Human ESC–derived retinal epithelial cell sheets potentiate rescue of photoreceptor cell loss in rats with retinal degeneration

See allHide authors and affiliations

Science Translational Medicine  20 Dec 2017:
Vol. 9, Issue 421, eaai7471
DOI: 10.1126/scitranslmed.aai7471

Growing cell sheets for retinal transplantation

Loss of retinal pigment epithelial (RPE) cells is responsible for severe vision impairments in retinal degenerative disorders, such as retinitis pigmentosa. Stem cell–derived RPE cells have shown promise as a cell therapy for treating retinal degeneration in preclinical studies, but long-term positive outcomes require efficient engraftment and survival of the transplanted cells. Taking a step in this direction, Ben M’Barek et al. have developed a tissue-engineered product comprising human embryonic stem cell (hESC)–derived RPE cells that were grown as sheets on a biological scaffold. Transplantation of these hESC-RPE cell sheets into a rat model of retinal degeneration resulted in greater cell engraftment and survival and improved visual acuity compared to hESC-RPE cells transplanted as cell suspensions. These tissue-engineered cell sheets may improve the efficacy of cell therapy for treating retinal degenerative diseases.


Replacing defective retinal pigment epithelial (RPE) cells with those derived from human embryonic stem cells (hESCs) or human-induced pluripotent stem cells (hiPSCs) is a potential strategy for treating retinal degenerative diseases. Early clinical trials have demonstrated that hESC-derived or hiPSC-derived RPE cells can be delivered safely as a suspension to the human eye. The next step is transplantation of hESC/hiPSC-derived RPE cells as cell sheets that are more physiological. We have developed a tissue-engineered product consisting of hESC-derived RPE cells grown as sheets on human amniotic membrane as a biocompatible substrate. We established a surgical approach to engraft this tissue-engineered product into the subretinal space of the eyes of rats with photoreceptor cell loss. We show that transplantation of the hESC-RPE cell sheets grown on a human amniotic membrane scaffold resulted in rescue of photoreceptor cell death and improved visual acuity in rats with retinal degeneration compared to hESC-RPE cells injected as a cell suspension. These results suggest that tissue-engineered hESC-RPE cell sheets produced under good manufacturing practice conditions may be a useful approach for treating diseases of retinal degeneration.


Human pluripotent stem cells including human embryonic stem cells (hESCs) and human-induced pluripotent stem cells (hiPSCs) are promising cell sources for regenerative medicine. The rationale for their use involves their differentiation into the cell type of interest that has been damaged or injured, and several clinical trials are underway (1, 2). Retinal pigment epithelial (RPE) cells derived from hESCs or hiPSCs are transplanted as a cell suspension rather than as a more physiologically relevant three-dimensional (3D) cell sheet. After the safety results are obtained from the first clinical trials for treating retinal dystrophies (1, 2), the next challenge will be to transplant hESC/hiPSC-derived RPE cell sheets into the subretinal space with the goal of providing longer-term survival and function of transplanted RPE cells (3, 4).

RPE cells form a specialized epithelial cell layer in the retina that is critical for photoreceptor cell homeostasis and survival (5). Several diseases lead to loss of RPE cells with major consequences for vision. Retinitis pigmentosa is a clinically and genetically heterogeneous group of inherited retinal dystrophies. Multiple gene mutations have been shown to cause retinitis pigmentosa and each mutation induces a specific phenotype affecting either the RPE cells or photoreceptor cells or both (6, 7). Age-related macular degeneration is a pathological condition leading to the degeneration of RPE cells with a complex etiology comprising both genetic and environmental factors (810). Age-related macular degeneration is one of the leading causes of blindness and an increasing global burden with 196 million patients projected to suffer from this disease worldwide by 2020 (11).

Currently, there is a crucial lack of specific treatments for these diseases and limited therapeutic options (8, 12). Cell replacement therapy has been proposed as a strategy to replace dead or damaged retinal cells with cells derived from different sources including mesenchymal stem cells, peripheral or fetal RPE cells, or hESCs/hiPSCs (1321). hESCs and hiPSCs have the capability to self-renew and could be used as an unlimited source of retinal cells for treating retinal dystrophies (16, 19). Recent results have indicated that hESCs/hiPSCs can be differentiated into RPE cells spontaneously after removal of fibroblast growth factor 2 (FGF-2) or by using methods for obtaining nonadherent cells (2225).

