.11, respectively, in each eye. Intravitreous injections were performed with a Harvard pump microinjection apparatus and pulled glass micropipets, as previously described. 14 Each micropipet was calibrated to deliver 1 L of vehicle containing 10 9 or 10 8 particles on depression of a foot switch. The mice were anesthetized, pupils were dilated, and under a dissecting microscope, the sharpened tip of the micropipet was passed through the sclera just behind the limbus into the vitreous cavity, and the foot switch was depressed. Subretinal injections were performed using a contact lens, which allowed visualization of the retina during the injection. The pipet tip was passed through the sclera posterior to the limbus and was positioned just above the retina. Depression of the foot switch caused the jet of injection fluid to penetrate the retina resulting in fairly uniform blebs that confirmed that the vector had been deposited in the subretinal space. Measurement of the Sizes of Laser-Induced CNV Lesions Ten days after vector injection, the area of CNV at each rupture site was measured in choroidal flatmounts. 17 Mice were anesthetized and perfused with 1 ml phosphate-buffered saline containing 50 mg/ml fluorescein-labeled dextran ( average molecular weight; Sigma, St. Louis, MO), as previously described. 3 The eyes were removed and fixed for 1 hour in 10% phosphate-buffered formalin. The cornea and lens were removed, and the entire retina was carefully dissected from the eyecup. Radial cuts (four to seven, average five) were made from the edge to the equator and the eyecup was flatmounted in aqueous mounting medium (Aquamount; BDH, Poole, UK) with the sclera facing down. Flatmounts were examined by fluorescence microscopy (Axioskop; Carl Zeiss, Thornwood, NY), and images were digitized using a three-color charge-coupled device (CCD) video camera (IK- TU40A; Toshiba, Tokyo, Japan) and a frame grabber. Image-analysis software (Image-Pro Plus; Media Cybernetics, Silver Spring, MD) was used to measure the total area of hyperfluorescence associated with each burn, corresponding to the total fibrovascular scar. The areas within each eye were averaged to give one experimental value, and mean values were calculated for each treatment group and compared by Student s unpaired t-test. Transgenic Mice with Increased Expression of VEGF in Photoreceptors Transgenic mice with VEGF expression driven by the rhodopsin promoter develop subretinal neovascularization, due to expression of VEGF in photoreceptors beginning at approximately postnatal day (P)6. 2 At P21, a cohort of transgene-positive mice were killed, and the amount of neovascularization in each retina was measured, as described later. The remainder of the mice received no injection or an intravitreous injection of 10 9 particles of AdNull.11 or AdPEDF.11. At P28, the mice were killed, and the amount of subretinal neovascularization was quantified as previously described. 3 Briefly, mice were anesthetized and perfused with 1 ml phosphate-buffered saline containing 50 mg/ml fluorescein-labeled dextran. The eyes were removed and fixed for 1 hour in 10% phosphate-buffered formalin. The cornea and lens were removed, and the entire retina was carefully dissected from the eyecup, radially cut from the edge of the retina to the equator in all four quadrants, and flatmounted in aqueous medium with photoreceptors facing upward. The retinas were examined by fluorescence microscopy at 200 magnification, which provides a narrow depth of field, so that when the focus is on neovascularization on the outer surface of the retina, the remainder of the retinal vessels are out of focus allowing easy delineation of the neovascularization. The outer edge of the retina, which corresponds to the subretinal space in vivo, is easily identified, and therefore the focal plane was standardized from slide to slide. Images were digitized using a three-color CCD video camera and a frame grabber. Image-analysis software (Image-Pro, Media Cybernetics, Inc.) was used to delineate each of the lesions and calculate the number in each retina, the area of each lesion, and the total area of neovascularization per retina, as previously described. 