Defects in the photoreceptor-specific gene encoding aryl hydrocarbon receptor-interacting protein-like 1 (AIPL1) are clinically heterogeneous and present as Leber Congenital Amaurosis, the severest form of early-onset retinal dystrophy and milder forms of retinal dystrophies such as juvenile retinitis pigmentosa and dominant cone-rod dystrophy. [Perrault, I., Rozet, J.M., Gerber, S., Ghazi, I., Leowski, C., Ducroq, D., Souied, E., Dufier, J.L., Munnich, A. and Kaplan, J. (1999) Leber congenital amaurosis. Mol. Genet. Metab., 68, 200–208.] Although not yet fully elucidated, AIPL1 is likely to function as a specialized chaperone for rod phosphodiesterase (PDE). We evaluate whether AAV-mediated gene replacement therapy is able to improve photoreceptor function and survival in retinal degeneration associated with AIPL1 defects. We used two mouse models of AIPL1 deficiency simulating three different rates of photoreceptor degeneration. The Aipl1 hypomorphic (h/h) mouse has reduced Aipl1 levels and a relatively slow degeneration. Under light acceleration, the rate of degeneration in the Aipl1 h/h mouse is increased by 2–3-fold. The Aipl1−/− mouse has no functional Aipl1 and has a very rapid retinal degeneration. To treat the different rates of degeneration, two pseudotypes of recombinant adeno-associated virus (AAV) exhibiting different transduction kinetics are used for gene transfer. We demonstrate restoration of cellular function and preservation of photoreceptor cells and retinal function in Aipl1 h/h mice following gene replacement therapy using an AAV2/2 vector and in the light accelerated Aipl1 h/h model and Aipl1−/− mice using an AAV2/8 vector. We have thus established the potential of gene replacement therapy in varying rates of degeneration that reflect the clinical spectrum of disease. This is the first gene replacement study to report long-term rescue of a photoreceptor-specific defect and to demonstrate effective rescue of a rapid photoreceptor degeneration.
Inherited retinal dystrophies are one of the most common causes of blindness in the Western world, affecting approximately 1 in 3000 individuals. To date, there are no effective treatments for this heterogeneous group of conditions that are caused by defects in any one of over 185 different genes, 134 of which have now been identified (www.sph.uth.tmc.edu/Retnet/). A majority of the defects are in genes that are expressed in either photoreceptors or the retinal pigment epithelium (RPE). There is considerable variation in the severity of these conditions. The most severe are recessive conditions resulting from loss of function; photoreceptor defects tend to result in more severe disease than that caused by RPE defects. This is exemplified by Leber Congenital Amaurosis (LCA), a genetically heterogeneous, autosomal recessive retinal degenerative disease which is the most severe form of inherited retinopathy and the commonest cause of congenital blindness in children accounting for 10–18% of cases (1,2). LCA is characterized by markedly impaired vision or complete blindness and near absence of electrical responses to light within the first year of life.
Over the past two decades, major advances in understanding disease processes in animal models of human retinopathy and the development of efficient retinal gene transfer using either adeno-associated virus (AAV) or lentivirus-based vector systems have led to robust proof of concept studies of gene replacement therapy (3–8). Better outcomes have generally been seen in models of RPE defects (6,9–11) probably due to the fact that gene delivery to RPE cells is more efficient and widespread and more importantly, the photoreceptor cells in these conditions are inherently healthy. The culmination of these developments is the commencement of three human ocular gene therapy trials, treating patients with LCA, caused by mutations in an RPE-specific gene, RPE65 (12–14). Although the patients involved in these trials were predominantly adults and have established retinal degeneration, the preliminary results suggest some improvement in visual function is possible (12,13), paving the way for additional clinical trials of gene therapy for a variety of inherited retinal degenerations.
In general, gene replacement therapy for photoreceptor-specific defects has not been very effective at slowing severe photoreceptor degeneration characterized by fast progression, such as the rd1 mice. AAV-based vectors have been shown to be the most efficient vectors for gene transfer to the retina (15,16). A number of AAV serotypes are able to transduce photoreceptors efficiently although AAV2/5 and particularly AAV2/8 appear to be the most efficient (17). The first studies using an AAV2/2 vector in rds mice did not alter rate of retinal degeneration despite the restoration of structure and function—50% of the photoreceptors are still lost by 8 weeks (3,18). More success was achieved in RPGRIP deficient mice following AAV2/2-mediated gene replacement therapy (7). This is possibly due to a requirement for lower levels of transgene expression in RPGRIP deficient mice or because the structural abnormality in rds mice was too severe to correct effectively. The speed of disease progression is likely to be important in determining the potential for successful therapy. The window of opportunity in a disease model is created by the balance between time for vector-mediated expression and rate of photoreceptor degeneration. In models of rapid degeneration such as the rd1 mouse, gene replacement therapy using AAV2/2 vectors is rather unlikely to be successful since most of the photoreceptors are lost before maximal transgene expression can take place. The rd1 mouse is a well-characterized model of autosomal recessive retinitis pigmentosa and has a very fast retinal degeneration. It is homozygous for a null mutation in the Pde6b gene encoding the ß-subunit of rod photoreceptor cGMP phosphodiesterase which results in the loss of function of ß-PDE. Photoreceptor degeneration begins at P10 and is complete by 4 weeks. Despite a number of attempts to treat the rd1 mice by gene replacement therapy using a variety of vectors (19–22), effective rescue has proved elusive. However, gene replacement therapy has been more effective in rd10 mice, a hypomorphic Pde6b mutant with a missense mutation in exon 13. The mutation causes partial loss of PDE activity and a milder phenotype than the mutation in rd1 mice. Loss of photoreceptor cells begins at P16 and is complete by P60. Dark rearing slows the degeneration by a further 4 weeks and allows the formation of outer segments that are not normally formed in either rd10 or rd1 mice. Pang et al. (23) showed it was possible to preserve photoreceptors and retinal function following gene replacement therapy using an AAV2/5 vector. However, animals were dark reared until vector administration at P14 and then dark reared for a further 2 weeks after treatment. They were then exposed to light for 1 week before the effect of treatment was assessed.