Although injection of an RPE cell suspension into the human eye is relatively easy and has been accomplished in current clinical trials, it is not clear whether cells in suspension engraft efficiently into the retina or how long they survive (1, 2, 26, 27). New systems to deliver RPE cells as a preformed epithelial sheet are under development. This approach requires selection of a substrate on which RPE cells will organize and that will be part of the transplanted material. Different types of substrates have been proposed, including collagen or synthetic biomaterials (20, 27, 28). Another approach is to transplant hESC/hiPSC-derived RPE cell sheets without any supporting artificial substrate (29, 30). Although previous RPE cell sheet approaches improved graft survival after transplantation into athymic nude rats (27), the functional benefit compared to injection of an RPE cell suspension has not been demonstrated. A direct comparison of RPE cell suspensions versus RPE cell sheets without substrate reported similar restoration of vision in a rat model of retinitis pigmentosa (29).

Here, we used a human amniotic membrane (hAM) as a substrate for RPE cells derived from hESCs. Compared to synthetic biomaterials, hAM presents advantages for RPE cells as it has a basement membrane and therefore provides the RPE cells with an environment that mimics the in vivo retinal environment. In addition, hAM is permeable to nutrients, and its use in ophthalmology for treating corneal ulceration has demonstrated its low immunogenicity and its anti-inflammatory and antimicrobial properties (3134). There is a well-developed supply chain for hAM from donor patients, as well as good manufacturing practice (GMP) manufacturing sites that include cryopreservation and quality control processes (31). Thus, hAM is an attractive alternative to synthetic biomaterials as a scaffold for the formation of RPE cell sheets. We have developed a tissue-engineered product composed of hESC-derived RPE cells cultured on a hAM substrate. We transplanted this product into the subretinal space of rats with photoreceptor cell degeneration. Using behavioral, electrophysiological, and histological tools, we compared functional outcomes with those of hESC-derived RPE cells injected as a single-cell suspension.


Generation of a clinical-grade feeder-free hESC line and its differentiation into RPE cells

A clinical-grade hESC line RC-9 was grown on U.S. Food and Drug Administration–approved human feeder cell layers according to GMP until passage 10 and then was grown under feeder-free culture conditions (35). At passage 25, the RC-9 hESCs were frozen and banked. Throughout the culture process and after final banking, this hESC line was extensively characterized, ensuring quality control and freedom from viruses and bacteria according to international guidelines (fig. S1 and table S1) (36).

RPE cells were obtained by spontaneous differentiation of confluent hESC cultures (Fig. 1A) (25). Briefly, hESCs were grown to confluence, after which FGF-2 was removed from the culture medium. Pigmented patches that appeared were mechanically harvested and replated (fig. S2). Quantitative reverse transcription polymerase chain reaction from three representative batches of homogeneous hESC-RPE monolayers at passage 1 demonstrated mRNA expression of typical RPE markers, such as RPE-specific 65-kDa protein (RPE65), paired box 6 (PAX6), bestrophin 1 (BEST1), and microphthalmia-associated transcription factor (MITF) relative to undifferentiated cells. As expected, mRNA expression of two pluripotency markers POU class 5 homeobox 1 (POU5F1) and nanog (NANOG) has stopped (Fig. 1B). In addition, hESC-RPE cells exhibited a classical RPE cobblestone shape and showed polarized localization of proteins. Zonula occludens-1 (ZO-1; a tight junction-associated protein), EZRIN (a protein localized in RPE microvilli), and the proto-oncogene tyrosine-protein kinase MER (MERTK; involved in RPE cell phagocytosis) were found at the apical membrane, and BEST1 (a calcium-activated chloride-ion channel) was found at the basolateral membrane (Fig. 1, C to E, and fig. S3, B to D). The purity of hESC-RPE cells was quantified by counting the percentage of cells expressing PAX-6 and MITF, two markers used as an index of RPE identity (5, 29). PAX-6 and MITF were detected in 97.9 ± 0.7% and 97.6 ± 0.7% of hESC-RPE cells, respectively, and the markers were co-expressed in 96.5 ± 0.5% of hESC-RPE cells (figs. S3A and S4). We also combined this evaluation with fluorescence-activated cell sorting (FACS) and showed that 99.3 ± 0.2% of cells expressed cytokeratin (fig. S5), another RPE cell epithelial marker (3739). FACS analysis of hESC-RPE cells stained for the pluripotency markers LIN28 (an RNA-binding protein that regulates stem cell renewal) and TRA1-81 (translin-like protein; a protein lost upon differentiation) showed that the number of residual pluripotent stem cells capable of forming teratomas in vivo was lower than the detection limit. Compared to predifferentiation stage hESCs, hESC-RPE cells at passage 1 after differentiation showed increased number of cells expressing the melanogenesis marker tyrosinase-related protein 1 (TYRP1) (hESCs, 0.0%; hESC-RPE cells, 94.5%) and a decrease in the number of cells expressing the pluripotency marker LIN28 (hESCs, 93.7%; hESC-RPE cells, 0.025%) (Fig. 1F). We then accurately quantified the presence of residual undifferentiated hESCs in the final hESC-RPE cell population by evaluating the lower limit of detection. We identified a detection threshold of 0.05% for hESC contamination of hESC-RPE cells by mixing ARPE-19 cells, a human RPE cell line, with decreasing concentrations of hESCs from 50 to 0.05% (fig. S6, A and B). No hESCs were detected in hESC-RPE cell populations in any of the batches tested (fig. S6C). Together, our data indicated that our culture process had yielded a pure hESC-derived RPE cell population.

