Tropisms of AAV for Subretinal Delivery to the Neonatal Mouse Retina and Its Application for In Vivo Rescue of Developmental Photoreceptor Disorders

Background Adeno-associated virus (AAV) is well established as a vehicle for in vivo gene transfer into the mammalian retina. This virus is promising not only for gene therapy of retinal diseases, but also for in vivo functional analysis of retinal genes. Previous reports have shown that AAV can infect various cell types in the developing mouse retina. However, AAV tropism in the developing retina has not yet been examined in detail. Methodology/Principal Findings We subretinally delivered seven AAV serotypes (AAV2/1, 2/2, 2/5, 2/8, 2/9, 2/10, and 2/11) of AAV-CAG-mCherry into P0 mouse retinas, and quantitatively evaluated the tropisms of each serotype by its infecting degree in retinal cells. After subretinal injection of AAV into postnatal day 0 (P0) mouse retinas, various retinal cell types were efficiently transduced with different AAVs. Photoreceptor cells were efficiently transduced with AAV2/5. Retinal cells, except for bipolar and Müller glial cells, were efficiently transduced with AAV2/9. Horizontal and/or ganglion cells were efficiently transduced with AAV2/1, AAV2/2, AAV2/8, AAV2/9 and AAV2/10. To confirm the usefulness of AAV-mediated gene transfer into the P0 mouse retina, we performed AAV-mediated rescue of the Cone-rod homeobox gene knockout (Crx KO) mouse, which exhibits an outer segment formation defect, flat electroretinogram (ERG) responses, and photoreceptor degeneration. We injected an AAV expressing Crx under the control of the Crx 2kb promoter into the neonatal Crx KO retina. We showed that AAV mediated-Crx expression significantly decreased the abnormalities of the Crx KO retina. Conclusion/Significance In the current study, we report suitable AAV tropisms for delivery into the developing mouse retina. Using AAV2/5 in photoreceptor cells, we demonstrated the possibility of gene replacement for the developmental disorder and subsequent degeneration of retinal photoreceptors caused by the absence of Crx.


Introduction
Functional analysis of the genes expressed in the mammalian retina is essential for understanding the molecular basis for human retinal development and disease. Recent progress in techniques of comprehensive analysis of gene expression using microarray and next generation sequencing make it possible to obtain many candidate genes that are possibly associated with retinal development and disease [1,2]. Although in vivo analysis of candidate genes using transgenic and/or knockout mice is beneficial in revealing the in vivo functions of the genes, it is still expensive, timeconsuming, and requires a great deal of skill. Therefore, a rapid and convenient method of in vivo gene transfer would be beneficial to the field. To transduce a gene into the mouse retina, in vivo electroporation and virus-mediated gene transfer are currently the most available methods. In vivo electroporation is a method in which plasmid DNA is incorporated into retinal tissue by highvoltage pulses. This method efficiently transduces DNA into rod photoreceptor cells, but much less efficiently into bipolar, amacrine, and Müller glial cells. Moreover, cone photoreceptor, horizontal, and ganglion cells are barely transduced by in vivo electroporation [3]. For virus-mediated transduction, retrovirus, lentivirus, adenovirus, and adeno-associated virus (AAV) have been developed as vehicles for retinal gene transfer. In particular, AAV has many advantages for retinal gene transfer, including high transduction efficiency in non-dividing cells, long-term transgene expression, and low-toxicity. AAV is a non-pathogenic parvovirus, which consists of single-stranded DNA covered with capsid proteins. Each AAV serotype is different in the capsid structure, which leads to different tropisms and transduction efficiencies. Twelve serotypes have currently been used as a vehicle for in vivo gene transfer (AAV2/1-AAV2/12). AAV tropisms for gene transduction into several murine organs and tissues, including the retina, are different according to developmental stage (neonatal or adult) [4,5]. The previous studies on AAV serotype tropism in subretinal injections into the adult mouse retina revealed that retinal pigment epithelium (RPE) cells are efficiently transduced with AAV2/1, and RPE and photoreceptor cells are efficiently transduced with AAV2/2, AAV2/5 [6,7], and AAV2/8 [7]. However, detailed AAV tropisms for transduction into the developing mouse retina have not been reported.
In the current study, we examined the tropism of seven AAV serotypes (AAV2/1, AAV2/2, AAV2/5, AAV2/8, AAV2/9, AAV2/10, and AAV2/11) by subretinal injection into the P0 mouse retina. We revealed that AAV can transduce encoded genes into various retinal cell types in the developing mouse retina. In addition, to validate the usefulness of AAV-mediated gene transfer into the developing mouse retina, we performed AAV-mediated rescue of Crx KO mice. CRX is a transcription factor that is predominantly expressed in photoreceptor cells and is essential for photoreceptor maturation [8,9,10]. We previously reported that Crx KO mice exhibit a total lack of outer segment formation, an absence of both scotopic and photopic electroretinograms (ERG), and progressive photoreceptor degeneration [10]. Our AAVmediated rescue experiment led to a partial restoration of morphological and functional characteristics in the Crx KO retina. In humans, the mutations of Crx are associated with three forms of retinal degeneration, including cone and rod dystrophy (CORD) [11,12,13], retinitis pigmentosa (RP) [13], and Leber congenital amaurosis (LCA) [13,14], all of which can lead to vision loss. Thus, our results also provide a clue to the suitability of gene therapy for development disorders and degeneration of the retina in humans.