3 Identification of Apoptotic Cells by TUNEL Apoptotic cells were detected by TdT-dUTP terminal nick end-labeling (TUNEL). Ten days after the intraocular injection of vector and 24 days after rupture of Bruch s membrane, eyes were rapidly removed and frozen in optimum cutting temperature embedding compound (OCT; Miles Diagnostics, Elkhart, IN). Frozen serial sections (10 m) were cut through the entire extent of each burn. Sections were fixed in 1% paraformaldehyde for 10 minutes at room temperature and stained by a kit (ApopTag Fluorescein Red; Intergen, Purchase, NY) used according to the manufacturer s instructions with minor modifications. The sections were also histochemically stained with biotinylated Griffonia simplicifolia lectin B4 (GSA, Vector Laboratories, Burlingame, CA) which selectively binds vascular cells. After TUNEL staining, slides were incubated for 30 minutes in 10% normal porcine serum in 0.05% Tris-buffered saline (TBS; ph 7.6). Slides were incubated for 2 hours at room temperature with fluorescein-labeled GSA. After rinsing with 0.05 M TBS, slides were mounted with aqueous medium. RESULTS Effect of Adenoviral Vector-Mediated Expression of PEDF in Rho/VEGF Mice In rho/vegf transgenic mice, expression of VEGF in photoreceptors begins at approximately P6, increases to a steady state level by about P14, and continues at that level throughout adulthood. 2 The total area of subretinal neovascularization per retina steadily increases between P18 and P45, the oldest age examined. 3 In agreement with this previous study, 3 at P21 rho/vegf mice had extensive neovascularization (Fig. 1A) that appeared even greater at P28 (Fig. 1B). Transgenic mice that

3 2430 Mori et al. IOVS, July 2002, Vol. 43, No. 7 FIGURE 1. Intraocular injection of AdPEDF.11 caused regression of ocular neovascularization. At P21, a cohort of rho/vegf transgenic mice were perfused with fluorescein-labeled dextran and killed, and the amount of subretinal neovascularization was assessed by examination of retinal flatmounts by fluorescence microscopy. Focusing on the outer border of the retina allowed delineation of numerous areas of hyperfluorescence partially surrounded by RPE cells (A, arrows). Also at P21, the remainder of the transgenic mice were divided into three groups and were either given no injection (B) or an intravitreous injection of 10 9 particles of AdNull.11 (C) or AdPEDF.11 (D). One week later at P28, the mice were perfused with fluorescein-labeled dextran, and subretinal neovascularization was assessed on retinal flatmounts. Mice that had not received any injection showed numerous areas of hyperfluorescence associated with pigmented cells (B, arrows). Mice that had been given an intravitreous injection of AdNull.11 also showed several areas of hyperfluorescence (C), although it appeared that the number was somewhat fewer than that in uninjected P28 mice. Mice that had been given an intravitreous injection of AdPEDF.11 showed an occasional small hyperfluorescent spot (D, arrow), but many fewer than in (A C). There were pigmented cells, suggesting that the neovascularization had regressed, leaving pigmented cells behind. Adult C57BL/6 mice had laser-induced rupture of Bruch s membrane at three locations in each eye. Fourteen days after laser treatment, a cohort of the mice were perfused with fluorescein-labeled dextran and the area of CNV was measured by image analysis of choroidal wholemounts (E). The remainder of the mice were divided into seven treatment groups (F L), and 10 days after treatment, they were perfused with fluorescein-labeled dextran, and the area of CNV was measured by image analysis of choroidal wholemounts. Fourteen days after rupture of Bruch s membrane, there were large areas of CNV at rupture sites (E). Twenty-four days after rupture of Bruch s membrane, mice that received no treatment (F), an intravitreous injection of 10 9 particles of AdNull.11 (G), or a subretinal injection of 10 8 particles of AdNull.11 (H, I), all showed areas of CNV that appeared similar to or somewhat larger than the baseline level at 14 days after laser (E). In mice that received subretinal injections of AdNull.11, the size of CNV lesions appeared similar, regardless of whether the rupture site was inside the bleb caused by the subretinal injection (H) and therefore had direct exposure to the vector, or the rupture site was outside the bleb (I). Mice that received an intravitreous injection of 10 9 particles of AdPEDF.11 (J) or a subretinal injection of 10 8 particles of AdPEDF.11 (K, L) showed less CNV than the baseline level at 14 days after laser treatment (E) or in the AdNull.11-treated eyes 24 days after laser treatment (G I). Note the irregular borders of the CNV 10 days after intravitreous injection of AdPEDF.11 (J) with a ring