Introduction

Gene therapy provides a means of addressing protein deficiencies or excesses that contribute to diseases. There are several types of ocular diseases that can be treated by gene transfer, and the eye has characteristics that are potentially advantageous. It is a small compartment, which can decrease the amount of virus needed and enhance transduction efficiency. One important application is gene replacement for mutated genes that code for proteins needed for the normal functioning of photoreceptors or retinal pigmented epithelial (RPE) cells.^1,2,3,4 Gene transfer provides a way to replace the defective gene product and potentially prevent retinal degeneration. Another application is to transduce cells within the eye to produce a survival factor, resulting in a high intraocular level of the survival factor in an attempt to rescue photoreceptors from mutation-induced cell death.^5,6,7,8 A third application is to transduce intraocular cells to produce an antiangiogenic protein to suppress ocular neovascularization.^{9,10,11,12,13,14}

There is great potential for each of these gene therapy applications, but there are also theoretical pitfalls. Two areas of concern are the possibility of vector-related toxicity and uncertainties associated with prolonged transgene expression. Adenoviral (Ad) vectors have several advantages, such as relative ease of production, their ability to accept large gene cassettes, and transduction of a variety of noncycling cells resulting in high expression levels. However, Ad vectors induce an immune response that results in inflammation and mediates destruction of transduced cells. Fortunately, there is little transduction of retinal neurons by Ad vectors, but there is still the possibility of bystander damage.^15,16,17 Adeno-associated viral (AAV) vectors and lentivirus vectors induce little immune response and mediate long-term expression.^{18,19,20,21,22,23} On the one hand, this is an advantage, but on the other the effects of excessive and/or prolonged expression of various transgenes in retinal cells are unknown. Both AAV and HIV vectors transduce photoreceptors and transduction of photoreceptors may be particularly risky, because even increased expression of wild-type rhodopsin or peripherin/rds can result in degeneration of photoreceptors.^24,25 For gene replacement for loss of function photoreceptor mutations, transduction of photoreceptors is unavoidable, and may require inducible promoter systems that provide precise control of transgene expression. For the other applications, the particular cells transduced are irrelevant provided sufficient transgene product reaches needed locations within the eye. Targeting of cells that are not critical for visual function is a prudent approach. One way to do this is to perform periocular injections of vector to transduce connective tissue cells on the outside of the eye. In this study, we tested the feasibility of using this approach to treat ocular neovascularization with AdPEDF.11.

Materials and methods

Adenoviral vectors expressing -galactosidase or PEDF

Production and quantification of type 5 adenoviral vectors that express -galactosidase or human PEDF from a cytomegalovirus (CMV) immediate early promoter expression cassette have been previously described.^10,17,26 The vectors were deleted for E1A, E1B, E3 and either contain the intact E4 region (AdLacZ.10) or are deleted for the E4 region (AdPEDF.11). A vector that does not express transgene product (AdNull.11) was used as a control.

Periocular injection of vectors or PEDF protein

Mice were treated humanely in strict compliance with the Association for Research in Vision and Ophthalmology statement on the use of animals in research. Recombinant PEDF was produced and purified as previously described.¹⁰ Adult or neonatal C57BL/6 or Balb/C mice were anesthetized and a 32 gauge needle of a Hamilton syringe was inserted under the conjunctiva of one eye and 3 l containing 5 10⁹ AdLacZ.10 was injected. Mice were euthanized at 1, 4, or 14 days after injection. Periocular injections of AdPEDF.11 or AdNull.11 were carried out in untreated adult C57BL/6 mice immediately after laser-induced rupture of Bruch's membrane, or in neonatal mice with oxygen-induced ischemic retinopathy on day 3 of oxygen treatment, postnatal day (P) 10. Recombinant PEDF protein¹⁰ was dissolved at a concentration of 0.6 g/l in buffer containing 150 mM sodium chloride and 20 mM sodium phosphate, pH 7.4. Adult C57BL/6 mice were given an injection of 5 l containing 3 g of PEDF under the conjunctiva of one eye. The mice were euthanized 2, 6, and 24 h after injection and ocular frozen sections were immunohistochemically stained for PEDF as described below.

Assessment of LacZ expression

At the 4 day time point, ocular and orbital sections were stained for LacZ, and at all three time points liver sections were stained for LacZ. Eyes, orbital tissue, or liver were frozen in optimal cutting temperature embedding compound (OCT; Miles Diagnostics, Elkhart, IN, USA). Frozen section of (10 m) were rinsed in PBS and reacted overnight with 1 mg/ml of X-gal (5-bromo-4-chloro-3-indolyl galactopyranoside, Sigma) in a solution containing 5 mM K₃Fe(CN)₆, 5 mM K₄Fe(CN)₆3H₂O, 1 mM MgCl₂ in PBS. Sections were postfixed for 15 min in 0.5% glutaraldehyde in PBS and washed in PBS.