Infection Efficiencies of Seven AAV Serotypes
To quantitatively assess the tropism of seven AAV serotypes in the mouse retina, we measured the infection efficiencies for each retinal cell type. Infection efficiency is calculated by the percentage of cells expressing mCherry out of retinal cell-specific markerpositive cells ( Fig. 2A). Photoreceptor and horizontal cells were efficiently transduced with AAV2/1 (Fig. 2B, Rod: 65.563.9% in the middle area, Cone: 80.864.5% and horizontal: 55.669.2%). The AAV2/1-injected retinas also exhibited the highest efficiency for Müller glial cells among seven serotypes (Fig. 2B, 47.965.8%). AAV2/2 and AAV2/8 showed similar transduction patterns with each other, and horizontal and ganglion cells were transduced mainly with these serotypes (

AAV-mediated Rescue Experiment for the Crx KO Retina
To validate the usefulness of AAV-mediated gene transfer into the developing mouse retina, we performed an AAV-mediated rescue experiment for the Crx KO mice. CRX is a transcription factor which plays a crucial role in photoreceptor maturation through photoreceptor gene transactivation [1,8,9,10]. We generated an AAV2/5 vector expressing Flag-tagged Crx cDNA under the control of the Crx 2kb promoter to drive specific expression in photoreceptor cells (AAV2/5-Crx2kb-Flag-Crx = AAV-Crx) [17,18] (Fig. 3A). We injected AAV-Crx subretinally into Crx whole retinas (photoreceptor-side-up) two weeks after subretinal injection with PBS or one of each of the seven serotypes of AAV-CAG-mCherry. (C-I) Fluorescence images of mCherry expression two weeks after subretinal injection into the P0 mouse retina with AAV2/1-(C), AAV2/2-(D), AAV2/5-(E), AAV2/8-(F), AAV2/9 (G), AAV2/10-(H), and AAV2/11-(I), CAG-mCherry. Scale bar represents 1 mm (B) and 50 mm (C-I). ITR: inverted terminal repeat, RPE: retinal pigment epithelium, ONL: outer nuclear layer, INL: inner nuclear layer, GCL: ganglion cell layer. doi:10.1371/journal.pone.0054146.g001 KO retinas at P0. To confirm the Flag-Crx expression in the retina, we performed an RT-qPCR analysis using RNA from the whole retina at three weeks after injection. We observed significant expression of Flag-Crx mRNA in the Crx KO retinas treated with AAV-Crx (Fig. 3B). We further analyzed FLAG-CRX expression in control and AAV-Crx treated retinas by western blotting using an anti-FLAG antibody. We detected a 38 kDa FLAG-CRX band in the AAV-treated Crx KO retinal lysates (Fig. 3C). To determine whether FLAG-CRX is expressed in photoreceptor cells, we performed an immunostaining of the control Crx KO and AAV-Crx-injected Crx KO retinas with the anti-FLAG antibody. FLAG signals were predominantly detected in photoreceptor cells in AAV-Crx-treated Crx KO retinas (Fig. 3D). Non-specific signals were also detected in the RPE and blood vessels in the INL in both control and AAV-Crx-treated Crx KO retinas. This is very likely due to autofluorescence from the RPE and to the reaction of the anti-mouse secondary antibody to endogenous mouse antibodies in the blood vessels (Fig. 3D). AAV-Crx-mediated FLAG-CRX expression was widely distributed throughout the AAV-Crxtreated Crx KO retinas (Fig. 3E, F).

