Gene Therapy for Detached Retina by Adeno-Associated Virus Vector Expressing Glial Cell Line–Derived Neurotrophic Factor

Abstract

purpose. To examine the protective effect of glial cell line–derived neurotrophic factor (GDNF) on retinal detachment (RD)–induced photoreceptor damage by using gene delivery.

methods. Gene delivery to photoreceptors was achieved by subretinal injection of recombinant adeno-associated virus expressing GDNF (rAAV-GDNF) in the right eyes and AAV expressing Escherichia coli LacZ (rAAV-LacZ) in the left eyes of Lewis rats. RD in bilateral eyes was induced with subretinal injection of high-density vitreous substitute in the temporal retina 3 weeks after gene delivery. The synthesis and accumulation of GDNF within the retina was monitored 3 weeks after RD by immunohistochemistry and enzyme-linked immunosorbent assay (ELISA), respectively. The rescue of photoreceptors was evaluated by monitoring the preservation of the thickness of photoreceptor outer segment (OS) and outer nuclear layer (ONL). Apoptosis in the photoreceptors was studied using the TdT-dUTP terminal nick-end labeling (TUNEL) method 2 days after RD. Müller cell activity was checked using the immunohistochemistry with glial fibrillary acidic protein (GFAP) antibody 28 days after RD.

results. Gene delivery was demonstrated by immunohistochemical study. The results of ELISA confirmed that high levels of neurotrophic factors were produced in retinas. Photoreceptor OS degeneration and the gradual shortening of the ONL were noted after RD in all the eyes. However, rAAV-GDNF–treated eyes retained longer OS than rAAV-LacZ–treated eyes 7 (P = 0.012) and 28 days (P = 0.008) after RD. ONL was also longer in rAAV-GDNF–treated eyes than in rAAV-LacZ–treated eyes 7 (P = 0.012) and 28 days (P = 0.008) after RD. GDNF-treated eyes had statistically less apoptotic cells than control eyes in photoreceptor layer (P = 0.043). Subretinal proliferation of Müller cells was suppressed in the GDNF-treated group, indicating less scar formation.

conclusions. GDNF is a potential factor that can protect photoreceptors from degeneration. In addition to preserving the OS and ONL structures, GDNF may exert its protective action by preventing the apoptosis of photoreceptors after RD. GDNF gene therapy may be a valuable adjuvant to current treatments in certain complicated forms of RD.

Retinal detachment (RD) is a common cause of human visual impairment, especially in patients with high myopia. ¹ Standard methods of treatment include laser photocoagulation, pneumatic retinopexy, scleral buckling, and vitrectomy. Recent advances in surgical instruments and chemical agents such as perfluorocarbon liquids have greatly improved surgical outcome and reduced complication rates. ² Because photoreceptors continue to undergo cell loss after RD, surgical outcomes are not always satisfactory. Degeneration of the outer segment (OS) was considered to be the primary effect of detachment and incomplete regeneration the most likely cause of continued visual deficits after successful reattachment surgery. ³ Therefore, it is important to preserve the integrity of photoreceptors. Well-preserved photoreceptors may offer a better chance of achieving better visual acuity after surgery. 

Various agents, such as ciliary neurotrophic factor (CNTF), basic fibroblast growth factor (bFGF), pigment epithelium-derived factor (PEDF), and brain-derived neurotrophic factor (BDNF) have successfully achieved rescue of photoreceptors both morphologically and functionally. ^{4 5 6 7 8 9 10} These have all been shown to enhance photoreceptor survival in animal models of retinitis pigmentosa or RD. Long-term maintenance of high levels of therapeutic peptide in vivo may be critical to the rescue of photoreceptors. However, due to the short half-lives of these agents, repeated intravitreous injection is necessary, which is likely to cause undesired side effects. The delicate tissue and the difficulty in gaining access make repeated subretinal injection of recombinant protein impractical. With the recent advances in gene therapy, genes have been delivered into eyes, and therapeutic proteins are synthesized inside the retina to rescue photoreceptors. ^{11 12 13}  

Glial cell line–derived neurotrophic factor (GDNF) has been shown to prolong the survival of the dopaminergic neuron both in vitro ¹⁴ and in vivo. ¹⁵ Studies have also found that GDNF in rat brain can protect axotomy-induced neuronal degeneration ¹⁶ and can protect nigral dopamine neurons against 6-hydroxydopamine toxicity in vivo. ¹⁷ Recent investigations have shown that intracerebroventricular and intraparenchymal administration of GDNF potentially protects the cerebral hemisphere from damage induced by middle cerebral artery (MCA) occlusion. ¹⁸ In addition, the increase in nitric oxide that accompanies an MCA infarction is blocked almost completely by GDNF. ¹⁸ In the studies involving retina, GDNF induces histologic and functional protection of rod photoreceptors in the rd/rd mouse. ¹⁹ GDNF promotes the survival of axotomized retinal ganglion cells in adult rats. ^{20 21} Rod OS maintenance is enhanced in the presence of bFGF, CNTF, and GDNF in vitro. ²² From these observations, we hypothesize that GDNF may also protect photoreceptors from the injury caused by RD. 