Fig. 1 Characterization of hESC-RPE cells.

(A) Timeline of the differentiation protocol for generating retinal pigment epithelial (RPE) cells from human embryonic stem cells (hESCs). (B) Quantitative reverse transcription polymerase chain reaction analysis was performed to measure mRNA expression of RPE markers [paired box 6 (PAX6), retinal pigment epithelium–specific 65-kDa protein (RPE65), bestrophin 1 (BEST1), and microphthalmia-associated transcription factor (MITF)] and pluripotency markers [POU class 5 homeobox 1 (POU5F1) and nanog (NANOG)] in three hESC-RPE cell differentiation batches at passage 1. Expression is presented relative to expression in undifferentiated hESCs. (C to E) Confocal images (maximal projections of zx planes) of hESC-RPE cells after immunostaining for the markers ZO-1 (zonula occludens-1) and BEST1 (C), PAX-6 and EZRIN (D), and MERTK and BEST1 (E). Nuclei were counterstained with DRAQ5 (white). Scale bars, 10 μm. (F) Evaluation by flow cytometry of the number of tyrosinase-related protein 1 (TYRP1)–positive and LIN28-positive cells before hESCs and after differentiation into RPE cells (hESC-RPE cells). Four hESC-RPE cell batches were tested in total. (G) Representative images of hESC-RPE cells after 3 hours of exposure or no exposure to fluorescein isothiocyanate (FITC)–labeled pig photoreceptor cell outer segment (FITC-POS; green). Nuclei were counterstained with 4′,6-diamidino-2-phenylindole (DAPI) (white). Scale bars, 10 μm. (H) Quantification of vascular endothelial growth factor (VEGF) secreted by hESC-RPE cells at different time points during culture using an enzyme-linked immunosorbent assay. Values plotted are means ± SD.

A crucial function of native RPE cells is their ability to phagocytose the shed outer segments of photoreceptor cells (5, 40). We demonstrated that hESC-RPE cells were able to phagocytose pig fluorescein isothiocyanate (FITC)–labeled photoreceptor cell outer segments (Fig. 1G). Using a permeable support to provide independent access to both sides of a hESC-RPE cell layer, we also showed that, similar to native RPE cells, hESC-RPE cells secreted vascular endothelial growth factor (VEGF) preferentially at the basal membrane (Fig. 1H). The hESC-RPE cells obtained with our protocol were banked using GMP-compliant cryopreservation media.

Characterization of tissue-engineered hESC-RPE cell sheets

To obtain hESC-RPE cell sheets for transplantation, we used a clinical-grade decellularized hAM as a biological scaffold. This hAM contained a basement membrane, which mimicked Bruch’s membrane, and could support growth of RPE cell primary cultures (41, 42). RPE cells derived from hESCs using our GMP-compliant protocol were seeded onto the basement membrane side of the decellularized hAM. After 4 weeks in culture, pigmented epithelial cells that expressed classical RPE markers, such as TYRP1 and MITF, appeared (Fig. 2, A to C). The apical location of ZO-1 demonstrated that the hESC-RPE cell sheets formed tight junctions (fig. S7, A and B). The localization of EZRIN at the apical side indicated the presence of microvilli (fig. S7, C to E) (43, 44). Expression of the basolateral marker BEST1 indicated that ion channels had formed in the correct location (fig. S7, F and G). The expression and polarization of RPE cell markers in hESC-RPE cell sheets were similar to those for hESC-RPE cells cultured on coated culture dishes (Fig. 1 and fig. S3). We also demonstrated the ability of hESC-RPE cell sheets grown on a hAM scaffold to secrete VEGF (fig. S7H). Ultrastructural studies using transmission and scanning electron microscopy confirmed that hESC-RPE cell sheets cultured on a hAM scaffold showed the correct location on the apical side of microvilli, cilia, tight junctions, and melanosomes; the cell nucleus was located on the basal side of the hESC-RPE cell sheet (Fig. 2, D and E, and fig. S8). Finally, hESC-RPE cell sheets grown on the hAM scaffold were able to phagocytose photoreceptor cell outer segments in the same way as hESC-RPE cells cultured on coated culture dishes (fig. S9, A to D; movie S1; and Fig. 1G). Together, these results showed that the hAM scaffold enabled hESC-RPE cells to form well-organized epithelial sheets that showed the correct localization of key RPE proteins.

Fig. 2 In vitro assessment of hESC-RPE cell sheets.