Immunohistochemistry of Crx KO Retinas Treated with AAV-Crx
We further analyzed the expression of RHODOPSIN, GNAT1, S-OPSIN, and M-OPSIN in AAV-Crx-treated Crx KO retinas by immunostaining. The RHODOPSIN protein level was slightly increased in AAV-Crx-injected Crx KO retinas ( Fig. 4A-D), while GNAT1 signals were markedly increased ( Fig. 4E-H). Similarly, S-OPSIN and M-OPSIN signals were markedly increased ( Fig. 4I-P). Consistent with the results of the RT-qPCR analysis shown in Figure 3G-Q, the levels of these molecules also increased in AAV-Crx-treated Crx KO retinas. In addition, GNAT1, S-OPSIN, and M-OPSIN signals appeared to be localized in outer segments in AAV-Crx-injected Crx KO retinas ( Fig. 4E-P). RHODOPSIN and GNAT1 are localized in the rod outer segment and S-OPSIN and M-OPSIN are localized in the cone outer segment. The Crx KO retina lacks outer segment formation [10,20]. This immunohistochemical data suggests that the defect of outer segment formation in the Crx KO retina was partially restored by the subretinal injection of AAV-Crx into Crx KO retinas.
We further examined the morphology of the outer segment in the control Crx KO and AAV-Crx-injected Crx KO retinas in detail by transmission electron microscopy at fifteen weeks after AAV treatment (Fig. 4Q-S). We observed no photoreceptor cells in control retinas because of severe photoreceptor degeneration in the Crx KO retina (Fig. 4Q). In contrast, outer segments containing disk lamina were observed in AAV-Crx-injected Crx KO retinas (Fig. 4R, S). These data showed that the subretinal delivery of AAV-Crx into P0 Crx KO retinas partially restored outer segment formation in the Crx KO mice.