of surrounding pigmented cells (arrowheads) that may indicate a region occupied by neovascularization before regression. In mice that received subretinal injections of AdPEDF.11, the size of CNV lesions appeared similar, regardless of whether the rupture site was inside (K) or outside (J), but the former tended to less hyperfluorescence, with most of it limited to large vascular structures (arrows), whereas the latter showed morphology more like that after intravitreous injections (J) with a ring of surrounding pigmented cells (L, arrowheads). Bar, 100 m. received an intravitreous injection of 10 9 particles of AdNull.11 on P21 had extensive neovascularization at P28, but it appeared to be somewhat less than that in uninjected P28 mice (Figs. 1C versus 1B). Transgenic mice that received an intravitreous injection of 10 9 particles of AdPEDF.11 on P21, had much less neovascularization than uninjected P28 mice (Figs. 1D versus 1B). There were RPE cells present, but almost no hyperfluorescence was associated with the RPE cells (1D, arrow), which suggests that neovascularization had been present and regressed, leaving RPE cells without hyperfluorescence in their wake. This interpretation was supported by quantitative analysis (Fig. 2). The total area of neovascularization per retina was significantly greater at P28 than at P21 in uninjected mice. Mice that received an intravitreous injection of 10 9 particles of AdNull.11 on P21 and were then killed at P28 showed less neovascularization than P28 uninjected mice, but no significant difference from P21 uninjected mice. This indicates that the vector itself has some antiangiogenic activity in the transgenic model, as was shown in a previous study. 14 However, mice that received an intravitreous injection of 10 9 particles of AdPEDF.11 on P21, had significantly less neovascularization at P28 than P28 AdNull.11-injected mice or P21 uninjected mice. This confirms that intraocular expression of PEDF caused regression of subretinal neovascularization in rho/vegf transgenic mice. Regression of CNV Fourteen or 24 days after laser treatment in adult C57BL/6 mice, there was extensive CNV at sites of rupture of Bruch s membrane (Figs. 1E and 1F, respectively). Mice that received an intravitreous injection of 10 9 particles of AdNull days after laser treatment and were killed 10 days later had large areas of CNV at sites of rupture of Bruch s membrane (Fig. 1G) that looked very similar to those in uninjected mice. The same was true of mice that received a subretinal injection of 10 8

4 IOVS, July 2002, Vol. 43, No. 7 Effect of PEDF on Neovascularization 2431 large vessels (Fig. 1K, arrows) and no small vessels. A possible explanation is that the small vessels regressed, leaving only a few larger vessels. This lesion was located within the subretinal bleb caused by the subretinal injection of AdPEDF.11. The lesion shown in Figure 1L is also from a mouse that was given a subretinal injection of 10 8 particles of AdPEDF days after laser treatment, but the rupture site was outside the region of the bleb. Compared with lesions in Ad.Null-injected or uninjected eyes, the lesion was smaller and had a morphology consisting of a small region of hyperfluorescence surrounded by hyperpigmentation, similar to that of lesions in many eyes that received an intravitreous injection of AdPEDF.11 (Fig. 1J). Measurement of the areas of CNV at rupture sites showed that eyes that received either an intravitreous or subretinal injection of AdPEDF.11 on day 14 after laser treatment, had significantly less neovascularization on day 24 than eyes injected with AdNull.11 or uninjected eyes (Fig. 3). Eyes that received a subretinal injection of AdPEDF.11 also had significantly less neovascularization than the baseline amount at day 14, suggesting that regression of CNV had occurred. Within each of the groups, there was little variability in the amount of CNV, in agreement with observations in previous studies using this model. 