Development of a Gene Therapy Vector for RDH12-Associated Retinal Dystrophy

Abstract

Early-onset severe retinal dystrophy (EOSRD) is a genetically heterogeneous group of diseases resulting in serious visual disability in children. A significant number of EOSRD cases, often diagnosed as Leber congenital amaurosis (LCA13), are associated with mutations in the gene encoding retinol dehydrogenase 12 (RDH12). RDH12 is a member of the enzyme family of short-chain dehydrogenases/reductases. In the retina, RDH12 plays a critical role in reducing toxic retinaldehydes generated by visual cycle activity that is required for the light response of the photoreceptor cells. Individuals with RDH12 deficiency exhibit widespread retinal degeneration impacting both rods and cones. Although Rdh12-deficient (Rdh12^−/−) mice do not exhibit retinal degeneration, functional deficits relevant to visual cycle function can be demonstrated. In the present study, we describe the development and preclinical testing of a recombinant adeno-associated viral (rAAV) vector that has the potential for use in treating EOSRD due to RDH12 mutations. Wild-type and Rdh12^−/− mice that received a subretinal injection of rAAV2/5 carrying a human RDH12 cDNA driven by a human rhodopsin-kinase promoter exhibited transgene expression that was stable, correctly localized, and did not cause retinal toxicity. In addition, administration of the vector reconstituted retinal reductase activity in the retinas of Rdh12^−/− mice and decreased susceptibility to light damage associated with Rdh12 deficiency, thus demonstrating potential therapeutic efficacy in an animal model that does not exhibit a retinal degeneration phenotype. These findings support further efforts to develop gene replacement therapy for individuals with RDH12 mutations.

Introduction

The visual pigments rhodopsin and cone opsins use the vitamin A analog 11-cis retinal to absorb light and initiate signaling responses by the photoreceptor cells.¹ Light causes 11-cis retinal to isomerize to all-trans retinal that stabilizes the signaling conformations of the visual pigments. Following the decay of these species, all-trans retinal is released from bleached rhodopsin and cone opsins and recycled to replenish the supply of 11-cis retinal chromophore. Although essential for phototransduction, the chemical nature of the vitamin A-dependent visual cycle creates the potential for cellular toxicity. When visual cycle reactions are inefficient or disrupted, retinaldehydes and retinaldehyde-condensation products can accumulate, resulting in damage to the photoreceptor cells and retinal pigment epithelium.^2–4 This can occur as a result of mutations affecting key visual cycle proteins associated with inherited retinal degeneration or as a result of the cumulative effects of light exposure that may contribute to the pathology of age-related macular degeneration.⁵

Retinol dehydrogenase 12 (RDH12), also known as short-chain dehydrogenase/reductase family 7C member 2, belongs to a family of enzymes that use NADPH to reduce multiple substrates, including 11-cis retinaldehyde and all-trans retinaldehyde.⁶ Although widely expressed outside the eye, in the retina, RDH12 localizes specifically to the inner segments of rod and cone photoreceptor cells.^7,8 Loss-of-function mutations in the gene encoding RDH12 are estimated to cause 3.4–10.5% of autosomal recessive retinal dystrophy diagnosed as Leber congenital amaurosis (LCA) or early-onset severe retinal dystrophy (EOSRD).^9–12 At young ages, children with RDH12-associated retinal dystrophy exhibit severely decreased visual acuity, visual field constriction, and loss of recordable electroretinogram (ERG) responses, consistent with observed macular atrophy and peripheral pigment accumulation.^11,13,14

Unlike individuals with biallelic RDH12 mutations, Rdh12-knockout (Rdh12^−/−) mice exhibit normal retinal histology without apparent retinal degeneration even at advanced ages.^7,8,15 Although overall retinoid levels in Rdh12^−/− and wild-type mice are similar, retinal homogenates from the knockout mice have decreased capacity to reduce both all-trans retinal and 11-cis retinal and have increased levels of the lipofuscin component A2E.⁸ Rdh12^−/− mice also exhibit increased light damage susceptibility.^7,16 Thus, despite the lack of a retinal degeneration phenotype, Rdh12^−/− mice exhibit quantitative measures of RDH12 loss-of-function suitable for use in evaluating therapeutic outcomes and pertinent to the disease in affected individuals.