The Improvement of Retinal Function in the Crx KO Retina Treated with AAV-Crx
To evaluate the effect of the AAV-Crx injection on retinal function, we performed electroretinogram (ERG) recordings. Crx KO mice exhibit flat scotopic and photopic ERG responses at P30, resulting from the defect of formation of outer segments and considerable loss of phototransduction molecules in the Crx KO mice [10]. At 6 and 15 weeks after AAV treatment, we measured ERG responses from three control and three AAV-Crx-treated eyes. The mice used for ERG recordings were independently prepared between the two time points. We first tried to record scotopic ERG responses from the AAV-treated mouse eyes in a conventional way [21,22], but no positive response was detected. Therefore, we then measured photopic ERG responses with one hundred stroboscopic flashes of 1.0 log cd-s/m 2 . The ERG responses obtained under these conditions majorly reflect cone photoreceptor functions. Consistent with our previous observation [10], all of the control eyes exhibited completely flat responses at both 6 and 15 weeks after AAV treatment (Fig. 5A top, Table 1). In contrast, two of three AAV-Crx-treated eyes showed significant photopic a-and b-waves at both 6 and 15 weeks after AAV treatment (Fig. 5A bottom, Table 1). One of the AAV-Crx-treated Crx KO mice did not show a detectable ERG photopic response. Although this mouse eye widely expressed FLAG-CRX, the eye was much smaller than the other two mouse eyes and was severely damaged histologically (data not shown), probably due to subretinal injection damage, resulting in physiological dysfunction. This result shows that the subretinal injection of AAV-Crx into Crx KO retinas partially restored the physiological function of Crx KO photoreceptor cells.
Progressive photoreceptor degeneration is also observed in the Crx KO mice [10]. We examined photoreceptor degeneration in the retinas fifteen weeks after AAV treatment by immunostaining using an anti-FLAG antibody (Fig. 5B). Although there was substantial photoreceptor cell death in both control and AAV-Crx-RHODOPSIN (RHO), Cone: S-OPSIN, Horizontal: CALB1, Bipolar: CHX10, Amacrine: PAX6, Ganglion: BRN3B, Mü ller: S100b). According to retinal cell types, the numbers of marker-positive and marker/mCherry double-positive cells were counted for calculation of infection efficiency (n = 3 from three different mice). Cells were counted in the central area of the retina for calculating the infection efficiencies of cone photoreceptor, bipolar, and Mü ller glial cells, and in the central, middle and peripheral areas of the retina for calculating the infection efficiencies of rod photoreceptor and amacrine cells. Infection efficiency was calculated using the formula indicated in Figure   treated Crx KO retinas, AAV-Crx-treated Crx KO retinas had a thicker outer nuclear layer than that of control retinas (Fig. 5B top  panels). Remarkably, most of the surviving photoreceptor cells in AAV-Crx-treated Crx KO retinas expressed FLAG-CRX (Fig. 5B  bottom panels). This observation suggests that photoreceptor cells without FLAG-CRX expression died and that AAV-Crx inhibited photoreceptor cell death in the Crx KO retinas. To confirm this result, we performed immunostaining with an anti-active-caspase-3 antibody, an apoptosis marker, three weeks after AAV treatment (Fig. 5C, D). We observed a significant reduction of apoptotic cell numbers in AAV-Crx-treated Crx KO retinas (Fig. 5C, D). These results show that subretinal injection of AAV-Crx prevented photoreceptor cell death to some extent in the Crx KO retina.