The human aryl hydrocarbon receptor-interacting protein-like 1 (AIPL1) gene encodes a 384 amino acid protein that incorporates three consecutive tetratricopeptide repeat motifs and shares close structural homology with the FK506-binding protein family of specialized chaperones (24). These features collectively suggest a role for AIPL1 as a molecular chaperone involved in retinal protein folding or cellular translocation (25,26). In patients, AIPL1 defects are associated with LCA as well as with cone-rod dystrophy and juvenile retinitis pigmentosa (27,28). We have used two mouse models of AIPL1 deficiency. One model, the Aipl1−/− mouse has a targeted disruption of the Aipl1 gene and produces no Aipl1 (29). The second model, the Aipl1 h/h mouse, has a targeted disruption of intron 2 that results in reduced levels of Aipl1 mRNA and a 20–25% reduction in the levels of Aipl1 (30). Pathogenesis of AIPL1-associated retinal degeneration has been linked to cGMP PDE. In parallel to AIPL1 deficiency, all three subunits of the cGMP-PDE holoenzyme (α, ß and γ) were found to be reduced post-transcriptionally in both mouse models, suggesting that AIPL1 may have a role in the post-translational processing of cGMP-PDE. With complete loss of photoreceptors by 3 weeks, the retinal degeneration in the Aipl1−/− mouse is even faster than rd1 mouse. Photoreceptor loss is considerably slower in the Aipl1 h/h hypomorphic mutant. Retinal degeneration begins after 12 weeks of age and by 6–8 months over half of photoreceptor cells are lost. Constant light exposure further accelerates the degeneration in the Aipl1 h/h mouse so that photoreceptor cell loss is almost complete by 5 months of age. In these models, normal retinal histology and photoreceptor morphology including outer segments are present at birth and most of photoreceptor cells are lost after the retina has terminally differentiated.
The aim of this study is to develop an effective gene therapy strategy to treat retinal dystrophies due to AIPL1 deficiency. Given the variable severity of disease seen in patients with AIPL1 defects, we assess the efficacy of gene replacement therapy using AAV2/2 and AAV2/8 vectors to treat Aipl1−/− and Aipl1 h/h mice that simulate different rates of retinal degeneration. We assess the effect of gene transfer using light and electron microscopy, immunohistochemistry and by electroretinography. This study is the first to describe long-term rescue of a photoreceptor-specific defect following gene replacement therapy. It also the first to describe the use of an AAV2/8 vector to treat a retinal degeneration and demonstrate effective rescue of a rapid photoreceptor degeneration.
Verification of Aipl1 transgene expression
The murine Aipl1 cDNA was PCR amplified from murine retinal cDNA and cloned between into an AAV2 plasmid backbone with a CMV promoter and SV40 polyadenylation site. The plasmid construct pD10/CMV-Aipl1 and the viral vector AAV2-CMV-Aipl1 were tested in vitro to ensure that they were functional. Transfection using the plasmid construct and transduction using a recombinant AAV serotype 2 vector, AAV2-CMV-Aipl1, were performed in 293T cells. Western blot analysis showed a single, appropriate band of 38 kDa (Fig. 1A). We proceeded to test the function of the viral vector in the Aipl1 h/h mouse model. Messenger RNA was isolated from Aipl1 h/h retinas 28 weeks after unilateral subretinal injection of AAV2-CMV-Aipl1 into three animals. Quantitative PCR and paired analysis of the samples showed that the mean relative level of Aipl1 expression was more than nine times higher than in untreated eyes, although it should be noted that this difference might in part be due to an increased loss of photoreceptor cells in the untreated eyes. The mean relative level in treated eyes at 28 weeks post-injection was approximately half that seen in wild-type animals (data not shown).
AAV-mediated Aipl1 gene replacement therapy results in increased levels of PDE and restores localization to the outer segment
AAV2-CMV-Aipl1 was injected subretinally into the superior and inferior hemispheres of the right eyes of 4-week-old Aipl1 h/h animals (n = 16), leaving the contralateral left eye untreated to serve as an internal control for inter-animal variation and test-retest variability. As additional controls, a number of Aipl1 h/h mice (n = 5) received subretinal injections of PBS and wild-type C57B/6 (n = 6) received subretinal injections of AAV2-CMV-Aipl1. Retinas were analysed at 28 weeks after injection (Fig. 1B). In wild-type mice, Aipl1 was found mainly in photoreceptor inner segments, but some protein was present in the cell body and synaptic spherules. In uninjected or mock (PBS)-injected Aipl1 h/h retinas, Aipl1 immunofluorescence was much reduced and there was marked thinning of the outer and inner nuclear layers and disruption of the retinal architecture. Following injection of AAV2-CMV-Aipl1, there was a substantial increase in Aipl1 immunofluorescence in appropriate cellular locations. Immunofluorescence was also seen in the RPE layer because AAV2 transduces both photoreceptor cells and RPE and a ubiquitous promoter was used to drive transgene expression.
Since it had previously been suggested that Aipl1 affected rod phosphodiesterase (PDE) levels (29,30), we investigated the effect of rAAV-mediated Aipl1 gene replacement on rod PDE expression and localization in photoreceptors 28 weeks after vector administration (Fig. 1C–E). Confocal imaging of Aipl1 and β-PDE immunofluorescence in untreated Aipl1 h/h retinas revealed not only reduced Aipl1 immunofluorescence but also reduced levels of β-PDE which was localized to the inner segments rather than outer segments (Fig. 1C). In eyes that received subretinal AAV2-CMV-Aipl1, substantially increased Aipl1 immunofluorescence was accompanied by increased immunofluorescence for ß-PDE. More interestingly, a shift in the subcellular localization of β-PDE was seen in treated retinas; ß-PDE immunostaining was restored to the outer segments of photoreceptors (Fig. 1D). The relative amount and localization of Aipl1 and PDE in treated eyes were similar to that seen in wild-type sections (Fig. 1E). These results indicated that not only the amount of PDE but also its subcellular localization is dependent on Aipl1.
Maintenance of photoreceptor function in Aipl1 h/h mice following AAV-mediated gene replacement therapy
To evaluate the effect of AAV-mediated Aipl1 expression on retinal function, dark-adapted, scotopic ERG recordings were obtained from 16 Aipl1 h/h animals at regular intervals following subretinal injection of AAV2-CMV-Aipl1 when the animals were 4 weeks old. ERGs were recorded simultaneously from the treated and untreated eyes of each animal. Representative ERG recordings from a single Aipl1 h/h mouse at 20 weeks post-injection (Fig. 2A) and ERG waveforms at 28 weeks post-injection (Fig. 2B) showed that ERG amplitudes in the treated eye were substantially higher compared with the untreated fellow eye. Oscillatory potentials, which represent post-synaptic neuronal activity in the inner retina, were more obvious in the treated eye, but were reduced in the untreated eye. A series of ERG recordings obtained from the treated and untreated eyes of another Aipl1 h/h animal at various follow-up time points following subretinal injection of AAV2-CMV-Aipl1 is shown in Figure 2C. Over the 28 week period, the untreated eye showed a steady decline in a-wave and b-wave amplitude, whereas the treated eye showed stabilization of amplitude and waveform with differences apparent from 12 weeks post-injection onwards (Fig. 2C). At 28 weeks, the ERG response from the untreated eye was almost flat, whereas the treated eye maintained a substantial ERG response, indicating preservation of photoreceptor function.