Assessment of PEDF expression

Expression of PEDF was assessed by immunohistochemistry or enzyme-linked immunoabsorbant assay (ELISA). Four days after periocular injection of AdPEDF.11 or AdNull.11, mice were killed and tissues were dissected, snap-frozen in liquid nitrogen, and stored at -80°C for ELISA or eyes were frozen in OCT. Immunohistochemical staining for PEDF was done using a previously characterized rabbit anti-human PEDF antibody by a described technique.¹⁰ Briefly, 10 m frozen sections were fixed with 4% paraformaldehyde for 30 min, washed with 0.05 M Tris-buffered saline (TBS), incubated in methanol/H₂O₂ for 10 min at 4°C, and washed with TBS. Specimens were blocked with 10% normal goat serum (NGS) in 0.05 M TBS for 30 min at room temperature to prevent nonspecific binding. The slides were incubated with a 1 : 50 dilution of rabbit anti-human PEDF antibody in 1% NGS/0.05 M TBS overnight at 4°C in a humidified chamber. After warming to room temperature, the sections were washed with 1% NGS/0.05 M TBS and incubated with biotinylated goat anti-rabbit antibody for 30 min. After washing, the slides were incubated in streptavidin-phosphatase and developed with HistoMark Red (Kirkegaard and Perry, Gaithersburg, MD, USA) according to the manufacturer's instructions. Sections were dehydrated and mounted with Cytoseal. For controls, anti-PEDF antibody was incubated for 3 h at room temperature with a five-fold molar excess of PEDF prior to incubation on tissue sections.

For ELISAs, specimens were mechanically homogenized with mortar and pestle, sonicated in PBS/100 M PMSF using a Branson 2510 sonicator (Branson, Danbury, CT, USA), and microfuged. Total protein was measured in supernatants using the Bio-Rad Protein Assay (Bio-Rad, Hercules, CA, USA) and concentrations of PEDF were measured using a two antibody sandwich ELISA in 96-well immunoplates (Immulon-IV, Dynatech). Each well was coated with 100 l of 0.1 M NaHCO₃ containing 2 g/ml of a mouse monoclonal anti-human PEDF antibody (Chemicon, Temecula, CA, USA) for 16 h at 4°C. The plates were washed 3 times with PBS/0.05% Tween-20 and blocked for 1 h at 37°C with 200 l of PBS/10% FBS. After washing, tissue samples or recombinant human PEDF²⁷ standards (1–1000 ng/ml) were added. After 1.5 h at room temperature, plates were washed and then incubated for 1 h at room temperature with 100 l of rabbit anti-human PEDF polyclonal antibody conjugated to horseradish peroxidase (0.5 g/ml in PBS/10% FBS). The wells were washed and a TMB substrate/peroxide mixture was added for 25 min. The reaction was stopped with 100 l of 0.5 M sulfuric acid and the plate was read with a microplate reader. The limit of sensitivity of the assay was 0.1 ng/ml.

Mouse model of laser-induced choroidal neovascularization

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 three locations in each eye of each mouse as previously described.²⁸ Briefly, laser photocoagulation (75 m spot size, 0.1 s duration, 120 mW) was delivered using the slit-lamp delivery system and a hand-held cover slide as a contact lens. Burns were performed in the 9, 12, and 3 o'clock positions 2–3 disk 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,²⁸ so only burns in which a bubble was produced were included in the study. After photocoagulation treatment, one eye was given a periocular injection of 5 10⁹ or 1 10⁹ particles of AdPEDF.11 and the other eye was given a periocular injection of 5 10⁹ particles of AdNull.11 or no injection. After 2 weeks, mice were killed and the amount of choroidal neovascularization was assessed.

Measurement of the sizes of laser-induced CNV lesions

Two weeks after laser treatment, the sizes of CNV lesions were measured in choroidal flat mounts.²⁹ Mice used for the flat mount technique were anesthetized and perfused with 1 ml of phosphate-buffered saline containing 50 mg/ml of fluorescein-labeled dextran (2 10⁶ average MW, Sigma, St Louis, MO, USA) as previously described.³⁰ The eyes were removed and fixed overnight in 10% phosphate-buffered formalin. The cornea and lens were removed and the entire retina was carefully dissected from the eyecup. Radial cuts (4–7, average 5) were made from the edge to the equator and the eyecup was flat mounted in Aquamount with the sclera facing down. Flat mounts were examined by fluorescence microscopy on an Axioskop microscope (Zeiss, Thornwood, NY, USA) and images were digitized using a three color CCD video camera (IK-TU40A, Toshiba, Tokyo, Japan) and a frame grabber. Image-Pro Plus software (Media Cybernetics, Silver Spring, MD, USA) was used to measure the total area of hyperfluorescence associated with each burn, corresponding to the total fibrovascular scar.

Mouse model of oxygen-induced ischemic retinopathy

Ischemic retinopathy was produced in C57BL/6 mice by a method described by Smith et al.³¹ Seven-day-old (P7) mice and their mothers were placed in an airtight incubator and exposed to an atmosphere of 753% oxygen for 5 days. Incubator temperature was maintained at 232°C, and oxygen was continuously monitored with a PROOX model 110 oxygen controller (Reming Bioinstruments Co., Redfield, NY, USA). At P10, mice were given a periocular injection of 5 10⁹ particles AdPEDF.11 in one eye and 5 10⁹ particles of AdNull.11 or no injection in the other eye. After injection, the mice were returned to oxygen for two more days and then put in room air. At P17, the mice were killed, and eyes were rapidly removed and frozen in OCT embedding compound.