Recent advances highlight the promise of adeno-associated virus (AAV)–mediated gene replacement therapy for the treatment of retinopathies caused by loss-of-function mutations.¹⁷ Studies focused on another visual cycle gene, encoding the retinoid isomerohydrolase RPE65, were the first to show restoration of visual function in affected individuals following subretinal administration of AAV2-based vectors carrying RPE65 cDNA.^18–21 A recombinant AAV2/2-RPE65 vector, with the trade name Luxturna™, recently became the first FDA-approved gene therapy for an inherited disease.²² For RPE65 mutations,^23,24 vision loss and retinal degeneration result from inability of the RPE to synthesize the chromophore of the visual pigments, 11-cis retinal.²⁵ RPE65 has served as an exemplary test case for therapeutic strategies designed to restore visual function by increasing chromophore levels, including pharmacological intervention.²⁶

In contrast, loss-of-function mutations affecting genes necessary for the recycling phase of the visual cycle do not decrease chromophore levels but instead lead to increased levels of toxic retinoid compounds.²⁷ In the case of mutations in RDH12, widespread photoreceptor cell death at young ages likely results from increased exposure to retinaldehydes and short-chain aldehydes,²⁸ due to loss of retinal reductase activity in photoreceptor inner segments.²⁹ Accordingly, studies of therapeutic intervention in RDH12-associated retinal dystrophy have the potential to broadly inform the development of therapeutic strategies to improve the recycling phase of the visual cycle.

To initiate studies of the potential of AAV-mediated gene therapy for RDH12 loss-of-function, we generated a recombinant AAV2/5 vector carrying human RDH12 cDNA under the control of a human photoreceptor cell–specific promoter. Conditions were established that achieved stable expression of the RDH12 transgene that did not result in retinal damage and restored retinal reductase activity in Rdh12^−/− mice. These findings represent an important advance in efforts to establish the feasibility of gene replacement therapy for inherited retinal degeneration caused by RDH12 mutations.

Materials and Methods

Recombinant adeno-associated viral vector construction and purification

Human RDH12 was amplified from human retinal cDNA by polymerase chain reaction (PCR) using primers designed to encompass the entire RDH12 coding region, and the product was cloned and sequenced as described previously.⁹ The RDH12 cDNA was inserted into the multiple cloning site of the parental pAAV-hGRK1p.hrGFP vector, and the resulting pAAV-hGRK1p.hRDH12 construct was packaged into AAV. The AAV2/5 pseudotyped vector was generated by bipartite transfection: (1) AAV vector plasmid encoding the gene of interest and (2) AAV helper plasmid encoding AAV Rep proteins from serotype 2 and Cap proteins from serotype 5, and adenovirus helper functions into 293T cells. The transfection and purification were performed using a protocol as published.^30,31

Two days after transfection, cells were lysed by repeated freeze and thaw cycles. After initial clearing of cell debris, the nucleic acid component of the virus producer cells was removed by Benzonase treatment. The recombinant AAV vector particles (rAAV2/5-hGRK1p.hRDH12) were purified by affinity chromatography using an AVB Sepharose column (GE Healthcare), washed in 1 × phosphate-buffered saline (PBS), and concentrated to a volume of 100–150 mL using Vivaspin 4 (10 kDa) concentrators (Vivaproducts).

Vector titers were determined using quantitative real-time PCR amplification (qPCR).³² Endotoxin contamination of plasmids, crude vector lysates, and purified vectors was measured by Pyrotell-T kinetic turbidimetric assessment (Associates of Cape Cod) following the manufacturer's protocols. Endotoxin levels were below 2.5 EU/mL, which is lower than the U.S. FDA and European Pharmacopoeia specifications per dose.³³ Protein contamination was assessed by SYPRO™ Ruby (Thermo Fisher) staining of protein blots. A total of 10¹¹ viral particles, as determined by qPCR, were run on a 10% polyacrylamide gel together with a reference sample and a size marker, until clear separation between the 75 and 100 kDa bands was achieved. Gels were fixed in 50% methanol and 7% acetic acid for 30 min, stained overnight with 0.1% SYPRO^TM Ruby in fixation solution, and washed in 10% methanol and 7% acetic acid for 30 min. Proteins were imaged under UV illumination.

Animals and subretinal injections

All experimental procedures complied with the regulations of the University of Michigan Institutional Animal Care and Use Committee (IACUC) and conformed to the guidelines on the care and use of animals adopted by the Association for Research in Vision and Ophthalmology (Rockville, MD). C57BL/6J and BALB/cJ mice were obtained commercially (The Jackson Laboratory). Rdh12^−/− mice were generated as described previously and bred on mixed C57BL/6J background to generate pigmented animals homozygous for Rpe65-Met⁴⁵⁰ or on a BALB/cJ background to generate albino animals homozygous for Rpe65-Leu⁴⁵⁰.¹⁵ All mouse strains were confirmed to be Crb1^rd8 negative.³⁴

The mice were maintained under cyclic light conditions (12-h light–dark, <20 lux). Young adult mice of both genders in equal numbers were used for injections. Mice were placed under general anesthesia with an intraperitoneal injection of ketamine (80 mg/kg; Par Pharmaceuticals) and xylazine (10 mg/kg; VetOne). Pupils were dilated with topical phenylephrine (2.5%; Paragon BioTeck) and tropicamide (1.0%; Akorn). Under an ophthalmic surgical microscope, a small incision was made through the cornea adjacent to the limbus using the tip of a 30-gauge hypodermic needle. A 34-gauge blunt needle fitted to a gas-tight Hamilton syringe was inserted through the incision behind the lens and pushed through the retina. Mice received a subretinal injection of 1.3–2 μL each to produce bullous retinal detachment in the superior or inferior hemisphere within the nasal quadrant and covering approximately one-third of the retina.

