Repair of Retinal Degeneration following Ex Vivo Minicircle DNA Gene Therapy and Transplantation of Corrected Photoreceptor Progenitors

Abstract

The authors describe retinal reconstruction and restoration of visual function in heritably blind mice missing the rhodopsin gene using a novel method of ex vivo gene therapy and cell transplantation. Photoreceptor precursors with the same chromosomal genetic mutation were treated ex vivo using minicircle DNA, a non-viral technique that does not present the packaging limitations of adeno-associated virus (AAV) vectors. Following transplantation, genetically modified cells reconstructed a functional retina and supported vision in blind mice harboring the same founder gene mutation. Gene delivery by minicircles showed comparable long-term efficiency to AAV in delivering the missing gene, representing the first non-viral system for robust treatment of photoreceptors. This important proof-of-concept finding provides an innovative convergence of cell and gene therapies for the treatment of hereditary neurodegenerative disease and may be applied in future studies toward ex vivo correction of patient-specific cells to provide an autologous source of tissue to replace lost photoreceptors in inherited retinal blindness. This is the first report using minicircles in photoreceptor progenitors and the first to transplant corrected photoreceptor precursors to restore vision in blind animals.

Graphical Abstract

Barnea-Cramer et al. show retinal reconstruction and rescue of vision in heritably blind mice following ex vivo minicircle or viral gene correction and cell transplantation of photoreceptor precursors. Converging cell and gene therapies may be applied in future studies toward ex vivo correction of patient-specific cells to replace lost photoreceptors in hereditary neurodegeneration.

Introduction

Photoreceptor degeneration is the leading cause of blindness in many inherited retinopathies, including retinitis pigmentosa (RP). A central goal of human therapy is to develop methods to safely and robustly treat patients afflicted with genetic degenerative disorders. Cell replacement studies in mice have shown improvement in visual function following transplantation of photoreceptor precursors in animals with partial1, 2, 3, 4, 5 or complete6, 7, 8 degeneration of the outer nuclear layer (ONL). In the former, the recently described mechanism of cytoplasmic transfer may restore functional proteins to host cells.9, 10, 11, 12 In humans, tissue for retinal replacement therapy may be sourced from autologous induced pluripotent stem cells (iPSCs)¹³ or allogeneic stem cell lines obtained from stem cell banks. In RP patients, patient-specific iPSCs may provide an autologous source of cells to replace the lost photoreceptor cells of the retinal ONL. However, in order to provide sustained results, the use of patient-derived iPSCs would require the disease-causing gene mutation to be corrected before cells are transplanted. Transplantation of genetically corrected photoreceptors has not yet been reported, and it is not known if ex vivo corrected photoreceptors, which were missing functional properties prior to treatment, would be able to survive, mature, and function within the damaged retina. A successful method for ex vivo gene therapy and attainment of visual improvement through transplantation of rescued cells may provide an important stepping stone for translational research in humans.

In vivo, gene therapy has been immensely successful for retinal disease. The most commonly used vectors in retinal therapy are recombinant adeno-associated virus (rAAV) vectors, as these represent a safe and efficient strategy for gene delivery to non-dividing cells. A recent study additionally shows de novo regeneration of rods following AAV gene transfer to Müller glia in mice with partial retinal degeneration.¹⁴ Despite its advantages, AAV gene delivery still faces limitations, particularly the restricted packaging size and possible immune reactions against the viral capsid.¹⁵ Non-viral vectors may be beneficial for gene therapy, as they have low immunogenicity and a low risk of insertional mutagenesis; are easy to produce on a large scale; and, most relevantly, have a large packaging capacity that allows delivery of large transgenes and entire genomic DNA fragments. Currently, the majority of non-viral delivery methods are not applicable for clinical gene delivery due to low efficiency or toxicity.^16,17 Administration of gene therapy to photoreceptor cells ex vivo instead of in vivo can overcome some of the barriers and extracellular challenges of gene delivery and, thus, provides opportunities for long-term in vivo assessment of non-viral therapeutic approaches. Plasmid transgene expression is invariably suppressed over time,¹⁸ as the high unmethylated CpG content in the plasmid bacterial backbone prompts the silencing of episomal transgene expression.¹⁹ Minicircles (MCs), are plasmid derivatives devoid of a bacterial backbone. MCs are produced as circular expression cassettes and reduced to the minimal size required for transgene expression and, thus, are more adept to achieve sustained gene expression.20, 21, 22, 23 The most obvious characteristic of minicircle vectors is their small size compared to that of plasmids. The size of MCs is an advantage for overcoming obstacles on the way to gene expression. Small vectors have better bioavailability characteristics than larger ones; thus, MC DNA molecules have been proposed to be better suited for gene transfer and expression¹⁸ than plasmids, as a reduction in size is strongly associated with increased expression.^24,25 The size of MCs is reduced in comparison to AAV vectors as well, and may prove advantageous in bioavailability for gene transfer. An additional advantageous feature of MC DNA is its structure. The supercoiled structure remains unchanged through recombination, and it has been proposed that supercoiled plasmids are superior to linear plasmids in transduction efficiency.^22,23 Supercoiled molecules are better suited to reach the perinuclear region and prevail in a higher intracellular concentration compared to linear plasmids.²³

Previous studies have suggested that MCs are superior to plasmid DNA vectors for gene delivery, as they offer increased and persistent gene expression compared to plasmid DNA both in vitro and in vivo.^22,26, 27, 28, 29, 30 MCs have been shown to be capable of sustained expression for months, with expression levels that are eventually 10- to 1,000-fold higher than their corresponding plasmids.^19,31,32 To date, MC technology has not been used in ocular gene therapy or as a method for ex vivo gene therapy.

The high packaging capacity of MCs may be relevant for the efficient replacement of large retinal degeneration genes, such as ABCA4, USH2A, and RP1, which are untreatable with AAV.

To gain insights into the therapeutic applicability of photoreceptor-precursor cells, which have been genetically modified by minicircle DNA, we aimed to model the clinical scenario of RP patients receiving a transplant of treated tissue to replace degenerated photoreceptors. We treated and transplanted completely non-functional murine rod photoreceptor precursors sourced from donor mice into adult blind mice with the same genetic mutation. Specifically, we treated Rho^−/− donor rod photoreceptor cells ex vivo with human RHO gene therapy and transplanted treated photoreceptors into adult blind Rho^−/− mice. We assessed MCs alongside rAAV gene delivery as a potential therapeutic strategy for ex vivo correction of photoreceptor cells. The use of late-stage Rho^−/− mice, which have almost no remaining ONL, represents the clinically relevant time point of transplantation in blind RP patients.^6,7 Successful transplantation of ex vivo genetically repaired rod precursors may provide insight into autologous cell transplantation in RP and other neurodegenerative diseases.