Recombinant adeno-associated virus (rAAV) vectors provide a highly efficient gene delivery system that can facilitate long-term transduction, and the system has been used in a wide variety of gene therapy studies. ^{23 24} Recently, we reported the effective suppression of experimental arthritis and the reduction of infarction size induced by cerebral ischemia by rAAV-based gene approaches. ^{25 26 27} In gene therapy involving eyes, we have successfully suppressed choroidal neovascularization by rAAV-expressing angiostatin. ²⁸ Moreover, the potential of rAAV vectors in the gene therapy of ocular tissues has been implicated by the delivery of marker genes by this vector, which achieves long-term and stable gene expression in retinal tissues. ^{29 30 31}  

In this study, to establish the potential of GDNF as a protective agent for RD-induced photoreceptor damage, rAAV carrying the GDNF gene was injected into the subretinal space and then RD was induced. We observed obvious and lasting protection of photoreceptors by this approach. 

Materials and Methods

Generation of rAAV-GDNF

rAAV encoding rat GDNF cDNA was constructed in previous work, ²⁷ by a three-plasmid cotransfection system, as previously described. ²³ The rAAV was purified twice by cesium chloride ultracentrifugation as described. ²³ Titers of rAAV-GDNF and rAAV-LacZ were determined by dot blot hybridization, with GDNF and LacZ DNA, respectively, used as probes. 

Animals and Transfection of Virus

Lewis rats weighing 250 to 300 g were used. The animals were handled in accordance with the ARVO statement for the Use of Animal in Ophthalmic and Vision Research. Rats were anesthetized with intramuscular injection of 0.15 mL/kg of an equal-volume mixture of 2% lidocaine (Xylocaine; Astra, Södertälje, Sweden) and 50 mg/mL ketamine (Ketalar; Parke-Davis, Morris Plains, NJ). After rats were anesthetized, pupils were dilated with 1% tropicamide (1% Mydriacyl; Alcon Laboratories, Hempstead, UK) and then eyes were gently protruded with a rubber sleeve. The eyes were then covered with sodium hyaluronate (Healon; Pharmacia and Upjohn, Uppsala, Sweden) and a transparent disc, which allowed the fundus to be well visualized under a surgical microscope. Then a 90° peritomy was made in the temporal quadrant, a sclerotomy was made 1 mm behind the limbus with the tip of a 27-gauge needle. A 33-gauge blunt-tip needle (Hamilton, Reno, NV) was inserted tangentially toward the posterior pole of the eye and 3 μL of viral suspension containing 1.1 × 10¹⁰ viral particles—equal to 1.1 × 10⁸ infectious units or 1.1 × 10⁷ transduction units—was injected. ³² The needle was left in the subretinal space for 1 minute after injection to reduce the degree of reflux. Subretinal injection was confirmed by identifying RD in the temporal quadrant. Similarly, the contralateral eye was injected with rAAV-LacZ serving as the control. 

Experimental Retinal Detachment

The animals were prepared and anesthetized as described 3 weeks after subretinal delivery of viral particles. After pupils were dilated and eyeballs were protruded with a rubber sleeve, RD was created with subretinal injection of sodium hyaluronate (Healon-GV, Pharmacia and Upjohn) as described elsewhere, with modification. ³³ Briefly, the sclera was punctured at the temporal equator with a 30-gauge needle, and the needle was slowly advanced into the subretinal space. The injection of sodium hyaluronate reproducibly produced RD in approximately half of the retina. RD was confirmed in every animal by surgical microscope. Reproducible and circumscribed blebs of RD were created that remained essentially stable during the time of study (4 weeks). If a subretinal hemorrhage was noted during the surgical procedure, the animal was discarded to prevent interference of study results. 