The rAAV2/5-hGRK1p.hRDH12 vector and PBS controls were administered to one eye of each animal and the other eye was noninjected. For initial optimization, a range of vector concentrations was tested, resulting in doses of 1.3 × 10⁷ to 1.5 × 10⁹ viral genomes (vg). For analysis of outcomes, the doses used are indicated in the figure legends. Cohorts consisted of three to eight mice at 5–8 weeks old, and studies were performed at least in triplicate. Mice were euthanized by anesthesia overdose by intraperitoneal injection of ketamine (180 mg/kg) and xylazine (20 mg/kg), followed by bilateral pneumothorax.

Immunohistochemistry and immunoblotting

For immunohistochemistry, mice were euthanized and eyes were scored for orientation and enucleated. For freeze-substitution³⁵ processing of rAAV2/5-hGRK1p.hRDH12-injected eyes and controls, whole eyes were flash-frozen in dry ice-cooled isopentane for 30 s, transferred to dry ice-cooled methanol containing 3% glacial acetic acid, incubated at −80°C for 48 h and then overnight at −20°C, and embedded in paraffin. Sections (6 μm) were obtained and deparaffinized, and antigens retrieved in 1 mM EDTA, 0.05% Tween 20, pH 8.0, at 90°C for 30 min. Sections were permeabilized with PBS plus 0.3% Triton X-100 (PBS-T); blocked with 1% bovine serum albumin, 10% normal goat serum in PBS-T; incubated with primary antibodies overnight at 4°C; and then incubated with fluorophore-conjugated secondary antibodies for 1 h at room temperature. Washed sections were coverslipped using ProLong Gold gel mount containing 4′,6-diamidino-2-phenylindole (DAPI; Thermo Fisher) and imaged using a Leica DM6000 fluorescence microscope.

For immunoblotting, retinal homogenates were prepared from enucleated eyes in hypotonic lysis buffer containing 0.1% Triton X-100 and protease inhibitors, then subjected to sodium dodecyl sulfate–polyacrylamide electrophoresis, and transferred onto nitrocellulose. The membranes were blocked, incubated with primary antibody overnight, washed, incubated with alkaline phosphatase–conjugated secondary antibody, washed, and developed using 5-bromo-4-chloro-3′-indolyphosphate p-toluidine and nitro-blue tetrazolium chloride (Sigma).

Primary antibodies used were: a mouse anti-RDH12 monoclonal (2C9) specific for the human protein¹⁵; a rabbit anti-Rdh12 polyclonal (CSP) specific for the mouse protein¹⁵; a mouse anti-RHO monoclonal (1D4)³⁶; a rabbit anti-RHO polyclonal³⁷; a rabbit anti-red/green cone opsin polyclonal (Millipore); and a mouse anti-GAPDH monoclonal (Thermo Fisher).

Optical coherence tomography

Mice were anesthetized and pupils dilated as described for subretinal injections. Systane lubricant eye drops (Alcon) were given for corneal hydration. Optical coherence tomography (OCT) was performed with a spectral domain OCT system (Bioptigen Envisu R2200 SD-OCT system). The camera lens was adjusted to acquire images from eyes without physical contact. A volume analysis centered on the optic nerve head was performed. The volume size was 1.4 × 1.4 mm.

Quantification of retinoid content

Levels of all-trans retinal and 11-cis retinal in mouse eyes were determined, as described previously.¹⁵ In brief, mice at 6 weeks postinjection were dark-adapted overnight, transferred to dim red light, euthanized via anesthesia overdose, and the eyes were enucleated and frozen in liquid N_2. Each eye was homogenized in 1 mL chloroform:methanol:hydroxylamine (2 M, 3:6:1) on ice, sonicated with a microtip probe (30 times for 1 s) on ice, and then incubated at room temperature for 2 min. Chloroform (200 μL) and water (240 μL) were added and mixed by vortexing, and the phases separated by centrifugation. The lower phases were collected, dried under argon, and dissolved in hexane.

Retinoids were identified and quantified by normal-phase high-performance liquid chromatography (HPLC) using a Waters Alliance separation module and photodiode array detector with a Supelcosil LC-31 column (25 cm × 4.6 mm × 3 μm) developed with 5% 1,4-dioxane in hexane. Peaks were identified by comparison to retention times of standard compounds and evaluation of wavelength maxima and quantified by comparison of peak areas at 347 and 351 nm for syn and anti 11-cis retinal oxime, respectively (generated from 11-cis retinal; National Eye Institute), and at 357 and 361 nm for syn and anti all-trans retinal oxime, respectively (generated from all-trans retinal; Sigma).

