Function of MYO7A in the Human RPE and the Validity of Shaker1 Mice as a Model for Usher Syndrome 1B

Abstract

Purpose.: To investigate the function of MYO7A in human RPE cells and to test the validity of using shaker1 RPE in preclinical studies on therapies for Usher syndrome 1B by comparing human and mouse cells.

Methods.: MYO7A was localized by immunofluorescence. Primary cultures of human and mouse RPE cells were used to measure melanosome motility and rod outer segment (ROS) phagocytosis and digestion. MYO7A was knocked down in the human RPE cells by RNAi to test for a mutant phenotype in melanosome motility.

Results.: The distribution of MYO7A in the RPE of human and mouse was found to be comparable, both in vivo and in primary cultures. Primary cultures of human RPE cells phagocytosed and digested ROSs with kinetics comparable to that of primary cultures of mouse RPE cells. Melanosome motility was also comparable, and, after RNAi knockdown, consisted of longer-range fast movements characteristic of melanosomes in shaker1 RPE.

Conclusions.: The localization and function of MYO7A in human RPE cells is comparable to that in mouse RPE cells. Although shaker1 retinas do not undergo degeneration, correction of mutant phenotypes in the shaker1 RPE represents a valid preclinical test for potential therapeutic treatments.

Usher syndrome is an autosomal recessive deaf-blindness disorder classified clinically by three subtypes. Usher syndrome type 1 is the most severe, with profound congenital deafness, followed by progressive retinal degeneration. ¹ Cochlear implants are now being used to correct the hearing of children with Usher 1, ² but there is no treatment of their ensuing blindness. Approximately half the cases of Usher 1 are classified as Usher 1B, ^3–5 which is caused by loss-of-function mutations in the MYO7A gene. ⁶  

Replacement gene therapy is a feasible approach to prevent retinal degeneration in Usher 1B. In a proof-of-principle study, it was shown that mutant retinal phenotypes in shaker1 mice could be corrected by the injection of lentiviral-MYO7A into the subretinal space. ⁷ Although shaker1 mice carry mutations in the orthologue of the Usher 1B gene ^8,9 and those used in this gene therapy study carried a null allele of Myo7a, ^9–11 shaker1 mice do not undergo retinal degeneration. ¹² In this respect, they resemble murine models of other subtypes of Usher 1 ^13–16 but are in contrast to patients with Usher 1. Shaker1 mice do have a number of mutant retinal phenotypes, the clearest of which is the mislocalization and defective motility of melanosomes in the retinal pigment epithelium (RPE). ^17,18 These mutant phenotypes can be used to test the functionality of MYO7A that is expressed after gene therapy transduction of retinal cells. ⁷ Nevertheless, these phenotypes have not been demonstrated in the retinas of patients with Usher 1B; hence, it is not clear how relevant they are to the human disease. 

In the present study, we tested whether primary cultures of human and mouse RPE cells were comparable with respect to MYO7A localization and MYO7A-related cell biology. In particular, we tested whether the human RPE cells exhibited the same MYO7A-dependent motility of melanosomes described for mouse RPE and, thus, whether defective melanosome localization and motility was a relevant phenotype for testing efficacy in preclinical animal studies of gene therapy for Usher 1B blindness. 

Materials and Methods

Immunolabeling of Retinal Sections

A postmortem human eye (kindly provided by Mary Rayborn, Cole Eye Institute, Cleveland Clinic, Cleveland, OH) from an 89-year-old donor was dissected and immersion fixed in 4% formaldehyde in 0.1 M cacodylate buffer approximately 5 hours after enucleation. Pieces of the eyecup were processed for cryosectioning, as described previously. ¹⁸ Sections (12-μm thick) were collected, mounted on microscope slides (Fisher Scientific, Pittsburgh, PA), and immunolabeled with the affinity-purified MYO7A polyclonal antibody pAB2.2, ¹⁹ as described previously. ¹⁸ Fluorescence micrographs were collected using a laser scanning confocal microscope (FV1000; Olympus, Tokyo, Japan). False-color merged images and image montages were assembled using a graphics editing program (Photoshop CS3; Adobe, Mountain View, CA) on a laptop computer (MacBook Pro; Apple, Cupertino, CA). 