Quantitation of retinal neovascularization

Frozen sections (10 m) of eyes from injected and uninjected mice were histochemically stained with biotinylated Griffonia simplicifolia lectin B4 (GSA, Vector Laboratories, Burlingame, CA, USA), which selectively binds to vascular cells. Slides were incubated in methanol/H₂O₂ for 10 min at 4°C, washed with 0.05 M Tris-buffered saline, pH 7.6 (TBS), and incubated for 30 min in 10% normal porcine serum. Slides were incubated for 2 h at room temperature with biotinylated GSA, and after rinsing with 0.05 M TBS they were incubated with avidin coupled to peroxidase (Vector Laboratories) for 45 min at room temperature. After being washed for 10 min with 0.05 M TBS, slides were incubated with diaminobenzidine (Research Genetics, Huntsville, AL, USA), to give a brown reaction product. Some slides were counterstained with eosin, and all were mounted with Cytoseal (Stephens Scientific, Riverdale, NJ, USA).

To perform quantitative assessments, 10 m serial sections were cut through the entire eye starting with sections that included the iris root on one side of the eye and proceeding to the iris root on the other side. Every 10th section, roughly 100 m apart, was stained with GSA, examined with an Axioskop microscope, and images were digitized using a three CCD color video camera and a frame grabber. Image-Pro Plus software was used to delineate GSA-stained cells on the surface of the retina and their area was measured. For plotting the area measurements for the figure, the mean of the measurements from each eye was used as a single experimental value.

Statistical analysis

In mice with laser-induced rupture of Bruch's membrane or ischemic retinopathy, the effect of periocular injection of AdPEDF.11 on the amount of neovascularization was evaluated. For the laser-induced choroidal neovascularization model, the analysis included up to three neovascularization area measurements per eye. Some eyes had less than three, because whenever the laser was not associated with the formation of a bubble, that particular lesion was excluded. All measurements from either eye of a mouse were assumed to be exchangeable when modeling correlation structure, and were assumed to be subject to nonerror variability due only to treatment for experiments in which different treatments were administered to each eye of a mouse. We have previously demonstrated that this assumption is valid.³² An overall test for treatment effect was first performed using analysis of variance (ANOVA). If the overall test indicated a significant treatment effect, individual treatments were compared with null vector injection using analyses that adjusted for multiple comparisons including Tukey's test, Bonferroni (Dunn) test, and the general linear model (GLM) with Dunnett's test. The level established for statistical significance was a P-value <0.05 (two-tailed test). All analyses were performed using SAS software (SAS Institute, Inc., Cary, NC, USA).

Results

Periocular injection of AdLacZ.10 results in prominent LacZ expression around the eye

Four days after periocular injection of 5 10⁹ particles of AdLacZ.10, there was prominent staining for LacZ in the lateral rectus muscle (asterisks), orbital fat and the connective tissue around the site of injection (Figure 1a). There was little or no staining in other quadrants of the orbit, but staining of the sclera and episclera often extended all the way around the globe (arrow). Occasionally, sections adjacent to the site of injection in which there was strong staining of periorbital tissue also showed faint staining of retinal cells (Figure 1b, arrowheads), indicating that a small amount of vector had penetrated the sclera. The frequent staining of episclera in the absence of other periocular tissues in areas remote from the injection site (Figure 1c, arrow) suggests a tissue plane between Tenon's capsule and episclera that facilitates spread of virus and a differential adenoviral tropism for episcleral tissue. Even when there was intense staining of surrounding periocular tissue, there was no staining in the optic nerve, suggesting that the optic nerve sheath prevented viral penetration (Figure 1d). Blue staining was observed in the lacrimal gland in orbits injected with virus (Figure 1e, panel 1), in the contralateral orbits (panel 3), and in uninjected mice (panel 2), suggesting some sort of endogenous -galactosidase-like activity, rather than viral transduction. High magnification views showed intense LacZ staining in muscle fibers (Figure 1f, asterisk) and the episclera (arrow). Other than the lacrimal gland, there was no staining of periocular tissue in orbits contralateral to vector injection (Figure 1g–i). There was no staining in liver sections 1 day after injection (Figure 1j), but 4 days after injection each liver evaluated showed a few LacZ-positive cells (Figure 1k). At 14 days after injection, staining was no longer detected in the liver (Figure 1l).

Figure 1.

