Live fluorescent retinal imaging at different points following subretinal injection in 3 dogs and 1 primate. The dog D4 was injected with recombinant adeno-associated virus type 2/adeno-associated virus 2 capsid cytomegalovirus promoter and green fluorescent protein genome (rAAV-2/2.CMV.gfp), dog D5 with rAAV-2/4.CMV.gfp, dog D6 with rAAV-2/5.CMV.gfp, and the primate Mac5 with rAAV-2/4.CMV.gfp. mpi indicates months postinjection.
Fundus photographs and fluorescein angiograms of the eye expressing green fluorescent protein at 36 months postinjection for dog D4 and at 24 months postinjection for dogs D5 and D6. Early- and late-phase fluorescein angiograms are shown. The arrow indicates a window defect at the needle intrusion site.
Fundus photograph, fluorescein angiograms (top row), and indocyanine green angiograms (bottom row) of the eye expressing green fluorescent protein in the primate Mac5 at 18 months postinjection. Early and late phases of both angiograms are shown. The arrow indicates a window defect at the needle intrusion site.
Bilateral full-field electroretinographic recordings in dog D4 (A) and primate Mac5 (B) at 36 months and 18 months postinjection, respectively. The top traces are the injected right eyes. The bottom traces are the control left eyes. The top 2 recordings are low- and high-intensity scotopic responses, but the bottom 2 recordings show photopic responses (responses to light-adapted single-flash and 30-Hz flicker stimuli, respectively). μV indicates microvolt; ms, milliseconds.
Le Meur G, Weber M, Péréon Y, Mendes-Madeira A, Nivard D, Deschamps J, Moullier P, Rolling F. Postsurgical Assessment and Long-term Safety of Recombinant Adeno-Associated Virus–Mediated Gene Transfer Into the Retinas of Dogs and Primates. Arch Ophthalmol. 2005;123(4):500-506. doi:10.1001/archopht.123.4.500
To evaluate, in dogs and primates, the short-term effects of subretinal injection and the safety of long-term recombinant adeno-associated virus (rAAV)–mediated transgene expression with respect to retinal morphology and function.
Subretinal delivery of rAAV (serotype 2, 4, or 5) was performed unilaterally in 14 beagles and 9 macaques. Postsurgical condition was evaluated during a 2-month follow-up study. Three dogs and 1 primate were examined for the long-term study. Green fluorescent protein expression was monitored by fluorescent retinal imaging. Retinal anatomy and function were assessed by angiography and electroretinography, respectively.
Transgene expression was observed in 20 of 23 subretinally injected animals (both with and without vitrectomy). We did not detect an inflammatory response in any of the 23 treated subjects. In the long-term study, transgene expression was detected at the latest points evaluated: 36 months for the rAAV-2–injected dog, 24 months for the rAAV-4 and rAAV-5 dogs, and more than 18 months for the rAAV-4–injected primate. Angiography examinations were performed and showed no retinal abnormalities. Functional evaluation showed normal electroretinographic amplitude responses that were similar to those of the noninjected contralateral eyes.
Subretinal injection of the rAAV vector in dogs and primates is a safe procedure with no perioperative complications and a high rate of successful retinal gene transfer. The retinal anatomy and function remained unchanged, despite persistent transgene expression up to 36 months postinjection with rAAV-2, -4, or -5. Additionally, we observed no other adverse effects, such as tumor formation due to possible insertional mutagenesis. These short- and long-term studies on rAAV transgene expression using large animals are encouraging for the prospects of ocular gene therapy applications in humans.
These short- and long-term studies on rAAV transgene expression using large animals are encouraging for the prospects of ocular gene therapy applications in humans.
Significant progress in understanding the molecular basis of a variety of retinal diseases has promoted the development of gene therapy approaches to treat these diseases.