Results

Vector Design and Sustained Gene Expression in Ex Vivo Cultured Primary Retinal Cells

We used mice with a homozygous rhodopsin null mutation (Rho^−/−) as both host and donor animals. Due to the absence of the Rho gene, rod function is completely absent from birth,³³ and by approximately 3 months of age, a near-complete loss of photoreceptors occurs.³⁴ To identify donor cells efficiently following transplantation, we used Rho^−/− donor animals that also express GFP specifically in post-mitotic rod photoreceptors via the Nrl promoter (Rho^−/−, Nrl-GFP). We designed therapeutic vectors driven by the human photoreceptor-specific rhodopsin kinase (RHOK, also known as GRK1) promoter, to deliver the rhodopsin gene to photoreceptors in cis with a red fluorescent reporter (DsRed) (Figure 1A). Co-expression of the endogenous Nrl-GFP and the exogenous DsRed in treated Rho^−/−, Nrl-GFP cells would indicate transduction of rod cells. Because ex vivo retinal gene therapy has not yet been described, we decided to use a recombinant mutant AAV, rAAV2/2(Y444F), which is already in practice in retinal research³⁵ alongside the novel MC vector. We designed the construct (~3 kb) to be appropriate for the packaging capacity of AAV (under ~4.7 kb) and used the same plasmid to produce both therapeutic vectors (Figures 1B and 1C).

Figure 1.

Vector Design and In Vitro Transgene Expression

(A) The transgene construct contains a rhodopsin kinase (RHOK) promoter, RHO coding sequence, an internal ribosome entry site (IRES), and a DsRed fluorescent marker. (B) The recombinant, capsid mutant AAV vector- rAAV2/2(Y444F) was produced by cloning the construct into a plasmid backbone containing inverted terminal repeats (ITRs) of AAV2 on either side. (C) Minicircles (MCs) are produced by site-specific recombination, separating a parental plasmid into two parts: (1) the minicircle that includes the expression cassette and (2) the bacterial backbone, which is then subject to degradation. (D and E) MC-treated (D) and rAAV2/2(Y444F)-treated (E) photoreceptors in vitro, co-expressing GFP from the endogenous Nrl promoter and exogenous DsRed. Scale bars, 10 μm. (F) In vitro mRNA expression was assessed by quantitative real-time PCR. Compared to untreated Rho^−/− retinal cells, upregulation of the Rhodopsin gene was found in cells treated by MC and by rAAV2/2(Y444F) (for both, ***p < 0.001). Both gene therapy groups differed from WT expression (p < 0.001), but no difference was found between the two treatment groups (p = 0.39). One-way ANOVA, F(3, 24) = 53.4, p < 0.001, Tukey’s correction for multiple comparisons. The in vivo Rho expression level at P13 was higher than that in WT neonate retina harvested at P0–P3 and maintained in vitro for 14 days, t = 6.18, ***p < 0.001. n = 7 in each group. Results are expressed relative to Rho in dissociated WT retinal cells.

As MCs had not been previously studied in retinal cells, we first assessed patterns of in vitro retinal-cell transfection of wild-type (WT) cells with CMV.GFP.SV40polyA MC DNA (Figures S1A and S1B) and compared MCs to a conventional CMV.CBA.GFP plasmid (Figure S1C). We achieved sustained expression of protein using MC DNA for 7 days. Plasmid DNA, in comparison, peaked at 72 h and produced transient expression, with GFP expression lost by day 7. We then studied the potential for sustained expression of rhodopsin transgenes in vitro in rod progenitors. We dissociated retinas from neonate Rho^−/−, Nrl-GFP mice (postnatal day [P]0–P3) and treated retinal cells in vitro with either rAAV2/2(Y444F) or MC vectors carrying the RHOK.RHO.IRES.DsRed expression cassette. Following gene delivery, we observed DsRed expression co-localized with GFP in rod precursor cells in both groups (Figures 1D and 1E), with DsRed onset occurring 12 h following gene delivery with MC and 3 days following rAAV2/2(Y444F) delivery. DsRed expression did not differ between MC (61% ± 7% standard error of the mean [SEM]) and rAAV 2/2 (72% ± 6.5% SEM) treatments 14 days post-transfection (t = 2.01, p = 0.077, ns, unpaired t test). To evaluate the efficiency of both vectors in delivering rhodopsin to cells, we assessed mRNA expression in unsorted cells by quantitative real-time PCR 14 days post-transfection. We found a significant difference in rhodopsin expression between MC-treated, rAAV2/2(Y444F)-treated, untreated Rho^−/−, and WT cultured cells (F(3, 24) = 53.4, p < 0.001, one-way ANOVA; n = 7 per group; Tukey’s correction) (Figure 1F), with upregulation of rhodopsin compared to untreated Rho^−/− in both MC-treated (p < 0.001) and rAAV2/2(Y444F)-treated (p < 0.001) cells. MC and rAAV2/2(Y444F) vectors reached 60% ± 7% (mean ± SEM) and 44% ± 8% (mean ± SEM) of WT rhodopsin levels, respectively (both different from WT, p < 0.001). The two treatments did not differ in rhodopsin expression level (p = 0.39). To understand the relationship between photoreceptor maturation and rhodopsin expression, we further compared Rho expression of WT cells in vitro (obtained at post-natal days [P]0–P3, cultured 14 days) to that of cells obtained from mice at P13. Rho expression in cultured WT retinal cells was significantly lower than that in whole retinas of P13 WT mice (t = 6.18, p < 0.001); possibly because dissociated photoreceptor cells in vitro do not develop outer segments (OSs), in which Rho is sequestered.

Reconstruction of the Host ONL by Ex-Vivo-Treated Rods

We performed transplantation experiments to assess the possibility of cell replacement using ex-vivo-treated photoreceptors. We dissociated retinas from Rho^−/−, Nrl-GFP neonatal mice (P1–P3). To enrich cultures in rod precursors prior to transplantation, we used magnetic activated cell sorting (MACS) against the CD73 rod cell surface marker, as previously described;^36,37 cultures were enriched to 87.2% ± 7.1% (mean ± SEM) GFP-positive rod precursor cells (Figure S2A).

The RHOK.RHO.IRES.DsRed gene replacement was delivered to rod-enriched Rho^−/−, Nrl-GFP cultures by means of either rAAV2/2(Y444F) or MC vectors. A third group of Rho^−/−, Nrl-GFP cells received no treatment and were cultured and sorted alongside the treatment groups for sham transplantation. We transplanted cells on day 3 of culture, after verifying DsRed expression in the rod precursor cells of both treatment groups. We dissociated rod-enriched cultures again immediately prior to transplantation to obtain a single-cell suspension and transplanted 2 × 10⁵ cells into the subretinal space of 3-month-old Rho^−/− hosts (Figures S2B–S2G).