Reporter gene or PEDF expression after periocular injection of adenoviral vectors. In all, 13 adult C57BL/6 mice were given a 3 l injection containing 5 10⁹ particles of AdLacZ.10 in one eye and killed 1 (n=4), 4 (n=4), or 14 days (n=5) after injection. Periocular and ocular expression of LacZ was assessed on day 4 and expression was evaluated in the liver on days 1, 4, and 14. (a) Orbital sections through regions adjacent to the site of injection showed strong staining for LacZ in the rectus muscle (asterisk), connective tissue, and fat. The episclera was often stained all the way around the globe (arrow). (b) High magnification view of the left side of the section from (a) shows faint LacZ staining in the retina (arrow head), indicating that a small amount of virus was able to pass through the sclera and transfect some retinal cells. There is strong staining in the episclera (arrow) and rectus muscle (above the asterisk). (c) High magnification view of the right side of the section from (a) shows strong staining in the episclera (arrow). (d) A section through the optic nerve (ON) shows strong staining of adjacent sclera and connective tissue, but no staining of the nerve. (e) Lacrimal gland from injected orbits (panel 1), contralateral orbits (panel 3), and orbits of uninjected mice (panel 2) all had a faint blue coloration suggesting endogenous -galactosidase-like activity rather than transduction of lacrimal gland. (f) High magnification shows staining of muscle fibers (asterisk) and episclera (arrow). (g) A section through a contralateral orbit and adjacent region of brain (B) and sinus (S) separated from the orbit by bone (red arrowheads) shows no staining for LacZ. (h and i) High magnification views of the contralateral orbit shown in (g) demonstrate the absence of LacZ staining in connective tissue (asterisk) and sclera (arrows). (j) Liver sections showed no evidence of LacZ staining 1 day after periorbital injection of AdLacZ.10. (k) Liver sections typically showed a few LacZ-positive cells (arrowheads) 4 days after injection. (l) LacZ-positive cells were no longer seen in liver sections 2 weeks after injection. Immunohistochemistry for PEDF was carried out on ocular sections 4 days after periocular injection of 5 10⁹ particles of AdPEDF.11 or control vector encoding soluble Flt-1 (AdFlt-1.10). (m) Red reaction product indicates PEDF in the retina and orbital tissue (OT). The arrowhead indicates the surface of the retina, INL indicates the inner nuclear layer, and ONL indicates the outer nuclear layer. At high magnification (p) staining is seen in the sclera (arrow). (n) Incubation of primary antibody with PEDF eliminates staining in the retina (arrowhead shows surface) and orbital tissue (OT). At high magnification, there is no staining in the sclera (q). (o) Four days after injection of AdFlt-1.10, there is no staining for PEDF in the retina (arrowhead) or orbital tissue (OT). At high magnification, no staining in the sclera is detected (r).

Full figure and legend (250K)

Periocular injection of AdPEDF.11 results in localization of PEDF in periocular and intraocular tissues

Four days after periocular injection of 5 10⁹ particles of AdPEDF.11, there was immunohistochemical staining for PEDF in the retina, periocular tissue, and optic nerve (Figure 1m and p). Staining within the retina and optic nerve, which showed minimal and no transduction with AdLacZ.10, respectively, implies diffusion of secreted PEDF to these sites. Incubation of primary antibody with PEDF eliminated the staining, indicating specificity for PEDF (Figure 1n and g). Four days after injection of an adenoviral vector coding for the extracellular portion of vascular endothelial growth factor receptor-1, there was no staining for PEDF in the retina or periocular tissue (Figure 1o and r).

In several mice, ocular tissues were carefully dissected from ipsilateral and contralateral orbits and eyes 4 days after periocular injection of 5 10⁹ particles of AdPEDF.11. ELISAs for PEDF showed significantly higher levels of PEDF in orbital tissue, sclera, choroid, retina, and anterior segment from the injected side compared to the contralateral side (Figure 2). Levels of PEDF were particularly high in sclera, choroid, and anterior segment. The relatively low PEDF level measured in orbital tissue is likely because transduction is limited to a relatively small percentage of cells within the orbit, because the lacrimal gland occupies a substantial volume and is not transduced. The elevated levels in the retina corroborate the immunohistochemical findings and indicate that PEDF is able to pass through the choroid and outer blood–retinal barrier and enter the retina.

Figure 2.

Levels of PEDF in ocular tissues after periocular injection of AdPEDF.11. Adult C57BL/6 mice were given a periocular injection of 5 10⁹ particles of AdPEDF.11 on one side, and after 4 days the mice were killed and ocular and orbital tissues were dissected from both sides. Levels of PEDF were measured in tissue homogenates as described in Materials and Methods. The bars show the mean (s.e.m.) level of PEDF for the following number of experimental values for each tissue: serum, 4; orbital tissue, 8; optic nerve, 8; sclera, 8; choroid, 7; retina, 16; anterior segment, 8. Black bars show values on the side of the injection and gray bars show values from the contralateral side. For serum, the gray bars are values from uninjected mice. Statistical comparisons were made by a paired t-test.