11,14,15,18 FIGURE 2. Image analysis demonstrated decreased neovascularization in rho/vegf transgenic mice after expression of PEDF. Several litters of transgene-positive rho/vegf mice were divided into four groups: group 1 mice (no injection, P21) were killed on P21 (n 14), group 2 mice (no injection, P28) were killed on P28 (n 16), group 3 mice received an intravitreous injection of 10 9 particles of empty vector (AdNull.11) on P21 and were killed on P28 (n 18), and group 4 mice received an intravitreous injection of 10 9 particles of vector containing a PEDF expression construct (AdPEDF.11) on P21 and were killed on P28 (n 15). The mice were killed by perfusion with fluoresceinlabeled dextran, retinal flatmounts were prepared, and the total area of perfused new blood vessels at the outer border of the retina was measured using fluorescence microscopy and computerized image analysis. Group 4 had significantly less neovascularization than the baseline level in group 1 (*P 0.05 by unpaired t-test for groups with similar variance) demonstrating PEDF-induced regression of neovascularization. Group 4 had significantly less neovascularization than group 3( P 0.05 by unpaired t-test for groups with unequal variance), and group 3 was not significantly different from group 1, indicating that the regression of neovascularization was due to expression of PEDF and not to some other effect of the vector. particles of AdNull days after laser treatment and were killed 10 days later. This was true regardless of whether the rupture site was outside the region of retina that was detached by the subretinal injection (Fig. 1H) or within the area of the bleb (Fig. 1I). In contrast, mice that received an intravitreous injection of 10 9 particles of AdPEDF days after laser treatment and were killed 10 days later had smaller areas of CNV (Fig. 1J) than AdNull.11-injected or uninjected mice. Note that the morphology of the hyperfluorescent area was unusual, in that the borders were irregular and there was prominent surrounding hyperpigmentation (1J, arrowheads). The appearance was consistent with regression of CNV, leaving hyperpigmentation in the region in which involution of neovascularization occurred. Hyperpigmentation surrounding an irregular area of hyperfluorescence was not the only morphology that suggested regression in treated mice. Another one is illustrated in Figure 1K which shows a lesion 24 days after rupture of Bruch s membrane in a mouse that received a subretinal injection of 10 8 particles of AdPEDF days after laser treatment. The area of hyperfluorescence is smaller than those in AdNull.11- injected or uninjected mice, and it consists of a few relatively Apoptosis in Cells within CNV Lesions Fourteen days after laser-induced rupture of Bruch s membrane, mice received no injection, an intravitreous injection of 10 9 particles of AdNull.11 or AdPEDF.11, a subretinal injection of vehicle alone, or 10 8 particles of AdNull.11 or AdPEDF.11. Ten days later, the mice were killed, ocular sections were labeled with GSA, and TUNEL staining was performed on adjacent sections or sections were double labeled with GSA and TUNEL. Uninjected eyes showed a few TUNEL-stained cells in the retina overlying laser-induced CNV, but none within the CNV (not shown). The same was true of eyes that received a subretinal injection of vehicle alone (Figs. 4A, 4D, 4G) or an intravitreous injection of AdNull.11 (not shown). In contrast, TUNEL staining in eyes that received an intravitreous injection of AdPEDF.11 showed a few labeled cells within CNV lesions (not shown) and eyes that received a subretinal injection of AdPEDF.11 showed numerous labeled cells in CNV lesions that also stained with GSA, indicating that they were dying vascular cells (Figs. 4C, 4F, 4I). Prominent TUNEL staining of photoreceptor cells was also present in regions of retina that had been detached by subretinal injection of AdPEDF.11. Eyes that had received subretinal injection of AdNull.11 also showed prominent TUNEL labeling of photoreceptors in areas of retina that had been detached by the injection, but no staining of cells in CNV lesions (Figs. 4B, 4E, 4H). This suggests that the death of retinal neurons is attributable to the adenoviral vector, and apoptosis of vascular cells in CNV lesions is due to PEDF. DISCUSSION Antiangiogenic treatments could be useful for management of several types of disease. This is particularly true of eye diseases, because neovascularization is the underlying problem that is responsible for most cases of new blindness in developed countries. 6,19 Recently, several agents have been demonstrated to decrease the amount of neovascularization that occurs in various models of ocular neovascularization, when the agents are given before the onset of angiogenic stimuli Such treatments, when given prophylactically, might reduce the incidence of neovascularization in high-risk patients, but they may be less helpful in patients who already have established ocular neovascularization.

