Previous studies have tested gene replacement therapy in RPE65-deficient dogs using recombinant adeno-associated virus 2/2 (rAAV2/2), -2/1 or -2/5 mediated delivery of the RPE65 gene. They all documented restoration of dark-and light-adapted electroretinography responses and improved psychophysical outcomes. Use of a specific RPE65 promoter and a rAAV vector that targets transgene expression specifically to the RPE may, however, provide a safer setting for the long-term therapeutic expression of RPE65. Subretinal injection of rAAV2 pseudotyped with serotype 4 (rAAV2/4) specifically targets the RPE. The purpose of our study was to evaluate a rAAV2/4 vector carrying a human RPE65cDNA driven by a human RPE65 promoter, for the ability to restore vision in RPE65 À/À purebred Briard dogs and to assess the safety of gene transfer with respect to retinal morphology and function. rAAV2/4 and rAAV2/2 vectors containing similar human RPE65 promoter and cDNA cassettes were generated and administered subretinally in eight affected dogs, ages 8-30 months (n ¼ 6 with rAAV2/4, n ¼ 2 with rAAV2/2). Although fluorescein angiography and optical coherence tomography examinations displayed retinal abnormalities in treated retinas, electrophysiological analysis demonstrated that restoration of rod and cone photoreceptor function started as soon as 15 days post-injection, reaching maximal function at 3 months postinjection, and remaining stable thereafter in all animals treated at 8-11 months of age. As assessed by the ability of these animals to avoid obstacles in both dim and normal light, functional vision was restored in the treated eye, whereas the untreated contralateral eye served as an internal control. The dog treated at a later age (30 months) did not recover retinal function or vision, suggesting that there might be a therapeutic window for the successful treatment of RPE65 À/À dogs by gene replacement therapy.

We previously described chimeric recombinant adeno-associated virus (rAAV) vectors 2/4 and 2/5 as the most efficient vectors in rat retina. We now characterize these two vectors carrying the CMV.gfp genome following subretinal injection in the Wistar rat, beagle dog, and cynomolgus macaque. Both serotypes displayed stable GFP expression for the duration of the experiment (6 months) in all three animal models. Similar to the AAV-2 serotype, AAV-2/5 transduced both RPE and photoreceptor cells, with higher level of transduction in photoreceptors, whereas rAAV-2/4 transduction was unambiguously restricted to RPE cells. This unique specificity found conserved among all three species makes AAV-2/4-derived vectors attractive for retinal diseases originating in RPE such as Leber congenital amaurosis (RPE65) or retinitis pigmentosa due to a mutated mertk gene. To provide further important preclinical data, vector shedding was monitored by PCR in various biological fluids for 2 months post-rAAV administration. Following rAAV-2/4 and -5 subretinal delivery in dogs (n = 6) and in nonhuman primates (n = 2), vector genome was found in lacrymal and nasal fluids for up to 3-4 days and in the serum for up to 15-20 days. Overall, these findings will have a practical impact on the development of future gene therapy trials of retinal diseases.