Full figure and legend (26K)

Recombinant PEDF injected periocularly enters into the retina

The demonstration of intraocular PEDF after periocular injection of AdPEDF.11 suggests that it may also be possible to deliver PEDF to the eye by periocular injection of PEDF protein. To test this, mice were given a periocular injection of 5 l containing 3 g of recombinant PEDF and euthanized at 2, 6, or 24 h after injection. Immunohistochemical staining showed a strong signal for PEDF in the periocular tissue and sclera at 2, 6, and 24 h after injection (Figure 3a–c). There was little staining for PEDF in the retina 2 or 6 h after injection, but by 24 h after injection, there was an unequivocal signal for PEDF. This signal was competed out by preincubation of the primary antibody with PEDF (Figure 3d). Eyes not given a periocular injection showed no staining for PEDF (Figure 3e). An eye given a periocular injection of AdPEDF.11 was used as a positive control and showed staining for PEDF in periocular tissue and faint staining in the retina (Figure 3f).

Figure 3.

Recombinant PEDF protein diffuses into the eye from the periocular space. Adult C57BL/6 mice were given a periocular injection of 5 l containing 3 g of recombinant PEDF in one eye and euthanized 2, 6, or 24 h after injection. Ocular frozen sections were immunohistochemically stained for PEDF. Red reaction product demonstrates high levels of PEDF in the orbit and sclera 2 (a), 6 (b), or 24 h (c) after injection. There is little staining for PEDF in the retina 2 or 6 h after injection, but there is unequivocal staining in the retina 24 h after injection (c). Incubation of primary antibody with PEDF eliminated all but faint background staining in the episclera (d). Staining for PEDF in control, uninjected eyes showed only faint background (e). A section from a Balb/C mouse given a periocular injection of 5 10⁹ particles of AdPEDF.11 showed red reaction product in the orbit and sclera and faint staining in the retina.

Full figure and legend (84K)

Periocular injection of AdPEDF.11 suppresses choroidal neovascularization

C57BL/6 mice had laser-induced rupture of Bruch's membrane at three locations in each eye and then received a periocular injection of 5 10⁹ or 1 10⁹ particles of AdPEDF.11 on the right side and a periocular injection of 5 10⁹ particles of AdNull.11 or no injection on the left side. Eyes that had not received any periocular injection (n=32 eyes, 76 rupture sites) showed large areas of choroidal neovascularization at Bruch's membrane rupture sites (Figure 4a and b). The size and appearance of choroidal neovascularization was very similar in eyes that received a periocular injection of 5 10⁹ particles of AdNull.11 (n=17 eyes, 34 rupture sites; Figure 4c and d). In contrast, eyes that received a periocular injection of 1 10⁹ (n=17 eyes, 34 rupture sites; Figure 4e and f) or 5 10⁹ (n=33 eyes, 78 rupture sites; Figure 4g and h) particles of AdPEDF.11 had small areas of choroidal neovascularization. This impression was confirmed by image analysis, which showed that periocular injection of 1 10⁹ or 5 10⁹ particles of AdPEDF.11 resulted in significantly smaller areas of choroidal neovascularization than periocular injection of 5 10⁹ particles of AdNull.11 (Figure 4i). An overall treatment effect was identified by ANOVA (P=0.0016). An adjustment for multiple comparisons using Tukey's test or the Bonferroni test showed significant differences (P<0.05) for Null vector versus 5 10⁹ AdPEDF.11 or Null vector versus 1 10⁹ AdPEDF.11. Results from the GLM procedure with the Dunnett's Test were consistent with the ANOVA test.

Figure 4.

Periocular injection of AdPEDF.11 inhibits choroidal neovascularization. Bruch's membrane was ruptured by laser photocoagulation at three locations in each eye of adult C57BL/6 mice. Mice were then given no injection (a and b) or a periocular injection of 5 10⁹ particles of AdNull.11 (c and d) in the left eye and, in the right eye, periocular injection of 1 10⁹ (e and f) or 5 10⁹ particles of AdPEDF.11 (g and h). After 14 days, mice were perfused with fluorescein-labeled dextran, and choroidal neovascularization was measured by image analysis of choroidal flat mounts (a, c, e, and g; arrowheads show the outer border of choroidal neovascularization). Some eyes were used for frozen sections (b, d, f, h), which were stained with biotin-labeled Griffonia simplicifolia lectin, which selectively binds to vascular cells. The binding was visualized with HistoMark red and sections were counterstained with Contrast Blue. The arrows show the lateral borders of choroidal neovascularization in each section. No injection (a and b) or periocular injection of 5 10⁹ particles of AdNull.11 (c and d) resulted in large areas of choroidal neovascularization. Periocular injection of 1 10⁹ (e and f) or 5 10⁹ particles of AdPEDF.11 (g and h) resulted in small areas of choroidal neovascularization. (i) Each bar represents the mean (s.e.m.) area of choroidal neovascularization measured by image analysis for periocular injection of 1 10⁹ (18 eyes, 42 rupture sites) or 5 10⁹ particles of AdPEDF.11 (33 eyes, 78 rupture sites), or 5 10⁹ particles of AdNull.11 (17 eyes, 34 rupture sites). * P<0.05 by Tukey's test, Bonferroni test, and Dunnett's test.