Enzyme-Linked Immunosorbent Assay

The ELISA was used to measure the production of GDNF in the retina. ELISA was undertaken 3 weeks after viral transfection. After rats were anesthetized, the eyeballs were enucleated. Cornea, lens, and vitreous were removed, and the retinas of rats were dissected for ELISA. Retina was sonicated on ice in a lysis buffer (200 μL) containing a proteinase inhibitor cocktail (Complete Mini; Roche Diagnostics, Mannheim, Germany). An acid-treatment method was adopted to enhance detection of GDNF. ³⁴ Briefly, homogenized tissue samples were treated by adding 1 N HCl until the pH was less than 3.0, which was confirmed by spotting an aliquot on litmus paper. The samples were incubated for 15 minutes at room temperature and then neutralized by adding 1 N NaOH, and pH was confirmed to be approximately 7.6. After centrifugation for 10 minutes, the supernatant was used in the assay. The wells of a 96-well ELISA plate were coated with anti-rat GDNF antibody (50 μg/mL, 100 μL/well; R&D Systems, Minneapolis, MN) in 0.1 M buffered saline (pH 8.2) for 6 hours at room temperature. After three washes with phosphate-buffered saline (PBS), the plate was blocked with 3% milk at 4°C overnight. The wells were then washed three times with PBS. Samples were incubated in wells for 5 hours at room temperature (100 μL/well). The wells were then washed three times with PBS and biotin-conjugated anti-GDNF antibody (400 ng/mL, 100 μL/well; R&D Systems) was added to each well and incubated for 2 hours at room temperature. After three washes with PBS, streptavidin horseradish peroxidase (1:1000 dilution, 100 μL/well; HRP conjugate, Zymed Laboratories, South San Francisco, CA) was added to each well and incubated for 30 minutes at room temperature in the dark. After three washes with PBS, tetramethylbenzidine (TMB; 100 μL/well; American Society for Clinical Laboratory Science, Mansfield, MA) was added as a substrate waiting for color change and then the reaction was terminated with 2 N H₂SO₄ as a stop solution. The plate was then read with a microplate reader (Spectra MAX 250; Molecular Devices, Sunnyvale, CA) at 450 nm absorbance. Recombinant rat GDNF (R&D Systems) was serially diluted ranging from 0 pg/mL to 2000 pg/mL in 3% milk to make a standard curve. 

Morphology and Electron Microscopy Study

For orientation, temporal sclera was sutured with stitches before enucleation. After enucleation, cornea, lens, and vitreous were removed. A midline sagittal cut was made in the eyecup, crossing the optic disc. The temporal retina was fixed in 2.5% glutaraldehyde in sodium phosphate buffer for 2 hours. The tissue was then fixed in phosphate-buffered osmium tetroxide (1%) for 1 hour and embedded in Spurr resin. For quantifying the length of photoreceptor OS and ONL, retinas approximately 100 μm temporal to the midline sagittal cut were sectioned and sampled. Measurements were taken at an interval of 100 μm in the section and combined to obtain an average OS and ONL length. Slides were coded so that the observers did not know the treatments. The retina was sectioned at 0.5 μm, counterstained with toluidine blue, and examined by light microscopy (Eclipse E800; Nikon, Osaka, Japan) and, in some cases, sectioned at 90 nm, by electron microscopy (JEM-2000 EX; JEOL, Tokyo, Japan). 

In Situ TUNEL Labeling

Apoptosis of photoreceptors usually occurs 1 to 3 days after the creation of RD. ³⁵ Therefore, eyeballs were enucleated 2 days after RD for TUNEL analysis. Eyeballs were sectioned along a perpendicular plane close to the optic nerve head, then fixed in 4% paraformaldehyde at 4°C overnight, embedded in paraffin, and sectioned at 5 μm. Retinas were sampled approximately 100 μm from the optic nerve for TUNEL-positive photoreceptors. TUNEL was performed using an apoptosis TdT DNA fragment detection kit (TdT FragEL; Oncogene, Darmstadt, Germany) according to the manufacturer’s instructions. Briefly, paraffin sections were deparaffinized in xylene and rehydrated through a graded series of alcohol. The samples were treated with proteinase K for 15 minutes at room temperature and washed in PBS. Endogenous peroxidase was quenched by incubating the sections with 3% H₂O₂ for 5 minutes at room temperature and washed in PBS. The sections were incubated with optimized ratio of biotin-labeled and unlabeled deoxynucleotides, terminal deoxynucleotidyl transferase (TdT) and 20% of 5× cacodylate buffer in a moist chamber for 1.5 hour at 37°C and washed in PBS. Peroxidase-conjugated streptavidin was added for 30 minutes at room temperature and washed with PBS. Diaminobenzidine was used as a chromogen. The results were viewed with a microscope (Eclipse E800; Nikon) after mounting in mounting gel. 