Adeno-associated virus type 2 (AAV-2) is a human parvovirus that has gained increasing attention because of its successful use, in many different organisms, as a gene transfer vector.1,2 The AAV-2 genome consists of a 4.7–kilo-base pair (kbp) single-stranded DNA molecule, which consists of 2 open reading frames, rep (replication) and cap (capsid), flanked by 2 145-bp inverted terminal repeats. Recombinant AAV-2 (rAAV-2) vectors used for gene therapy are derived from the wild-type virus by deleting the entire viral coding region (rep and cap) and replacing it with a reporter or therapeutic gene. More recently, 7 other rAAV serotypes (rAAV-1, -3, -4, -5, -6, -7, and -8) have been isolated and cloned.3- 8
Recombinant AAV vectors are among the most efficient vehicles for treatment of retinal diseases because the tropisms and transduction patterns of these vectors lead to efficient and stable gene transfer in retinal pigment epithelial (RPE), photoreceptor, and ganglion cells.9- 18 As a consequence, an increasing number of strategies for the molecular treatment of retinal disease relies on rAAV vectors. To date, rAAV-mediated gene therapy studies have involved gene transfer to spontaneous and genetically engineered animal models of retinal diseases. The diseases that have been studied include neovascular diseases (age-related macular degeneration, diabetic retinopathy, and retinopathy of prematurity) and retinal degenerative diseases (retinitis pigmentosa and Leber congenital amaurosis). In rodents, successful treatment of neovascularization and retinal degeneration has been demonstrated.19- 23 In dogs, 2 studies have reported the restoration of vision in a canine model of Leber congenital amaurosis.24,25 Although restoration of vision was observed in some Briard dogs using AAV-2–mediated delivery of the canine RPE65 gene, rAAV serotypes that lead to exclusive transgene expression in RPE, such as rAAV-4,18 might be a better candidate for safe and stable long-term gene transfer in humans. As shown previously, rAAV-2 and rAAV-5 are the most appropriate vectors to treat retinal degeneration where ganglion cells or photoreceptors are targeted.12,14- 16,18,26 Although the therapeutic potential of these novel rAAV-mediated treatments appears promising for retinal degenerative diseases, such as retinitis pigmentosa or Leber congenital amaurosis, additional safety studies should be carefully conducted and evaluated in large animals before proceeding to clinical application in humans. Evaluation of rAAV-mediated gene transfer into the eyes of large animals is relevant to the future clinical development of human applications: the ocular anatomy of large animals is more similar to humans than that of either the mouse or the rat. Nonhuman primates possess ocular anatomic features virtually identical to those of humans. Its components are of similar proportion and the retina possesses a macula. Also, the surgical procedures for vector delivery and the amount of injected vector would be similar in dogs and primates to what is projected for use in humans. In addition, dogs or primates have greater immunological and biological similarity to humans, which is critical in determining how well vectors are tolerated before clinical trials can proceed in humans.
In this report, we have evaluated the postsurgical condition and the long-term safety of rAAV-2, -4, and -5–mediated transgene expression with respect to retinal morphology and function in dogs and primates. Because direct examination of the green fluorescent protein (GFP) expression in the retina can be easily done on anesthetized animals using fluorescence fundus photography, we used vectors encoding for the GFP gene. Following a single subretinal injection of either rAAV-2, -4, or -5, each carrying a cytomegalovirus promoter and green fluorescent protein (CMV.gfp) genome, we examined 3 dogs and 1 primate for up to 36 months. Green fluorescent protein expression in these animals was monitored by fluorescent retinal imaging. The retinal morphology and function of these long-term transgene expressing animals were evaluated by angiography and electroretinography (ERG), respectively.
All rAAV vectors carried a CMV.gfp cassette flanked by the AAV-2 inverted terminal repeats and encapsidated in an AAV-2, -4, or -5 capsid, yielding rAAV-2/2CMV.gfp, rAAV-2/4CMV.gfp, or rAAV-2/5CMV.gfp, respectively. The rAAV-2/2, rAAV-2/4, and rAAV-2/5 vectors were produced as previously described.27- 29 The rAAV titer was determined by dot blot and is expressed as vector genomes (vg) per milliliter.30 The titers were 1.5×1012 vg/mL, 4×1012 vg/mL, and 2×1012 vg/mL for rAAV-2/2, rAAV-2/4, and rAAV-2/5, respectively.
We purchased the dogs (D1-D14) and macaques (Mac1-Mac9) from the Centre d’Elevage du Domaine des Souches (Mezilles, France) and from BioPrim (Baziège, France), respectively. All animals were cared for in accordance with the Association for Research in Vision and Ophthalmology (Rockville, Md) statement about the use of animals in ophthalmic and vision research. Subretinal injections were performed via a transvitreal approach under isoflurane gas anesthesia as previously described.18 The protocol was approved by the Institutional Animal Care and Use Committee of the University of Nantes.