Full figure and legend (162K)

Periocular injection of AdPEDF.11 does not significantly reduce retinal neovascularization due to ischemic retinopathy

In mice with ischemic retinopathy, periocular injection of 5 10⁹ particles of AdPEDF.11 did not cause significant reduction of retinal neovascularization compared to eyes that did not receive an injection or those that received a periocular injection of 5 10⁹ particles of AdNull.11 (Figure 5). ANOVA showed no evidence of a treatment effect.

Figure 5.

Periocular injection of AdPEDF.11 does not inhibit ischemia-induced retinal neovascularization. Litters of C57BL/6 mice were placed in 75% oxygen at P7. At P10, they were given a periocular injection of 5 10⁹ particles of AdPEDF.11 in one eye and no injection in the other eye (n=6 mice), or a periocular injection of 5 10⁹ particles of AdPEDF.11 in one eye and a periocular injection of 5 10⁹ particles of AdNull.11 in the other eye (n=7 mice). The mice were put back in 75% oxygen until P12 and then removed to room air. At P17, the mice were killed and the amount of retinal neovascularization was quantified as described in Materials and Methods. There was no significant difference in the amount of retinal neovascularization between eyes that received a periocular injection of AdPEDF.11 in one eye and no injection in the fellow eye (AdPEDF, shaded bar), no injection in either eye (no injection, shaded bar), or a periocular injection of AdPEDF.11 in one eye (AdPEDF, open bar) and a periocular injection of AdNull.11 in the fellow eye (AdNull, open bar).

Full figure and legend (37K)

Discussion

Gene therapy provides a novel way to achieve sustained local delivery of therapeutic proteins to the eye. However, the safety and feasibility of intraocular injections of vectors in humans are unknown. Intraocular injections themselves carry a low risk of infection, retinal detachment, hemorrhage, and cataract, but until repeated intraocular injections of antiangiogenic agents are examined in the context of clinical trials, the magnitude of the risks will not be known. Vector-related toxicity may not be as much of a problem as once thought, even for adenovirus,³³ but the consequences of any toxicity are likely to be more significant after intraocular injection compared to injections outside the eye, because damaged retinal neurons cannot be replaced. Also, highly specialized retinal neurons may be more easily damaged from excessive production of transgenes compared to connective tissue cells.^24,25

In this study, we tested the hypothesis that therapeutic proteins can be delivered to the retina and choroid after periocular injection of vector. Periocular injection of AdLacZ.10 showed that connective tissue cells in the orbit were strongly transduced. Episcleral tissue was particularly prone to transduction and often showed LacZ expression in areas remote from the injection site extending all around the globe, suggesting that virus may spread well in the sub-Tenon's space and/or Ad vectors have a strong tropism for episclera. Periocular injection of AdPEDF.11 resulted in significantly increased levels of PEDF in the retina, choroid, sclera, anterior segment, and orbital connective tissue compared to the same structures from the contralateral side. Immunohistochemistry showed staining for PEDF in the ipsilateral, but not the contralateral retina. Rare transduced cells were seen in the liver, indicating that a small amount of vector entered the circulation, but PEDF levels in the serum were at the limit of detection and were not different from serum levels in uninjected mice. This suggests that the exposure of vector and transgene to extraocular tissues is small and that the transgene is provided to the retina and choroid primarily by local diffusion and not through the circulation. Therefore, periocular cell-derived-PEDF, a 50 kDa protein, is able to pass through mouse sclera and gain access to the choroid and retina. Furthermore, the levels of intraocular PEDF attained after periocular injection of AdPEDF.11 are sufficient to suppress choroidal neovascularization strongly. Additional studies are needed to determine if PEDF can diffuse through sclera in larger animals and, if so, there is a good chance that choroidal diseases may be treated by periocular injection of vectors encoding secreted proteins of 50 kDa or less. However, it should be noted that the orbit is not an immune privileged space and, while adenovirus has been used for proof of concept, investigations with other vectors that do not incite an immune response will be useful.

Periocular injection of AdPEDF.11 did not cause inhibition of retinal neovascularization in mice with ischemic retinopathy. Although levels of PEDF in the retina were elevated, they were roughly 10-fold lower than levels of PEDF in the choroid. The level of PEDF in the retina may have been below a critical threshold required for an antiangiogenic effect and hence the lack of effect on retinal neovascularization. Therefore, the prospect for treating retinal diseases with periocular gene transfer is not as good as the prospect for treating choroidal diseases.

In addition to demonstrating that PEDF produced by periocular cells is able to enter the mouse eye and inhibit choroidal neovascularization, we have found that periocular injection of recombinant PEDF also results in diffusion of PEDF through the sclera into the choroid and retina. If this can also be done in larger animals, then any way of achieving high periocular levels of appropriate-sized therapeutic proteins may be considered for the treatment of retinal and choroidal diseases. Biodegradable or nondegradable sustained release implants could be sutured to the sclera or injectable formulations for sustained release of proteins could be considered. Alternatively, osmotic pumps could be used to deliver proteins to the periocular space. Using this approach, Ambati et al³⁴ demonstrated that a 150 kDa protein, an anti-intercellular adhesion molecule-1 monoclonal antibody, was able to penetrate the sclera and inhibit VEGF-induced stimulation of myeloperoxidase activity in the choroid and the retina. If only a very brief exposure of the protein is needed, periocular injection of a protein solution could be done.