Discussion
The main goal of this study is to establish a method for AAVmediated retinal gene transfer into developing retinas for in vivo analysis of retinal genes and to apply the method to the rescue of retinal degeneration. Through the transduction of AAV-CAG-mCherry into the developing mouse retina, we showed that various retinal cell types were transduced differently with each of seven serotypes of AAV. We quantitatively analyzed infection efficiencies of these AAV serotypes. In addition, we demonstrated the usefulness of AAV transduction into the developing mouse retina for the rescue of the retinal degeneration of Crx KO mice. These results will be useful to researchers who perform in vivo analysis of retinal genes using AAV.
We subretinally injected seven serotypes of AAV into the P0 mouse retina. We chose P0, because the retina at this stage is still developing and amenable to surgical treatment by subretinal injection. Serotypes AAV2/5 and AAV2/9 exhibited remarkable and efficient infection. Photoreceptor cells were selectively and efficiently transduced with AAV2/5. Retinal cells, except for Müller glial cells and bipolar cells, were efficiently transduced with AAV2/9. We observed efficient transduction into horizontal and/ or ganglion cells with AAV2/1, AAV2/2, AAV2/8, AAV2/9 and AAV2/10. These data suggests that in vivo gene transfer by subretinal injection of AAV into the mouse retina enables us to analyze gene function in retinal cell types that are difficult or impossible to transduce by in vivo electroporation (cone photoreceptor, horizontal, and ganglion cells) [3]. To perform an analysis specifically in a certain cell type in the retina, a cell type-specific promoter of relatively short length for each cell type will need to be developed in the future.
In the current study, the infection efficiencies of AAV into bipolar and Müller glial cells were low in all tested serotypes, although AAV2/1 efficiently targeted Müller glial cells. These two retinal cells undergo differentiation at the latest stages during development [21]. In contrast, retinal cells, which are generated at embryonic stages or at early postnatal stages (photoreceptor, horizontal, amacrine and ganglion cells), were transduced with all analyzed serotypes. These results suggest that the AAV serotypes tested in this study are more infectious to differentiated retinal cells than to retinal progenitor or differentiating cells. Since more than 100 AAV serotypes have been isolated [22], AAV serotypes which are capable of efficiently infecting bipolar and Müller glial cells may be found in the future. Our observations on AAV tropism in the developing retina in the current study may expand the application of AAV for in vivo transduction to retinal cell types, such as cone photoreceptors, horizontal cells, and ganglion cells, which are barely transduced by in vivo electroporation.The AAV tropisms shown in the current study were different from those for the adult mouse retina in previous reports. In subretinal transduction into the adult retina AAV2/1 mainly targets RPE cells, and AAV2/2, 2/5, and 2/8 mainly target both RPE and photoreceptor cells [6,7]. However, based on our results using P0 mice, each AAV serotype targeted several retinal cell types. This observation is consistent with the results reported by Surace et al. [5], in which they showed the difference of tropisms for transduction into fetal, neonatal and adult retinas using AAV2/ 1-, 2/2-, and 2/5-CMV-EGFP. There are two possible explanations for the shift of AAV tropism between the developing retina and the mature retina. The first is that the expression or abundance of AAV receptors in host retinal cells may change during retinal development. The second is that injected virus particles may more easily diffuse across the retina due to the smaller size of the P0 retina and dynamic cell migration during retinal development. The differences of AAV tropisms between neonatal and adult mice are also observed in the aorta, liver and kidney [4]. This suggests that the time point of transduction of an AAV vector during development is important to appropriately perform gene transfer to a cell type of interest.
We attached the Crx 2kb promoter to the AAV vector for the rescue experiment on the Crx KO retina. We previously reported that this promoter directs specific gene expression in developing and mature rod and cone photoreceptor cells [17,18]. Expression analysis by RT-qPCR and immunohistochemistry showed improved expression of both rod-specific genes (Rhodopsin and Gnat1) and cone-specific genes (S-opsin and M-opsin) in AAV-Crx-treated Crx KO retinas (Fig. 3G-Q and 4A-P). This indicates that the Crx 2kb promoter successfully drove the Crx expression in both rod and cone photoreceptor cells. Thus the Crx 2kb promoter can be used with AAV to drive expression specifically in photoreceptor cells.
AAV-Crx modestly but significantly improved the photopic ERG responses, but we were not able to detect scotopic ERG responses in AAV-treated eyes. In order to detect relatively weak ERG responses in AAV-Crx-treated Crx KO eyes, we stimulated the mouse eyes with one hundred stroboscopic flashes (1.0 log cds/m 2 ). However, under these conditions, it is very difficult to measure scotopic ERG responses because the intermittently repeated flashing lights abrogate the dark adaptation that is necessary for scotopic ERG recordings. Because of such technical difficulties, we were not practically able to measure scotopic ERG responses in our current study. AAV-Crx also partially restored outer segment formation in the Crx KO mice. This result not only confirms that AAV transduction into the P0 retina is a useful method for in vivo retinal gene transfer but also suggests that gene AAV-treated Crx KO retinas from three different mice respectively. An anti-FLAG antibody was used to detect FLAG-CRX. ACTB (b-actin) was used as a loading control. (D-F) Immunostaining of the Crx KO retinas treated with AAV-Crx or PBS using an anti-FLAG antibody. The distribution of FLAG-CRX expression in control and AAV-Crx treated Crx KO retinas (E, F). Enlarged images in white boxes (a-c and d-f) in Figure 3E and F, respectively. Scale bar represents 50 mm (D) and 1 mm (E, F). (G-Q) Expression analyses of eleven genes related to human retinal diseases three weeks after treatment (Control Crx KO retinas: n = 3 from three different mice and AAV-treated Crx KO retinas: n = 4 from four different mice). Rhodopsin (G), Gnat1 (H), Sopsin (I), M-opsin (J), Pde6g (K), Slc24a1 (L), Rdh12 (M), Rpgrip1 (N), Nrl (O), Cabp4 (P), and Fscn2 (Q). Control retinas were injected with PBS (Vehicle). Rpl4 was used for normalization. Primers for qPCR were listed in Table S1. Error bar represents the SD from the means of three control retinas and four treated retinas. ** p,0.01, * p,0.05. ITR: inverted terminal repeat, RPE: retinal pigment epithelium, ONL: outer nuclear layer, INL: inner nuclear layer, GCL: ganglion cell layer. doi:10.1371/journal.pone.0054146.g003 In addition, this is the first report on AAV-mediated rescue for mice with mutations of a transcription factor regulating photoreceptor development and related to human photoreceptor degeneration. Nrl, Nr2e3, and Otx2 also play crucial roles in photoreceptor development through transcriptional regulation of photoreceptor genes [1,23,24,25]. The mutations of these genes in humans are associated with several types of retinal degeneration (RetNet: https://sph.uth.tmc.edu/retnet/home.htm). Our current result suggests that human retinal degenerations caused by the mutations in these transcription factors can be restored by AAVmediated gene therapy. This possibility will be examined by AAVmediated rescue of mice with these gene mutations in the future.