5 2432 Mori et al. IOVS, July 2002, Vol. 43, No. 7 FIGURE 3. Image analysis demonstrated decreased CNV at sites of Bruch s membrane rupture after expression of PEDF. Bruch s membrane was ruptured at three sites in each eye of adult C57BL/6 mice, which were divided into six groups: group 1 mice (no injection, day 14) were killed after 14 days (n 10 mice, 70 CNV lesions), group 2 mice (no injection, day 24) were killed after 24 days (n 10 mice, 79 CNV lesions), group 3 mice received an intravitreous injection of 10 9 particles of empty vector (AdNull.11) on day 14 and were killed on day 24 (n 8 mice, 41 CNV lesions), group 4 mice received a subretinal injection of 10 8 particles of AdNull.11 on day 14 and were killed on day 24 (they were subdivided into two groups: CNV lesions that were within the region of the bleb caused by the subretinal injection [AdNull.11/subretinal/in, n 17] and CNV lesions that were outside the region of the bleb [AdNull.11/subretinal/out, n 17]), group 5 mice received an intravitreous injection of 10 9 particles of vector containing a PEDF expression vector (AdPEDF.11) on day 14 and were killed on day 24 (n 10 mice, 41 CNV lesions), and group 6 mice received a subretinal injection of 10 8 particles of AdPEDF.11 on day 14 and were killed on day 24 (they were subdivided into two groups: CNV lesions that were within the region of the bleb caused by the subretinal injection [AdPEDF.11/subretinal/in, n 33] and CNV lesions that were outside the region of the bleb [AdPEDF.11/subretinal/out, n 34]). The mice were killed by perfusion with fluorescein-labeled dextran, choroidal flatmounts were prepared, and the total area of CNV was measured using fluorescence microscopy and computerized image analysis. Both subgroups of group 6 mice that received subretinal injection of AdPEDF.11 had significantly less neovascularization than the baseline level in group 1 (*P 0.05 by unpaired t-test for groups with similar variance) demonstrating PEDF-induced regression of CNV. Group 5 also had significantly less CNV than group 3 ( P 0.05 by unpaired t-test for groups with equal variance), and group 3 had significantly more CNV than group 1 (P 0.04), indicating that the regression of neovascularization was due to expression of PEDF and not to some other effect of the vector. We have previously shown that intraocular injection of a PEDF expression construct packaged in adenoviral vectors results in increased expression of PEDF in the eye and reduces the amount of ocular neovascularization that occurs in three different models. 14 In this study, we demonstrated that in two of the models, rho/vegf transgenic mice and CNV due to rupture of Bruch s membrane, PEDF gene transfer in eyes with already-established neovascularization caused regression of the neovascularization. Regression studies are not feasible in the model of oxygen-induced ischemic retinopathy, because spontaneous regression of the neovascularization occurs. This is the first demonstration of a pharmacologic treatment that causes regression of ocular neovascularization and could be applied in the many patients who have established ocular neovascularization. Eyes injected with AdPEDF.11 showed prominent TUNEL staining in vascular cells within the CNV lesion. The staining was much greater in eyes that had received a subretinal injection versus an intravitreous injection, and subretinal injections resulted in greater involution of the neovascularization. This suggests that PEDF promotes programmed cell death in cells that make up new vessels. TUNEL staining was present in some retinal neurons after subretinal injection of AdPEDF.11, but it was equally present in eyes that received a subretinal injection of AdNull.11 and was absent in eyes that received a subretinal injection of vehicle. TUNEL staining was limited to areas of retina that had been detached by the subretinal injection of vector. Therefore, increased apoptosis of retinal neurons does not seem to be an effect of