To confirm stable gene expression from both MC and AAV vectors, we examined host mice 3 months following transplantation (at approximately 6 months of age). In both treatment groups (n = 8 per group), we identified a homogeneous layer of cells in the subretinal space of host mice, with the majority of cells expressing fluorescent protein at levels detectable in vivo by scanning laser ophthalmoscopy (SLO) (shown later in Figure 3A). Histologically, the new layer was correctly localized in the subretinal space between the residual retina and the retinal pigment epithelium (RPE) of host adult Rho^−/− mice (shown later in Figure 3B). Endogenous GFP signal in cells of the reconstructed layer confirmed the cells’ identity as transplanted donor Rho^−/−, Nrl-GFP cells and co-expression of exogenous DsRed provided a marker for successful ex vivo gene transfer (Figure 2C). The thickness and localization of the graft (Figure 3D) was comparable to the WT ONL (Figure 3E). The grafted ONL was in contact with the host retina but without an obvious OS layer pointing toward the RPE. The possibility of erroneously identifying host photoreceptors as transplanted cells resulting from cytoplasmic fusion with GFP-positive donors9, 10, 11, 12 is reduced by the near absence of an ONL in host animals (see Figures S3A and S3B for images of adult Rho^−/− retina compared to WT retina).

Figure 3.

In Vivo Rhodopsin Expression and Maturation in Treated Donor Cells

(A–C) RHO expression was found in MC-treated (A and B) and rAAV2/2(Y444F)-treated (C) GFP/DsRed-positive donor cells residing in the subretinal space of host Rho^−/− mice following transplantation. Cells extended slender inner segments with short OS structures expressing rhodopsin. (D–F) Phosphodiesterase 6 beta (PDE6β; D), retinal OS membrane protein 1 (ROM1; E), and G protein subunit alpha transducin 1 (GNAT1; F) were expressed in OS structures of cells treated by MC gene therapy, providing indication of rod cell maturation in grafted cells. (G) The pan-photoreceptor marker recoverin was expressed in the cell cytoplasm, co-localizing with cytoplasmic GFP, confirming cell identity as photoreceptors, and providing a control for localization of OS proteins. White arrowheads indicate protein expression; outlined arrows indicate staining of a non-GFP photoreceptor in the thin remaining host ONL. Scale bars, 50 μm.

Figure 2.

Reconstruction of the Adult Rho^−/− ONL by Treated Rod Precursor Cells

(A) In vivo image of the retina obtained by scanning laser ophthalmoscopy (SLO) 3 months post-transplantation of MC-treated cells. Transplanted GFP/DsRed-positive cells are observed in autofluorescence (AF) mode as white dots or clusters. NIR, near infra-red. (B) Three continuous images of an adult Rho^−/− retina transplanted with MC-treated cells. Scale bar, 100 μm. (C) Grafted cells co express endogenous Nrl-GFP and DsRed transgenes. (D) Reconstructed layer of GFP/DsRed cells residing in the subretinal space of the Rho^−/− host transplanted with MC-treated cells. (E) Age-matched Rho^+/+, Nrl-GFP retina. Grafted cells residing in the subretinal space did not extend a clear layer of segments toward the RPE. (F–H) Representative layers of MC-treated (F), rAAV2/2(Y444F)-treated (G), and sham-treated (H) grafts. n = 5 specimens per group. Scale bars, 50 μm. INL, inner nuclear layer; ONL, outer nuclear layer; RPE, retinal pigment epithelium; IS, inner segment; OS, outer segment. (I) Three months following transplantation, the proportions of GFP-positive surviving cells were 4.95% ± 0.39% and 3.26% ± 0.65% of cells in the MC and AAV treatment groups, respectively. The proportions of cells expressing both GFP and DsRed were 63.3% ± 9.5% and 86.3% ± 4.5% for the MC and AAV groups, respectively. In the sham transplantation group, 0.05% ± 0.008% of transplanted GFP-positive cells survived 3 months following transplantation. A difference was observed between the three treatment groups in number of surviving cells (F(2, 12) = 31.49, p < 0.0001, one-way ANOVA) and DsRed expression (F(2, 12) = 53.3, p < 0.0001, one-way ANOVA). Both MC and AAV gene therapy groups were significantly different from sham transplantation (both ps < 0.0001), but no difference was observed between the two treatment groups in the number of surviving cells (p = 0.058, ns) or percentage of DsRed-expressing cells (p = 0.063, ns). N = 5 per group; one-way ANOVA, Bonferroni correction for multiple comparisons.

To assess cell survival and transgene expression, we calculated the percentage of GFP- and DsRed-positive cells in retinal sections from the three experimental groups (Figures 2F–2H). As expected, we did not find any DsRed-positive cells in sham-transplanted animals, and we found a difference between the three experimental groups, both in number of grafted GFP-positive cells (F(2, 12) = 31.5, p < 0.0001, one-way ANOVA) and DsRed-positive cells in the subretinal space (F(2, 12) = 53.3, p < 0.0001, one-way ANOVA). In both MC and AAV gene therapy groups, however, more GFP-positive cells survived compared to sham transplantation (both ps < 0.0001). There was no significant difference between MC and AAV groups in the number of GFP-positive cells (p = 0.058) or in the percentage of these cells expressing DsRed (p = 0.063). Indeed, the majority of surviving GFP-positive cells also expressed DsRed (63% ± 9% and 86% ± 4% for MC and AAV groups, respectively), suggesting that transduction with the DsRed-tagged rhodopsin vector increased donor cell survival (Figure S3). These findings confirmed sustained transgene expression from both vectors 3 months following transplantation, despite a long-term reduction in the number of surviving GFP-positive cells between 1 month (Figure S4) and 3 months (Figure 2I) post-treatment as previously observed in the absence of immune suppression.⁶

In Vivo Maturation of Transplanted Cells

We detected immunoreactivity against rhodopsin in the donor cells of both the MC (Figures 3A and 3B) and rAAV2/2(Y444F) groups (Figure 3C) 3 months following transplantation. Since donor cells and host animals are homozygous for a null rhodopsin mutation (Rho^−/−), expression of rhodopsin was observed in approximately 60% of the surviving GFP- and DsRed-positive cells in both treatment groups and indicates production of the protein following successful gene transfer and replacement of the absent gene by WT RHO (see Figure S5 for the absence of RHO expression in age-matched Rho^−/−). Transplanted cells developed distinct but morphologically abnormal OSs, as evidenced by rhodopsin immunostaining restricted to short linear projections from one end of the GFP-expressing cell body (Figures 3A–3C). This is consistent with correct RHO protein folding and trafficking.

To assess the functional maturation of rod photoreceptors, we also assayed the pattern of the rod OS proteins phosphodiesterase 6 beta (Pde6b), retinal OS membrane protein 1 (Rom1), and G protein subunit alpha transducin 1 (Gnat1) 3 months after transplantation. Treated cells, while morphologically abnormal, had matured to express functional rod OS-specific proteins assayed (Figures 3D–3F; Figure S6), indicating that they may be capable of supporting vision in mice, as previously described in photoreceptors with structural deficits.^6,7,38 We used the pan-photoreceptor marker recoverin as a control for protein localization and found recoverin appropriately localized in the cell cytoplasm, confirming photoreceptor cell identity (Figure 3G).