At each time point assessed, there was considerable variation in the ERG amplitudes in the untreated eyes of individual Aipl1 h/h mice, indicating inter-animal variation in the rate of photoreceptor cell degeneration in this model (data not shown). Furthermore, the ERG results of the same cohorts of animals vary substantially between recording sessions, suggesting high inter-session variability. In view of this variation, paired t-tests were used to analyse the mean b-wave and a-wave amplitudes between treated and untreated eyes at the various time points from 4 to 28 weeks (Fig. 2D and E). Both the b-wave and a-wave amplitudes were significantly higher in the treated compared with the untreated eyes at 20, 24 and 28 weeks post-injection (P ≤ 0.05). There was an initial increase in ERG amplitudes from birth until adulthood (8 weeks) that reflected normal development in mice. Four weeks after vector administration, there was a reduction in b-wave amplitudes in treated eyes that was probably due to injection-related trauma. However, from 20 weeks post-injection onwards, b-wave amplitudes were significantly higher in treated compared with untreated eyes (P ≤ 0.05), a difference which remained consistent up to the final time point of 28 weeks post-injection. The mean b-wave amplitude in treated eyes (n = 15; mean = 295.3 ± 13.6 µV) at 20 weeks post-injection was ∼20% higher that in untreated eyes (n = 15, mean = 243.6 ± 15.6 µV). Analysis of mean a-wave amplitudes similarly showed that treated eyes maintained higher amplitudes compared with untreated eyes, this difference was statistically significant (P ≤ 0.05) from 20 weeks post-injection onwards up to the final time point of 28 weeks post-injection (Fig. 2E). At 20 weeks post-injection, the mean a-wave amplitude in treated eyes (n = 15; mean = 109.6 ± 20.4 µV) was ∼38% higher than that in untreated eyes (n = 15; mean = 79.2 ± 33.8 µV).
In the group of PBS-injected animals, no significant differences were seen in the b-wave amplitudes between injected and uninjected eyes (Supplementary Material, Fig. S1). No significant differences were found in the b-wave amplitudes in treated and untreated eyes of wild-type C57B/6 mice which received a unilateral subretinal injection of AAV2-CMV-Aipl1, indicating that over expression of Aipl1 in photoreceptors and RPE did not affect retinal function (Supplementary Material, Fig. S2A).
AAV-mediated gene replacement therapy increases photoreceptor survival in Aipl1 h/h mice
The effects of AAV-mediated Aipl1 expression in the same group of Aipl1 h/h mice were evaluated by assessing the retinal morphology. Semi-thin sections of treated and untreated Aipl1 h/h retinas were taken at 52 weeks after unilateral injection of AAV2-CMV-Aipl1. Treated eyes had received double subretinal injections of the therapeutic vector in the superior and inferior hemisphere of the eye. Prior to embedding, the eyes were orientated so that sections would be sagittally orientated to show the superior and inferior treated areas of the retina. In treated eyes, the retina had considerably more photoreceptor cell nuclei and the photoreceptor outer segments were longer and more densely packed (Fig. 3A). The preservation of photoreceptor cells was evident throughout the treated retina as seen by the preservation of the outer nuclear layer around the whole of the sagittal section. Untreated retinas showed substantial loss of photoreceptor cell nuclei and outer segments and this was also seen as a thinning of the outer nuclear layer throughout the sagittal section of the eye (Fig. 3A). Ultrastructurally, differences were seen in morphology of photoreceptor outer segments. The number of photoreceptor outer segments (OS) in untreated Aipl1 h/h retina was substantially reduced, markedly shortened and disorganized. The OS disk membranes were also disorganized and less tightly packed and there was loss of the normal laminar arrangement (Fig. 3B). Vacuolar inclusions containing debris material was present in the inner segments of the retina (Fig. 3C). The contact between the RPE and photoreceptor OS was also abnormal and reduced. Intervening vacuoles could also be seen between OS tips and the RPE. In contrast, in treated Aipl1 h/h retina, photoreceptor OS were elongated and the membranous OS disks had a regular and densely packed arrangement (Fig. 3D), which was similar to that seen in wild-type mice (Fig. 3E). The photoreceptor OS in treated retina maintained close contact with the RPE and showed normal interdigitation with the microvilli of the RPE (Fig. 3F).
Morphometric analysis of treated and untreated eyes of Aipl1 h/h mice at 28 weeks post-injection (Fig. 3G) showed that the mean number of photoreceptor cell nuclei in treated eyes (n = 7) was found to be 41% higher than in the untreated eyes (mean photoreceptor cell count in treated eyes = 479.3 ± 16.6 versus untreated eyes = 339.3 ± 8.2; P = 0.0003) (Fig. 3H). The preservation of photoreceptors in treated eyes was consistent in all of the seven animals assessed, with a significantly greater photoreceptor nuclei count in the treated eye compared with the fellow untreated eye (P < 0.05) (Fig. 3I).
The outer nuclear layer was quantified in five wild-type C57B/6 mice that received unilateral subretinal injection of AAV2-CMV-Aipl1 using the same method. Supplementary Material, figure S2B shows confocal images of an injected and uninjected eye from a single wild-type animal. Paired t-test showed no significant differences between the mean photoreceptor cell count of wild-type eyes that received subretinal injection of AAV2-CMV-Aipl1 and uninjected eyes (Supplementary Material, Fig. S2C), indicating that there was no detrimental effects on histology or photoreceptor survival.
AAV-mediated gene replacement therapy using a human cDNA results in effective treatment of Aipl1 h/h mice
In order to evaluate a therapeutic construct that might be used in clinical studies, we substituted the mouse Aipl1 cDNA within the expression cassette with a human AIPL1 cDNA. Aipl1 h/h mice (n = 10) received subretinal injections of AAV2-CMV-AIPL1 when they were 4 weeks old and were analysed as before using morphological assessments and ERG recording. Follow up, however, included longer time points, up to 50 weeks post-injection (Fig. 4A–C). Paired analysis of photoreceptor cell counts between treated and untreated eyes showed that treated eyes had significantly higher (73% higher) photoreceptor cell count than untreated eyes at 50 weeks post-injection (mean photoreceptor cell count in treated eyes=463.5 ± 19.8 versus untreated eyes = 267.4 ± 13.9, P = 0.00006, n = 6) (Fig. 4A and B). Treated eyes that received subretinal injection of AAV-CMV-AIPL1 maintained significantly higher b-wave ERG amplitudes up to 50 weeks post-injection (P = 0.03) (Fig. 4C). At the final time point of 50 weeks, the mean b-wave amplitude in treated eyes (n = 10, mean = 194.6 ± 25.2 µV) was 57% higher than the mean b-wave amplitude in untreated eyes (n = 10, 123.98 ± 20.42 µV). AAV-CMV-AIPL1 treated eyes showed significantly higher a-wave amplitudes than untreated eyes between 16 and 50 weeks post-injection (P < 0.05) (Fig. 4D).