Retinal reductase assays

Mouse retina homogenates were assayed for retinal reductase activity at 6 weeks postinjection, as described previously.⁸ Briefly, light-adapted mice were euthanized, eyes enucleated, and each dissected retina was homogenized individually in 125 μL of 0.25 M sucrose, 25 mM Tris-acetate, pH 7, 1 mM dithiothreitol. The homogenates were centrifuged at 1,000 g for 5 min to remove unbroken cells, and the supernates were sonicated with a microtip probe (30 times for 1 s) on ice. Protein concentrations were determined by a modification of the Lowry procedure,³⁸ and RDH12 expression was evaluated by Western analysis. Similar samples were pooled from three to five mice, and 20 μg of each pooled lysate was assayed in triplicate in buffer containing 200 μm all-trans retinal and 200 μm NADPH in HEPES buffer (pH 7). Reactions were incubated for 0–45 min at 37°C, and retinoids were extracted in chloroform–methanol. All-trans-retinol formation was quantitated using normal-phase HPLC analysis with comparison to standards (Sigma), as described previously.

ERG recordings

ERG analysis was performed as described previously using the Espion e2 recording system (Diagnosys).³⁹ In brief, mice were anesthetized and pupils dilated as described for subretinal injections. Body temperature was maintained at 37°C with a heating pad. Corneal ERGs were recorded from both eyes using gold wire loops with 0.5% tetracaine topical anesthesia and a drop of 2% methylcellulose for corneal hydration. A gold wire loop placed in the mouth was used as reference, and a ground electrode was on the tail.

The ERG protocol consisted of recording dark-adapted (scotopic) responses to brief white flashes (−2.31 log cd.s.m⁻² for rod isolated B-waves; 1.09 log cd.s.m⁻² for rod–cone combined response). Light-adapted (photopic) ERGs were recorded after 10 min of adaptation to a white 32 cd.m⁻² rod-suppressing background in response to 1.09 log cd.s.m⁻² intensity flashes. Responses were amplified at 1,000 gain at 1.25 to 1,000 Hz and digitized at a rate of 2,000 Hz. A notch filter was used to remove 60 Hz line noise. Responses were computer-averaged and recorded at 3- to 60-s intervals depending upon the stimulus intensity.

Light-induced retinal damage

Mice received a subretinal injection of rAAV2/5-hGRK1p.hRDH12, or an equal volume of PBS, and contralateral eyes were not injected. Beginning 6 weeks later, ERG analysis was performed and scotopic responses were quantified as described previously. The following week, mice were dark-adapted overnight, their pupils dilated with tropicamide (1.0%), and then were placed in a light box in individual clear trays. The mice were exposed to 5,000 lux for 2 h and then were returned to vivarium housing (12 h light–dark) for 7 days, after which ERG analysis was repeated. The percent ERG response remaining after light damage was calculated for each eye, and the averages for each cohort were calculated.

Statistical analysis

Graphed data show mean values ± standard error of the mean (Prism; GraphPad). For analysis of retinoid content, after performing D'Agostino–Pearson omnibus normality test and Bartlett's test for equal variance, comparison among multiple groups was made using one-way analysis of variance. For analysis of the RDH activity, after subjecting the data to F-test to determine the appropriate t-test parameters, comparison of initial rates was made using two-tailed t-tests. For analysis of ERG amplitudes, comparison of the responses in treated eyes and untreated eyes was made using two-tailed paired t-tests. In all tests, significance is indicated as: *p < 0.05; **p < 0.01; ***p < 0.001.

Results

rAAV2/5-mediated expression of human RDH12 in mouse photoreceptors

Our efforts to develop an RDH12 gene therapy vector focused on the use of AAV2 pseudotyped with AAV5 capsids (AAV2/5), as vectors of the AAV5 serotype have previously been shown to mediate efficient transduction of photoreceptor cells in the murine and primate retina.^40,41 A human RDH12 cDNA was cloned into an AAV2 construct downstream of a human rhodopsin kinase (GRK1) promoter that drives photoreceptor-specific expression (Fig. 1A).^42–44 After packaging with AAV5 capsid, the rAAV2/5-hGRK1p.hRDH12 vector was produced in transduced cells, purified, and assessed on protein gels (Fig. 1B). Wild-type (C57BL/6J) and Rdh12^−/− (knockout) mice at 5–8 weeks of age received a subretinal injection of rAAV2/5-hGRK1p.hRDH12 at a range of doses (1.3 × 10⁷ to 1.5 × 10⁹ vg).