Animals
For the evaluation of the tropism of AAVs in vivo, we used ICR mice (Charles River). The Crx KO mice were generated as described in our previous study [10]. All procedures conformed to the ARVO statement for the Use of Animals in Ophthalmic and Vision Research, and these procedures were approved by the Institutional Safety Committee on Recombinant DNA Experiments and the Animal Research Committee of Osaka Bioscience Institute (approval ID 10-401) and Institute for Protein Research, Osaka University (approval ID 24-05-0). Mice were housed in a temperature-controlled room with a 12-hour light/dark cycle. Fresh water and rodent diet were available at all times.

AAV Production
AAV was produced by triple transfection of an AAV vector plasmid, an adenovirus helper plasmid, and an AAV helper plasmid (pAAV2/1, pAAV-RC [Agilent technologies], pXR5, pAAV2/8, pAAV2/9, pAAV2/rh10, and pEEV2/11) into AAV-293 cells by the calcium phosphate method. The cells were harvested at 72 hours after transfection, and lysed by four freeze-and-thaw cycles. The supernatant was collected by centrifugation, and treated with Benzonase nuclease (Novagen) to eliminate cellular DNA/RNA and excess plasmid DNAs. This virus preparation was used for subretinal administration. A titer of each AAV (in vector genomes (VG)/mL) was determined by qPCR using SYBR GreenER Q-PCR Super Mix (Invitrogen) and Thermal Cycler Dice Real Time System Single MRQ TP870 (Takara). The primers used for AAV titrations are listed in Table S1. The titers of all serotypes of AAV-CAG-mCherry used in this study were adjusted to approximately 2610 12 VG/mL. The titer of AAV2/5-Crx2kb-Flag-Crx used in the Crx KO rescue experiment is 2.6610 12 VG/mL.

Subretinal Injection
Subretinal injection of AAV was performed as described elsewhere [3,29]. P0 mice were anesthetized by chilling on ice, the eye was opened by cutting along the fused junctional epithelium where the two eyelids come together, and a small incision was made with a 30-gauge needle in the sclera near the junction with the cornea. 0.4 mL of an AAV preparation was injected into the subretinal space through the incision using an Ito micro syringe (Ito Corporation) with a 33-gauge blunt-ended needle under a dissecting microscope. Fast Green dye was added to AAV preparations at a final concentration of 0.1% as a tracer to confirm that the AAV preparations were injected into the subretinal space [29]. For histological analyses, we only used retinas in which the dye in the AAV preparation was confirmed to be evenly distributed and which were without severe damage caused by the injection process.

Immunostaining
For immunohistochemistry, 14 mm thick retina sections were washed twice in phosphate-buffered saline (PBS), and permeabilized with 0.1% Triton X-100 (wt/vol) in PBS, then incubated with PBS containing 4% donkey serum (vol/vol) for 1 h to block samples. The samples were incubated with a primary antibody at 4uC overnight. After washing with PBS, these samples were incubated with secondary antibodies at 25uC for 1 hour. In the current study, we also used the following primary antibodies: anti-

Quantification of Infection Efficiency
AAV-CAG-mCherry-injected retinas were co-immunostained with cell type-specific markers for each retinal cell shown above and an anti-mCherry antibody. We counted the number of marker-positive cells and marker/mCherry double-positive cells according to cell types for the calculation of infection efficiencies.