Immunohistochemistry

Samples were fixed in 4% paraformaldehyde for 2 hours after removal of cornea and lens. Then they were put into 30% sucrose (in PBS) overnight at 4°C. This was followed by embedding in optimal cutting temperature (OCT) compound and sectioning in a microtome cryostat (CM1900; Leica, Wetzlar, Germany). The sections were placed on slides that had been coated with 1% gelatin and 0.1% chromium potassium sulfate (Sigma, St. Louis, MO) in distilled water to promote adhesion of the sections to the glass surface. Samples were blocked with 1% goat serum and 1% bovine serum albumin for 30 minutes after washing in PBS. For the detection of GDNF, samples were incubated with rabbit polyclonal antibody recognizing GDNF (1:200 dilution; Santa Cruz Biotechnology, Santa Cruz, CA) using a commercial kit (LSAB 2; Dako, Carpinteria, CA). For the detection of Müller cells, samples were incubated with rabbit anti-glial fibrillary acidic protein (GFAP) antibody (Dako), using donkey anti-rabbit IgG conjugated to the fluorochrome Cy3 (Jackson ImmunoResearch, West Grove, PA) as a secondary antibody. The results were viewed with a fluorescence microscope (Eclipse E800; Nikon). 

Statistical Analysis

The Wilcoxon signed rank test was used to test for statistical difference in the length of OS and the thickness of ONL, as revealed by histology between rAAV-GDNF– and rAAV-LacZ–injected eyes. The test was also used to tell the difference in production of GDNF and TUNEL-positive photoreceptors between rAAV-GDNF– and rAAV-LacZ–injected eyes. The Mann-Whitney test was used determine the difference in production of GDNF between naïve and rAAV-GDNF–injected eyes. The results were computed by computer (SPSS ver. 9.0; SPSS Inc, Chicago, IL). The data are presented as the mean ± SD. P < 0.05 was considered significant. 

Results

Subretinal Injection of rAAV-GDNF Leading to Synthesis of GDNF in Photoreceptors

An rAAV vector delivering the rat GDNF gene was produced as described in the Materials and Methods section. Gene delivery to photoreceptors was achieved by subretinal injection of rAAV-GDNF in the right eyes and rAAV-LacZ in the left eyes of five Lewis rats. Three weeks after gene delivery, the eyeballs were harvested, fixed, sectioned, and reacted to antiserum recognizing GDNF. Prominent GDNF signal was identified in the photoreceptors over the temporal retina (Fig. 1B) . GDNF signal was found in the cytoplasm in ONL. Sometimes, the OS also showed positive staining. In the control eye, no expression of GDNF was identified (Fig. 1A) . These results confirmed the synthesis of GDNF in rAAV-GDNF–transduced cells. 

GDNF Levels in Retina Transfected with rAAV-GDNF

To estimate the levels of GDNF that can be achieved in transduced eyes, tissue of the entire retina was harvested and processed for ELISA. Fourteen Lewis rats received a subretinal injection of rAAV-GDNF in the right eye and rAAV-LacZ in the left eye. A separate group of six rats received a subretinal injection of PBS in the right eye. GDNF levels in retinas of bilateral eyes of both groups were determined 3 weeks after subretinal injection and are shown in Figure 2 . The amounts of GNDF in normal eyes without injection were at background levels (9.50 ± 11.36 pg/mL). In fact, GDNF levels were undetectable in three samples. GDNF became detectable in eyes with subretinal injection of PBS (53.33 ± 23.57 pg/mL), and the levels were higher than in naïve eyes. This suggests the induction of GDNF synthesis by subretinal injection. In eyes with rAAV-LacZ injection (67.86 ± 43.91 pg/mL), GDNF levels were similar to those in eyes with PBS injection (53.33 ± 23.57 pg/mL), indicating that rAAV-LacZ did not further stimulate synthesis of GDNF. The most important finding was that the amounts of GDNF in eyes transfected with rAAV-GDNF (366.71 ± 229.16 pg/mL) were approximately five times those in eyes transfected with rAAV-LacZ (67.86 ± 43.91 pg/mL; P = 0.001). This indicates that cells transduced with rAAV-GDNF synthesized high level of GDNF in the retina. 