AAV2/8 mediated gene replacement therapy using a murine cDNA results in effective rescue of light-accelerated retinal degeneration in Aipl1 h/h mice
In order to assess the efficacy of gene replacement therapy in the context of a more rapid degeneration, we accelerated the photoreceptor cell loss in Aipl1 h/h mice by exposing animals to continuous light. We had previously found that Aipl1 h/h mice kept under constant light exhibited a much faster course of degeneration (our unpublished observation). To ensure that we obtained more rapid onset of gene expression from the AAV vector, we produced AAV2/8-CMV-Aipl1 by packaging the therapeutic construct using AAV8 capsid. A group of Aipl1 h/h mice (n = 23) received unilateral subretinal injections of AAV2/8-CMV-Aipl1 when they were 4 weeks old. The mice were placed under constant illumination 1 week after the subretinal injections. Although the retinal degeneration progressed at a faster rate, photoreceptor rescue was still obtained. When the group of AAV2/8-treated Aipl1 h/h mice that underwent light acceleration was compared with the AAV2/2-CMV-Aipl1 treated mice, Aipl1 h/h mice that did not undergo constant light exposure, larger differences between treated and untreated eyes in terms of functional and morphological outcome measures (Figs 5 and 6) were obtained in the light exposed animals. ERGs were recorded simultaneously from the treated and untreated eyes of each animal. Figure 5A shows a light intensity series at 21 weeks post-injections. At this stage, the treated eye maintains substantial ERG amplitudes and shows increasing responses elicited by the increasing light stimuli, while ERG responses are lost in the untreated eye. Normally shaped waveforms, similar to that of wild-type mice, were recorded from treated eyes while from untreated eyes, the ERG was virtually undetectable (Fig. 5B). ERGs recorded at the various follow-up time points from a single procedure animal in this group are shown in Figure 5C. Over time, the ERG amplitudes in the untreated eye decreased rapidly and disappeared completely by 21 weeks, while the ERG response in the treated eye was maintained throughout the follow-up period, suggesting that photoreceptor survival was prolonged following treatment with AAV2/8-CMV-Aipl1. The apparent improvement in the ERG amplitudes between weeks 5 and 14 was likely to be due to variation occurring between recording sessions. Paired statistical analysis of the mean ERG b-wave amplitudes showed that b-wave amplitudes in treated eyes were significantly higher than untreated eyes up to 21 weeks post-injection (Fig. 5D). At 21 weeks post-injection, the mean b-wave amplitude in treated eyes (n = 20; mean = 87.3 ± 5.4 µV) was 110% higher that in untreated eyes (n = 20, mean = 41.5 ± 4.2 µV). Analysis of the mean ERG a-waves showed that treated eyes also had significantly higher a-wave amplitudes compared with untreated eyes (Fig. 5E). Statistically significant differences were seen at earlier time points from 5 weeks post-injection onwards up to the last time point of 21 weeks post-injection. At 21 weeks post-injection, mean ERG a-wave in treated eyes was 16.8 ± 2.3 µV (n = 20) and in untreated eyes it was 8.9 ± 2.8 µV (n = 20).
There were many more photoreceptor cells in treated eyes compared with untreated controls (Fig. 6A). The mean number of photoreceptor nuclei in treated eyes (n = 7) was 123% higher than in untreated eyes (mean photoreceptor cell count in treated eyes=202.1 ± 20.2 versus untreated eyes = 90.64 ± 23.0; P = 0.0007) at 21 weeks post-injection (Fig. 6B). In each individual animal that was assessed, significant photoreceptor preservation was seen in the treated eye compared with contralateral control (Fig. 6C). In order to assess the extent of photoreceptor preservation, we analysed sagittally orientated semithin retinal sections and found ONL preservation throughout the whole of the eye (Fig. 6D), indicating that the effect of treatment was not localized. Untreated eyes showed complete loss of the ONL by 21 weeks. Electron microscopy of AAV2/8-CMV-Aipl1 treated eyes showed that although outer segments were present, these were shortened compared with animals which did not undergo light exposure and the membranous disks were less well organized (Fig. 6E). Photoreceptor inner segments appeared to be normal in appearance (Fig. 6F). In untreated eyes, no photoreceptor cells could be seen (Fig. 6G); the inner nuclear layer and bipolar cells were seen to lie adjacent to the RPE, with intervening vacuoles or debris-containing spaces (de) (Fig. 6H).
To investigate the effect of AAV8-mediated Aipl1 expression on PDE levels, retinal homogenates of treated and untreated eyes were obtained from the AAV2/8-CMV-Aipl1 injected Aipl1 h/h mice at 21 weeks after treatment. Western blot analysis of paired retinal homogenates from treated Aipl1 h/h mice and wild-type controls showed that PDE was present in treated mutant eyes and severely reduced or absent from untreated eyes (Fig. 6I). Quantification of the immunoblots indicated that PDE levels in treated eyes were significantly higher than in untreated eyes (P = 0.02). At 21 weeks post-injection with light acceleration, the mean level of PDE in treated eyes was 65% of wild-type levels while in untreated eyes, PDE was 10% of wild-type (Fig. 6J). Since treated eyes at this time point contained approximately half the normal number of photoreceptor cells, following Aipl1 gene replacement therapy, the level of PDE in each cell might be similar to that in wild-type photoreceptors if we adjust for the number of cells.