RPE Cell Culture

Primary human fetal RPE (HFRPE) cells were isolated and cultured on filter supports in a custom-formulated medium, based on Chee's essential medium (CEM), called replacement CEM, as described. ²⁰ The tenets of the Declaration of Helsinki were followed, and prior informed consent was obtained from all donors. For the analysis of phagocytosis and melanosome motility, HFRPE cells were replated onto laminin-coated 24-well transwell filters (Corning, Corning, NY) or glass-bottom dishes (Matek, Ashland, MA), respectively. HFRPE cells were allowed to redifferentiate and repigment for 1 month before analysis. 

Early-passage (P4) ARPE-19 cells ²¹ were kindly provided by Sassan Azarian (from a stock that originally came from the American Type Culture Collection, Manassas, VA) and were propagated on plastic in DMEM/Ham's F12 (Invitrogen, Carlsbad, CA) supplemented with 10% fetal bovine serum with penicillin and streptomycin (Invitrogen). For all experiments, passage number was kept at less than P20. 

Mouse primary RPE cells were isolated from 12-day-old C57BL/6 mice. Animals were maintained on a 12-hour light/12-hour dark cycle with 30 to 80 lux fluorescence lighting during the light phase. Procedures complied with institutional animal care guidelines and with the ARVO Statement for the Use of Animals in Ophthalmology and Vision Research. Isolated mouse RPE cells were cultured on 24-well transwell filters (Corning) or glass-bottom dishes (Matek), as described previously. ^18,22  

HFRPE and primary mouse RPE cells were fixed with 4% formaldehyde in 0.1 M PBS and were labeled with MYO7A pAb2.2 and FITC-phalloidin. For Western blot analysis, RPE cells were solubilized in NP-40 lysis buffer (150 mM NaCl, 1% NP-40, 50 mM Tris-HCl, pH 8.0). Proteins were separated by SDS-PAGE and transferred to transfer membrane (Immobilon-P; Millipore, Billerica, MA). Blots were immunolabeled with MYO7A pAb2.2 and anti-alpha tubulin mAb B-512 (Sigma, St. Louis, MO). 

RNA was isolated from ARPE19 cells, using a mini kit (RNeasy; Qiagen, Valencia, CA). reverse transcriptase (Superscript II; Invitrogen, Carlsbad, CA) and oligo(dT) primers were used for first-strand cDNA synthesis from poly(A) RNA templates. RT-PCR was performed using the following primer pairs: MYO7A forward, 5′-CTCCAGAAAGCCCTGAGAAAC-3′; reverse, 5′-CTCCTCGAACTTGACCCTG; B2M forward, 5′-ATGTCTCGCTCCGTGGCCTTA-3′; reverse, 5′- TAGAAAGACCAGTCCTTGCTG-3′. 

Generation of shRNA Constructs

Short hairpin interfering RNAs (shRNAs) targeted to mouse, rat, or human MYO7A were designed and cloned according to the manufacturer's instructions using an oligonucleotide design program (siRNA Designer; Promega, Madison, WI) and a cloning system (siSTRIKE U6 Hairpin; Promega). shRNA oligonucleotides used were as follows: anti-MYO7A (human) forward, 5′-ACC GTG GAC AAG ATG TTT GGC TCT TCC TGT CAA GCC AAA CAT CTT GTC CAC TTT TTC-3′; reverse, 5′-TGC AGA AAA AGT GGA CAA GAT GTT TGG CTT GAC AGG AAG AGC CAA ACA TCT TGT CCA-3′; anti-Myo7a (rat) forward, 5′-ACC GTA TAC CAA CAA GAA GAT ACT TCC TGT CAT ATC TTC TTG TTG GTA TAC TTT TTC-3′; reverse, 5′-TGC AGA AAA AGT ATA CCA ACA AGA AGA TAT GAC AGG AAG TAT CTT CTT GTT GGT ATA-3′; anti-Myo7a (mouse) forward, 5′-ACC GAA GCA GCT GAC TGA CAA TCT TCC TGT CAA TTG TCA GTC AGC TGC TTC TTT TTC-3′; reverse, 5′-TGC AGA AAA AGA AGC AGC TGA CTG ACA ATT GAC AGG AAG ATT GTC AGT CAG CTG CTT-3′. 