We next assessed connectivity between the graft and host retina and found the presence of the synaptic vesicle membrane protein synaptophysin in the border region between host INL and grafted cells (Figures 4A and 4B), in a pattern consistent with donor rods forming synaptic terminals in close proximity to the host inner retina.^6,7,39 Synaptophysin staining in untreated adult WT Rho^+/+, Nrl-GFP retina and Rho^−/−, Nrl-GFP retina is presented for comparison (Figure S7). We examined glial fibrillary acidic protein (GFAP) to determine whether a glial barrier had formed between the host retina and transplanted cells. Most donor-derived Müller glia were removed by MACS prior to transplantation, since they do not express CD73; therefore, Müller cells observed enveloping the cell mass would arise predominantly from the host. In both treatment groups, we found glial processes extended into the mass of grafted cells without clear formation of a horizontal glial barrier (Figures 4C and 4D). A complete glial barrier would impede interactions of donor cells with the host retina, but some Müller cells are required for the function of photoreceptors; thus, integration and support of the graft by host glia may contribute to viability of the newly formed ONL.

Figure 4.

Graft Host Connectivity

(A and B) Synaptophysin was appropriately localized between the host INL and MC-treated (A) or AAV-treated (B) donor cell mass. Dashed lines outline donor rods. Scale bars, 10 μm. (C and D) Host Müller cells are identified by glial fibrillary acidic protein (GFAP). Müller cell processes extended from the host INL into the graft without formation of a clear horizontal glial barrier in MC-treated (C) and rAAV2/2(Y444F)-treated (D) cells. White arrowheads indicate protein expression. Scale bars, 50 μm.

Recovery of Visual Function

We saw evidence of improved visual function using three independent tests conducted in all animals of the experimental groups (MC, rAAV2/2(Y444F), and sham) 3 months following transplantation. We assessed visual function before surgery in the same animals to stratify experimental groups (Figure S8) and evaluate discrete functional restoration. We tested untreated age-matched WT and Rho^−/− mice alongside treatment groups to validate test protocols and provide reference behavior.

We first tested the pupil light reflex (PLR),^1,6,40, 41, 42 a brainstem reflex triggered by photic input, to examine whether light sensitivity was restored in the retina and, if so, whether signals from the retina could be relayed toward the brain. We calculated the PLR of experimental groups by normalizing animals’ pupil area following a rod-specific light stimulus (for light stimulus selection, see Figure S9) to the area prior to stimulus (Figure 5A). Groups differed in their pupil light response following transplantation (F(2, 18) = 14.78, p < 0.001, one-way ANOVA), with an improvement over sham transplantation in both MC-treated (p < 0.0001) and rAAV2(/2Y444F)-treated (p < 0.05) animals. Improvement in PLR was greater in the MC-treated group compared to the rAAV2(Y444F)-treated group (p < 0.05) (Figure 5B; Video S1).

Figure 5.

Improvement of Visual Function and Behavior

(A) Representative images of pupils before and after light stimulation. (B) The three treatment groups differed in PLR after light exposure, F(2, 18) = 14.78, p < 0.001, with mean pupil area significantly reduced in the rAAV2/2(Y444F) group (*p = 0.043) and the MC-treatment group (***p < 0.0001). The MC-treatment group improved more than the rAAV2/2(Y444F) group (*p < 0.05). (C) The BLA arena. (D) The three groups differed in the amount of time spent in the dark compartment, F(2, 18) = 4.97, p < 0.05). Sham-transplanted mice avoided light less than MC-treated (*p = 0.036) and rAAV2/2(Y444F)-treated mice (*p = 0.047). There was no difference between the two gene therapy groups (ns, not significant). (E) The OMR arena. (F) Following unilateral transplantation to the left eye, tracking in the clockwise direction (driven by the treated eye) was observed more often in both MC-treated mice (t = 4.05, **p = 0.007, paired t test) and rAAV2/2(Y444F)-treated mice (t = 4.5, **p = 0.004, paired t test) but not in sham-treated mice (t = 0.47). The number of head tracks elicited by the treated eye was different between the three groups, F(2, 18) = 8.77, p < 0.01, so that sham-treated mice tracked the grating less than MC- and rAAV2(Y444F)-treated mice (for both, **p < 0.017). Dashed lines represent the mean response of age-matched WT and untreated Rho^−/− mice. One way ANOVA, Bonferroni correction, n = 7 per group. Error bars represent ± 1 SEM.

Video S1. Pupil Light Reflex (PLR)

Mice in the MC treatment groups respond with a grater PLR response than mice in the sham transplantation group (p<0.0001). The full PLR response of the two groups following a light stimulus is presented in consecutive images in the movie.

We assessed behavioral light avoidance (BLA)^6,7,43, 44, 45 to evaluate the effect of retinal reconstruction on visually guided behavior (see Figure 5C for schematic of the BLA arena). Mice are nocturnal animals and have a tendency to avoid light exposure in open environments⁴⁶ if they are able to detect the light. Mice in the MC treatment group (p < 0.05) and rAAV2/2(Y444F) group (p < 0.05) showed greater light avoidance than mice in the sham-treated group (F(2, 18) = 4.97, p < 0.05, one-way ANOVA). There was no difference between the two treatment groups (p = 0.58) (Figure 5D; Video S2). The number of transitions an animal makes between the light and dark compartments of the BLA arena is commonly used as a measure of anxiety during this test.^46,47 The groups did not differ in number of transitions (F(2, 18) = 0.53, one-way ANOVA) (Figure S9), excluding the contribution of anxiety levels to the difference observed between groups.

Video S2. Behavioral Light Avoidance (BLA)

The camera is placed above the illuminated compartment of the BLA arena and animals can be seen moving between the lit and dark compartments (Speed x8). Mice in the MC treatment group avoided light more than mice in the sham treated group (p<0.05).

Finally, we used the optomotor response (OMR)^4,48, 49, 50, 51 assay to examine whether treated animals could respond to a dynamic stimulus (Figure 5E). Because an optomotor head tracking response is contingent on optokinetic nystagmus,^50,52 the directionality of head tracking indicates the eye from which the head track is derived.⁴⁹ As mice received a unilateral transplantation of treated cells, we could use the direction of OMR head tracks elicited in each animal to specifically compare treated and untreated eyes.

We found that treated mice of both groups responded to the rotating grating by head tracking (Figure 5F), with significantly more head tracks elicited by the treated eye compared to the untreated eye in rAAV2/2(Y444F)-treated mice (t = 4.5, p < 0.01, paired t test) and MC-treated mice (t = 4.05, p < 0.01, paired t test), but not in sham transplanted mice (t = 0.47, ns, paired t test). Head tracking elicited by stimulation of the treated eye differed between the three groups (F(2, 18) = 8.77, p < 0.01, one-way ANOVA), with an increased number of head tracks by animals in both gene therapy groups (both ps < 0.01). No difference was observed in number of head tracks per trial between MC- and rAAV2/2(Y444F)-treated mice.