If a 50 kDa protein can pass through mouse sclera and achieve therapeutic levels in the choroid and retina, smaller molecules should be able to do so with even greater facility. Proof of principle has been provided for implantation of sustained release devices in the vitreous cavity.³⁵ Our data suggest that implantation of sustained release devices in the periocular space could be considered as an alternative, and would have the advantages of a lower risk of endophthalmitis and retinal detachment, and easier retrieval and/or replacement.

Our study raises several questions that should be addressed in the future. What is the upper size limit for proteins for local diffusion into the choroid and retina from the periocular space? Do other characteristics of proteins play a role in their diffusability? In vitro studies have suggested that molecular radius may be a more important determinant than molecular weight and that 150 kDa globular proteins with a molecular radius of 5.23 nm are able to penetrate the sclera.³⁶ One would predict that proteins that bind to one or more components of the extracellular matrix would have less ability to gain access to the choroid and retina when expressed or released in the periocular space, but this should be tested. Will periocular injection of other vectors that usually result in longer duration transgene expression, such as adeno-associated viral vectors or lentiviral vectors, result in a similar pattern and level of expression and provide long-term treatment for chronic retinal and choroidal diseases with a single injection? We plan to address these questions in future studies.

References

1. Bennett J et al. Photoreceptor cell rescue in retinal degeneration (rd) mice by in vivo gene therapy. Nat Med 1996; 2: 649–654. | Article | PubMed | ISI | ChemPort |
2. Kumar-Singh R, Farber DB. Encapsidated adenovirus mini-chromosome-mediated delivery of genes to the reina: application to the rescue of photoreceptor degeneration. Hum Mol Genet 1998; 7: 1893–1900. | Article | PubMed | ISI | ChemPort |
3. Takahashi M, Miyoshi H, Verma IM, Gage FH. Rescue from photoreceptor degeneration in the rd mouse by human immunodeficiency virus vector-mediated gene transfer. J Virol 1999; 73: 7812–7816. | PubMed | ISI | ChemPort |
4. Acland GM et al. Gene therapy restores vision in a canine model of childhood blindness. Nat Genet 2001; 28: 92–95. | Article | PubMed | ISI | ChemPort |
5. Cayouette M, Gravel C. Adenovirus-mediated gene transfer of ciliary neurotrophic factor can prevent photoreceptor degeneration in the retinal degeneration (rd) mouse. Hum Gene Ther 1997; 8: 423–430. | PubMed | ISI | ChemPort |
6. Cayouette M et al. Intraocular gene transfer of ciliary neurotrophic factor prevents death and increases responsiveness of rod photoreceptors in the retinal degeneration slow mouse. J Neurosci 1998; 18: 9282–9293. | PubMed | ChemPort |
7. Fong-Qi L et al. Long-term protection of retinal structure but not function using rAAV.CNTF in animal models of retinitis pigmentosa. Mol Ther 2001; 4: 461–472. | Article | PubMed | ChemPort |
8. Lau D et al. Retinal degeneration is slowed in transgenic rats by AAV-mediated delivery of FGF-2. Invest Ophthalmol Vis Sci 2000; 41: 3622–3633. | PubMed | ChemPort |
9. Mori K et al. Inhibition of choroidal neovascularization by intravenous injection of adenoviral vectors expressing secretable endostatin. Am J Pathol 2001; 159: 313–320. | PubMed | ISI | ChemPort |
10. Mori K et al. Pigment epithelium-derived factor inhibits retinal and choroidal neovascularization. J Cell Physiol 2001; 188: 253–263. | Article | PubMed | ISI | ChemPort |
11. Lai C-C et al. Suppression of choroidal neovascularization by adeno-associated virus vector expressing angiostatin. Invest Ophthalmol Vis Sci 2001; 42: 2401–2407. | PubMed | ISI | ChemPort |
12. Honda M et al. Experimental subretinal neovascularization is inhibited by adenovirus-mediated soluble VEGF.flt-1 receptor gene transfection: a role of VEGF and possible treatment for SRN in age-related macular degeneration. Gene Ther 2000; 7: 978–985. | Article | PubMed | ISI | ChemPort |
13. Lai C-M et al. Inhibition of angiogenesis by adenovirus-mediated sFlt-1 expression in a rat model of corneal neovascularization. Human Gene Ther 2001; 12: 1299–1310.
14. Bainbridge J et al. Inhibition of retinal neovascularization by gene transfer of soluble VEGF receptor sFlt-1. Gene Ther 2002; 9: 320–326. | Article | PubMed | ChemPort |