Effective rescue of Aipl1−/− mice following AAV2/8-mediated gene replacement therapy
Since an AAV2/8 vector proved effective for mediating gene replacement therapy in the light accelerated Aipl1 h/h model, we proceeded to evaluate the efficacy of AAV2/8-CMV-Aipl1 mediated gene replacement therapy in the Aipl1−/− mouse model that had an extremely fast degeneration. A group of Aipl1−/− mice (n = 12) received unilateral subretinal injections of AAV8-CMV-Aipl1. Because of the early onset of the degeneration, we treated animals when they were 12 days old (P12). Assessments of retinal function and immunohistochemistry were performed 16 days post-injection when the mice were 28 days old (Fig. 7). In untreated Aipl1−/− mice, the ONL was completely absent, but in treated mice we observed a thick ONL with inner segments containing Aipl1 and outer segments containing PDE. In untreated eyes, no Aipl1 or PDE immunofluorescence was observed. Semithin sections of retina from 4-week-old mice showed that the treatment resulted in substantial preservation photoreceptor cells with long, densely packed outer segments (Fig. 7B). In contrast, no photoreceptor cells could be seen in untreated eyes. ERG recordings were normally extinguished by P18 in the Aipl1−/− mouse. While untreated eyes had flat ERG tracings, treated Aipl1−/− eyes exhibited good scotopic and photopic ERG amplitudes 16 days after treatment (Fig. 7C). The difference in mean scotopic ERG b-wave amplitudes between treated and untreated eyes was statistically significant (P = 0.03; mean b-wave treated = 243.5 ± 67.2 µV, mean b-wave untreated = 6.25 ± 8.9 µV, n = 4). At 6 weeks post-injection, treated eyes maintained significantly higher ERG b-wave amplitude compared with untreated eyes (P = 0.0006; mean b-wave treated = 234.7 ± 51.1 µV, mean b-wave untreated = 25.5 ± 6.4 µV, n = 6). These findings indicated that early treatment using AAV8-CMV-Aipl1 was able to significantly improve photoreceptor function and delay retinal degeneration in the Aipl1−/− mouse. ERG analysis of Aipl1−/− mice injected with AAV2/2 therapeutic vector did not show any evidence of preservation function or photoreceptor rescue (Fig. 7D). These mice were also injected at P12 and neither the treated nor the untreated eyes showed any ERG responses at P28 (16 days post-injection).
In this study, we have shown that AIPL1-associated retinal degeneration is amenable to gene replacement therapy. We have demonstrated efficacy in animal models of AIPL1 deficiency that exhibit three different rates of degeneration. We demonstrate efficacy first in the hypomorphic Aipl1 h/h model and then in increasingly rapid degenerations, specifically the light accelerated Aipl1 h/h model and the Aipl1−/− model. This study demonstrates the effective long-term rescue of a photoreceptor-specific defect and the most effective rescue to date of a rapid retinal degeneration.
In general, gene-based treatments for autosomal recessive retinal dystrophies require not only restoration of function to affected cells but also the prevention of photoreceptor cell death. In this study, we demonstrate that we are able to correct the cellular defect. Following subretinal delivery of Aipl1 replacement genes packaged in either AAV2 or AAV8 vectors, there is increased production of Aipl1 in photoreceptor inner segments and this led to increased levels of PDE and the restoration of normal PDE translocation from the inner to the outer segments. As a result, retinal degeneration is significantly slowed. We observe improved photoreceptor cell survival, preservation of outer segment morphology and stabilization of retinal function. Both therapeutic constructs, one containing the murine Aipl1 cDNA and the other containing the human AIPL1 cDNA are similarly effective. This is not surprising since AIPL1 is highly conserved between species; there is 96% similarity and 86% identity between the murine and human cDNAs.
AAV-mediated gene replacement therapy results in a significant preservation of photoreceptors and retinal function in Aipl1 h/h mice 1 year after treatment—the latest time point examined. This is the most sustained rescue of a photoreceptor-specific gene defect reported to date. The most sustained rescue reported previously was following AAV-mediated gene replacement therapy in an RPGRIP deficient mouse model, in which photoreceptors were preserved for 5 months (7). Photoreceptors are still present in untreated Aipl1 h/h mice that are 12 months old and the duration of rescue probably reflects the mild phenotype and slow rate of degeneration. Retinal degeneration is more severe in Aipl1 h/h mice kept under constant light, with loss of 84% of photoreceptors by 5 months. This can be attributed to a heightened sensitivity of the Aipl1 h/h mice to light stimuli, as WT mice kept under identical conditions show no light damage at all. Despite the more rapid cell loss, we still observe very significant preservation of photoreceptor cells and retinal function 5 months after gene replacement therapy using an AAV2/8. While rearing animals in continuous light accelerates photoreceptor loss in Aipl1 h/h mice, even faster retinal degeneration occurs in Aipl1−/− mice with all photoreceptors lost by the time animals are 3 weeks old. Following AAV2/8 mediated gene replacement therapy in Aipl1−/− mice, we have been able to achieve significant preservation of photoreceptor and retinal function for over 3 months, the last time point tested so far. This represents the most effective rescue to date of a rapid retinal degeneration.
AAV-mediated gene replacement therapy does not completely prevent loss of photoreceptor cells. At 50 weeks, AAV2-CMV-AIPL1 treated eyes from Aipl1 h/h mice has on average ∼15% fewer photoreceptor cells than normal wild-type mice, while untreated eyes has ∼50% fewer photoreceptor cells. Although gene replacement therapy in Aipl1−/− mice using AAV2/8 is remarkably effective, at 8 weeks after vector administration, treated eyes contain ∼10% fewer photoreceptors than wild-type eyes (data not shown). There may be several reasons why photoreceptor degeneration proceeds despite treatment. First, only two-thirds of the retina (at most) has been transduced. Secondly, even earlier intervention than postnatal day 12 might be necessary for maximal therapeutic effect. In this regard, it is noted that AAV8 could take 3–4 days to turn on the expression of the therapeutic gene. By then, mice will be over 15 days of age when substantial cell loss is known to have occurred. Thirdly, expression level from the CMV promoter might not have been optimal for AIPL1.
In these experiments, we injected a single eye of each animal with therapeutic vector leaving the contralateral uninjected eye to serve as an internal control. Injection of a control vector into the contralateral eye would allow assessment of therapeutic transgene expression in isolation. However, the main aim of this study was to determine the overall therapeutic benefit of subretinal injection of an AAV vector expressing Aipl1.
Interestingly, preservation of photoreceptors in Aipl1 h/h mice is better than might have been expected given the relatively modest difference in ERG amplitudes between treated and untreated eyes. This is most likely due to the peculiar ERG phenotype of the Aipl1 h/h mice. It has previously been shown that Aipl1 h/h rods have a higher sensitivity to light compared with normal rod photoreceptor cells, and require fewer photons to elicit a response (30,31). They manifest supranormal ERG when young, presumably because of a larger than normal dark current due to higher free cGMP levels. In this regard, the Aipl1 h/h mice are reminiscent of rod heterozygotes, which are similarly haploinsufficient for PDE. The restoration of functional AIPL1 in Aipl1 h/h rods should, therefore, result in correction of this abnormality so that photoresponses become more similar to that of wild-type photoreceptors with a reduction in the altered sensitivity of Aipl1 h/h rods. Hence, it is likely that the relatively modest difference in ERGs compared with difference in photoreceptor preservation seen after treatment is partly due to a reduction in ERG amplitudes as a result of a reduction in photoreceptor hypersensitivity following gene transfer.