Western Blot Analysis
Western blot analysis was performed as described previously [27]. The membrane was incubated with an anti-FLAG antibody (1:1000, F1804, Sigma). The membrane was then incubated with a horseradish peroxidase-conjugated goat antibody against mouse IgG (1:10000, Zymed). For the secondary immunoreaction, the PVDF membrane was incubated with WB Stripping Solution (Nacalai Tesque) to remove antibodies, and blocked again with 5% skim milk (wt/vol) in TBS. Further immunoblots were performed using a mouse antibody against b-actin (ACTB, 1:5000, Sigma).

RT-qPCR
The mouse retinas were harvested and dissected at 3 weeks after injection. Total RNA (1 mg) from the retina was isolated using TRIzol reagent (Invitrogen) and converted to cDNA using Superscript II Reverse Transcriptase (Invitrogen). Real time PCR was performed using SYBR GreenER Q-PCR Super Mix (Invitrogen) and Thermal Cycler Dice Real Time System Single MRQ TP870 (Takara) according to the manufacturer's instructions. Quantification was performed by Thermal Cycler Dice Real Time System software version 2.0 (Takara). The primer sequences used for qPCR are listed in Table S1.

ERG Recordings
Mice were anesthetized with an intramuscular injection of 80 mg/kg ketamine and 16 mg/kg xylazine. Pupils were dilated with topical 0.5% tropicamide and 0.5% phenylephrine HCl, and the mice were placed on a heating pad for the duration of the ERG recordings. ERGs were recorded with a gold wire loop placed on the cornea anesthetized with 1% tetracaine. A gold wire electrode was placed on the sclera 1 mm from the temporal limbus as the reference electrode. The mice were placed in a Ganzfeld bowl and 100 stroboscopic stimuli of 1.0 log cd-s/m 2 (PS33 Plus; Grass Telefactor) were averaged with a repetition rate of 1 sec to record the ERGs. Signals were amplified and bandpass filtered between 1 and 1000 Hz (Power Lab; AD Instruments, Castle Hill, mice showed no ERG response, while two of the three treated eyes from different mice showed significant responses. (B) Immunostaining of FLAG-CRX in the retinas at fifteen weeks after AAV-Crx treatment. (C and D) Immunostaining of active-caspase-3 and FLAG-CRX in Crx KO retinas three weeks after AAV-Crx treatment (C). The number of active-caspase-3-positive cells (D, Control retinas: n = 3 from three different mice and AAV-treated Crx KO retinas: n = 4 from four different mice). Arrowheads indicate active-caspase-3-positive cells. Scale bar represents 50 mm. Error bar represents the SD from the means of three control retinas and four treated retinas. * p,0.05. RPE: retinal pigment epithelium, ONL: outer nuclear layer, INL: inner nuclear layer, GCL: ganglion cell layer. doi:10.1371/journal.pone.0054146.g005 Table 1. Quantitative analysis of ERG amplitudes in control and AAV-Crx treated eyes. Australia). Amplitudes of both a-and b-waves were quantified from photopic ERG responses.

Transmission Electron Microscopy
Specimens for transmission electron microscopy were prepared in the following manner. Eyes were enucleated from anaesthetized mice. Following the removal of anterior segment, each posterior eyecup was fixed with 2% glutaraldehyde and 2% paraformaldehyde in a cacodylate-based buffer adjusted at pH 7.4. After fixation with 1% osmium tetraoxide for 90 min, the retinas were dehydrated through a graded series of ethanol (50%-100%) and nbutylglycidylether. Finally, they were embedded in epoxy resin. Ultrathin sections were cut on an ultramicrotome (Ultracut E, Reichert-Jung, Vienna, Austria), and stained with uranyl acetate and lead citrate. Retinas were observed by transmission electron microscope (1200EX, JEOL, Japan).

Statistical Analysis
Statistical significance was calculated with a Student's t test. A value of p,0.05 was taken to be statistically significant. Data are presented as means 6 SD.