Figure 1.

Expression of human RDH12 in mice that received subretinal injection of rAAV2/5-hGRK1p.hRDH12. (A) Schematic of pAAV-hGRK1p.hRDH12 in which a human RDH12 cDNA is cloned downstream of a human rhodopsin kinase (GRK1) promoter and in between inverted terminal repeat sequences derived from the AAV2 genome. (B) SYPRO™ Ruby-stained gel of purified AAV2/5 preparations showing VP1, 2, and 3 capsid proteins. Lanes: L, Bio-Rad Precision Plus Dual Color Standard; 1, AAV2/5 reference sample (10¹⁰ vg); 2, rAAV2/5-hGRK1p.hRDH12 (10¹¹ vg). (C, D) RDH12 expression at 6 weeks after subretinal injection of rAAV2/5-hGRK1p.hRDH12 (1.4 × 10⁹ vg), evaluated using antibodies specific for mouse Rdh12 (mRdh12) or human RDH12 (hRDH12), as indicated. (C) Western analysis of retinal lysates from noninjected C57BL/6J mice (anti-mRdh12), rAAV2/5-hGRK1p.hRDH12-injected Rdh12^−/− mice (anti-hRDH12), and PBS-injected Rdh12^−/− mice (anti-hRDH12). GAPDH was a loading control. (D) Immunohistochemical analysis of endogenous mouse Rdh12 (red; anti-mRdh12) shows localization to the IS, ONL, and OPL in retinal sections from C57BL/6J mice, but not in Rdh12^−/− mice. Immunohistochemical analysis of recombinant human RDH12 (green; anti-hRDH12) resulting from injection of rAAV2/5-hGRK1p.hRDH12 shows localization to the IS, ONL, and OPL in both C57BL/6J and Rdh12^−/− mice. Immunoreactivity in the OPL corresponds to RDH12 present in the inner most region of the photoreceptor inner segments. Dots of immunoreactivity scattered across the inner retina represent cross-reactivity with blood vessels seen in nonperfused tissue with some preparations of anti-hRDH12. DAPI staining of nuclei (blue). Phase-contrast images are shown to the left. Images of retina sections from noninjected mice evaluated by immunohistochemical analysis with anti-hRDH12 are shown to the right. Scale bars, 50 μm. DAPI, 4′,6-diamidino-2-phenylindole; GCL, ganglion cell layer; hGRK1, human rhodopsin kinase promoter; hRDH12, human RDH12 cDNA; INL, inner nuclear layer; IPL, inner plexiform layer; IS, inner segments; ITR, inverted terminal repeat; ONL, outer nuclear layer; OPL, outer plexiform layer; OS, outer segments; PBS, phosphate-buffered saline; polyA, Simian virus 40 polyadenylation signal; rAAV, recombinant adeno-associated virus; RDH12, retinol dehydrogenase 12; RPE, retinal pigment epithelium; SD/SA, Simian virus 40 splice donor/splice acceptor site; vg, viral genomes.

At 6 weeks postinjection of 1.4–1.5 × 10⁹ vg, Western blots of retinal proteins using an antibody specific for human RDH12 (2C9) showed stable expression of the recombinant protein in injected Rdh12^−/− mice, whereas an antibody specific for mouse Rdh12 (CSP) detected the endogenous mouse protein in noninjected wild-type mice (Fig. 1C). In mice that received the same dose of rAAV2/5-hGRK1p.hRDH12, immunohistochemical analysis of human RDH12 expression showed colocalization with endogenous mouse Rdh12 in photoreceptor inner segments in injected wild-type mice and a similar pattern of expression in injected Rdh12^−/− mice (Fig. 1D).

OCT analysis of mice that received a subretinal injection of rAAV2/5-hGRK1p.hRDH12 showed minor injection site scarring, but no significant alterations in laminar structure across a distance of 1,000 μm on both sides of the optic nerve head (Fig. 2A). Immunohistochemical analysis of rhodopsin and red/green opsin showed no evidence of mislocalization or decreased expression indicative of photoreceptor toxicity. Co-staining confirmed their correct localization relative to RDH12 in both rod and cone photoreceptor cells of wild-type and Rdh12^−/− mice that received a subretinal injection of rAAV2/5-hGRK1p.hRDH12 (Fig. 2B).

Figure 2.