15. Rakoczy PE et al. Recombinant adenovirus-mediated gene delivery into the rat retinal pigment epithelium in vivo. Austr NZ J Ophthalmol 1998; 26: S56–S58.
16. Anglade E, Csaky KG. Recombinant adenovirus-mediated gene transfer into the adult rat retina. Curr Eye Res 1998; 17: 316–321. | Article | PubMed | ChemPort |
17. Mori K et al. Intraocular adenoviral vector-mediated gene transfer is increased in proliferative retinopathies. Invest Ophthalmol Vis Sci 2002; 43: 1610–1615. | PubMed | ISI |
18. Flannery JG et al. Efficient photoreceptor-targeted gene expression in vivo by recombinant adeno-associated virus. Proc Natl Acad Sci USA 1997; 94: 6916–6921. | Article | PubMed | ChemPort |
19. Bennett J, Duan D, Engelhardt JF, Maguire AM. Real-time, noninvasive in vivo assessment of adeno-associated virus-mediated retinal transduction. Invest Ophthalmol Vis Sci 1997; 38: 2857–2863. | PubMed | ISI | ChemPort |
20. Ali RR et al. Adeno-associated virus gene transfer to mouse retina. Hum Gene Ther 1998; 9: 81–86. | PubMed | ISI | ChemPort |
21. Bennett J et al. Stable transgene expression in rod photoreceptors after recombinant adeno-associated virus-mediated gene transfer to monkey retina. Proc Natl Acad Sci USA 1999; 96: 9920–9925. | Article | PubMed | ChemPort |
22. Guy J, Qi X, Muzyczka N, Hauswirth WW. Reporter expression persists 1 year after adeno-associated virus-mediated gene transfer to the optic nerve. Arch Ophthalmol 1999; 117: 929–937. | PubMed | ChemPort |
23. Miyoshi H, Takahashi M, Gage FH, Verma IM. Stable and efficient gene transfer into the retina using an HIV-based lentiviral vector. Proc Natl Acad Sci USA 1997; 94: 10319–10323. | Article | PubMed | ChemPort |
24. Olsson JE et al. Transgenic mice with a rhodopsin mutation (Pro23His): a mouse model of autosomal dominant retinitis pigmentosa. Neuron 1992; 9: 815–830. | Article | PubMed | ISI | ChemPort |
25. Sarra GM et al. Gene replacement therapy in the retinal degeneration slow (rds) mouse: the effect on retinal degeneration following partial transduction of the retina. Hum Mol Genet 2001; 10: 2353–2361. | Article | PubMed | ChemPort |
26. Brough DE et al. A gene transfer vector-cell line system for complete functional complementation of adenovirus early regions E1 and E4. J Virol 1996; 70: 6497–6501. | PubMed | ISI | ChemPort |
27. Duh EJ et al. Pigment epithelium-derived factor suppresses ischemia-induced retinal neovascularization and VEGF-induced migration and growth. Invest Ophthalmol Vis Sci 2002; 43: 821–829. | PubMed | ISI |
28. Tobe T et al. Targeted disruption of the FGF2 gene does not prevent choroidal neovascularization in a murine model. Am J Pathol 1998; 153: 1641–1646. | PubMed | ISI | ChemPort |
29. Edelman JL, Castro MR. Quantitative image analysis of laser-induced choroidal neovascularization in rat. Exp Eye Res 2000; 71: 523–533. | Article | PubMed | ISI | ChemPort |
30. Tobe T et al. Evolution of neovascularization in mice with overexpression of vascular endothelial growth factor in photoreceptors. Invest Ophthalmol Vis Sci 1998; 39: 180–188. | PubMed | ISI | ChemPort |
31. Smith LEH et al. Oxygen-induced retinopathy in the mouse. Invest Ophthalmol Vis Sci 1994; 35: 101–111. | PubMed | ISI | ChemPort |
32. Ando A et al. Nitric oxide is proangiogenic in retina and choroid. J Cell Physiol 2001; 191: 116–124. | Article |
33. Rasmussen HS et al. Safety of intravitreal delivery of adenovirus in cynomolgus monkeys. Invest Ophthalmol Vis Sci (suppl) 2002: 43: 1289.
34. Ambati J et al. Transscleral delivery of bioactive protein to the choroid and retina. Invest Ophthalmol Vis Sci 2000; 41: 1186–1191. | PubMed | ISI | ChemPort |
35. Musch DC et al. Treatment of cytomegalovirus retinitis with a sustained-release ganciclovir implant. The Ganciclovir Impant Study Group. N Engl J Med 1997; 337: 83–90.
36. Ambati J et al. Diffusion of high molecular weight compounds through sclera. Invest Ophthalmol Vis Sci 2000; 41: 1181–1185. | PubMed | ISI | ChemPort |

Acknowledgements

Supported by grants from Michael Panitch, the National Eye Institute (EY05951, EY12609, K08EYB420, and core grant P30EY1765), the Juvenile Diabetes Foundation (PG), Knights Templar (PG and AD), Research to Prevent Blindness (a Lew R Wasserman Merit Award (PAC), a Career Development Award (EJD), and an unrestricted grant), the BMA John William Clark Award (AD), GenVec, Inc., the Ruth and Milton Steinbech Foundation, and Dr and Mrs William Lake. PAC is the George S and Dolores Dore Eccles Professor of Ophthalmology and Neuroscience.