PEDF, but rather a consequence of adenoviral vectors injected in the subretinal space. As noted in a previous study, 14 adenoviral vectors themselves had some preventive effects on neovascularization in VEGF transgenic mice, but not on CNV. Unlike injection of AdPEDF.11, injection of AdNull.11 did not cause regression of neovascularization in either model, nor did it result in apoptosis in cells within the neovascular lesions. Therefore, it is likely that the mechanism of the null vector effect is different from that of PEDF. Recently, Stellmach et al. 13 demonstrated that systemic injection of recombinant PEDF prevents the development of retinal neovascularization in mice with ischemic retinopathy. Eyes of mice treated with PEDF showed increased TUNEL staining of retinal vascular cells within the retina and in the presence of PEDF, cultured endothelial cells also showed increased TUNEL staining. These data suggest that PEDF promotes apoptosis in endothelial cells, and the authors surmised that endothelial cells participating in neovascularization are differentially sensitive, although no proof of that contention was provided. Our data support the hypothesis of Stellmach et al., because as illustrated in Figure 4I, there are many TUNELstained cells within the CNV, and the feeder vessels from the underlying choroid, but no TUNEL staining of the retinal vascular cells that are not involved in the neovascularization. Also, choroidal vascular cells in areas not underlying CNV showed no TUNEL staining. Therefore, promotion of apoptosis in endothelial cells in new vessel sprouts may underlie the ability of PEDF to prevent neovascularization, and promotion of apoptosis in established new vessels may underlie the ability of PEDF to cause regression. It is important to determine why normal quiescent vessels are resistant to PEDF-induced cell death and whether long-standing neovascularization acquires resistance. This study provides the first demonstration of a drug or protein that causes regression of ocular neovascularization. Scatter photocoagulation causes regression of retinal neovascularization without directly destroying it, 20 but the mechanism by which it induces regression is unknown. Scatter photocoagulation decreases oxygen demand and increases oxygen supply to ischemic inner retina, thereby decreasing expression of VEGF 21 ; but, is decreased expression of VEGF sufficient to cause regression of established retinal neovascularization? VEGF acts as a survival factor for newly formed vessels, but established vessels are less dependent on VEGF. 22 Therefore, there is reason to hypothesize that scatter photocoagulation may have an effect in addition to decreasing expression of VEGF that contributes to regression of retinal neovascularization. Because hyperplasia of RPE cells occurs in laser scars and PEDF is normally produced by RPE cells, 23,24 it is possible that scatter photocoagulation results in higher levels of PEDF in the eye that, combined with decrease expression of VEGF, leads to

6 IOVS, July 2002, Vol. 43, No. 7 Effect of PEDF on Neovascularization 2433 FIGURE 4. Injection of AdPEDF.11 resulted in apoptosis of cells within CNV. Bruch s membrane was ruptured at three sites in each eye of adult C57BL/6 mice, and after 14 days, the mice were given a subretinal injection of vehicle (A, D, G), 10 8 particles of AdNull.11 (B, E, H), or 10 8 particles of AdPEDF.11 (C, F, I). Five days after injection, adjacent ocular frozen sections were stained with fluorescein-labeled GSA, which selectively binds vascular cells (A C; arrowheads: superior border of the CNV), TUNEL (D F), or both GSA and TUNEL (G I). Eyes injected with vehicle showed rare TUNEL-positive cells in the retina, but none in CNV (D, G). Eyes injected with AdNull.11 showed many TUNEL-stained cells in the outer nuclear layer in regions where the bleb from the subretinal injection had been located, but no TUNEL staining in the CNV (E, H). Eyes injected with PEDF.11 showed