Electroretinography (ERG) analysis was conducted to compare MC and rAAV2 Y444F treatments with sham treatment and uninjected age-matched Rho^−/− mice (Figure S11). Under scotopic conditions, aimed to assess rod function, no difference was detected in the a-wave or b-wave response in the treated eye of experimental groups and corresponding eye of untreated Rho^−/− mice (Figure S11A). The absence of improvement in scotopic ERG response was expected, as mice lose scotopic b-wave ERG by 10 weeks,^33,34,53 and following transplantation at 12–14 weeks, a very large number of integrated rod cells would be required to restore this response. The relatively low number of radially integrated cells may more reliably allow assessment of vision by the functional and behavioral methods described earlier. However, investigation of cone photoreceptor involvement in visual improvement was required, as a difference in cone function may imply that cells are affecting transplantation by means of cone neuroprotection rather than rod regeneration. Following light adaptation, mice of different treatment groups showed no difference in amplitude of a-wave (F = 0.65, p = 0.59, ns) or b-wave (F = 0.73, p = 0.53, ns) response (Figure S11B) or in response to a flickering light stimuli (Figures S10C and S10D). Because rod cells require a period of time for restoration of function following a flash of light, the response to the flicker stimulus would selectively represent cone sensitivity.

To further examine residual host cone photoreceptors, we histologically examined the retinas of treated Rho^−/− mice using staining against cone-arrestin. While some residual host cones were present in adult Rho^−/− mice³⁴ of all three experimental groups, no difference was observed in cone number between groups (n = 7 per group; F = 0.61 p = 0.43, one-way ANOVA) (Figure S12). Furthermore, we did not observe any cone arrestin protein in transplanted cells, corroborating previous findings that visual improvement in the treatment groups was independent of residual host cone function or donor cone photoreceptors.

Discussion

We report retinal reconstruction and restoration of visual function by photoreceptor cells following ex vivo gene therapy and show highly efficient and sustained transgene expression in photoreceptor cells by use of MC DNA. Transplantation of Rho^−/− cells, which had been treated ex vivo with both viral rAAV2/2(Y444F) and non-viral MC RHO gene therapy, reconstructs the degenerate retina of the adult Rho^−/− mouse. Transplanted cells mature to express the previously missing rhodopsin gene, as well as other mature photoreceptor markers. The data collectively support the notion that transplanted Rho^−/− photoreceptor cells are functional following RHO gene replacement. Treated cells are able to restore functional and behavioral light responses to previously blind animals, while similar improvements were not seen with unmodified Rho^−/− photoreceptor sham injections. Functional experiments were designed to isolate the specific response of donor rod cells and avoid the contribution of remaining host cone photoreceptors or intrinsically photosensitive retinal ganglion cells.

We show de novo reconstruction of the photoreceptor layer in mice that had almost no photoreceptors remaining at the time of transplantation. The thickness and localization of the transplanted cell layer are comparable to those of a WT retina and are consistent with findings from a previous study using Rho^+/+ donor cells in rd1 mice, which have end-stage retinal degeneration and also showed positive pupillometry improvements.⁶ The OSs of the newly formed nuclear layer did not assume WT cell morphology or orientation toward the RPE. However, it has been previously shown that photoreceptor cells derived from healthy mice do not develop normal morphology post-transplantation in the severely degenerate retina but are still able to improve functional vision.^6,7 Moreover, photoreceptor cells that lack a fully formed OS were recently shown to support vision in mice,³⁸ challenging the notion that normal photoreceptor structure is required in order to improve vision by cell transplantation.

While functional and behavioral measures, including PLR, BLA, and OMR showed improvement in vision, electro retinal function did not show improvement of rod function in treated animals. The lack of response in photopic conditions indicates the loss of detectable cone cell function in animals of all groups and provides evidence that the responses observed in previous testing can be reliably attributed to transplanted treated rods and not to the neuroprotection of host cones. Importantly, the absence of a rod-ERG response is not indicative of absent circuit formation by transplanted cells. Since the ERG response is a vector response resulting from the summation of synchronous electrical changes arising from many individual cells, a very large number of radially aligned photoreceptors is needed to achieve an a-b wave. In humans, RP patients lose their ERG response early on, at a time when they still maintain excellent visual acuity, and so being without an ERG response does not equate to having no vision.⁵⁴

The rhodopsin gene was the first gene linked to RP,^55,56 it holds extensive structural and functional importance, and it has been central in animal models of RP, as without this gene, no rod function is possible. The transplantation of healthy, Rho-expressing, photoreceptor progenitor cells in the Rho^−/− mouse has been previously studied,^1,5 showing that Rho^+/+ cells are able to improve visual function in the partially degenerate retina. In order to qualify the results of the present study, we used protocols corresponding to those previously published when transplanting healthy cells in Rho^−/−1 and other mice with severe retinal degeneration^6,7 and found that the results obtained following transplantation of ex-vivo-treated cells in this study were similar in magnitude to the improvement attained following transplantation of healthy cells. Similarities in the percentage of pupil constriction at a corresponding time point (1 s following light stimulus of 10¹⁵ photons/cm²/s), as well as behavioral measures such as light avoidance and improved OMR consistently show that treatment in this study provides amelioration that is in line with that achieved by WT cells.

In previous studies, healthy stem cells^57,58 and retinal cells⁵⁹ had been treated ex vivo to deliver growth and transcription factors to the retina to promote neuroprotection of remaining retinal cells in retinal disease. The present study sought to investigate the transplantation of ex vivo MC-treated cells at a stage in which no functioning rod photoreceptors were present in the host, and host cone photoreceptor function was vastly diminished, representing a relevant time point for transplantation in blind patients with RP.^6,7 While clinical trials using human iPSC-derived RPE are underway for the treatment of age-related macular degeneration (AMD),¹³ there are no proposed photoreceptor-replacement therapies for RP to date. However, encouraging results from a clinical trial in which a subretinal electronic retina was transplanted in RP patients60, 61, 62 suggest that, even in the case of severely degenerated retina, transplantation of light receptors may prove successful in restoring functional vision to blind individuals (for a review of the similarities between the subretinal electronic retina and potential photoreceptor transplantation therapies, see Cramer and MacLaren⁶³).

Translating results from animal studies to the prospective improvement in patients pose a few challenges, the greatest of which is the question of whether a statistically significant improvement in these tests would prove to be clinically significant for patients. The different tests show that treatment in this study was able to repair different aspects of the visual pathway. Improved PLR shows that light impulses transmitted to the retina could travel to the brainstem to feed into appropriate neural pathways and produce an arc reflex and consensual constriction of the contralateral eye. Restored light avoidance shows that mice were now able to detect and respond to light, and the OMR test provides evidence of a behavioral response to a dynamic stimulus. The visual acuity of the OMR drum grating seen by the treated mice is 0.1 cycle per degree (cpd). In humans, normal visual acuity is 30 cycles per degree—equivalent to 20/20 on the Snellen eye chart. Therefore 0.1 cpd is equivalent to 20/6,000. Although this is low, a recent clinical study involving patients from our center receiving an electronic retinal implant showed that patients who were previously blind from RP achieving 0.1 cpd had useful vision for localizing and identifying objects,⁶² which is consistent with our observations in the mice.