For lentiviral expression, the human-specific shRNA was resynthesized to include a 5′ ApaI and a 3′ EcoRI site and was cloned into the entrance vector pENTRANS4RiG2-hU6 (Gateway; Invitrogen) after digestion with ApaI and EcoRI (NEB). The shRNA expression cassette, including a downstream CMV-GFP reporter cassette (Hs.U6-Hs.MYO7AshRNA-CMV-EGFP), was then cloned into the lentiviral destination vector pDA-LV using LR clonase according to the manufacturer's instructions (Invitrogen) to generate the lentiviral expression vector pLV-U6-Hs.MYO7AshRNA-CMV-EGFP. pENTRANS4RiG2-hU6 and pDA-LV were kindly provided by Xian-Jie Yang. 

Transient Knockdown of MYO7A in ARPE19 and Primary Human Fetal RPE Cells

ARPE19 cells were transiently transfected using a liposome-based reagent (Lipofectamine 2000; Invitrogen), with the psiSTRIKE-neomycin or -hMGFP vector (Promega) containing the human, mouse, or rat MYO7A-specific shRNAs. Cells were transfected in parallel with the empty vector as a negative control for RNA-mediated silencing. Transfected cells were selected in growth medium containing G418 (600 μg/mL) for 1 week before they were replated at low density in G418-containing medium (300 μg/mL). These cell populations were passaged at least three times and then maintained at confluence for 7 days before Western blot analysis or RT-PCR. 

Primary human fetal RPE cells were grown to confluence on 24-well transwell filters (Corning) and then transfected with either pSTRIKE.MYO7AshRNA.NEO or empty vector, using transfection reagent (FuGENE HD; Roche, Indianapolis, IN) according to the manufacturer's instructions. 

Lentiviral Expression of MYO7A shRNA in Primary Human Fetal RPE Cells

Primary HFRPE cells grown on glass-bottom dishes or transwell filters were incubated with 50 μL LV.U6.MYO7AshRNA-CMV-EGFP (1 × 10⁸ TU/mL) in replacement CEM containing 6 μg/mL hexadimethrine bromide (Polybrene; Sigma) at 37°C for 3 hours. After this time, fresh replacement CEM was added to a final volume of 1 mL, and cells were incubated for an additional 24 hours before the growth medium was replaced with fresh virus-free medium without hexadimethrine bromide (Polybrene). Cells were analyzed for MYO7A expression and melanosome motility 7 days after infection. 

Melanosome Motility Measurements

Melanosome motility data sets were recorded (1 frame · s⁻¹ for 5 minutes) using bright-field time-lapse microscopy, from primary mouse RPE cells, and from untreated or lentiviral MYO7A shRNA-treated human fetal RPE cells, as described previously. ¹⁸ Kymograph traces and displacement measurements were extracted for 20 to 30 melanosomes per treatment using the multiple kymograph function in ImageJ software (developed by Wayne Rasband, National Institutes of Health, Bethesda, MD; available at http://rsb.info.nih.gov/ij/index.html) running on a laptop computer (MacBook Pro; Apple). 

ROS Phagocytosis and Digestion

The digestion of ingested mouse ROS by primary mouse or human fetal RPE cells grown on filters was assayed as described. ²³ Human or mouse RPE cells were incubated with 5 × 10⁶ ROS per well for 20 minutes, washed with cold Dulbecco's PBS (Invitrogen) to remove unbound ROS, and incubated for a further 30 or 60 minutes before quantification of remaining ingested phagosomes selectively labeled with the polyclonal O1 opsin antibody. ¹¹ The total number of ingested ROS and the number of DAPI-positive nuclei per field were counted in images recorded from five randomly selected fields-of-view per treatment. 

Results

Immunofluorescence of MYO7A in Human Retina

The distribution of MYO7A in postmortem human retinas was examined by DIC and immunofluorescence confocal microscopy. In low-magnification micrographs, MYO7A is evident primarily in the apical RPE (Figs. 1A–C). At higher magnification, much of the immunolabeling appears to colocalize with melanosomes in the RPE (Figs. 1D–F). The pattern of MYO7A expression identified in the human retina, and colocalization with melanosomes in the RPE, is identical with that described previously for Myo7a in mouse RPE. ¹⁸ This comparable distribution supports the hypothesis that MYO7A has similar functions within the human and mouse RPE. 

Figure 1.