We monitored the GFP expression in live animals with fluorescence retinal imaging using a Canon UVI retinal camera (Lheritier SA, Saint-Ouen-l’Aumône, France) connected to a digital imaging system (Lhedioph Win Software; Lheritier SA). We examined the retinas at monthly intervals following injection. Identical experimental conditions and parameters were used for fluorescence fundus photography at each point.
The pupils were dilated 20 minutes before anesthesia with tropicamide (Ciba Vision Faure, Novartis Pharma SAS, Annonay, France) and phenylephrine hydrochloride (10% Neosynephrine, Novartis Pharma SAS). Animals were anesthetized as described earlier. Fluorescein angiography and indocyanine green angiography were imaged using a Canon UVI retinal camera connected to a digital imaging system (Lhedioph Win Software). We injected intravenously 0.1 mg/kg of 10% fluorescein sodium solution (Ciba Vision Ophthalmics, Blagnac, France). The photograph sequence was started after injection with the fluorescein exciter and barrier filters in place until dye elimination. We injected intravenously 0.5 mg/kg of indocyanine green dye (infracyanine; Laboratoires SERB, Paris, France). Late-phase angiograms were obtained 15 minutes after injection.
Retinal function was tested using simultaneous bilateral flash photopic and scotopic ERG. We recorded ERGs in a standardized fashion, according to International Society for Clinical Electrophysiology of Vision protocols,31 using a computer-based system (Neuropack μ MEB-9102K; Nihon-Kohden, Tokyo, Japan) and contact lens electrodes (ERGjet; Universo Plastique SA, Le Crêt-du-Locle, Switzerland). Bandpass filter cut-off frequencies were 1 Hz and 100 Hz for all measurements. Analysis time was 100 milliseconds. The a-wave amplitude was measured from the baseline to the a-wave peak, and the b-wave amplitude was measured from the a-wave peak to the b-wave peak, while flicker amplitude was measured peak to peak. Each uninjected eye served as an intra-individual control.
Subretinal delivery of rAAV serotypes -2, -4, or -5 encoding for GFP was performed unilaterally in 14 beagles (D1-D14) and in 9 macaques (Mac1-Mac9) (Table 1). Total volumes of 60 μL to 100 μL of vector were injected into the right eye of each animal. For all the dogs injected, a 44-gauge cannula was inserted through a sclerotomy and advanced through the vitreous, creating a single retinotomy, before injecting the vector, which caused a retinal bleb. For D14, the retina did not detach properly during the subretinal injection, and therefore no transgene expression could be detected. In all other dogs treated (13 [93%]), we made the following observations: retinas spontaneously flattened by 24 to 48 hours postinjection, no inflammatory response was detected by fundus photography within 2 months postinjection, and transgene expression was observed by live fluorescence imaging within 1 week, persisting to the last point. Two primates (Mac1 and Mac2) were injected using the same transvitreal approach. For both macaques, the retina reattached as soon as the cannula was removed, resulting in the vector suspension being immediately released into the vitreous. As expected, primates Mac1 and Mac2 never displayed transgene expression. For the 7 other macaques (Mac3-Mac9), in which a vitrectomy was performed before subretinal injection, retinas spontaneously flattened by 24 to 48 hours postinjection, no inflammatory response was detected, and transgene expression was observed. In all successfully injected retinas, the transduced area seen by live fluorescence imaging matched exactly the retinal detachment created by the injection. The GFP expression gradually increased over time to reach a maximal level at approximately 60 days postinjection.18 Although most of the animals expressing GFP were killed for histology and biodistribution studies, dogs D4, D5, and D6 and macaque Mac5 were kept for the long-term study.
Subretinal injection of rAAV-2/2CMV.gfp, rAAV-2/4CMV.gfp, and rAAV-2/5CMV.gfp was performed within the tapetal retinas of dogs D4, D5, and D6, respectively. We chose the tapetal retina because RPE cells are not pigmented in this region. In Mac5, we performed the subretinal injection of rAAV-2/4CMV.gfp outside the macula. We monitored retinal morphology and GFP expression with color fundus photographs for up to 36 months (D4), 24 months (D5, D6), or 18 months (Mac5) (Figure 1). Noticeably, when compared with live fluorescent imaging obtained in the dog treated with rAAV-2/4CMV.gfp, the GFP signal was less intense in primate Mac5. This result is not surprising because in primates, RPE cells are strongly pigmented, resulting in partial fluorescence quenching. The GFP expression levels in the retinas of the 3 dogs and the primate remained remarkably stable during the respective monitoring periods.