We recently published a robust method for the differentiation of human iPSCs into photoreceptor progenitors and have shown that these human photoreceptor progenitors are able to mature and reconstruct the severely degenerated retina, We recently published a robust method for the differentiation of human iPSCs into photoreceptor progenitors and have shown that these human photoreceptor progenitors are able to mature and reconstruct the severely degenerated retina, restoring a degree of visual function to blind mice.⁷ Other studies have provided further accounts of visual improvement in mice following transplantation of healthy murine or human stem-cell-derived photoreceptors,^2,5,8,64 showing that high-quality methods for photoreceptor progenitor differentiation have been achieved. Previous transplantation studies, results of recent in vitro studies showing gene correction in RP-patient-derived photoreceptors (https://static-content.springer.com/esm/art%3A10.1038%2Fsrep19969/MediaObjects/41598_2016_BFsrep19969_MOESM1_ESM.pdf),65, 66, 67, 68, 69 and the results from clinical trials of transplanted electronic retina,⁶¹ taken together with the novel results of this study, provide evidence that cells can be differentiated, corrected, and transplanted to improve vision and that even completely blind patients maintain a functional pathway for vision that may be used or activated following transplantation.⁶³

The present study provides the first account of retinal treatment by MC DNA gene delivery and is the first study to use ex vivo gene therapy in combination with cell transplantation to rescue functional vision in mice.

The use of MCs may provide substantial advance in ex vivo delivery of large genes to photoreceptors and, thus, enable the treatment of currently incurable ocular diseases. MCs are a transformational technology, as they have no size limitations or risks (unlike viral gene delivery). Expression levels in vitro by MC delivery of rhodopsin and functional improvement attained following the transplantation of MC-treated cells were similar to those attained by rAAV2 Y444F. This was a somewhat unforeseen outcome, as in a majority of studies, the efficiency of non-viral gene delivery to photoreceptor cells in vivo has been reported to be lower than that of viral delivery methods.^16,17,70

The unique properties of MCs make them optimal for gene transfer and may explain the efficient transfection achieved by MCs compared to previously studied non-viral vectors. MCs do not contain a bacterial backbone and, thus, eliminate bacterial immunogenic unmethylated CpG motifs⁷¹ that are involved in the silencing of episomal transgene expression.¹⁹ Because they do not have the plasmid bacterial elements, they are also more likely to persistently express the transgene in a clinical setting. The elimination of the bacterial backbone also reduces the size of the DNA to be transferred into cells,⁷² increasing their bioavailability. Relevant to the results obtained with MC DNA are recent reports of sustained and efficient gene expression in the neonatal retina and, to a lesser extent, in the adult retina, by use of compact DNA nanoparticles.73, 74, 75 Further experimentation comparing these small and efficient non-viral vectors may be useful for further application in retinal gene therapy.

In future studies, MC DNA could be used as a vehicle to replace plasmids for improved ex vivo and in vitro delivery of CRISPR-Cas9 with appropriate guide RNAs to edit specific patient mutations. These findings may provide a new means for treating retinal degeneration in patients and increase feasibility of ex vivo gene therapy for an array of scientific and medical applications by providing an efficient and sustainable method for ex vivo gene transfer to cells.

Materials and Methods

Animals

Rhodopsin knockout mice (Rho^tm1Phm, referred to as Rho^−/−) were generously provided by Jane Farrar (Trinity College, Dublin, Ireland) and were used as hosts in transplantation studies. Mice were all female and, at the time of subretinal transplantation, were between 12 and 14 weeks old. Age-matched adult WT C57BL/6 mice were provided by the Biomedical Sciences division of the University of Oxford. Tissue for retinal culture and transplantations was taken from Rho^−/−, Tg(Nrl-EGFP) mice (referred to hereinafter as Rho^−/−, Nrl-GFP). These were obtained by crossing Rho^−/− with Tg(Nrl-l-EGFP)Asw/Tg(Nrl-l-EGFP)Asw mice (a kind gift of Anand Swaroop, National Eye Institute, Bethesda, MD, USA), followed by backcrossing of progeny, Rho^+/−,Tg (Nrl-EGFP)^+/+, to the parental Rho^−/− line.

Animals were housed under a 12-h:12-h light (<100 lux):dark cycle, with food and water available ad libitum. All procedures were performed under the approval of local and national ethical and legal authorities and in accordance with the Association for Research in Vision and Ophthalmology statements on the care and use of animals in ophthalmic research. A sample size of 7 mice per group was selected, consistent with that reported in comparable publications in the field, and was calculated at 80% according to pilot experiments (n = 9 per group in pilot experiments). 21 Rho^−/−, Nrl-GFP mice were allocated between the three experimental groups. The groups were stratified according to baseline results in three measures of visual function to exclude baseline variance between groups. 7 WT mice and 7 untreated Rho^−/−, Nrl-GFP mice were tested alongside experimental groups to provide behavioral controls for transplantation.

For subretinal injections and in vivo imaging procedures, animals were anesthetized by intraperitoneal injection of a mixture of medetomidine (Dormitor, 1 mg/mL; Pfizer, Sandwich, UK) and ketamine (Ketaset, 100 mg/mL; Fort Dodge, Southampton, UK), and pupils were fully dilated with a mixture of 1% tropicamide and 2.5% phenylephrine hydrochloride eye drops (both from Bausch & Lomb, Kingston-Upon-Thames, UK). Anesthesia was reversed by intraperitoneal injection of atipamezole (brand name, Antisedan; 2 mg/kg body weight) in sterile water. Following treatment, animals of the different experimental groups were co-housed.

Cell Culture

Animals at P0–P3 were sacrificed by decapitation. Eyes were immediately extracted, and retinas were dissected free from surrounding tissues in 0.01 M sterile phosphate-buffered saline (PBS) at room temperature (RT). Retinal dissociation was performed with the Worthington Papain Dissociation System (Worthington Biochemical, Lakewood, NJ, USA). Cells were plated at a seeding density of approximately 2 × 10⁵ cells per well in 24-well plates in 500 μL complete neuronal growth medium (Neurobasal-A) supplemented with B27 (2%), N2 (1%), L-glutamine (0.8 mM), penicillin (100 U/mL), and streptomycin (100 μg/mL) (all from Invitrogen, Paisley, UK).⁷⁶ Cells were maintained at 34°C, and media were changed on in vitro day 2 and every second day thereafter. In vitro experiments were replicated in at least 6 wells for each experimental condition (3 pooled retinas or 3 × 10⁶ dissociated cells per well).

Prior to transplantation, photoreceptor precursors were lifted from culture wells by light pipetting and dissociated a second time to ensure a single-cell suspension for transplantation.