 

Immunolocalization of MYO7A in the human retina by confocal microscopy. (A) DIC, (B) immunofluorescence, (C) merged image with MYO7A label in green. OS, outer segments; IS, inner segments; ONL, outer nuclear layer; OPL, outer plexiform layer; IPL, inner plexiform layer; INL, inner nuclear layer; GC, ganglion cells. (D–F) High-magnification micrographs of the RPE cell layer from (A) to (C). (D, E, arrows) Cells that appear to be red blood cells; they emit light autofluorescence that does not represent MYO7A immunolabeling. Scale bar: (C) 100 μm; (E) 25 μm.

MYO7A Expression, Phagocytosis, and Melanosome Motility in Cultured Human RPE Cells

The expression and distribution patterns of both endogenous MYO7A and human MYO7A resulting from lentivirus transduction have been characterized in differentiated cultures of primary mouse RPE cells. ⁷ Native MYO7A was associated with the actin-rich cortex, apical processes, and a subset of melanosomes. Lentivirus-MYO7A infection of MYO7A-null shaker1 RPE cells effected similar MYO7A expression levels and localization. The introduced human MYO7A was also able to compensate functionally for the loss of murine MYO7A; it rescued defects in the clearance of ingested photoreceptor outer segments and in the transport of melanosomes, which have been shown to be caused by the loss-of-function of MYO7A. ^18,23,24 Here we have compared the distribution and function of endogenous human MYO7A in human RPE cells with those of mouse MYO7A in mouse RPE cells. In differentiated primary cultures of both human fetal (HFRPE) and mouse RPE cells, MYO7A was found in apically distributed puncta and was colocalized with the cortical F-actin network (Figs. 2A, 2B). In HFRPE cells, MYO7A labeling was more intense at the cortex and was localized to larger apical puncta than seen in primary mouse RPE, most likely as a result of the increased differentiation of the human cells, which typically demonstrate more defined apicobasal polarity and higher transepithelial resistance than primary mouse RPE cells. ²⁰  

Figure 2.

 

MYO7A in cultured human fetal RPE cells and comparison with primary mouse RPE cell cultures. (A) Distribution of MYO7A. Confocal micrographs at the level of the apical region (at the base of the apical processes) and the cell body region (passing through the apical part of each nucleus) from confluent primary human fetal RPE cells, labeled with anti-MYO7A pAb (red in merge), FITC-phalloidin, and DAPI. Schematic diagrams illustrate the positions of the optical sections. (B) Confocal micrographs of the cell body region (comparable to lower images in A) of confluent primary mouse RPE cells (mRPE) and primary human fetal RPE cells (HFRPE) labeled with anti-MYO7A pAb (red) and FITC-phalloidin (green). Scale bars, 25 μm. (C, D) Kymograph plots (illustrating displacement over time) of individual melanosome movements from (C) primary mouse RPE cells and (D) primary human fetal RPE cells. Examples of tracks from stationary melanosomes are highlighted in blue, and tracks from melanosomes undergoing short rapid movements (as indicated by lateral displacements in the track) are highlighted in orange. (E) Graph comparing the rate of degradation of ingested mouse ROS between primary mouse RPE cells (○) and primary human fetal RPE cells (■).

To compare MYO7A function in differentiated human and mouse RPE cells, we analyzed the ingestion and degradation of purified mouse ROSs using an immunofluorescence assay and the dynamics of melanosome transport using time-lapse bright-field microscopy. In both assays, HFRPE cells were found to behave in a manner similar to that of primary mouse RPE cells. No significant difference was seen in the number of ingested ROSs or in the rate of ROS degradation between human and mouse RPE cultures (Fig. 2C). Kymograph analysis of melanosome dynamics in mouse (Fig. 2D) and human (Fig. 2E) RPE demonstrated that, in both cases, pigment granules were highly constrained in their movements. Displacements, when they occurred, were short and fast. Even the velocities of these rapid constrained movements showed no significant difference between HFRPE and mouse primary RPE cells (MRPE, 1.68 ± 0.43 μm · s⁻¹; HFRPE, 1.59 ± 0.36 μm · s⁻¹), strongly suggesting a MYO7A-dependent melanosome transport system in HFRPE comparable to that previously identified in mouse RPE. ^18,24–26  

RNAi Knockdown of MYO7A in Cultured Human RPE Cells

To study the effect of loss-of-function of MYO7A in human RPE cells, shRNA species were designed to specifically knock down the human, rat, or mouse MYO7A transcripts. To confirm functionality and specificity, ARPE-19 cells were stably transfected with pSTRIKE.shRNA.NEO expressing the human-, mouse-, or rat-specific shRNAs. After selection, levels of MYO7A protein expression were assayed by Western blot analysis (Fig. 3A, top). The expression of MYO7A transcripts was also assayed by RT-PCR (Fig. 3A, bottom). For both the protein and the mRNA, specific and effective knockdown of MYO7A expression was observed only with the human targeted shRNA construct and not with the mouse- or rat-specific shRNAs or in cells stably expressing just the plasmid backbone without an shRNA. ARPE-19 cells stably expressing the human MYO7A-targeted shRNA had undetectable levels of MYO7A expression at the level of the protein and the message. 