Morphology and Quantitative Analysis of the Length of OS and the Thickness of the ONL

Shortening and degradation of the OS and ONL are regarded as the representative changes in the structure of photoreceptors during RD. ^{36 37} In this study, the protective effect of GDNF is in its ability to prevent the shortening and degradation of OS and ONL. The representative retina morphologies examined 7 (Figs. 3B 3C 3D) and 28 (Figs. 3E 3F 3G) days after RD are shown in Figure 3 . OS and IS showed signs of severe degeneration both 7 and 28 days after RD (Figs. 3B 3E) compared with eyes without RD (Fig. 3A) . Similar to other reports, photoreceptor cell loss was patchy but obvious at both time points. In RD, eyes with rAAV-LacZ injection (Figs. 3C 3F) , photoreceptor cell loss and degeneration was also obvious. In the retinas treated with rAAV-GDNF, however, the morphology of photoreceptors was much better preserved (Figs. 3D 3G) than in retina treated with rAAV-LacZ injection. 

In this study, electron microscopy was used to reveal the protection by GDNF at the ultrastructure level. Electron micrographs of photoreceptor OS from retinas receiving rAAV-GDNF (Figs. 4C 4E) and rAAV-LacZ (Figs. 4B 4D) and detached for 7 (Figs. 4B 4C) and 28 (Figs. 4D 4E) days are shown. In rAAV-LacZ–injected eyes experiencing RD, the OS appeared as shortened and disorganized fragments 7 days (Fig. 4B) after RD. OS had completely degenerated by 28 days of detachment (Fig. 4D) . This was entirely different from the OS of a normal retina (Fig. 4A) . The degradation changes induced by RD were partially prevented by viral vector delivery of the GDNF gene. The rAAV-GDNF–injected eyes consistently had both longer and better-organized OS at 7 (Fig. 4C) and 28 (Fig. 4E) days after RD than did rAAV-LacZ–injected eyes. 

To further document the effect of GDNF gene delivery, we compared the length of OS and the thickness of ONL of eyes without viral injection with that in eyes injected with rAAV-LacZ and eyes injected with rAAV-GDNF. The length of OS and thickness of ONL of normal eyes were derived from four rats without any treatment. The naïve rats were approximately 10 weeks old, having similar weight (around 300 g) with rats with viral injection. In the measurements of the length of OS (Fig. 5A) , no significant difference (P = 0.453, n = 8) was noted between rAAV-GDNF–treated eyes (15.82 ± 4.31 μm, 90.40% ± 24.63% of normal eyes) and rAAV-LacZ–treated eyes (12.18 ± 2.79 μm, 69.60% ± 15.96% of normal eye) 1 day after RD. However, there was a significant difference (P = 0.012, n = 16) between rAAV-GDNF–treated eyes (11.03 ± 2.27 μm, 63.04% ± 12.96% of normal eyes) and rAAV-LacZ–treated eyes (4.49 ± 1.48 μm, 25.64% ± 8.44% of normal eyes) 7 days after RD. After RD for 28 days, the length of OS was significantly longer (P = 0.008, n = 16) in rAAV-GDNF–treated eyes (10.78 ± 1.97 μm, 61.60% ± 11.28% of normal retina) than in rAAV-LacZ–treated eyes (4.01 ± 0.91 μm, 22.92% ± 5.20% of normal retina). 

In the measurements of the thickness of ONL (Fig. 5B) , results were similar to those noted in the measurements of OS. There was no statistical difference (P = 0.99, n = 8) between rAAV-GDNF–treated eyes (37.57 ± 13.60 μm, 93.93% ± 34.00% of normal eyes) and rAAV-LacZ–treated eyes (36.46 ± 7.04 μm, 91.15% ± 17.60% of normal eyes) 1 day after RD. However, the retina retained thicker (P = 0.012, n = 16) ONL layers in rAAV-GDNF–treated eyes (38.49 ± 7.17 μm, 96.22% ± 17.93% of normal eyes) than in rAAV-LacZ–treated eyes (22.74 ± 4.94 μm, 56.85% ± 12.35% of normal eyes) 7 days after RD. One month after RD, there was still better preservation (P = 0.008, n = 16) of ONL layers in rAAV-GDNF–treated eyes (32.03 ± 11.20 μm, 80.08% ± 28.00% of normal eyes) than in rAAV-LacZ–treated eyes (19.38 ± 2.27 μm, 48.45% ± 5.68% of normal eyes). The results are summarized in Figure 5 . 