MACS

A MACS separation system and Anti-Rat IgG MicroBeads (Miltenyi Biotec, Bergisch Gladbach, Germany) were used as previously described.³⁷

Gene Therapy Vector Generation

A common plasmid vector was designed to include necessary elements for the delivery and expression of the human RHO gene in rod cells. The bicistronic expression cassette included the RHOK (also known as the GRK1) promoter, the human RHO sequence, and the red fluorescent reporter gene DsRed, which can be isolated by standard confocal interference filters from the endogenous GFP in donor cells. PCR primers were designed manually (see PCR Primers for Gene Therapy Vector Generation section) and manufactured by Sigma Aldrich (Dorset, UK). The plasmid was sequenced for the presence of the full RHOKpr.RHO.IRES.DsRed construct using primers as detailed earlier to ensure that no mutations were created during the cloning process (sequencing was performed by Source Bioscience, Oxford, UK). The resulting plasmid was then used in both AAV and MC vector production.

rAAV2/2(Y444F) capsid mutant vector was generated by site-directed mutagenesis of surface-exposed tyrosine residues on AAV capsid VP3 as previously described.³⁵ Vectors were purified by iodixanol gradient centrifugation, and titer was confirmed by PCR to the RHO sequence region. MC vector was generated from the RHOK.RHO.IRES.DsRed plasmid described earlier. Production of MC vector was performed by PlasmidFactory (Bielefeld, Germany).

Ex Vivo Gene Therapy Delivery

Both vectors were delivered to cultured cells 24 h after retinal dissociation.

Dissociated retinas were plated at a seeding density of approximately 3 × 10⁴ per well in 96-well plates with 180 μL complete media.

For rAAV2/2(Y444F) delivery, 24 h after dissociation, media were changed, and 10 μL rAAV2(Y444F) was added to cells (titer, 1 × 10⁵ viral genomes [vg]/mL). Media were changed every 48 h thereafter, and cells were imaged daily to detect GFP/DsRed expression.

For MC DNA delivery, media were changed to remove antibiotics prior to transfection, and primary cell cultures were transfected by NanoJuice Transfection Reagent and NanoJuice Transfection Booster (both from Novagen) at a concentration of 1 μg DNA:2 μL NanoJuice reagent:2 μL NanoJuice booster, added to cells drop-wise. For each well, transfection reagents were mixed with 20 μL serum-free media and incubated at RT for 5 min. Subsequently, transfection reagents were combined with plasmid or MC DNA and incubated for a further 15 min at RT.

PCR Primers for Gene Therapy Vector Generation

Gene expression was measured by quantitative real-time PCR, as detailed (see Quantitative Real-Time PCR Conditions and Primers), using a rhodopsin primer pair designed to amplify both mouse and human rhodopsin mRNA.⁷⁷

For RHOK: forward, 5′-GTCAATTGCTCGGTACCGGGCCCCAGAAG-3′; and reverse: 5′-GTGCCATTCATGGTGGTCCCGGGGCTGACACAG-3′. For RHO: forward, 5′-ACGGTACCACCACCATGAATGGCACAGAAG-3′; and reverse, 5′-GGGAGAGGGGCTTAGGCCGGGCCACCTG-3′. For IRES.DsRed: forward, 5′-GACCCCAGGATGTAGGCCCCTCTCCCTCCC-3′; and reverse, 5′-AAAGCGGCCGCCTACAGGAACAG-3′.

Quantitative Real-Time PCR Conditions and Primers

Quantitative real-time PCR was performed on the Applied Biosystems 7500 Real-Time PCR System.

1 μL cDNA was added to each well of the quantitative real-time PCR plate with 9 μL SYBR Green PCR Master Mix as follows: 5 μL iTaq, 1 μL forward primer (2 μM), 1 μL reverse primer (2 μM), 2 μL dH₂O.

For rhodopsin: forward, 5′-AGCAGCAGGAGTCAGCCACC-3′; and reverse, 5′-CCGAAGTTGGAGCCCTGGTG-3′. For Arp: forward, 5′-GATCATCCAGCAGGTGTTTGAC-3′; and reverse, 5′-GTGTACTCAGTCTCCACAGACAATG-3′. For beta-actin: forward, 5′-GCTGTGCTATGTTGCTCTAGACTTC-3′; and reverse, (5′-CATAGAGGTCTTTACGGATGTCAAC-3′.

Values obtained for the target genes were normalized to the geometric mean of the housekeeping genes acidic repeat protein (Arp) and beta-actin to reach a ΔCT (cycle threshold) value. Results were calculated by differences in gene expression between normalized target genes (ΔΔCT) and expressed as a relative quantity value (2 ^ΔΔCT).

Cell Transplantation

Transplantations were performed by subretinal injection under direct visualization as previously described.⁷ 2 μL diluted cells (approximately 2 × 10⁵) were transplanted unilaterally to the left eye, with the other uninjected eye acting as a control for some experiments.

SLO

In order to assess cell survival and retinal distribution of cells in vivo, imaging was performed using a confocal scanning laser ophthalmoscope (cSLO; Spectralis HRA, Heidelberg Engineering, Heidelberg, Germany). Imaging protocols adhered to those previously presented in adaptation of SLO imaging for mouse research¹⁵ and were acquired as previously described.⁷

PLR Assay

The rod-mediated PLR was assessed in unanesthetized animals by a custom pupillometry assay, as previously described,⁶ following dark adaptation of >12 h. Unanesthetized mice were held with the unoperated eye facing an infrared camera, while the operated eye was subjected to the light stimulus. For animals examined 4 weeks following transplantation, light stimuli were presented in ascending order of irradiance and controlled by neutral density filters, and animals were exposed to testing at a light intensity of 10¹⁴, 10¹⁵, or 10¹⁶ photons/cm²/s, with a period of at least 30 min of dark adaptation between recordings for each animal. For animals examined 3 months following transplantation, response was measured to a light intensity of 10¹⁵ photons/cm²/s. Pupil area 1 s after stimulus offset was normalized to the pre-stimulus pupil area. A time point between 0 and 2 s has been previously validated to isolate the response of classical photoreceptors over intrinsically photosensitive retinal ganglion cells (ipRGCs),^6,40 and a relatively early period within this time frame (1 s) was selected here to detect early-phase pupil constriction following dark adaptation, selectively representing the sensitivity of donor rod photoreceptors over residual host cone photoreceptors or ipRGCs. The researchers were blinded with regard to treatment during the test.

Behavioral Light Aversion (BLA) Assay

The BLA arena and assay were previously described.⁷ The amount of time spent by test animals in the dark compartment of the test area provided a measure of light avoidance. An LED array of green light centered at 510 nm was suspended over the test arena, closer to the wavelength of maximal rod sensitivity rather than that of ipRGCs (480 nm). A dim level of light was used (~10 lux at the arena center), which is below the saturation level of rods and, therefore, more adapted to rod response over that of cone photoreceptors or ipRGCs. The researcher was blinded with regard to treatment during the test. Data were recorded by a digital camera mounted above the lit chamber and calculated with ANY-Maze video tracking software (v.4.5; http://www.anymaze.co.uk/index.htm).