Expression of human RDH12 does not perturb mouse retinal structure or visual pigment localization. (A, B) C57BL/6J and Rdh12^−/− mice received a subretinal injection of rAAV2/5-hGRK1p.hRDH12 (1.5 × 10⁹ and 1.4 × 10⁹ vg, respectively) in one eye, and the other eye was not injected. (A) OCT analysis of retinal microstructure. Representative images from injected and noninjected contralateral eyes taken 6 weeks postinjection are shown. (B) Immunohistochemical analysis of rhodopsin and cone opsin expression in retinal sections from injected and noninjected contralateral eyes evaluated 16 weeks postinjection. Human RDH12 (green) and mouse rhodopsin (red) (left-side panels). Human RDH12 (green) and mouse red/green opsin (red) (right-side panels). DAPI staining of nuclei (blue). Scale bars, 50 μm. OCT, optical coherence tomography.

rAAV2/5-mediated expression of human RDH12 restores retinal reductase activity in Rdh12^−/− mouse retinas

Although Rdh12^−/− mice do not exhibit a retinal degeneration phenotype, the ability of their retinas to reduce exogenous retinaldehydes is significantly decreased,⁸ with residual activity attributed to the presence of other RDH isoforms.^6,45 To assay the retinal reductase activity in mice that received a subretinal injection of rAAV2/5-hGRK1p.hRDH12, relative to that in controls, retinal homogenates were assayed using HPLC analysis to quantitate the conversion of exogenous all-trans retinal to all-trans retinol (Fig. 3A). Retinas from Rdh12^−/− mice that received a subretinal injection of rAAV2/5-hGRK1p.hRDH12 (1.4 × 10⁹ vg) exhibited an average initial rate of all-trans retinol formation of 0.046 pmol min⁻¹ μg protein⁻¹, which was significantly increased relative to the average for noninjected Rdh12^−/− mice (0.013 pmol min⁻¹ μg protein⁻¹) and PBS-injected Rdh12^−/− mice (0.016 pmol min⁻¹ μg protein⁻¹) (p < 0.01). The average initial rate of all-trans retinol formation in noninjected wild-type control mice was 0.071 pmol min⁻¹ μg protein⁻¹. Since the injection bleb covered an estimated one-fourth to one-third of the retina (Fig. 3B), we estimated the level of the recombinant human protein in the treated area of retina to be, on average, approximately twice that of the endogenous mouse protein.

Figure 3.

rAAV2/5-hGRK1p.hRDH12 increases RDH12 activity in Rdh12-deficent mice. (A) HPLC analysis of retinal reductase activity in retinas from noninjected C57BL/6J mice, and from Rdh12^−/− mice that received a subretinal injection of rAAV2/5-hGRK1p.hRDH12 (1.3 × 10⁹ vg) or PBS, or were noninjected. At 6 weeks postinjection, all-trans retinol formation was quantitated in assays of retinal homogenates containing added NADPH and all-trans retinal. Each data point represents the mean ± standard error for a minimum of five independent experiments in which retinas from three to five mice were pooled and assayed in triplicate. , C57BL/6J; , noninjected Rdh12^−/−; , PBS-injected Rdh12^−/−; , rAAV2/5-hGRK1p.hRDH12-injected Rdh12^−/−. **p < 0.01. (B) Immunohistochemical analysis of human RDH12 expression in a whole eye cross section from an Rdh12^−/− mouse at 16 weeks rAAV2/5-hGRK1p.hRDH12 postinjection. Scale bar, 500 μm. Human RDH12 (green). DAPI stained nuclei (blue). HPLC, high-performance liquid chromatography.

rAAV2/5-mediated expression of human RDH12 does not significantly reduce chromophore levels in Rdh12^−/− mouse retinas

The capacity of RDH12 to reduce both all-trans retinal and the chromophore 11-cis retinal suggests that overexpression or mislocalization of human RDH12 has the potential to negatively impact visual cycle function. HPLC analysis of retinoid content was used to evaluate visual cycle function in Rdh12^−/− mice that received a subretinal injection of rAAV2/5-hGRK1p.hRDH12 (1.4 × 10⁹ vg), as well as in PBS-injected and noninjected wild-type controls (Fig. 4A). The levels of 11-cis retinal (mean ± standard error) in retinas of noninjected wild-type mice, and Rdh12^−/− mice from each treatment group, were not significantly different (C57BL/6J, 547 ± 11 pmol/eye; Rdh12^−/− noninjected, 503 ± 29 pmol/eye; Rdh12^−/− rAAV2/5-hGRK1p.hRDH12-injected, 453 ± 16 pmol/eye; Rdh12^−/− PBS-injected, 455 ± 26 pmol/eye) (p > 0.05) (Fig. 4B). In contrast, the levels of all-trans retinal were significantly decreased only in vector-injected Rdh12^−/− mice (C57BL/6J, 30 ± 5 pmol/eye; Rdh12^−/− noninjected, 44 ± 5 pmol/eye; Rdh12^−/− rAAV2/5-hGRK1p.hRDH12-injected, 29 ± 4 pmol/eye; Rdh12^−/− PBS-injected, 36 ± 4 pmol/eye) (p < 0.05) (Fig. 4C). Taken together, these findings show no evidence of visual cycle disruption resulting from rAAV2/5-hGRK1p.hRDH12-mediated expression of human RDH12 in mice and instead document a modest decrease in all-trans retinal potentially consistent with an in vivo therapeutic effect.