A number of studies have suggested various roles for AIPL1 in the retina. It was originally suggested that AIPL1 may have a role in development as a result of its putative interactions with NUB1 (32), but this is unlikely since both Aipl1−/− and Aipl1 h/h mice have normal retinal architecture prior to the onset of degeneration (29,30). Biochemical studies using a yeast two-hybrid screen suggested that AIPL1 may have a general role in enhancing the protein farnesylation in the retina. Among retinal proteins known to be farnesylated are PDE α-subunit, γ-transducin and rhodopsin kinase. However, this hypothesis does not appear to be supported by the finding that levels of rhodopsin kinase and transducin are unaffected in Aipl1−/− mice (29). Further assessment of candidate retinal proteins in the Aipl1−/− and Aipl1 h/h mice suggested that cGMP-PDE is the only protein that is reduced in levels, suggesting that cGMP-PDE is a specific client protein of AIPL1 (29,30) This study confirms that AIPL1 and cGMP-PDE are intricately linked. We find that cGMP-PDE, in particular the β-subunit of PDE is mislocalized to photoreceptor inner segments in untreated Aipl1 h/h eyes when the animals are examined at 28 weeks following subretinal injection in the other eye. In the contralateral treated eyes of these animals, the β-PDE signal is present in the photoreceptor outer segments suggesting that AIPL1 could indeed be a molecular chaperone ensuring PDE translocation or transport to the outer segments. The fact that photoreceptor cell loss in the Aipl1 h/h mouse proceeds slowly and outer segments are present for most of the degeneration makes it possible to compare and localize β-PDE expression in the outer segments of treated and untreated eyes. In the Aipl1−/− mouse, the retinal degeneration is too rapid for this analysis since most of the photoreceptors cells and outer segments have been lost in the untreated eye by the time AAV8-mediated expression of Aipl1 commences in the treated eye. Hence, Aipl1 may protect PDE subunits from proteosomal degradation or assist in the assembly or folding of the PDE holoenzyme whereby only properly folded protein is transported to the outer segments. The exact molecular mechanism and relationship between AIPL1 and cGMP-PDE remains to be elucidated. Precise regulation of cGMP synthesis and cGMP-PDE degradation is critical to the health of photoreceptors and mutations which disrupt the balance between the two, result in dysregulation and degeneration of these cells (33). Mutations in the PDE6B gene encoding the ß-subunit of cGMP-PDE lead to abnormal increases in cGMP and subsequent photoreceptor death. Mutations in retGC1 impair synthesis of cGMP, leading to a state equivalent to sustained photo-excitation and photoreceptor cell death (33–35). Similarly, one of the most important consequences of mutations in AIPL1 is likely to be the effect on cGMP-PDE levels, leading to photoreceptor dysfunction and rapid retinal degeneration.
This study describes the first use of AAV2/8 to treat a murine model of retinal degeneration. Compared with AAV2/2 and AAV2/5, AAV2/8 is much more efficient at transducing the retina (17) (36). This study evaluates the gene replacement therapy using AAV2/2 and AAV2/8 vectors and finds that the most effective rescue is achieved using AAV2/8. Superior results using this vector may result from its higher transduction efficiency and higher levels of maximal transgene expression. The Aipl1−/− mouse has a rapid retinal degeneration that is complete by 3 to 4 weeks of age. Since the onset of AAV2/2 mediated transgene expression occurs between 3 and 4 weeks following subretinal injection, it is not surprising that we do not see any evidence of photoreceptor rescue following treatment of Aipl1−/− mice using AAV2/2-CMV-AIPL1. The first signs of degeneration in the Aipl1 h/h mouse are seen at 12 weeks. By injecting AAV2/2 vectors into 4-week-old Aipl1 h/h mice, AAV2/2-mediated gene expression will be expected to commence by 7 to 8 weeks of age and maximal transgene expression by 16 weeks of age, allowing adequate time for transgene expression before significant photoreceptor degeneration occurs. Constant light exposure accelerates the Aipl1 h/h retinal degeneration by 2–3-fold so that the ONL is reduced to a single cell layer by 20–21 weeks. For effective treatment of this animal model, a vector with faster transduction kinetics is required. The onset of AAV2/8-mediated gene expression in photoreceptors is known to occur 3–4 days after subretinal delivery and maximal expression by 7 weeks (17). Effective rescue of the accelerated model using an AAV2/8 vector suggests that Aipl1−/− mice may be amenable to treatment. To ensure adequate transgene expression before all photoreceptor cells are lost, Aipl1−/− mice are treated when they much younger, 12 days old. The potential benefit of gene transfer has to be balanced against risks of retinal damage in younger mice. At P12, mice have a fully developed retina and have opened their eyes, thus facilitating subretinal injection. AAV2/8-mediated gene expression should occur by P15–16, at which stage there are still surviving photoreceptors. We observe photoreceptor rescue with AAV2/8 but not with titre-matched AAV2/2 in the Aipl1−/− mouse which has very rapid and severe photoreceptor degeneration. This suggests that AAV2/8 vector is more efficacious which may be due to its faster transduction kinetics and higher levels of expression in photoreceptor cells, properties which have been previously described (17) (36). However, since a different purification method was used to produce AAV2/8 vector, we cannot exclude the possibility that this might in part be responsible for the higher therapeutic efficacy of this AAV serotype.
In patients, AIPL1 defects are associated with LCA as well as with cone-rod dystrophy and juvenile retinitis pigmentosa (27,28). The variability in phenotype may be explained by the nature of the mutations. To date, 20 disease-causing mutations in AIPL1 have been reported (HGMD; www.hgmd.org). Of these, five are likely to lead to complete loss of AIPL1 function; four are nonsense mutations resulting in a severely truncated protein and one affects the splice site in intron 2 leading to a frame shift. The other mutations are missense mutations in the N-terminal or the tetratricopeptide domains and deletions in the C-terminal region. In vitro assays have shown that many of these mutations do not lead to the loss of AIPL1 expression and are unlikely to abolish protein function completely (32,37,38). It is also likely that there are other variants that give rise to altered protein with reduced function. In this study, we have assessed the potential for gene replacement therapy for AIPL1 deficiency, using mouse models with partial or absent function and different rates of retinal degeneration. Efficient rescue of Aipl1 h/h and Aipl1−/− mice suggests that patients with mutations that do not result in a complete loss of AIPL1 function might respond particularly well to gene replacement therapy and that treatment may also be effective in LCA patients with completely absent AILP1 function provided intervention occurs early in life.