Figure 3.

 

shRNA-mediated knockdown of MYO7A in human RPE cells. (A) Stable transfection of cultured ARPE-19 cells with anti-MYO7A shRNA-expressing plasmids demonstrated robust knockdown of MYO7A protein and mRNA after treatment with the human MYO7A-specific shRNA (Human). No knockdown of MYO7A was detected by Western blot analysis (WB) or RT-PCR after transfection with the mouse or rat Myo7a-specific shRNAs or with empty plasmid (Mock). Transient transfection of HFRPE cells with empty plasmid (Mock) or the human MYO7A-specific shRNA (Human) demonstrated knockdown of MYO7A protein expression in the MYO7A-shRNA treated cells at a level consistent with the low efficiency of transfection. (B) Lentiviral-mediated knockdown of MYO7A expression in primary HFRPE cells. (C, D) Melanosome motility in untreated HFRPE cells (C) and HFRPE cells infected with lentivirus expressing the human MYO7A specific shRNA (D). Top two images: micrographs of melanosomes that represent the first frame of Movie S1. Middle two images: paths along which the melanosomes moved during the period captured in the movie. Arrows: relatively long-range movements of three melanosomes in the MYO7A-shRNA-treated cell. Bottom two images: kymograph plots of movements of individual melanosomes, gathered from several different cells. (D) Larger lateral displacements indicate longer periods of fast movement for melanosomes in the MYO7A-shRNA treated cells. (E) Schematic of the human MYO7A-shRNA lentiviral expression construct.

Next, HFRPE cells were transiently transfected with pSTRIKE.shRNA.hMGFP expressing the human targeted shRNA. By assaying GFP expression 7 days after transfection, the efficiency of transfection was found to be 20% (data not shown). Western blot analysis, labeled with MYO7A antibodies, demonstrated a small reduction in the levels of expression (Fig. 3A, top) consistent with the low level of transient transfection. Because of the culturing conditions of the HFRPE cells, antibiotic selection of stable transformants was not a viable option. To increase the efficiency of transduction of the primary HFRPE cells and to allow accurate assessment of the human MYO7A shRNA-mediated knockdown, third-generation recombinant lentiviral vectors expressing the human-specific shRNA and a GFP reporter were produced as described previously (Fig. 3E). ⁷ Primary HFRPE cells grown on filters were transduced with 5 × 10⁶ TU per well and grown for 7 days. GFP expression indicated >90% of cells were transduced (data not shown) and Western blot analysis, labeled with MYO7A antibodies, indicated effective lentiviral-mediated knockdown (<10% of control levels) of MYO7A (Fig. 3B, right). Primary HFRPE cells grown on glass-bottom dishes were also transduced with the same human MYO7A-targeted, shRNA-expressing lentiviral vector under identical conditions. Transduced HFRPE cells were GFP positive (indicating expression of the human MYO7A shRNA) and were identified by fluorescence microscopy. They were used to obtain melanosome motility profiles via bright-field, time-lapse microscopy. Melanosome dynamics in treated HFRPE cells were studied 7 days after infection and were compared with uninfected HFRPE cells. Analyses of melanosome in infected and uninfected HFRPE cells demonstrated constrained movements in the untreated cells (Fig. 3C; Movie S1). In HFRPE cells treated with the lentiviral MYO7A shRNA, the melanosomes underwent longer displacements (Fig. 3D; Movie S1) at higher speeds, similar to transport kinetics previously described for melanosomes in MYO7A-null primary mouse RPE cells. ¹⁸ These results indicate that the role of MYO7A in moderating fast microtubule-based transport of RPE melanosomes is conserved between human and mouse. 