Prevention of Apoptosis by rAAV-GDNF Injection

TUNEL stain was used to investigate whether rAAV-GDNF injection can prevent RD-induced apoptosis. For TUNEL assay, a group of five rats were used. Three weeks after rAAV injection, RD was created. TUNEL assay was performed 2 days after RD. More TUNEL-positive photoreceptor cells were noted in rAAV-LacZ–injected eyes (Fig. 6C) , but only sparse TUNEL-positive cells were noted in rAAV-GDNF–injected eyes (Fig. 6D) . After counting apoptotic cells in the retina, we found less apoptotic cells in rAAV-GDNF–injected eyes (5.40 ± 3.44 cells/250 μm retina) than in rAAV-LacZ–injected eyes (26.20 ± 3.96 cells/250 μm retina; P = 0.043). In a different group of five rats, no viral vector was injected. RD was induced in right eyes only. No apoptotic cells were found in left eyes without RD. Apoptotic cells were most abundant in right eyes with RD (39.25 ± 7.89 cells/250 μm retina). Figure 6B shows a DNase-treated sample from an rAAV-GDNF–injected eye, which stained positive in a TUNEL assay and which as a positive control. Samples without TdT enzyme showed no TUNEL staining serving as a negative control (Fig. 6A) . 

Activity of Müller Cells Modulated by Injection of rAAV-GDNF

After RD, Müller cell proliferation is stimulated as part of the retina’s reparative process. GFAP is an excellent marker for the reactivity and hypertrophy of Müller cells. The animals were killed 28 days after inducement of RD. In normal naïve retina, this protein was concentrated in the end-foot region of the Müller cells bordering the vitreous cavity (Fig. 7A) . As expected, in eyes with rAAV-LacZ injection we observed a typical finding of GFAP distribution during the reparative stage after RD. There was prominent accumulation of GFAP in the retina and some labeled Müller cell processes were seen extending into the subretinal space (Fig. 7B) . In contrast, in rAAV-GDNF–injected eyes, subretinal proliferation was significantly suppressed (Fig. 7C) . This indicated a less-active reparative response and provided more evidence to indicate that GDNF can prevent RD-induced retinal damage. 

Discussion

This is the first report that subretinal injection of rAAV-GDNF can achieve gene expression in rat eyes and significantly protect photoreceptors from RD-induced damage. In this study, the gene delivery was demonstrated by immunohistochemistry. The results of ELISA confirmed that high levels of neurotrophic factors were produced in the retina. Histology analysis revealed better maintenance of the OS and ONL in eyes injected with rAAV-GDNF, and there were fewer apoptotic cells. The activity of subretinal proliferation by Müller cells after RD was suppressed by the effect of GDNF. Whether gene therapy could be a good adjuvant to present therapies requires further investigation. However, our results provide important evidence indicating the potential of this approach. 

The results in this study can be applied to the prevention of RD-induced damage in some complicated forms of RD. It has been demonstrated that recurrence or failure of reattachment is common in cases of trauma, cases in children, cases of giant retinal tears, cases accompanied with advanced proliferative vitreoretinopathy, and cases of combined tractional and rhegmatogenous RD. ^{38 39 40 41 42} Multiple surgeries are therefore necessary for the repair of RD. ^{38 39 43} Because photoreceptors undergo cell loss after RD, the surgical outcomes are usually not satisfactory under those circumstances. In such cases, the preservation of photoreceptors is of prime importance. Injection of rAAV-GDNF may provide protection to photoreceptors in cases of recurrent RD or failure of reattachment. Another application may be in the field of macular translocation—an effective treatment for age-related macular degeneration with subfoveal neovascularization. ⁴⁴ During the surgery, an artificially created RD is performed, and the macula is repositioned in another area with healthy retinal pigment epithelium, Bruch’s membrane, and choroid. The retina remains detached for several days until total absorption of subretinal fluid. Photoreceptor cells may be affected during the detachment period. ⁴⁵ Furthermore, the most common and serious postoperative complication of this surgery was RD. ^{44 46} Injection of rAAV-GDNF before macular translocation may augment resistance to the stress of RD for photoreceptors. 