To clinically monitor eventual chorioretinal alterations due to rAAV-mediated long-term transgene expression, we performed angiographies in the dogs and in the primate at different points after vector administration. As previously reported, hyperfluorescence or hypofluorescence detected by fluorescein and indocyanine green angiograms, respectively, is a sign of anatomical abnormalities such as RPE or choroidal atrophies.32,33 In dogs, the presence of the tapetum lucidum in the retina prevents any interpretation of indocyanine green angiograms; therefore, we performed only fluorescein angiography. No abnormalities were detected in fluorescein angiograms, as shown in the posterior fundus color photographs taken 36 months postinjection for dog D4 and 24 months postinjection for dogs D5 and D6 (Figure 2). In the 3 dogs, the transduced retinas had normal angiograms, except for the presence of a hyperfluorescent dot caused by a window defect at the needle intrusion site for dog D4. In the dogs D4 and D6, which both presented high levels of GFP expression, the GFP fluorescent signal became visible during the late phase of the angiogram because the fluorescein was gradually eliminated from the retinal and choroidal vasculature. This result was expected because the fluorescein and GFP have the same excitation and emission wavelengths. In the primate treated with rAAV-2/4CMV.gfp, the posterior fundus was normal at 18 months postinjection, and we observed no abnormalities in the fluorescein and indocyanine green angiograms (Figure 3). Similar to dog D4, the hyperfluorescent dot, a window defect at the injection site, is visible on the fluorescein angiogram. Together, these results indicate that long-term rAAV-mediated expression of a transgene in subretinally injected eyes does not alter retinal morphology.
The effect of long-term transgene expression following subretinal delivery of rAAV vectors on global retinal function in dogs was evaluated using flash photopic and scotopic ERG. The same investigator (G.L.) recorded the ERGs in a standardized fashion up to 36 months postinjection for dog D4, 24 months postinjection for dogs D5 and D6, and 18 months postinjection for primate Mac5. Normal patterns of ERG amplitudes were elicited in both the treated and untreated eyes of all 4 animals studied. The ERG recordings of D4 and Mac5 are shown in Figure 4. The summary of the ERG responses in the 3 treated dogs are reported as mean a-wave and b-wave amplitudes for each stimulus, rod ERG, ERG max, cone ERG, and flicker (Table 2). The ERG findings support the conclusion that normal rod and cone functions in dogs and primates are preserved for up to 36 months after efficient rAAV-mediated gene transfer and in the face of continual transgene expression.
In this study, we have found that subretinal rAAV delivery was possible using a transvitreal approach without vitrectomy in dogs. Unlike our findings in dogs, a pars plana vitrectomy was required in the primates before we could make a successful transvitreal subretinal injection. We have demonstrated that a single subretinal injection of rAAV vector into dogs or primates can sustain constitutive transgene expression in the retina for at least 36 months for AAV-2 and 24 months for AAV-4 and -5 in the dogs and 18 months for AAV-4 in the macaque. Monitoring GFP expression in the dogs treated with rAAV-2,
-4, and -5 and in the primate injected with rAAV-4 clearly demonstrated the stability of transgene expression level in the retina over time. To our knowledge, this ongoing study represents the longest reported transgene expression following rAAV injection into the retina. We have shown that several years of stable rAAV-2, -4, and -5–mediated gene transfer alter neither the structure nor the function of the retina.
Taking into account the 14 dogs and 7 primates (in which vitrectomy was performed), 20 of the 21 subretinally injected animals demonstrated an efficient gene transfer without perioperative complications. This result suggests that our surgical technique for subretinal injection in dogs and primates is both reliable and reproducible. We chose the volume of vector suspension delivered to create 1 single bleb easily visible with live fluorescence imaging. Gene replacement therapy for retinitis pimentosa due to a mutated MERTK gene23 or for Leber congenital amaurosis24,25 might require delivery of the vector to a larger area of the retina. The creation of a larger bleb or multiple blebs could cause adverse postoperative complications and has yet to be evaluated.