MATERIALS AND METHODS
Plasmid constructions and production of recombinant AAV2 and AAV8
The murine Aipl1 cDNA was PCR amplified from murine retinal cDNA using primers which have been designed to encompass the whole of the coding region. The PCR fragment obtained (1016 bases) was sequenced after cloning into pGem-T (Promega, Madison, WI, USA). The Aipl1 cDNA was cloned between the CMV promoter and the SV40 polyadenylation site of the construct AAV-CMV-GFP (39) to form the construct pD10/CMV-Aipl1 (total length 5739 bases). Recombinant AAV2 vector was generated from the murine construct as described (39). The human AIPL1 cDNA was also PCR amplified from human retinal DNA using primers which covered the whole coding region and included part of the untranslated regions at the 5′- and 3′-ends. The PCR fragment of 1215 bases was cloned into pGem-T (Promega Madison, WI, USA) and checked by sequencing. The AIPL1 cDNA was inserted into the parental plasmid AAV-CMV-GFP, replacing the GFP gene to generate the human construct pD10/CMV-AIPL1. The plasmids were packaged into AAV2 and AAV8 to generate two pseudotyped AAV viral vectors, AAV2-CMV-AIPL1 and AAV8-CMV-Aipl1, as described below.
Recombinant AAV serotype 2 vectors were produced from the constructs encoding murine Aipl1 and human AIPL1 by a method described previously, which uses a replicating amplicon pHAV7.3 containing the rep and cap genes, and PS1 helper virus (39). The constructs described above and the pHAV7.3 amplicon were used to transfect BHK cells by incubation with a mixture of three components, peptide6 ((K16)GACRRETAWACG) (an integrin-targeting peptide), plasmid DNA and lipofectin (Invitrogen Ltd., Paisley, UK) in OptiMEM. BHK cells were transfected with the mixture containing a total of 60 µg of plasmid DNA for 4 h, and followed by incubation with DISC-HSV (PS-1) as helper virus (39). After 36 h, at completion of the lytic cycle, the cells were collected, centrifuged and lysed by repeated freeze thawing. AAV2/2 was purified using a heparin column. Recombinant AAV serotype 8 vector was produced through a triple transient transfection method as described previously (40). The plasmid construct, AAV8 packaging plasmid and helper plasmid were mixed with 2.5 m calcium chloride and 2X HEPES buffered saline to form transfection complexes which was then added to 293T cells and left for 72 h. The cells were harvested, concentrated and lysed to release the vector. The vector was purified by ion exchange chromatography. Both of the AAV2/2 and AAV2/8 virus preparations were concentrated using Centricon 10 columns (Millipore, Bedford, MA, USA), washed in PBS and concentrated to a volume of 100–150 µl. Viral particle titres were determined by comparative dot-blot DNA prepared from purified viral stocks and defined plasmid controls. Purified vector concentrations used for all experiments were 1–2 × 1012 viral particles/ml.
Subretinal injections were performed on Aipl1 h/h mice at 4 weeks after birth. Surgery was performed under direct ophthalmoscopic control through an operating microscope. The tip of a 1.5 cm, 34-gauge hypodermic needle (Hamilton, Switzerland) was inserted tangentially through the sclera of the mouse eye, creating a self-sealing scleral tunnel wound. The needle tip was brought into focus in the subretinal space and 1.5 µl viral suspension was injected into the subretinal space to create a bullous retinal detachment around the injection sites. Two subretinal injections were performed for each mouse eye, one in the superior hemisphere and one in the inferior hemisphere. Aipl1−/− mice were injected at P12–13 days of age.
In vitro assessment of plasmid construct and viral vector and quantitative PCR
Functional assessment of the construct pD10/CMV-Aipl1 and expression of Aipl1 by AAV-CMV-Aipl1 was assessed in vitro. In a 24-well plate, 50 000 293T cells were seeded per well. One microgram of pD10/CMV-Aipl1 was used to transfect each well along with Lipofectin and peptide6 in the ratio described above, and 1 µl of AAV-CMV-Aipl1 (1011 vp/mL) was added to each well. Replicates of 12 were performed for plasmid transfection and virus infection, respectively. Transfected cells were collected 24 h later and infected cells were harvested at 48 h. The cells were lysed and the lysates pooled together. A standard western blot was performed using anti-AIPL1 antibody.
Total retinal RNA was isolated from treated and untreated Aipl1 h/h and from wild-type mice using TRIzol reagent (Invitrogen, Paisley, UK). Reverse transcription of RNA into cDNA was performed by using the Transcriptor First Strand cDNA Synthesis Kit (Roche Diagnostics, Burgess Hill, UK). Beta-actin was amplified as an internal standard. PCR primers for beta-actin quantitative PCR were 5′- GGAGGGGGTTGAGGTGTT and 3′- TGTGCACTTTTATTGGTCTCAAG. Primers for mAIPL1 quantitative PCR were 5′-GACGCTGCTGACCTCCAT and 3′- GCGGGACAACATAGGGTAGA. PCR was carried out for 40 cycles of 95°C (15 s), followed by 60°C (60 s).
Immunohistochemistry, morphometric analysis and immunoblotting
Wild-type eyes, treated and untreated Aipl1 h/h eyes were harvested; a cautery mark was placed on the limbus to aid orientation of the eyes, to ensure that quantification of photoreceptors and immunostaining were performed on equivalent areas on treated and untreated eyes. The eyes were embedded directly in Optimal Cutting Temperature (OCT) medium (RA Lamb, E. Sussex, UK), frozen in isopentane which had been precooled in liquid nitrogen and were stored at −80°C. Immunofluorescence for AIPL1 and ß-PDE were performed on freshly cut, frozen retinal sections which had been serially sectioned at thickness of 9–10 µm. Cryosection slides were air-dried, post-fixed briefly in 1%PFA and incubated in blocking buffer [10% normal goat serum (Dako Ltd.), 0.1% Tween and 3% BSA in TBS] for 1 h at room temperature. A polyclonal anti-AIPL1 antibody was diluted 1:500 in blocking buffer and slides were incubated overnight at 4°C with the anti-AIPL1 antibody (30). After washes in TBS, slides were incubated with the secondary antibody, Alexa Fluor® 488 goat anti-rabbit IgG conjugate (Molecular Probes, Leiden, The Netherlands) at dilution of 1:150 in TBS for 2 h. Slides were then washed in TBS before a second overnight incubation at 4°C with ß-PDE antibody purchased from Affinity BioReagents (Neshanic Station, NJ, USA), diluted at 1:500 in blocking buffer. Following overnight incubation, slides were washed in TBS and incubated with quaternary antibody, Alexa Fluor® 546 goat anti-rabbit IgG conjugate (Molecular Probes, Leiden, The Netherlands) at dilution of 1:150 in TBS for 2 h. Slides were washed again in TBS and counterstaining of the nuclei was performed using Hoechst 33342 diluted at 1:1000. Slides were mounted in fluorescent mounting medium (Fluormount, Dako) and sections examined by fluorescent microscopy and images captured on Leica DC 500 (Zeiss, Germany) digital camera.