GDNF, which belongs to the transforming growth factor-β superfamily, was first described as promoting the survival of dopaminergic neurons in vitro. ⁴⁷ GDNF is expressed in the rat retina from embryonic day (E) 15 to E19, mostly in the innermost layer. ⁴⁸ In the mouse embryo, it is expressed from E8.5 in the neuroectoderm surrounding the optic vesicle and later in the mesenchymal components of the developing eye. ⁴⁹ In other models of retinal degeneration or RD, GDNF or similar molecules: CNTF, bFGF, PEDF, and BDNF delay the degeneration and apoptotic death of photoreceptor cells. ^{4 5 6 7 8 9 10} Among those neurotrophic factors GDNF may be a particularly important neurotrophic factor, because GDNF can exert its neuroprotective effect even after the degeneration of photoreceptors. ¹⁹ Furthermore, in rats treated with either rAAV-GDNF or rAAV-LacZ, there were no obvious adverse morphologic effects in these eyes compared with naïve eyes. Some factors have been reported to produce retinal rosettes or folds, ⁷ neovascularization, ⁵⁰ and, specifically, minimal posterior subcapsular cataract ⁵¹ in the feline eye. In our study, no such effects were observed. 

The mechanisms through which GDNF acts to protect photoreceptors are still unknown. Yan et al. ²⁰ found that the effect of GDNF is receptor mediated. GDNF potently protects the cerebral hemispheres from damage induced by MCA occlusion. ¹⁸ The increase in nitric oxide that accompanied MCA occlusion and subsequent reperfusion is blocked almost completely by GDNF, suggesting its effect on the modulation of nitric oxide production. ¹⁸ Nicole et al. ⁵² described a novel mechanism for the neuroprotective effects of GDNF against N-methyl-d-aspartate (NMDA)–mediated neuronal death. They found that GDNF activates the mitogen-activated protein kinase (MAPK) pathway and modulates NMDA receptor activity, thus reducing the NMDA-induced calcium influx. GDNF attenuates the slowly triggered NMDA-induced excitotoxic neuronal death through a direct effect on cortical neurons. They also demonstrated that activation of an extracellular signal-regulated kinases (ERKs) pathway is necessary for GDNF-mediated reduction of the NMDA-induced calcium response. These effects could be responsible for the neuroprotective effect of GDNF in acute brain injury. 

In an in vitro assay, Carwile et al. ²² showed that GDNF, when used at relatively high concentrations, protects against the collapse of photoreceptor OS. The effect of GDNF appears to be dose dependent, and the concentration needed to be effective in neural tissue is much higher than the effective concentration of other growth factors. Another study also stressed that endogenous GDNF alone is not sufficient for the rescue of photoreceptors. ¹⁹ Therefore, only sufficient amounts of GDNF exert its neuroprotective effect. Our ELISA results demonstrated that sufficient amounts of GDNF were produced. This may be due to efficient retinal transfection by the rAAV. It is estimated that rAAV is up to 2000 times more efficient than adenovirus at transducing photoreceptor cells. ³⁰ This probably explains the high levels of GDNF synthesized in retina in our study. Although substantial amounts of GDNF were produced in rAAV-GDNF–transfected retina, a large standard deviation was measured in GDNF expression in the rAAV-GDNF–injected animals. There are several possible explanations. One of the possible mechanisms causing the variation of GDNF gene expression may be the reflux of AAV after subretinal injection. The eyes of rats are small and high intraocular pressure may be encountered after subretinal injection; therefore, there sometimes reflux occurs in this procedure. Secondly, variation of virus deposition in the retina may cause a variation in results. There may be an interanimal difference in the diffusion of virus in the subretinal space, which affects the number of cells infected. Therefore, the number of rods or cones transfected with AAV may be different among these eyes. Thus, a different amount of translated protein may be encountered, resulting in a large variation in ELISA test results. 

Although we present evidences of the validity of the concept that GDNF can rescue RD-induced photoreceptors damage, gene therapy for RD with rAAV-GDNF has its limitations. It is generally believed that it takes 10 to 14 days for a transduced gene to be expressed. ^{29 53} In clinical practice, this delay would probably cause irreversible damage to the retina. ^{37 54} Because rapid expression of the transduced gene is crucial for successful gene therapy, one promising solution would be to apply the newly developed modified rAAV vector, which shortens the incubation time by producing both plus-strand–containing AAV and minus-strand–containing AAV. Transduced DNA reanneals inside target cells and becomes double stranded and capable of transcription. This skips the replication process from single-strand viral DNA to double-strand DNA that causes a delay in gene expression (Xiao X, unpublished results, 2001). In our observation, genes transduced by this modified vector can be expressed within 48 hours after injection (Tsao et al., unpublished results, 2001). Lentivirus is another potential vector. Stable and efficient transfection of the retina has been achieved using different promoters. ^{55 56} With continuing improvement in these vectors, rapid transcription of the transgenes may be achieved in the future. 

In conclusion, photoreceptors can be protected from RD-induced damage by subretinal injection of rAAV-GDNF. Gene therapy with rAAV-GDNF may be a good adjunct to present treatments in complex types of RD. With the rapid development and improvement of viral vectors, more powerful and efficient vectors could be applied and may offer the same protection, even if vectors are injected after the onset of RD. 

 Supported by Chang Gung Memorial Hospital Medical Research Grant CMRP-347 and the Department of Health of Taiwan Grant DOH90-TD-1029.

 Submitted for publication January 7, 2002; revised May 1, 2002; accepted May 30, 2002.

 Commercial relationships policy: N.

 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.

 

Immunohistochemistry analysis detecting GDNF synthesis in rAAV-GDNF–injected eyes. Rat eyes were transduced with either rAAV-lacZ (A) or rAAV-GDNF (B). Immunohistochemistry analysis of retinas with antibodies recognizing GDNF was performed 3 weeks after rAAV transduction. IS, inner segment; ONL, outer nuclear layer; INL, inner nuclear layer; IPL, inner plexiform layer. Magnification, ×150.

Figure 2.

 

ELISA identifying high level of GDNF accumulation in retinal tissues after injection of rAAV-GDNF. Subretinal injection of rAAV-GDNF was performed in the right eyes and of rAAV-LacZ in the left eyes (n = 14). In a separate group of rats, subretinal injection of PBS was performed in the right eyes and no treatment in the left eyes (n = 6). Three weeks after rAAV transduction, tissue of the entire retina was harvested and processed for ELISA. *P < 0.05 compared with the rAAV-GDNF–transfected group. Bar, SD.

Figure 3.

 

Representative light micrographs demonstrating the protection of photoreceptor by rAAV-GDNF. Three weeks after rAAV transduction, RD was induced. Seven (B, C, D) and 28 (E, F, G) days after RD, eyeballs were harvested, sectioned, and stained. (A) Retina from an eye without RD or rAAV injection. (B, E) Retinas without virus injection. (C, F) Retinas of eyes with RD and rAAV-lacZ injection. (D, G) Retinas of eyes with RD and rAAV-GDNF injection. INL, inner nuclear layer; IPL, inner plexiform layer; IS, inner segment; ONL, outer nuclear layer; OPL, outer plexiform layer; OS, outer segment. Resin-embedded tissues, 0.5-μm section, toluidine blue staining. Magnification, ×200.

Figure 4.

 

Representative electron micrographs of photoreceptor OS showing the protection of photoreceptor by rAAV-GDNF. Three weeks after rAAV transduction, RD was induced. Seven (B, C) and 28 days (D, E) after RD, eyeballs were harvested, fixed, sectioned, and examined. (A) Retina without RD or rAAV injection. (B, D) Retinas of eyes with RD and rAAV-LacZ injection. (C, E) Retinas of eyes with RD and rAAV-GDNF injection. Magnification, ×4000.

Figure 5.

 

Average lengths of OS and outer nuclear layers in eyes receiving different treatment. The specimens shown in Figure 3 were subjected to the analysis of average length of OS (A) and depth of the ONL (B). Average length of OS and depth of the ONL in retina from the various treatment groups are shown as percentage of normal. Lengths of OS and ONL of normal eyes were derived from four normal naïve eyes. *P < 0.05 compared with rAAV-GDNF injected eyes. Bar, SD.

Figure 6.

 

Apoptosis in photoreceptors detected by TUNEL stain after retinal detachment for 2 days. Three weeks after rAAV transduction, RD was induced. Two days after the induction of RD, eyes were harvested, fixed, and subjected to TUNEL analysis. (A) Sample omitting TdT enzyme showed no TUNEL staining and served as the negative control. (B) A DNase treated sample from rAAV-GDNF injected eye stained positive in the TUNEL assay and served as the positive control. (C) rAAV-LacZ injected eyes. (D) rAAV-GDNF injected eyes. GC, ganglion cell layer; INL, inner nuclear layer; IPL, inner plexiform layer; ONL, outer nuclear layer. Magnification, ×150.

Figure 7.

 

Immunofluorescent micrographs showing activity in Müller cells. Three weeks after rAAV transduction, RD was induced. Immunohistochemical analysis of the retina with antibodies recognizing GFAP was performed 28 days after RD. (A) Retina of an eye without RD or rAAV injection. (B) Detached retina injected with rAAV-LacZ. (C) Detached retina injected with rAAV-GDNF. Magnification, ×200.

The authors thank Hwei-Chung Liu for excellent technical support in obtaining the electron micrographs.