From the perspective of gene therapeutics, we consider it essential to assess clinically the morphology of the retina to evaluate any possible adverse effects of long-term gene transfer. To this end, we performed fluorescein or indocyanine angiography in all the animals and detected no abnormalities or lesions. A recent study on rAAV-2.gfp–mediated gene transfer into the canine retina reported that 3 of the 8 subretinally injected animals developed a delayed-onset intraocular inflammatory response 4 to 6 weeks following vector delivery.34 Similarly, in a study testing gene replacement therapy using AAV-2/2–mediated delivery of a canine RPE65 gene into the retinas of Briard dogs affected by Leber congenital amaurosis,25 uveitis developed in 75% of the eyes treated with rAAV-2.rpe65. In both cases, the inflammation response was attributed either to the presence of a contaminant in the vector preparation or to an immune response to the expressed foreign protein (GFP or RPE65). Neither of these studies clinically assessed the retinas using angiography. It is known that during early stages of posterior uveitis, RPE may show focal areas of damage or may undergo atrophy, both detected by angiography.35 Therefore, clinical assessments of the retinal morphology over the transduced area in these 2 cohorts of dogs would have been of inestimable value to the field. Angiography examinations in dogs and primates are essential for any preclinical evaluation of retinal gene therapies.
Few studies have assessed the functional effects of rAAV vector delivery to the retina in normal animals. We found that rAAV-2, -4, or -5 transduced eyes expressing the transgene for several years all had intact retinal function. No ERG evidence of toxic effects was detected over the 36-month period for dog D4, over the 24-month period for dog D5 or dog D6, and over the 18-month period for primate Mac5. These findings are consistent with a study where normal ERG photoresponses were observed 4 months after the administration of rAAV-2.gfp in 3 of 4 primates.11 However, our data contrast with a study in which a significant reduction in global retinal function was observed at 8 weeks following subretinal delivery of rAAV-2.gfp in normal dogs.34 Assessing the function of the transduced part of the retina using multifocal ERGs would be the most accurate evaluation method. However, because animals are unable to fix their gaze on command as humans can, this examination is far more complicated than in humans. As a result, in animals, multifocal ERG recordings require simultaneous scanning-laser ophthalmoscope stimulation.36
A few years ago, a disturbing finding in an rAAV-2/2–mediated gene therapy experiment using a mouse model of the lysosomal storage disease mucopolysaccharidosis VII triggered the gene therapy community to reevaluate the safety of these vectors. In this study, some of the mice intravenously injected with the AAV-2/2 vector developed hepatocellular carcinoma at 8 months after vector administration.37 Results from quantitative polymerase chain reaction performed on the tumor samples could not determine if the tumors identified in these mice were a consequence of the rAAV treatment.38 This observation highlighted the need for long-term studies designed to address the tumorigenic potential of rAAV-based gene transfer vectors.
In summary, we have found that the subretinal injection of rAAV is a safe procedure that leads to efficient gene transfer in dogs and primates. We have shown the persistence of transgene expression over several years following the administration of rAAV-2, -4, or -5 and have further found no pathological effect on retinal anatomy or function. Our findings support the further development of rAAV serotypes 2, 4, and 5 for the clinical application of retinal gene therapy.
Correspondence: Fabienne Rolling, PhD, Institut National de la Santé et de la Récherche Médicale UMR U649, Centre Hospitalier Universitaire–Hotel Dieu, 30 Avenue J Monnet, 44035 Nantes CEDEX 01, France (firstname.lastname@example.org).
Submitted for Publication: February 20, 2004; final revision received October 4, 2004; accepted October 4, 2004.
Financial Disclosure: None.
Funding/Support: Vector Core (http://www.vectors.nantes.inserm.fr) at the University Hospital of Nantes (Nantes, France) is supported by the Association Française Contre les Myopathies (Evry, France), the Institut National de la Santé et de la Récherche Médicale (Paris, France), and the Fondation pour la Thérapie Génique en Pays de la Loire (Nantes). This work was also supported by the French Lions Club (Paris) and the Lions Clubs International Foundation (Oak Brook, Ill).
Acknowledgment: We thank Matthew Ellinwood and Anna Skulimowski for critical reading and editing.