For morphometric analysis, retinas were taken at the final time point of ERG analysis. Cryosections were cut at different levels through the whole horizontal extension of the retina and each eye was orientated so that each section spanned over the superior and inferior retina which are the sites of injection. Freshly cut, frozen sections at thickness of 10–12 µm were stained with propidium iodide (PI) to stain photoreceptor nuclei. To standardize measurements, three sagittally orientated central sections which passed through the optic nerve head were selected for the analysis. Images were captured from each side of the optic nerve head in each section, at approximately midpoint of the distance between the optic disc and the ora serrata. Sections from matched treated and untreated eyes were imaged at the same magnification using confocal microscopy (Zeiss, LSM 510). The nuclei in the outer nuclear layer in this area were counted by a single masked observer. The photoreceptor cell count for each eye was determined by averaging the cell counts from each confocal image of the eye. To control for inter-animal variation in the rate of degeneration, a paired t-test was performed between the photoreceptor cell counts in the treated and untreated eyes of each animal.
For immunoblotting, equal amounts (5–10 µg) of samples (cell lysates or retinal homogenates) were separated on 9% SDS–polyacrylamide gel and transferred by electroblotted to polyvinylidene difluoride membranes (Immobilion-P, Millipore). Membranes were blocked at room temperature for an hour and incubated overnight with primary antibodies, rabbit anti-Aipl1 antibody and rabbit ß-PDE antibody (both described above) and with mouse monoclonal anti-ß-actin antibody (Sigma) for loading controls at 16°C. After washing, the membranes were then incubated with peroxidase-conjugated secondary antibodies (goat anti-rabbit IgG and goat anti-mouse IgG, horseradish peroxidase conjugated antibody, Jackson Immunoresearch) for an hour at room temperature followed by further washes and developed by ECL substrates (ECL Plus Western Blotting Detection System, GE Healthcare).
Semithin sections and transmission electron microscopy
Prior to fixation, eyes for semithin sectioning were orientated by putting a suture on the nasal aspect of the eyes. Eyes were immersion fixed in 3% glutaraldehyde and 1% paraformaldehyde buffered to pH 7.4 with 0.07 M sodium cacodylate-HCL. After a minimum of 12 h of fixation, the cornea and lens were removed. The posterior segments were then osmicated for 2 h with a 1% aqueous solution of osmium tetroxide and dehydrated through ascending alcohols (50–100%, 10 min per step). After three changes of 100% ethanol, the specimens were passed through propylene oxide (2 × 20 min) and left overnight in a 50:50 mixture of propylene oxide and araldite. Following a single change to fresh araldite (5 h with rotation), the specimens were embedded and cured for 24 h at 60°C. Using the nasal suture, the eyes were embedded sagittally so that sectioning occurred in the vertical plane and the retinal sections contained the superior and inferior retina. Semithin (0.7 µm) and ultrathin (0.07 µm) sections were cut using a Leica Ultracut S microtome fitted with an appropriate diamond knife (Diatome histoknife Jumbo or Diatome Ultrathin). Following sequential contrasting with 1% uranyl acetate and lead citrate, the ultrathin section was analysed and photographed using a JEOL 1010 Transmission Electron Microscope operating at 80 kV. Semithin sections were stained with 1% toluidine blue and evaluated using a Leitz Diaplan microscope fitted with a Leica digital camera DC 500 for image capture.
Electroretinograms (ERGs) were recorded from both eyes of Aipl1−/− mice ∼2 weeks following subretinal injections (∼P28) and from Aipl1 h/h mice at 4 weekly intervals following subretinal injections. All animals were dark-adapted overnight (16 h) and anaesthetized with a single intraperitoneal injection of 0.15 ml of a mixture of Domitor (1 mg/ml medetomidine hydrochloride), and ketamine (100 mg/ml), and water in the ratio 5:3:42 prior to testing. The pupils were dilated using on drop of Tropicamide 1% and phenylephrine hydrochloride 2.5%. A single drop of 2% hydroxypropylmethylcellulose was placed on each cornea to keep it moistened and mice were then placed on a heated platform. ERGs were recorded using commercially available equipment (Toennies Multiliner Vision; Jaeger/Toennies, Germany for experiments in Figs 2 and 4; Espion E2; Diagnosys LLC, Massachusetts USA for experiments in Fig. 5) after corneal contact electrodes and midline subdermal reference and ground electrodes were placed. Bandpass filter cut-off frequencies were 1 and 300 Hz. Single flash recordings were obtained at light intensities increasing from 0.1 to 5000 mcds/m2 presented in a ganzfeld dome. Ten responses per intensity level were averaged with an interstimulus interval of 5 s (0.1, 1, 10 and 100 mcds/m2) or five responses per intensity with a 17 s interval (1000 and 3000 mcds/m2). Dark-adapted, rod dominated responses were elicited in the dark with 10 µs flashes of white light (1.37 × 105 cd/m2) presented at intervals of 1 min. Light-adapted, cone responses were elicited in the presence of a 41 cd/m2 rod-desensitizing white background with the same flashes (1.37 × 105 cd/m2) presented at intervals of 1 Hz.
For ERG analysis, the b-wave amplitude at 1000 mcds/m2 was used. The b-wave values (a-wave trough b-wave peak) of the treated (right) eye were paired with the untreated contralateral (left) eye to provide an internal control. This method controls for interanimal variance and test–retest variance present in rodent ERGs. For analysis of the ERG b-wave amplitudes and photoreceptor cell numbers, paired Student's t-tests were used to determine significance (P < 0.05).
H.V.T. is supported by Fonds National Suisse de la Recherche Scientifique, Roche Research Foundation and Holcim Stiftung zur Förderung der wissenschaftlichen Fortbildung. This work was also supported by grants from the European Union (AAVEYE), the UK Department of Health, National Institute of Health Research BMRC for Ophthalmology, NIH Grant EY10581 and Foundation Fighting Blindness.
We thank Drs Melanie Sohocki and Michael Dyer for providing the Aipl1−/− mice. M.H.T. is a Medical Research Council Clinical Training Fellow.
Conflict of Interest statement. None declared.