OMR

The animal’s response to a visual stimulus was assessed by measuring the OMR to a rotating grating as previously described.⁷ Mice display compensatory head movements in response to a moving repetitive stimulus pattern, with a smooth pursuit in the direction of the pattern movement followed by a fast saccadic movement in the opposite direction.⁵¹ Light stimulus was used as described earlier for the BLA assay. The researcher was blinded with regards to treatment during the test. Behavior was recorded by use of a digital camera mounted directly above the central platform. The number and direction of head tracks per trial were quantified manually by two independent scorers, who were blinded with regard to treatment, and averaged between the three experimental runs.

ERG

ERGs were obtained by adapting previously described procedures.³⁴ Mice were dark adapted overnight (at least 12 h) before the experiments. Mice were anesthetized, and their pupils were dilated as described earlier. Silver needle electrodes served as reference (forehead) and ground (tail), and gold wire ring electrodes served as active electrodes. Methylcellulose (Methocel; Ciba Vision, Wessling, Germany) was applied to eyes to ensure good electrical contact and to keep the eye hydrated during the procedure. The recording setup included a Ganzfeld bowl, a DC amplifier, and a computer-based control, and recordings were made using an Espion V5 system (Diagnosys, Cambridge, UK). ERGs were recorded from each eye separately after the mice were placed in the Ganzfeld bowl. Band-pass filter width was 1 to 300 Hz for single-flash and flicker-stimuli recordings. Single-flash and flicker recordings were obtained both under dark-adapted (scotopic) and light-adapted (photopic) conditions. Single-flash stimuli were presented with increasing intensities. Light intensity under scotopic conditions was gradually increased, with a delay between steps for rod sensitivity restoration. Stimulus intensity ranged from 10⁻⁴ cd.s/m² (candela seconds per square meter) to 10 cd.s/m², divided into four steps of 10⁻⁴, 10⁻², 1, and 10 cd.s/m². 16 responses were averaged with an inter-stimulus interval (ISI) of 5 s (for 10⁻⁴ and 10⁻² cd.s/m²) or 17 s (for 1 and 10 cd.s/m²). Flicker stimuli had an intensity of 1–10 cd.s/m², with frequencies of 6, 20, and 30 Hz. Light adaptation was performed with a background illumination of 30 cd.s/m² presented for 10 min to reach a stable level of the photopic responses.⁷⁸

Histology and Immunohistochemistry

Eyes were fixed in 4% paraformaldehyde (PFA; Thermo Fisher, Loughborough, UK) in PBS. The cornea was excised, the lens was removed, and eye cups were fixed in 4% PFA overnight. Fixed eyes were cryoprotected in a 10%–30% sucrose gradient and then washed in PBS, embedded in optimal cutting temperature (OCT) compound (Tissue-Tek, Sakura Finetek, the Netherlands), and frozen in liquid nitrogen. Cryosections (18 μm) were cut and affixed to poly-L-lysine-coated glass slides (Polysine; Thermo Scientific, Loughborough, UK).

Retinal sections were rinsed 3 × 5 min with 0.01 M PBS and permeabilized for 20 min with 0.3% Triton-X in PBS. Sections were then blocked for 1 h at RT in PBS and 0.1% Triton X-100 with 10% normal donkey serum before overnight incubation at 4°C with PBS containing 0.1% Triton-X, 2% normal donkey serum, and primary antibody. The following primary antibodies were applied: recoverin-specific antibody (Rb, 1:1,000, Millipore, ab5585), PDE6β-specific antibody (Rb, 1:100, Abcam, ab5663), rhodopsin-specific antibody (Rb, 1:1,000, Abcam, ab65694), GNAT1-specific antibody (Rb, 1:100, Abcam, ab74059), ROM1-specific antibody (Rb, 1:300, Abcam, ab220049), cone-arrestin-specific antibody (mouse: Rb, 1:1,000, Millipore, ab15282), synaptophysin-specific antibody (Rb, 1:200, Abcam, ab32127), and GFAP-specific antibody (Rb, 1:1,000, Abcam, ab7260). Following overnight incubation, slides were rinsed (3 × 5 min with PBS) and incubated for 2 h with secondary antibody (Alexa Fluor 635, donkey anti-rabbit immunoglobulin G [IgG], 1:250) in PBS containing 0.1% Triton-X, 2% normal donkey serum at RT. Slides were then rinsed (2 × 5 min with PBS), counterstained with 1:5,000 Hoechst 33342 (Molecular Probes, UK) in PBS for 5 min, and mounted using an antifade reagent (Prolong Gold, Invitrogen).

Confocal Microscopy

Retinal sections were viewed on a confocal microscope (LSM710; Zeiss, Jena, Germany). The fluorescence of Hoechst, GFP, DsRed, and Alexa Fluor 635 was excited using 350-nm UV, 488-nm argon, and 543-nm HeNe lasers, as appropriate. GFP-positive cells were first located using epifluorescence illumination, and then a series of XY optical sections (approximately 0.5-μm thickness) were taken in succeeding stacks to give XY projection images. Image processing was performed using ImageJ (v.1.47, NIH; https://imagej.nih.gov/ij/index.html).

Light Microscopy

For the purpose of quantitative analysis of GFP and DsRed expression in cultured cells and retinal sections, images were taken using the Leica DM IL inverted epifluorescence microscope and QImaging Retiga 2000R camera with QCapture Pro 7 software. Images were obtained using identical acquisition settings and exposure times for comparable plates or slides and were saved at a resolution of 1,200 × 1,600 pixels.

Cells were imaged in culture every 24–48 h. Average of GFP/DsRed cells per field of view was calculated as the mean number of cells counted per view on a 20× microscope objective, with 4 fields of view assessed per well to determine a mean cell count.

In retinal sections, the number of GFP- and DsRed-positive cells were counted in serial 18-μm non-overlapping sections through each eye. Cells were considered for analysis if they resided in the subretinal space or within the host residual outer nuclear layer.

Statistical Analyses

All measures are presented as mean and SEM, with a significance level of p < 0.05. The Shapiro-Wilks test was used to determine normality of distribution, and Levene’s test was used to assess homogeneity of variance in experimental groups. Means were compared using two-way analysis of variance (ANOVA) as appropriate, with Bonferroni post hoc test used for multiple comparisons or Tukey’s post hoc test for multiple comparisons if comparing more than three groups (α = 0.05). Paired or unpaired, two-tailed Student’s t test was performed as appropriate when comparing two groups. Statistical analyses were carried out using SPSS v.22 (IBM).

Data Availability

The datasets generated during and/or analyzed during this study are available from the corresponding author on reasonable request.

Author Contributions

A.O.B.-C., M.S., A.R.B., and R.E.M. conceived and designed the experiments. A.O.B.-C. and M.E.M. designed the Minicircle vector and generated the AAV vector. A.O.B.-C. performed in vitro experiments and data analysis. A.O.B.-C., S.D., and A.R.B. performed in vivo experiments and data analysis. A.O.B.-C., M.S., and D.F. performed the subretinal injections. A.O.B.-C., A.R.B., and R.E.M. wrote the paper.

Conflicts of Interest

The authors declare no competing interests.