many TUNEL-stained cells in the outer nuclear layer in regions where the bleb from the subretinal injection had been located, and also showed prominent TUNEL-staining in CNV and choroidal vessels feeding the CNV (F, I). High magnification (I, inset) showed crescent or donut-shaped TUNEL-stained chromatin typical of cells undergoing programmed cell death. Bar: 50 m; (I, inset) 25 m. regression of retinal neovascularization. Additional studies are needed to test this hypothesis. Unlike retinal neovascularization, for which scatter photocoagulation is a fairly effective treatment that often leads to long-term stability, current treatments for CNV are marginal and usually palliative. When CNV is well-localized and does not involve the center of the macula, it may be treated with confluent laser photocoagulation which destroys the new blood vessels and the overlying retina, resulting in a paracentral scotoma. When successful, this treatment can delay severe vision loss, but recurrences occur in roughly 60% of successfully treated patients over 5 years and almost always lead to severe loss of vision. 6 When CNV involves the center of the macula and has certain fluorescein angiographic characteristics, then photodynamic therapy may result in decreased hyperfluorescence due to damage to new vessels, but recurrences almost always occur, and visual results are disappointing. 25,26 Thus far, no treatment that does not directly ablate new vessels has been demonstrated to decrease CNV. Our data suggest that PEDF gene transfer could be an extremely promising approach for treatment of CNV. It may be possible to develop improved adenoviral vectors that provide long-term expression. In addition, adenoassociated viral (AAV) vectors provide another alternative for long-term expression, and recently we have demonstrated that intraocular injection of an AAV vector containing a PEDF expression construct inhibits CNV. 18 It is conceivable that prolonged increased intraocular expression of PEDF as is anticipated from AAVmediated gene transfer, could result in regression of CNV and prevent its regrowth. In addition, there are mounting data suggesting that PEDF promotes survival of photoreceptor cells. 27 Photoreceptor cell death and CNV both contribute to loss of vision in patients with age-related macular degeneration (AMD), a problem that has reached epidemic proportions in people over the age of The mechanism for the development of neovascular AMD is different from that in the model of laser-induced rupture of Bruch s membrane, and so the effects of PEDF could be different in neovascular AMD. However, it is reasonable to hypothesize that regardless of how CNV arises, PEDF may promote its regression. We look forward to future studies to see whether these promising results in animal models can be translated into new treatments that inhibit both the degenerational and neovascular components of AMD. References 1. Campochiaro PA. Retinal and choroidal neovascularization. J Cell Physiol. 2000;184: Okamoto N, Tobe T, Hackett SF, et al. Transgenic mice with increased expression of vascular endothelial growth factor in the retina: a new model of intraretinal and subretinal neovascularization. Am J Pathol. 1997;151: Tobe T, Okamoto N, Vinores MA, et al. Evolution of neovascularization in mice with overexpression of vascular endothelial growth

7 2434 Mori et al. IOVS, July 2002, Vol. 43, No. 7 factor in photoreceptors. Invest Ophthalmol Vis Sci. 1998;39: Tobe T, Ortega S, Luna L, et al. Targeted disruption of the FGF2 gene does not prevent choroidal neovascularization in a murine model. Am J Pathol. 1998;153: Klein R, Klein B. Vision disorders in diabetes. In: Group NDD, ed. Diabetes in America. 2nd ed. Washington, DC: National Institutes of Health; 1995: The Macular Photocoagulation Study Group. Argon laser photocoagulation for neovascular maculopathy: five year results from randomized clinical trials. Arch Ophthalmol. 