Discussion

Our results show that human fetal RPE cells, cultured as described previously, ²⁰ have the same distribution of MYO7A, parameters of melanosome motility, and rate of phagosome digestion as primary cultures of mouse RPE cells in culture. Moreover, when MYO7A is knocked down in the HFRPE cells by RNAi, melanosome motility is altered and is comparable to that in primary cultures of RPE cells form MYO7A-null mice. We show by immunofluorescence that the distribution of MYO7A in the normal human retina is the same as that in the mouse retina, with most MYO7A in the apical RPE, associated with melanosomes. 

It was reported that the distribution of MYO7A in the retina differs between human and rodent, with expression in the photoreceptor cells found only in human. ²⁷ However, later studies reported MYO7A in the connecting cilia of photoreceptor cells of rodents and humans. ^11,19,28,29 A mutant phenotype associated with the photoreceptor connecting cilia of MYO7A-null mice added further support to the localization of MYO7A in the normal connecting cilium. ^11,30 In the RPE, MYO7A was detected in the apical region by immunoelectron microscopy of adult human retinas. ¹⁹ Its distribution in this cell layer thus appeared to be consistent with that reported for rodents. ³¹ Quantification of immunogold-labeled MYO7A, observed with the use of electron microscopy, showed that most (but not all) MYO7A was associated with melanosomes in mouse, human, and pig RPE. ¹⁸ The immunofluorescence observations of the present study are consistent with these later studies. Our images (e.g., Fig. 1) appear similar to immunofluorescence images of MYO7A in rodent retinas (label of the photoreceptor connecting cilium cannot be resolved at this magnification), ^31,32 with immunolabel most concentrated in the apical region of the cell body though still detectable in other parts of the RPE cell, including the apical processes. 

Because of the unavailability of Usher 1B postmortem eyes containing regions of normal RPE, we have been unable to assess whether melanosome mislocalization is a characteristic of Usher 1B, as it is in the shaker1 retina. ¹⁷ Hence, we have investigated whether primary cultures of HFRPE cells have a mutant phenotype comparable to that of shaker1 RPE cells in primary culture, when MYO7A is knocked down. Before MYO7A knockdown, the human cells were comparable to the WT mouse RPE cells in terms of MYO7A localization, phagosome digestion, and melanosome motility. Importantly, removal of MYO7A generated the same response of increased faster, long-range movements of melanosomes in both human and mouse RPE cells. 

These observations indicate that MYO7A has the same function, with regard to melanosome motility in human and mouse RPE cells. In mouse RPE cells, MYO7A is required for the apical localization of melanosomes. ¹⁷ It appears to capture the melanosome from a microtubule motor and to move the organelle by binding to an exophilin, MYRIP. MYRIP in turn binds to RAB27A, which is associated with the melanosome membrane. ^18,24,25  

Gene replacement therapy for Usher 1B is expected to require targeting of both the photoreceptor and the RPE cells because the mutant phenotypes are cell autonomous. ³³ In preclinical studies with shaker1 mice, which, unlike Usher 1B patients, do not undergo retinal degeneration, ¹² correction of photoreceptor and RPE cellular phenotypes has been used to test for efficacy of the gene therapy. ⁷ The results of the present study indicate that correction of melanosome motility and localization is indeed a relevant preclinical measure for testing the efficacy of treating the RPE. 

Supplementary Materials

Movie S1 - 13.1 MB (QuickTime Movie) 

Time-lapse movies of melanosomes in control HFRPE cells (left), and HFRPE infected with human MYO7A-specific shRNA expressing lentivirus (right). 

Footnotes

 Supported by National Institutes of Health Grant EY07042 (DSW) and Core Grant EY00331 (DSW, DB) and by the National Neurovision Research Institute/Foundation Fighting Blindness. DSW is a Jules and Doris Stein Research to Prevent Blindness Professor. DB is the Dolly Green Professor of Ophthalmology.

Footnotes

 Disclosure: D. Gibbs, None; T. Diemer, None; K. Khanobdee, None; J. Hu, None; D. Bok, None; D.S. Williams, None

The authors thank Erin Legacki and Xiaodan Song for animal handling and husbandry, Mary Rayborn for fixing and shipping the human donor eye tissue, Xian-Jie Yang for the pENTRANS4RiG2-hU6 and pDA-LV, Sassan Azarian for the ARPE-19 cells, and the University of California at San Diego stem cell core group for use of the FV1000 confocal microscope.