Figure 4.

rAAV2/5-hGRK1p.hRDH12 does not significantly decrease steady-state levels of 11-cis retinal. HPLC analysis of retinoids in retinas from noninjected C57BL/6J mice, and from Rdh12^−/− mice that received a subretinal injection of rAAV2/5-hGRK1p.hRDH12 (1.4 × 10⁹ vg) or PBS, or were noninjected. At 6 weeks postinjection, following overnight dark adaptation, retinoids were extracted under dim-red light and quantified by HPLC analysis. (A) Representative chromatograms from each treatment condition: peaks for syn 11-cis retinal oxime, anti 11-cis retinal oxime, and syn all-trans retinal oxime are indicated. For groups of 10–14 mice for each treatment condition, total levels of (B) 11-cis retinal and (C) all-trans retinal were quantified and plotted as the mean ± standard error. *p < 0.05.

Visual function is maintained in mice expressing human RDH12

ERG of retinal electrical activity elicited by light stimuli of varying intensities was used to evaluate retinal function in wild-type and Rdh12^−/− mice at various times after a subretinal injection of rAAV2/5-hGRK1p.hRDH12 (1.4 × 10⁹ vg) or PBS. Analysis of scotopic and photopic ERG responses showed that expression of human RDH12 did not result in significant decreases in rod or cone function relative to noninjected eyes at 6 weeks postinjection (Fig. 5). Small decreases in ERG amplitudes detected in both wild-type and Rdh12^−/− mice at 6 weeks postinjection are likely due to the effects of retinal detachment. At times up to 54 weeks postinjection, there were no significant differences in ERG magnitudes between noninjected and rAAV2/5-hGRK1p.hRDH12-injected eyes of wild-type mice, or between noninjected and rAAV2/5-hGRK1p.hRDH12-injected eyes of Rdh12^−/− mice, as determined by pairwise comparisons (p > 0.10) (Supplementary Fig. S1). A decrease in ERG amplitudes over time was observed in all mice, reflecting the consequences of aging. Both wild-type and Rdh12^−/− mice that received a subretinal injection of rAAV2/5-hGRK1p.hRDH12 exhibited robust transgene expression and preserved retinal structure at 54 weeks postinjection (Supplementary Fig. S2A), with no evidence of mislocalization of rod and cone opsins indicative of photoreceptor cell death (Supplementary Fig. S2B). OCT analysis at 54 weeks postinjection showed good preservation of retinal microstructure (Supplementary Fig. S2C).

Figure 5.

rAAV2/5-hGRK1p.hRDH12 does not inhibit retinal function. Scotopic (rod-isolated and combined rod–cone) and photopic (cone-mediated) ERG responses recorded from noninjected C57BL/6J mice, and from Rdh12^−/− mice at 6 weeks postinjection of rAAV2/5-hGRK1p.hRDH12 (1.4 × 10⁹ vg) in one eye. Representative ERGs from each treatment group are shown. ERG, electroretinogram.

Human RDH12 protects against light damage in Rdh12^−/− mice

Previous studies have shown that Rdh12^−/− mice exhibit increased susceptibility to light damage.⁷ Increased light damage resulting from RDH12 deficiency is likely to be specifically linked to increased retinaldehydes and other short-chain aldehydes.^16,29 It is also known that a polymorphism in the mouse gene encoding Rpe65, the retinoid isomerohydrolase that generates 11-cis retinal, is a determining factor in susceptibility to light damage, which is greater in mice homozygous for leucine at amino acid residue 450 (Rpe65-Leu450) than in mice homozygous for methionine (Rpe65-Met450).⁴⁶

For studies of light damage susceptibility, Rdh12^−/− mice were bred onto the albino BALB/cJ background to generate knockout mice homozygous for Rpe65-Leu450. BALB/cJ and albino Rdh12^−/− mice received a subretinal injection of rAAV2/5-hGRK1p.hRDH12 (1.4 × 10⁹ vg) in one eye and received no treatment in the contralateral eye. At 6 weeks postinjection, ERG analysis of retinal activity was performed 1 week before, and 1 week after, subjecting the mice to light levels that cause significant retinal damage in the albino animals (5,000 lux for 2 h).

Under photopic conditions, B-wave amplitudes obtained after light damage were variable for both BALB/cJ and albino Rdh12^−/− mice. In contrast, under scotopic conditions, the percentage of rod-isolated and combined rod–cone activity remaining in eyes of albino Rdh12^−/− mice that received a subretinal injection of rAAV2/5-hGRK1p.hRDH12 was significantly greater than that remaining in untreated eyes (p ≤ 0.001) (Fig. 6). In addition, the eyes of BALB/cJ mice that received a subretinal injection of rAAV2/5-hGRK1p.hRDH12 did not exhibit significantly greater activity than that remaining in untreated eyes (p ≥ 0.2275). The very high level of significance obtained in pairwise comparisons of rod and rod–cone function remaining in treated and untreated Rdh12^−/− mice provides strong evidence of a protective role of rAAV2/5-hGRK1p.hRDH12 in reducing light damage susceptibility associated with RDH12 deficiency.

Figure 6.

rAAV2/5-hGRK1p.hRDH12 gene replacement therapy reduces light damage susceptibility in albino Rdh12-deficent mice. ERG analysis was performed, before and after exposure to 5,000 lux for 2 h, on wild-type BALB/cJ and albino Rdh12^−/− mice in which one eye received a subretinal injection of rAAV2/5-hGRK1p.hRDH12 (1.4 × 10⁹ vg), and the contralateral eye was not injected. Scotopic (rod-isolated and combined rod–cone) and photopic responses were quantified for groups of 10–13 mice. The percentage of the initial response remaining after light damage was calculated for injected and noninjected eyes. Mean outcomes ± standard error are shown for (A) albino Rdh12^−/− mice and (B) wild-type BALB/cJ mice, as well as the significance of the differences between injected and noninjected eyes calculated using two-tailed paired t-test analysis: ***p < 0.001, *p < 0.05.

Discussion

Multiple factors identify RDH12-associated retinal dystrophy as a compelling target for gene therapy, including its autosomal recessive mode of inheritance, prevalence in early-onset disease, severe disease burden, and effect on a well-defined biological pathway. However, a significant challenge for establishing therapeutic efficacy in preclinical studies is the absence of a retinal degeneration phenotype in Rdh12^−/− mice that is similar to that caused by RDH12 mutations in patients. Thus, our studies were designed to evaluate functional outcome measures that directly reflect the activity of the RDH12 transgene. Administration of rAAV2/5-hGRK1p.hRDH12 to Rdh12^−/− mice by subretinal injection was shown to result in stable expression of RDH12 that localizes to the photoreceptor cells, increases in vivo retinal reductase activity, and decreases light damage sensitivity. These findings provide strong support for the potential of gene supplementation to reconstitute RDH12 function in photoreceptor cells and decrease retinal degeneration in individuals with EOSRD due to RDH12-mutations.

As a member of the family of short-chain dehydrogenases/reductases, RDH12 exhibits broad substrate specificity that includes both 11-cis and all-trans retinaldehydes, as well as lipid peroxidation products.^6,16 It follows that RDH12 overexpression and/or mislocalization has the potential to disrupt photoreceptor cell homeostasis, in part, by skewing the retinol–retinaldehyde balance needed to maintain visual cycle function and photopigment stability. Notably, our studies in wild-type and Rdh12^−/− mice treated by subretinal injection of rAAV2/5-hGRK1p.hRDH12 evaluated over the course of at least one year using multiple forms of analysis showed no evidence of retinal damage or disruptions in retinoid metabolism. These findings identify rAAV2/5-hGRK1p.hRDH12 as a strong therapeutic candidate and represent a significant advance in the progress needed to move toward a phase I/II clinical trial.

Although the disease associated with RDH12 mutations results in early and severe retinal degeneration, recent natural history studies suggest that a therapeutic window of opportunity may exist for many patients. Despite that fact that ERG responses are extinguished at young ages, areas of preserved photoreceptors in the peripheral retina, observed by fundus examination and OCT,^12,47 likely contribute to visual function retained into adolescence, and in a few cases, to relatively older ages. A logical strategy would be to target these areas in an effort to decrease the rate of cell death and delay or prevent further loss of peripheral vision.

Macular atrophy is an early finding in nearly all patients,^{10–13,48,49} documented as young as 3 years of age, although some children retain areas of preserved outer nuclear layer and ellipsoid in the macula on OCT.^12,47,50,51 Preservation of useful central vision may be possible with early childhood intervention, using OCT as a guide. Thus, there is a critical need for early-stage patient identification through genetic screening, for which patient registries are likely to play an important role.

With the identification of rAAV2/5-hGRK1p.hRDH12 as a strong therapeutic candidate, coupled with advances occurring in patient identification and characterization, the realization of a phase I/II clinical trial of RDH12 gene replacement therapy may be close at hand.

Author Disclosure

D.A.T., A.J.S., and R.R.A. are inventors on a pending patent application on RDH12 gene therapy that has been filed by the University of Michigan and University College London. A.J.S. and R.R.A. have financial interests in MeiraGTx PLC; K.L.F., L.J., N.D.P., A.C., F.K.P., N.W.K., and A.T.F. have no competing financial interests.

Supplementary Material

Supplementary Figure S1

Supplementary Figure S2