1991;109: Aiello LP, Pierce EA, Foley ED, et al. Suppression of retinal neovascularization in vivo by inhibition of vascular endothelial growth factor (VEGF) using soluble VEGF-receptor chimeric proteins. Proc Natl Acad Sci USA. 1995;92: Robinson GS, Pierce EA, Rook SL, Foley E, Webb R, Smith LES. Oligodeoxynucleotides inhibit retinal neovascularization in a murine model of proliferative retinopathy. Proc Natl Acad Sci USA. 1996;93: Adamis AP, Shima DT, Tolentino MJ, et al. Inhibition of vascular endothelial growth factor prevents retinal ischemia-associated iris neovascularization. Arch Ophthalmol. 1996;114: Luna J, Tobe T, Mousa SA, Reilly TM, Campochiaro PA. Antagonists of integrin alpha-v beta-3 inhibit retinal neovascularization in a murine model. Lab Invest. 1996;75: Seo M-S, Kwak N, Ozaki H, et al. Dramatic inhibition of retinal and choroidal neovascularization by oral administration of a kinase inhibitor. Am J Pathol. 1999;154: Ozaki H, Seo M-S, Ozaki K, et al. Blockade of vascular endothelial cell growth factor receptor signaling is sufficient to completely prevent retinal neovascularization. Am J Pathol. 2000;156: Stellmach V, Crawford SE, Zhou W, Bouck N. Prevention of ischemia-induced retinopathy by the natural ocular antiangiogenic agent pigment epithelium-derived factor. Proc Natl Acad Sci USA. 2001;98: Mori K, Duh E, Gehlbach P, et al. Pigment epithelium-derived factor inhibits retinal and choroidal neovascularization. J Cell Physiol. 2001;188: Mori K, Ando A, Gehlbach P, et al. Inhibition of choroidal neovascularization by intravenous injection of adenoviral vectors expressing secretable endostatin. Am J Pathol. 2001;159: Dawson DW, Volpert OV, Gillis P, et al. Pigment epitheliumderived factor: a potent inhibitor of angiogenesis. Science. 1999; 285: Edelman JL, Castro MR. Quantitative image analysis of laser-induced choroidal neovascularization in rat. Exp Eye Res. 2000;71: Mori K, Gehlbach P, Yamamoto S, et al. AAV-mediated gene transfer of pigment epithelium-derived factor inhibits choroidal neovascularization. Invest Ophthalmol Vis Sci. 2002;43: Klein R, Klein BEK, Moss SE, Davis MD, DeMets DL. The Wisconsin Epidemiologic Study of Diabetic Retinopathy. II: prevalence and risk of diabetic retinopathy when age at diagnosis is less than 30 years. Arch Ophthalmol. 1984;102: The Diabetic Retinopathy Study Research Group. Photocoagulation treatment of proliferative diabetic retinopathy: clinical application of Diabetic Retinopathy Study (DRS) findings, DRS Report Number 8. Ophthalmology. 1981;88: Pournaras CJ, Tsacopoulos M, Strommer K, Gilodi N, Leuenberger PM. Scatter photocoagulation restores tissue hypoxia in experimental vasoproliferative microangiopathy in miniature pigs. Ophthalmology. 1990;97: Alon T, Hemo I, Itin A, Pe er J, Stone J, Keshet E. Vascular endothelial growth factor acts as a survival factor for newly formed retinal vessels and has implications for retinopathy of prematurity. Nat Med. 1995;1: Tombran-Tink J, Chader GG, Johnson SV. PEDF: a pigment epithelium derived factor with potent neuronal differentiating activity. Exp Eye Res. 1991;53: Steele FR, Chader GJ, Johnson LV, Tombran-Tink J. Pigment epithelium-derived factor: neurotrophic activity and identification as a member or the serine protease inhibitor gene family. Proc Natl Acad Sci USA. 1993;90: Husain D, Miller JW, Michaud N, Connolly E, Flotte TJ, Gragoudas ES. Effect of intravenous infusion using liposomal BPD verteporfin for photodynamic therapy of experimental choroidal neovascularization. Arch Ophthalmol. 1996;114: Schmidt-Erfurth U, Miller JW, Sickenberg M, et al. Photodynamic therapy with verteporfin for choroidal neovascularization caused by age-related macular degeneration: results of retreatments in a phase 1 and 2 study. Arch Ophthalmol. 1999;117: Jablonski MM, Tombran-Tink J, Mrazek DA, Iannoaccone A. Pigment epithelium-derived factor supports normal development of photoreceptor neurons and opsin expression after retinal pigment epithelium removal. J Neurosci. 2000;20: