Functional and Structural Recovery of the Retina after Gene Therapy in the RPE65 Null Mutation Dog

Abstract

purpose. To assess the efficacy of AAV-mediated gene therapy to restore vision in a large number of RPE65^−/− dogs and to determine whether systemic and local side effects are caused by the treatment.

methods. Normal RPE65 dog cDNA was subcloned into an rAAV vector under control of a cytomegalovirus promoter, and an AAV.GFP control vector was also produced with the titers 2 × 10¹² particles/mL and 2 × 10¹⁰ transducing U/mL, respectively. RPE65^−/− dogs, aged 4 to 30 months were treated with subretinal injections of the AAV.RPE65 and control vectors, respectively, in each eye, and three 24- to 30-month-old normal control dogs with the latter. Baseline and postoperative systemic and ophthalmic examinations, blood screenings, vision testing, and electroretinography (ERG) were performed. Two RPE65^−/− dogs were killed at 3 and 6 months after treatment for morphologic examination of the retinas.

results. RPE65^−/− dogs were practically blind from birth with nonrecordable or low-amplitude ERGs. Construct injections or sham surgeries were performed in 28 eyes; 11 were injected subretinally with the AAV.RPE65 construct. ERGs at 3 months after surgery showed that in the latter eyes, dark-adapted b-wave amplitudes recovered to an average of 28% of normal, and light adapted b-wave amplitudes to 32% of normal. ERG amplitudes were not reduced during a 6- to 9-month follow-up. No systemic side effects were observed, but uveitis developed in nine AAV.RPE65-treated eyes. No uveitis was observed in the eyes treated with the control vector. Immunocytochemistry showed expression of RPE65 in the retinal pigment epithelium (RPE) of AAV.RPE65-treated eyes. Fluorescence microscopy showed expression of green fluorescent protein (GFP) in the RPE and, to a lesser extent, in the neural retinas of AAV.GFP-treated eyes. Ultrastructurally, a reversal of RPE lipid droplet accumulation was observed at the AAV.RPE65 transgene injection site, but not at the site of injection of the control vector.

conclusions. In 10 of 11 treated RPE65^−/− eyes, gene transfer resulted in development of vision, both subjectively apparent by loss of nystagmus, and objectively recorded by ERG. Structurally, there was reversal of lipid droplet accumulation in the RPE. Uveitis developed in 75% of the transgene-treated eyes, a complication possibly due to an immunopathogenic response to the RPE65 molecule.

The RPE65 gene encodes a 61-kDa microsomal protein expressed almost exclusively in the retinal pigment epithelium (RPE). ^{1 2} The protein is associated with the smooth endoplasmic reticulum ¹ and is necessary for the synthesis of the 11-cis chromophore of photoreceptor cell visual pigments. ³ In the RPE65 knockout mouse, 11-cis retinoids are not formed, and there is an accumulation of all-trans retinyl esters in RPE. ³ This results in early-onset depression of the normal electroretinographic (ERG) responses and a slow but progressive retinal degenerative process. ^{1 4}  

Mutations in the human RPE65 gene underlie some forms of autosomal early childhood blindness including Leber congenital amaurosis and early-onset severe retinal dystrophy. ^{5 6 7 8} In addition, an autosomal recessively inherited retinal degenerative disease has been described clinically and morphologically in a strain of Swedish Briard dogs. ^{9 10 11} The molecular defect that underlies this canine disorder is a 4-bp deletion in the RPE65 gene. ^{12 13} There are close similarities between the clinical characteristics of the diseases resulting from RPE65 gene mutations in humans and in the dog. ¹⁴ The dog therefore is a valuable model for human retinal dystrophy. 

Dogs with the RPE65 null mutation are congenitally night blind, with severe visual deficits in daylight. Most, but not all cases show a rapidly oscillating nystagmus. Their resting pupillary size is slightly larger than in normal dogs. ¹⁰ ERGs are diagnostic at the age of 5 weeks. There are no, or barely recordable, scotopic responses and photopic ERG recordings are usually low, most clearly observed in 30-Hz flicker recordings. Ophthalmoscopic examination reveals normal fundus appearance until approximately 3 years of age, when there is a generalized vascular attenuation and a slight paling of the fundus. In some affected older animals grayish to white spots have been observed, mainly in the central tapetal and nontapetal fundus. The disease is slowly progressive, and morphologic studies have shown large lipoid-like inclusions in the RPE in the central fundus primarily, spreading peripherally with age. There is also disorganization of photoreceptor outer segments, followed by degenerative changes and later by rod and then cone loss, with a gradient going from the peripheral retina to the central parts with increasing age. ^{11 15}  

Recently, it has been reported that gene transfer using an rAAV-RPE65 construct in 3 RPE65 null mutation dogs restores functional vision. ¹⁶ The present study was undertaken to assess further the potential efficacy of AAV-mediated gene therapy in ameliorating the effects of the RPE65 gene mutation in dogs. 

Materials and Methods

Production and Characterization of rAAV Constructs

Total RNA was isolated from the RPE of a carrier Briard dog, by using extraction reagent (RNAzol B; Teltest, Friendswood, TX) and reverse transcribed with a kit (Retroscript; Ambion, Austin, TX). The normal RPE65 cDNA was amplified, by using oligonucleotide primers specific for canine RPE65 (forward 5′-TCC CCG CGG CTC GAG ATG TCC ATC CAA GTG GAG CAT C-3′, and reverse 5′-CCA TCG ATC TCT AGA TTA GGA TTT TTT GAA CAG TCC-3′) and subcloned into a cloning vector (pBluescript; Stratagene, La Jolla, CA) using restriction sites included in the primers. This construct was sequenced by automated dideoxy sequencing on (model CEQ 2000; Beckman Coulter, Fullerton, CA), to verify the authenticity of the RPE65 clone. The verified RPE65 cDNA was then subcloned into the pCI vector (Promega, Madison, WI) with the human cytomegalovirus promoter (hCMV) on its 5′ end and the late SV40 polyadenylation signal (PolyA) on its 3′ end. The pCI vector was transiently transfected into Cos-7 cells. After 48 hours, cells were harvested and analyzed for expression of RPE65 by immunoblot analysis. The cassette containing the hCMV, RPE65 and PolyA was removed and subcloned into the plasmid AAV2 vector, pSSV9, ¹⁷ between the two inverted terminal repeats (ITRs). A control construct was made by replacing the RPE65 cDNA with the green fluorescent protein (GFP) cDNA. The resultant subclones, pAAV.CMV.RPE65 and pAAV.CMV.GFP were verified by restriction analysis. Cesium chloride gradient-purified pAAV.CMV.RPE65 and pAAV.CMV.RPE65 DNAs were transfected into human embryonic kidney 293 (HEK293 cells; ATCC, Manassas, VA). The pAAV.CMV.GFP-transfected cells were examined for GFP expression by fluorescence microscopy, the pAAV.CMV.RPE65-transfected cells were harvested at 48 hours after transfection, and expression of RPE65 was assessed by Western blot analysis. The excision and replication of the recombinant AAV (rAAV) DNA from the two constructs was verified by Hirt analysis. ¹⁸ The same batch of DNA was then used to produce rAAV constructs. Large-scale production of the rAAV constructs for use in the preclinical trials on dogs was performed by heparin-affinity chromatography at the Vector Core Facility of University of North Carolina Gene Therapy Center (Chapel Hill, NC), according to published methods. ¹⁹ No specific quality control studies were performed on the AAV constructs. The titers of the rAAV.CMV.GFP and rAAV.CMV.RPE65 were 2 × 10¹⁰ transducing units (TU)/mL and 2 × 10¹² particles/mL, respectively. The rAAV viral constructs, resuspended in PBS, were stored at −70°C and thawed immediately before the subretinal injections. 

Animals

Twelve dogs homozygous for the RPE65 null mutation, aged 4 to 30 months, were included in the study, with three 24- to 30-month-old normal control dogs (Table 1) . All dogs underwent clinical baseline studies, including visual behavioral testing, ophthalmic examinations, and ERGs before the treatment. Similar postsurgery follow-up studies were performed at regular intervals. Two dogs were killed at 3 and 6 months after the initial gene transfer treatment, and their eyes used for morphologic studies. 

Fourteen of the dogs used for this study were Briard beagle dogs and one was an unrelated normal Labrador retriever. All procedures were approved by the Animal Care and Use Committee at the University of Missouri-Columbia. The investigation conformed to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

Ophthalmic Studies

Behavioral tests included assessment of the ability to navigate a maze and to follow a strong beam of light on the wall in a dimly lit room. Pupillary light reflexes were tested and indirect ophthalmoscopy (Welch Allyn, Skaneateles Falls, NY) and slit lamp biomicroscopy (model SL14; Kowa, Tokyo, Japan) were performed. Intraocular pressure was monitored by applanation tonometry (Tono-pen XL; Medtronic Solan, Jacksonville, FL). Fundus photographs were obtained using either a hand-held fundus camera (RC2; Kowa) or a confocal scanning laser ophthalmoscope (SLO; Retina Angiograph; Heidelberg Engineering, Heidelberg, Germany), the latter equipped with an argon blue laser (488 nm) and an infrared laser (835 nm) for fluorescence and fundus visualization, respectively. 

Surgical and Postoperative Treatment

All surgeries were performed under general anesthesia induced with propofol (6 mg/kg intravenously; Diprivan 1%; Astra Zeneca Pharmaceutical, LP, Wilmington, DE). The dogs were intubated, and the anesthesia was maintained with isoflurane (Abbott Laboratories, North Chicago, IL). Pupils were dilated with 1% atropine sulfate (Bausch & Lomb Pharmaceutical, Inc., Tampa, FL) and phenylephrine hydrochloride (Mydfrin 2.5%; Alcon Laboratories, Inc., Fort Worth, TX) eye drops. 

In most cases, both eyes were treated with subretinal injections: one eye with rAAV.RPE65 and the contralateral eye with rAAV.GFP. In total, 11 eyes were treated with subretinal injections of rAAV.RPE65. One eye was treated with intravitreous injection of the rAAV.RPE65 vector. Fourteen eyes were injected with rAAV.GFP in the subretinal space and one eye with rAAV.GFP in the vitreous cavity. Sham surgeries were performed in two eyes, using balanced salt solution (BSS; Alcon Laboratories Inc.) for the subretinal injections. Two eyes, contralateral to rAAV.RPE65-treated eyes, were not operated on. Thirty to 100 mL of any construct was injected subretinally and 100 to 200 mL of rAAV.GFP or rAAV.RPE65 was injected intravitreously (Table 1) . 

A lateral canthotomy was performed, and the conjunctiva and Tenon’s capsule were dissected to obtain access to the sclera. Two sclerotomies were performed on the temporal side of the eye, approximately 4 mm apart and 6 to 8 mm from the limbus. A fiberoptic light was inserted into the vitreous cavity through one sclerotomy and a custom-made micropipette through the other and guided toward the fundus under direct visualization through an operating microscope and a flat lens on the cornea. Vitrectomy was not performed. For subretinal injections, the inferonasal central part of the neuroretina was perforated, and a bleb was obtained with the injection of one of the constructs. The neuroretinal detachment encompassed approximately 25% to 30% of the total fundus area. After the subretinal deposition the micropipette and light guide were withdrawn and the sclerotomies sutured using 7-0 nylon (Vicryl; Ethicon, Piscataway, NJ). The conjunctiva, Tenon’s capsule and subcutaneous tissues were sutured with 7-0 nylon sutures and 5-0 silk sutures were used for skin closure. A subconjunctival injection of 2 mg/mL dexamethasone (Decadron Phosphate; 1 mg in 0.25 mL; Butler, Columbus, OH), was administered after surgery. 

Topical antibiotics (fucidic acid, Fucithalmic 1%; Lovens Kemiska Fabrik, Ballerup, Denmark) twice daily, and cycloplegics (atropine sulfate 1%; Bausch & Lomb Pharmaceutical) given twice daily, were instilled after surgery, usually for the first 10 days. Systemic treatment with 0.5 to 1 mg/kg corticosteroids (prednisone 20 mg; Roxane Laboratories Inc., Columbus, OH), in a single intradermal dose was routinely used for 2 weeks after surgery, and then tapered down. Topical treatment with anti-inflammatory agents was also given, (prednisolone acetate suspension 1%; Alcon Laboratories Inc.) three times daily during the first 2 weeks or longer, if necessary. 

Electroretinography

Retinal function was tested using simultaneous bilateral full-field flash ERG before surgery and two to six times after surgery in each dog, during a 6- (eight dogs) or 9-month (three dogs) follow-up period, except for one dog (Snoopy), for which the follow-up was 3 months. The first postoperative ERGs were performed 4 to 6 weeks after surgery. Before the recordings were performed, the dogs were dark adapted for 2 hours. General anesthesia, as described earlier, was induced in the dark under dim red light. Bilateral ERGs were obtained with a commercial Ganzfeld ERG system (ERG System TOR; Scanditronix, Uppsala, Sweden). Jet lenses (Universo Plastique, Grenchen, Switzerland) were cushioned on the cornea with methylcellulose (Methocel 2%; Ciba Vision, Munich, Germany) and platinum subdermal needle electrodes (Grass Instrumental Division; Astro-Med. Inc., West Warwick, RI) were used. The references were placed subcutaneously at the base of each ear and the ground at the occiput. The dogs were laid on their chests during the recordings and stay sutures were used to position the globes and eyelids. Scotopic ERGs were recorded with stimuli at −2.0-log cd-s/m² (low-intensity scotopic stimuli) and 0.6 log cd-s/m² (high-intensity scotopic stimuli). The dogs were then light adapted for 10 minutes by using white background light (37 cd/m²). Photopic ERG recordings were then obtained at 5.1, 30.1, and 50.1 Hz, using white light stimuli at 0.0 log cd-s/m². 

Study of Systemic and Local Effects of Treatment

General clinical examinations were performed from birth and regularly throughout the study. Serum chemistry profiles, complete blood counts, urinalyses, and serum and urine creatinine and protein ratios were determined before treatment and at monthly intervals thereafter. To determine B- and T-cell immune function, blood was collected after surgery for lymphocyte blastogenesis (LBT) testing using three different mitogens (concanavalin A [ConA], pokeweed mitogen [PWM], and phytohemagglutinin [PHA]) and repeated at biweekly intervals in all dogs included in the study. Mean test results were calculated from three replicate trials. Pre- and postvaccination titers (against parvo and adenovirus) were also measured as an additional test of immune competency. Data were analyzed by ANOVA with each dog serving as its own control. ELISA was performed in serum samples from all treated dogs and the two controls to measure antibodies against rAAV. 

Regarding the local ophthalmic effects, pre- and postoperative eye examinations were performed on daily during the first 2 weeks after surgery and thereafter once weekly for at least 4 weeks. Thereafter, the dogs were examined regularly as appropriate. If intraocular inflammatory responses were found (uveitis), these were treated symptomatically. Both systemic and topical anti-inflammatory medications were then administered in higher doses and frequency, as individually indicated. 

Morphologic Studies

Two of the dogs (Snoopy and Odie) were killed at 3 and 6 months after treatment, respectively. Eyes were enucleated, transected at the ora ciliaris retinas, and immediately placed in an electron microscopy (EM) fixative consisting of 1.25% glutaraldehyde, 2% paraformaldehyde, 0.13 M Na-cacodylate, and 0.13 mM CaCl₂ (pH 7.4). With the eyes immersed in the fixative, the corneas, irises, and lenses were removed. The vitreous was then extracted from each eye cup using a microvitrectomy instrument (Premier Microvit; Storz, St. Louis, MO). Immediately after removal of the vitreous, the eye cups were dissected to obtain regions both within and outside the areas that had been treated with the AAV vectors. For EM, one piece of each region was returned to the initial fixative and incubated at room temperature with gentle agitation for at least 2 hours. For fluorescence microscopy and immunohistochemistry, another piece of each region was transferred to a fixative consisting 4% paraformaldehyde, 50 mM sodium cacodylate, 8% sucrose (pH 7.4). The latter samples were incubated at room temperature with gentle agitation for 1 hour and then washed twice in 0.17 M sodium cacodylate (pH 7.4). 

Samples for electron microscopy were washed in 0.17 M sodium cacodylate (pH 7.4), postfixed with osmium tetroxide, and embedded in epoxy resin. Thin sections of the embedded tissues were cut and stained with uranyl acetate and lead citrate. The sections were examined and photographed with an electron microscope (model 1200EX; JEOL, Tokyo, Japan). 

For fluorescence microscopy and immunohistochemistry, tissue samples were embedded in optimal cutting temperature medium (Tissue Tek; Sakura Finetek, Torrance, CA) and were frozen on dry ice. To visualize GFP fluorescence, sections of the frozen tissues were cut at a thickness of 8 μm. The sections were mounted on glass slides in 0.17 M sodium cacodylate (pH 7.4) and were examined and photographed with a photomicroscope (Axiophot; Zeiss; Thornwood, NY) equipped with epi-illumination. Fluorescent emissions were stimulated with light from a 50-W high-pressure mercury vapor source. Examination and photography of the specimens were performed using an objective (Plan-Neofluor; Zeiss) with a 1.30 numerical aperture, a 395 to 440-nm bandpass exciter filter, an chromatic beam splitter (FT 460; Zeiss), and a barrier filter (LP 515; Zeiss). Photography was performed using ASA 100 film (EliteChrome; Eastman Kodak, Rochester, NY). 

For RPE65 immunolocalization, sections of the frozen tissue were cut at a thickness of 8 μm. The sections were mounted on slides (Superfrost/Plus; Fisher Scientific, Fair Lawn, NJ) and allowed to dry overnight. The slides were then washed for 15 minutes in phosphate-buffered saline (PBS: 137 mM NaCl, 2.7 mM KCl, 10 mM Na₂HPO₄, and 1 mM KH₂PO₄ [pH 7.4]) followed by incubation for 1 hour in blocking solution (0.5% bovine serum albumin, 0.2% Tween 20, 1 mg/mL sodium azide, 5% normal goat serum in PBS). After removal of the blocking solution, the sections were covered with an anti-RPE65 antibody that had been diluted 1:400 in the blocking solution. The sections were incubated with the primary antibody at 4°C overnight. After removal of the primary antibody solution, the sections were washed five times with PBS and were then incubated for 3 hours at room temperature in goat anti-rabbit IgG conjugated to a fluorescent dye (Alexa 568; Molecular Probes). For immunolabeling, the secondary antibody was diluted 1:150 in blocking solution. After incubation in the secondary antibody, the sections were washed with PBS and then water. The slides were air dried, and coverslips were mounted over the sections (Gel Mount; Biomedia Corp., Foster City, CA). Examination and photography of the sections were performed with fluorescence microscopy, as described for GFP visualization, except that the filters used consisted of a commercial fluorescence filter set (Rhodamine X; Chroma Technology Corp., Brattleboro, VT). This filter set transmits excitation light of 560 to 580 nm and emission light of 590 to 650 nm. 

Results

Characterization of rAAV Constructs

The integrity of pAAV.CMV.RPE65 and pAAV.CMV.GFP was verified by restriction enzyme analysis (data not shown). Strong GFP expression was detected in the cytoplasm of pAAV.CMV.GFP-transfected HEK293 cells from day 1 after transfection (Fig. 1A) . Western blot analysis of cell lysates demonstrated the presence of a protein band (approximately 61 kDa in size) corresponding to the size of the canine RPE65 protein in pAAV.CMV.RPE65-transfected HEK293 cells but not in pAAV.GFP-transfected HEK293 cells (Fig. 1B) . Hirt analysis showed the presence of 4- and 8-kb DNA bands (monomer and dimer replicative intermediates) on hybridization with a ³²P-labeled RPE65 probe, suggesting the successful replication of rAAV DNA (data not shown). 

Intraocular Surgery

Construct injection or sham surgery was performed without complication at the time of surgery in 28 eyes. Subretinal blebs covering 25% to 30% of the fundus were usually obtained. These blebs rapidly reduced in size and were usually not visible within 24 to 48 hours after surgery. In one case (Faithful), in which the bleb nearly encircled the optic nerve head, the retina was not fully reattached until after 6 weeks. 

Ophthalmic and Behavioral Studies

All RPE65^−/− dogs appeared to be blind from birth. As early as 4 to 5 weeks, when normal puppies start to use visual cues, it was apparent that the RPE65^−/− dogs were using other senses, such as hearing and smell, to compensate for severe visual impairment. At this age, the external aspects of the eyes were normal, but at age 6 to 7 weeks, bilateral nystagmus developed. 

Approximately 10 weeks after the subretinal treatment with rAAV-RPE65, the nystagmus disappeared in all 11 affected dogs. The nystagmus was, however, still prevalent in the dog (Checkers) that was treated by intravitreous rAAV.GFP injection, and in two RPE65^−/− dogs that were not included in the present study but were littermates to some of these 11 dogs. The treated fundi were mainly normal appearing except for in the 2.5-year-old dog (Bonus), in which there were bilateral slight generalized vascular attenuation and minute whitish specks in the central nontapetal fundus (Fig. 2A) . SLO clearly demonstrated the neuroretinal penetration site (Fig. 2B) and fluorescence photography showed an area of distinctive fluorescence in the left eye of this dog 3 months after treatment with rAAV.GFP (Fig. 2C) . 

Behavioral studies showed definite improvement of both day and night vision in 10 of the 11 dogs treated with subretinal injections of the rAAV.RPE65 gene construct. The one exception (Faithful) did not appear to have any increased visual capacity. In the 10 responsive dogs, vision appeared better in daylight than under dim light conditions. Preferential looking was observed from the side of the AAV.RPE65-treated eye. All 10 dogs appeared to be able to follow a strong beam of light on the wall in a dimly lit room. whereas Faithful continued to bump into objects both during light- and dark-adapted conditions and did not appear to see the light on the wall. This dog’s retina did not reattach within the normal time (described earlier) and in addition, uveitis developed 6 days after surgery, which subsided after intensive systemic and topical anti-inflammatory treatment. 

ERG Findings

Dark-adapted low- and high-intensity ERGs were mainly nonrecordable before treatment in the RPE65^−/− dogs. In the light-adapted state, however, low-amplitude responses were observed in some of the dogs, including single flash b-wave (2/12 dogs) and 30.1-Hz (6/12 dogs), but no 50.1-Hz, flicker responses. 

In all rAAV.RPE65 eyes treated with subretinal injections, the ERG amplitude responses improved. Statistically significant differences between AAV.RPE65-treated eyes and the contralateral control eyes were observed in the first six treated dogs (Snoopy, Odie, Perdita, Millie, Jip, and Buddy). Also, ERG b-wave thresholds were close to normal in these dogs (Bragadottir R, Lei, Narfström K, ARVO abstract 4596, 2002). Low- and high-intensity scotopic ERG responses were obtained in 9 of 11 AAV.RPE65, subretinally treated dogs. Photopic single-flash and 30- and 50-Hz flicker recordings were recorded in 8 of the 11 dogs. Table 2 illustrates the mean amplitudes of the affected dogs (n = 11) treated with subretinal injections of AAV.RPE65, before and after gene delivery at 10 to 12 weeks after surgery in all dogs. The dark-adapted b-wave maximum amplitudes recovered to an average of 28% of normal, and the light-adapted b-wave maximum amplitudes to 32% of normal. After the initial improvement, ERG responses were not reduced in the dogs available for the 6- or 9-month follow-up period (Figs. 3B 3C) . 

Systemic and Local Effects of Gene Transfer

All dogs were systemically healthy after surgery. There were no significant differences in serum chemistry profiles, complete blood counts, urinalyses, and urine protein and creatinine ratios before and after surgery (P > 0.05) in any of the dogs. Mild lymphopenia developed 1 month after surgery in the first six dogs that were treated with subretinal AAV.RPE65 injections but the pre- and postoperative differences were not statistically significant (P = 0.3387) (Caro ML, Estes DM, Cohn LA, Narfström K, ARVO Abstract 4594, 2002). The LBT results indicated normal T- and B-cell function (Fig. 4) . All dogs had positive postvaccination titers. ELISA indicated a twofold increase in antibody titer against rAAV. 

In the 28 surgically treated eyes, 9 showed postoperative intraocular inflammatory reactions (uveitis) on postoperative days 2 to 6 (Table 3) . 

Uveitis developed only in eyes treated with AAV.RPE65, which was graded from 1 to 4, depending on severity, of which one case (grade 4) was refractory to treatment (Fig. 5) . The other eight cases subsided after 4 to 12 weeks of systemic and/or topical anti-inflammatory treatment. 

Morphology

Expression of Recombinant Proteins.

In eyes of affected dogs treated with the rAAV.GFP vector, strong GFP fluorescence was observed in the RPE throughout the region of the retina where a subretinal bleb had been created during vector injection (Fig. 6A) . In some areas of this region, GFP expression was also observed in cells in the neural retina (Fig. 6B) . No GFP fluorescence was detected in either the RPE or the neural retina in regions of the eyes outside of the vector injection site. 

In the eye of the affected dog Snoopy that was treated with the rAAV.RPE65 vector, RPE65 immunofluorescence was observed predominantly in the RPE across much of the region of the retina where the vector had been injected (Fig 7A) . No distinct immunolabel was detected in the neural retina. In areas of the eyes outside the injection site, no RPE65 immunofluorescence was observed in either the RPE or the neural retina (Fig. 7B) . 

Ultrastructural Effects of rAAV-RPE65 Treatment.

In both affected dogs that were killed, the eye treated with the rAAV.GFP vector in the subretinal space, contained numerous large electron-lucent inclusions in the RPE. These inclusions had the characteristic features of lipid droplets. The inclusions were abundant within the treated region of the retina (Fig. 8A) , as well as outside these regions. In the eye of one dog (Snoopy) treated with the rAAV.RPE65 vector, the lipid droplets virtually disappeared from the area of the retina where the subretinal bleb had been created during vector injection (Fig. 8B) . However, in regions outside the injection site, the RPE still contained large numbers of these inclusions in both dogs that were examined (Fig. 9) . The disappearance of lipid droplets from the AAV.RPE65-treated region did not occur in the dog Odie. In both rAAV.GFP- and rAAV.RPE65-treated retinas, small degenerate remnants of the photoreceptor outer segments were present within and outside the vector injection site (Fig. 8A) . Treatment with rAAV.RPE65 did not result in any apparent recovery of outer segment morphology (Fig. 8B) at the 3- and 6-month follow-up ultrastructural study. In the AAV.RPE65-treated eyes, outer segment morphology was similar within and outside the treated regions. In addition, there was no appreciable difference in the appearance of the outer segments between the AAV.RPE65-and the AAV.GFP-treated eyes within the same animal (Fig. 5) . 

Discussion

In the present study, gene transfer in the RPE65^−/− dog by using an rAAV.RPE65 construct resulted in effective transduction of the RPE and reversed the disease-specific accumulation of lipid droplets in the RPE. Postoperative ERG recordings demonstrated a remarkable improvement of retinal function. With the normal canine RPE65 molecule introduced into the RPE cell of the mutant, it appears that, in the treated part of the retina, a normal visual cycle ²⁰ was restored. Throughout repeated clinical studies, including follow-up ERG recordings, retinal function improved as early as 4 weeks after treatment and appeared to persist for at least 9 months. The persistence of long-term function is an important new finding and shows that the transgene treatment has a stable therapeutic effect. Further electrophysiological and morphologic studies are currently under way for critical evaluation and correlation of the functional and structural properties of photoreceptors in rAAV.RPE65-treated eyes. 

The present study protocol did not allow us to judge whether rod and/or cone function profited from the AAV treatment, an issue recently raised in RPE65-knockout mice. ⁴ The specific improvement in daylight visual behavior and the marked recovery of ERG photopic responses, including those of 50.1-Hz flicker recordings in many of the dogs, suggest that cones may also have functionally improved. However, it cannot be excluded that the local functional recovery caused an overall increase in the sensitivity of previously desensitized rod responses in the treated area and possibly also in the surrounding area. 

Despite the improvement in retinal function indicated by ERG results in the dog Snoopy, no recovery in photoreceptor outer segment morphology was observed in this dog at 3 months after gene transfer, even though the accumulation of lipid droplets in the adjacent RPE was completely reversed. The ERG responses observed after treatment must therefore have been mediated solely by the degenerated photoreceptor outer segments present in the retina. The absence of apparent photoreceptor outer segment morphologic recovery in this dog may have been related to the persistent uveitis in the AAV.RPE65-treated eye. The uveitis was unilateral, however. Planned morphologic analysis of other treated dogs that showed substantial long-term recovery of retinal sensitivity with only transient or no uveitis should help determine whether uveitis may play a role in inhibiting restoration of normal outer segment morphology. 

The phenotype in this disease may vary between individuals but appears to be unusually advanced in both Snoopy and Odie, who were littermates and whose eyes were used for the postoperative ultrastructural studies at 3 and 6 months, respectively. Photoreceptor outer segments of both the AAV.RPE65- and AAV.GFP-treated eyes of the dogs degenerated severely, both within and outside the treated regions of the eye. This degeneration was most likely directly due to the genetic defect and not to the treatments. 

It remains to be determined whether eventual morphologic recovery of the photoreceptor outer segments accompanies correction of the molecular defect in the adjacent RPE and whether the amplitude of the long-term ERG recovery correlates with some degree of morphologic recovery of the photoreceptors. Given the relatively rapid turnover rate of the photoreceptor outer segments in normal animals, it was surprising that reversal of the disease phenotype in the RPE was not accompanied by recovery of outer segment morphology. It appears that there is a disease-related inhibition of outer segment morphogenesis that takes a long time to reverse. Functional recovery of the attenuated outer segments appeared to occur quickly, once the ability of the RPE to supply 11-cis retinoids was restored. This suggests that the abnormal-appearing outer segments in RPE65^−/− dogs have a normal complement of opsin and visual transduction cascade proteins. Outer segment morphogenesis, by contrast, is a complex process, the regulation of which is not completely understood. The regulatory mechanisms that control outer segment morphogenesis may respond very slowly to restored 11-cis retinoid availability. It will be of great interest to determine whether the dogs that showed persistent ERG recovery in the AAV.RPE65-treated eyes for over 9 months show any improvement in outer segment morphogenesis. 

The visual capacity of dogs is not easily judged objectively, especially in laboratory animals. We mainly used the maze test in our studies to test the dogs’ actual behavior and ability to avoid colliding with objects. We complemented this evaluation by using a strong light (from a portable slit lamp) directed toward the wall in a dimly lit room. With the dog sitting on a table, this test demonstrated clearly whether the dog visualized the moving light. The general behavior of the mutated dogs also gave a clue to visual capacity, which changed after gene transfer. Before treatment, the affected dogs were extremely cautious and did not direct their eyes toward visual cues. After gene transfer they became more alert and showed a distinct interest in their surroundings, such as a moving light. 

Nystagmus was found to appear at 6 to 7 weeks of age in all RPE65^−/− dogs but disappeared in both eyes approximately 10 weeks after gene transfer, even though the corrective construct was injected unilaterally. In contrast, this did not occur in an RPE65^−/− littermate that was injected intravitreously or in two untreated affected littermates. It appears that the newly induced visual processes in a localized region of the eye were enough to alleviate the nystagmus. Rapidly oscillating nystagmus is prevalent in children born with, or acquiring during the first year of age, severely reduced visual capacity due to various causes. It is not found, however, in completely blind children (Jakobsson P, Linkoping University Hospital, Sweden, personal communication, 2002). It appears that the nystagmus in the mutated dogs started at a time when mental and visual processes had matured and the dogs were struggling to see, perhaps with light perception only or severely reduced vision in bright light. The nystagmus disappeared approximately 10 weeks after subretinal gene transfer, indicating that the dogs’ visual capacity was increased. 

The RPE65^−/− dogs that were treated subretinally with the normal transgene spanned a wide age range: 4 to 30 months. Previous ultrastructural studies in RPE65 null mutation dogs have shown that there is a slowly progressive degeneration of the photoreceptors ^{11 15} observable as early as the age of 4 months. It was therefore surprising to find functional recovery in the older dogs, in which photoreceptor loss ought to be significant. It was obvious, however, that the photoreceptors that remained in the area of gene transfer regained function after treatment. Long-term follow-up morphologic studies are needed to investigate whether the degenerative disease process is halted by the rAAV.RPE65 injection or the apoptosis continues once it has been initiated, even after functional correction. 

Our data showed that uveitis developed in 75% of the rAAV.RPE65-treated eyes. Because neither sham surgery nor rAAV.GFP-treated eyes showed development of uveitis, the rAAV.RPE65 construct or the RPE65 component of the construct is likely to be the causative agent. Six cases were mild to moderate in inflammatory reaction but two cases were severe. One of these was refractory to treatment. This is a serious complication to the otherwise successful gene transfer strategy in treatment of the RPE65 null mutation dogs. It should be noted that, as in most experimental animal studies, no specific quality-control studies were performed for the AAV preparations. We can therefore not fully exclude the presence of a contaminant in one of the viral vectors. The uveitis may also be the result of an immune response to the RPE65 protein. Due to the mutation in the dogs, their immune systems would never have been exposed to this protein, so an immune tolerance for it would not have developed. RPE65 protein levels have not been measured in the eyes from normal, affected, or treated affected dogs. However, it is possible that the transduced RPE cells overexpress the protein, because the transgene is not under normal regulatory control. Although the RPE65 protein does not appear to be secreted from the RPE in normal animals, overproduction may result in leakage of the protein into the bloodstream where it would be exposed to the immune system and could be recognized as a foreign protein. The resultant immune response could be responsible for the observed uveitis. Variability in the severity of the uveitis may be related to viability between animals in the amount of RPE65 genes expressed. Analyses of anti-RPE65 antibody production in the treated dogs should indicate whether an immune reaction to this protein could underlie the uveitis. 

So far, at least eight ocular-specific molecules have been reported to be immunopathogenic. These include S-antigen (S-Ag), interphotoreceptor retinoid-binding protein (IRBP), rhodopsin, recoverin, phosducin, and melanin-associated antigen. ²¹ S-Ag and IRBP, especially, have been used for induction of experimental autoimmune uveitis in rats and serve as models for uveitic conditions in humans. Recently, it was shown that RPE65 is highly uveitogenic in three of four inbred strains of rat studied. ²² This finding suggests that this molecule could be involved in pathogenic autoimmunity also in the canine eye. This canine complication has implications for future clinical trials related to severe early-onset childhood blindness and Leber congenital amaurosis, as well as for clinical trails for gene therapy for many other inherited disorders. The absence of uveitis in some of the successfully treated eyes in this study indicates that this complication can be avoided. It is possible that by optimizing the degree of RPE65 gene expression achieved in the therapeutic treatment, restoration of visual function can be achieved without an inflammatory response. 

 Supported by the Foundation Fighting Blindness and Research to Prevent Blindness, Inc.

 Submitted for publication June 17, 2002; revised September 5 and October 9, 2002; accepted November 2, 2002

 Disclosure: K. Narfström, None; M.L. Katz, None; R. Bragadottir, None; M. Seeliger, None; A. Boulanger, None; T.M. Redmond, None; L. Caro, None; C.-M. Lai, None; P.E. Rakoczy, None

 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.

 

Table 1.

 

Dogs Included in the Study, Phenotype, and Age at Surgery and Death

Dog	Age at Surgery	Genotype	AAV.RPE65-Treated Eye	AAV.GFP-Treated Eye	BSS-Treated Eye	Age at Death
Milly	4 months	RPE65−/−	OD
Jip	4 months	RPE65−/−	OD	OS
Rex	4 months	RPE65−/−	OD	OS
Candy	4 months	RPE65−/−	OD	OS
Faithfull	4 months	RPE65−/−	OD	OS
Checkers	4 months	RPE65−/−	OD-intravitreous	OS
Buddy	4 months	RPE65−/−	OS
Odie	8 months	RPE65−/−	OD	OS		14 months
Snoopy	8 months	RPE65−/−	OD	OS		11 months
Perdita	1 year	RPE65−/−	OD	OS
Lady	1 year	RPE65−/−	OD	OS
Bonus	2.5 years	RPE65−/−	OD	OS
Jonna	2 years	RPE65−/+		OD	OS
Lufsen	2 years	RPE65+/+		OD	OS
Alonzo	2 years	RPE65+/+		OD

Figure 1.

 

(A) Fluoromicrograph showed expression of GFP in the HEK293 cells at 24 hours after transfection with pAAV.CMV.GFP. (B) Western blot analysis demonstrated expression of the canine RPE65 protein in cell lysates from HEK293 cells transfected with pAAV.CMV.RPE65 (lane 1), pAAV.CMV.GFP (lane 2), and untransfected control (lane 3).

Figure 2.

 

Scanning laser ophthalmoscope (SLO) images of the eyes of Bonus, 2.5 years old at the time of surgery. (A)Whitish specks (arrows) were visible in the central nontapetal fundus. (B) The subretinal penetration site (arrow) was visible in the inferior nasal fundus near a circular unrelated scar. (C) AAV.GFP fluorescence was seen in the area of the subretinal injection.

Table 2.

 

ERG Amplitudes before Surgery and at Follow-up

	Preoperative			10–12 Week Follow-up			Normal
		RPE65	Control	RPE65	Control
Scotopic b-wave (low intensity)	9.1 ± 1.0	8.8 ± 0.7	21.2 ± 4.9	5.1 ± 1.2	93 ± 39
Scotopic a-wave (high intensity)	5.1 ± 1.5	7.7 ± 1.8	19.3 ± 5.0	2.8 ± 0.6	85 ± 21
Scotopic b-wave (high intensity)	17.6 ± 4.8	15.1 ± 4.5	41.2 ± 9.4	3.9 ± 1.0	149 ± 26
Photopic a-wave	1.2 ± 0.4	1.2 ± 0.4	4.6 ± 1.4	2.0 ± 0.1	14 ± 2
Photopic b-wave	4.6 ± 1.3	3.9 ± 0.8	5.7 ± 1.3	2.2 ± 0.1	18 ± 4
30-Hz flicker	3.9 ± 0.5	3.3 ± 0.5	9.3 ± 1.7	2.0 ± 0	26 ± 6
50-Hz flicker	4.8 ± 1.0	6.3 ± 2.0	4.9 ± 1.3	3.9 ± 0.5	13 ± 3

Figure 3.

 

Actual, bilateral full-field electroretinographic (ERG) recordings in a control dog Jonna (A) and in an RPE65 null mutation dog Candy at 10-week (B) and 9-month (C) follow-ups. Top traces: right eye; bottom traces: left eye. Top two recordings: high- and low-intensity scotopic responses, respectively; bottom two recordings: photopic responses (responses to light-adapted single-flash and 30-Hz flicker stimuli, respectively).

Figure 4.

 

LBT with three different mitogens in all RPE^−/− dogs included in the study. Mean (n = 3 replicates) stimulation indices for each mitogen: (A) concanavalin A (Con A), (B) pokeweed mitogen (PWM), and (C) phytohemagglutinin (PHA) after treatment. Vertical bars indicate ±1 SD.

Table 3.

 

Ocular Complications

Dog	Dose(μL)	Uveitis	Onset (duration, days)	Severity
Odie	50*	+	4 (3)	1
Snoopy	60*	+	3 (90, then death)	4
Perdita	100*	−
Milly	100*	+	6 (3), 25 (3)	2
Buddy	60*	−
Jip	30*	−
Lady	70*	+	6 (4)	2
Rex	100*	+	4 (6)	2
Candy	95*	+	5 (5)	2
Checkers	200†	+	2 (4)	1
Bonus	100*	+	2 (3)	1
Faithful	60*	+	6 (31)	3

Figure 5.

 

Light microscopy of the retina of (A) the left control eye and (B) the AAV.RPE65-injected eye, 3 months after treatment in the RPE65^−/− dog Snoopy, showing (A) lipoid inclusions (arrows) in the retinal pigment epithelium (B) and the absence of lipoid inclusions but also the deposition of inflammatory cells (arrows) at the level of the internal limiting membrane, due to ongoing severe uveitis in this eye.

Figure 6.

 

Fluorescence micrographs of the retina from Snoopy, taken from the region of the eye treated with the AAV.GFP vector 3 months previously. Abundant GFP fluorescence was present in the RPE across the entire treated region of the retina (A, arrows). In some areas of the treated region, GFP-expressing cells were also observed in the neural retina (B, arrowheads).

Figure 7.

 

Fluorescence micrographs of the retina of Snoopy, taken from the eye treated with the AAV.RPE65 vector 3 months previously and immunostained for RPE65 protein localization. (A) The RPE in the area of the retina at the vector injection site showed abundant RPE65 immunostaining (A, arrow). No immunostaining was observed in the adjacent neural retina. (B) In areas of the eye outside of the injection site, no RPE65 immunostaining was observed in either the RPE (B, arrow) or the adjacent neural retina.

Figure 8.

 

Electron micrographs of the RPE and adjacent retina from the eyes of Snoopy. (A) Region of the retina from the left eye treated with the AAV.GFP vector. (B) Region of the retina from the right eye of the same dog treated with the AAV.RPE65 vector. Large lipid droplets (L) were abundant in the RPE of the left eye treated with the control vector, but these were absent in the AAV.RPE65-treated RPE. Despite the reversal of lipid droplet accumulation by AAV.RPE65 treatment in the right eye, photoreceptor outer segments (arrows) remained at the same degenerative stage in both eyes.

Figure 9.

 

Electron micrograph of the RPE from the right eye of Snoopy, treated with the AAV.RPE65 vector taken from a region of the eye outside the vector injection site. Lipid droplets (L) remained abundant in the RPE outside the treated region.

The authors thank Ginny Dodam and Jenny Garland for excellent technical assistance, Howard Wilson for valuable help with the graphics, and the University of North Carolina Gene Therapy Vector Core facility of R. Jude Samulski, PhD, for providing the SSV9 plasmid.