(A) Macroscopic photographic images of a hESC-RPE cell sheet after 6 weeks in culture. The right-hand image shows the hESC-RPE cells on the human amniotic membrane (hAM) scaffold. Scale bar, 50 μm. (B) Section of the hESC-RPE cell sheet illustrating the monolayer organization of hESC-RPE cells. Cells were stained for TYRP1 expression (red) and counterstained with DAPI (blue). The hAM scaffold is indicated. Scale bar, 50 μm. (C) hESC-RPE cells cultured on the hAM scaffold and stained for specific RPE markers. Top: ZO-1, green; MITF, red. Bottom: DAPI, blue; TYRP1, red. Scale bars, 50 μm. (D) Transmission electron microscopy image of the hESC-RPE cell sheet. Scale bar, 5 μm. (E) Scanning electron microscopy images of the hESC-RPE cell sheet showing hESC-RPE cells on the hAM scaffold at different magnifications. The bottom image (a magnification of the area indicated by the rectangle in the top image) shows the basement membrane and extracellular matrix fibers of the hAM scaffold. Scale bars, 5 μm (top) and 1 μm (bottom). Three different hESC-RPE cell batches were tested.

Development of a surgical method for transplanting hESC-RPE cell sheets into the subretinal space of the rat eye

Introducing a tissue-engineered hESC-RPE cell sheet into the subretinal space of the rat eye without damaging the recipient retina or the tissue-engineered product is technically challenging. We developed a grafting strategy to transplant the tissue-engineered product through a transscleral route while maintaining the polarity of the hESC-RPE cell sheet such that the RPE cells were in contact with the retina (Fig. 3A). We embedded the hESC-RPE cell sheet on a hAM scaffold into a GMP-compliant gelatin carrier to reduce friction between hESC-RPE cells and the injector. The gelatin was rigid at 4°C and liquefied at body temperature and, therefore, was suitable as a carrier for transplantation of the hESC-RPE cell sheet into the subretinal space of the rat eye (45). We then introduced a small piece of the tissue-engineered product (2 to 3 mm2 containing up to 5 × 104 hESC-RPE cells; Fig. 3B) into the head of an injection cannula (Fig. 3, B and C). The cannula could be opened to carefully position the tissue-engineered product into the subretinal space; when the cannula was closed, the tissue-engineered product rolled up. The transparency of the device enabled accurate positioning of the hESC-RPE cell sheet ready for injection (Fig. 3C and movie S2). A terminal deoxynucleotidyl transferase–mediated deoxyuridine triphosphate nick end labeling assay to detect DNA fragmentation during apoptosis was performed on the hESC-RPE cell sheet after extrusion from the injection device; the assay revealed no increase in hESC-RPE cell death compared to the tissue-engineered product before extrusion (fig. S10), validating the safety of the injection device for the delivery of the tissue-engineered product. Finally, optical coherence tomography (OCT) imaging after injection into the subretinal space of rat eyes in vivo revealed that the hESC-RPE cell sheet was correctly localized after the surgical procedure (Fig. 3D).

Fig. 3 hESC-RPE cell sheet transplantation and validation of the surgical method.

(A) Steps for the preparation and loading of the hESC-RPE cell sheet on the hAM scaffold into the injection device. (B) Image of the hESC-RPE cell sheet embedded in gelatin. A small piece (2 to 3 mm2) was cut for injection into the subretinal space of the rat eye (black rectangle). (C) Image showing loading of the hESC-RPE cell sheet (black rectangle) into the head of the injection device. (D) OCT analysis of the retina of a wild-type control rat and a dystrophic RCS rat after transplantation of the hESC-RPE cell sheet (red line, dystrophic + graft). (E to G) Representative immunofluorescence confocal microscopic images of sections of retina from athymic nude rats 10 days after transplantation with either a hESC-RPE cell sheet (left) or a hESC-RPE cell suspension (right). Sections were stained for human MTCO2 and human ZO-1 (E), human MERTK and human MTCO2 (F), or human MERTK and human collagen IV (G). Nuclei were visualized with a DAPI counterstain. Three rat retinas transplanted with hESC-RPE cell suspensions and four rat retinas transplanted with hESC-RPE cell sheets were analyzed. White box indicates regions that are enlarged. Images correspond to maximal projections of z stacks. Scale bars, 50 μm and 10 μm (higher magnification). IPL, inner plexiform layer; INL, inner nuclear layer; OPL, outer plexiform layer; ONL, outer nuclear layer; OLM, outer limiting membrane.

hESC-RPE cell sheets express MERTK in vivo

We grafted athymic nude rats with either a hESC-RPE cell suspension or the tissue-engineered hESC-RPE cell sheet and investigated the expression of RPE proteins in transplanted RPE cells 10 days after surgery. We found that grafted hESC-RPE cells, identified by specific staining for human cytochrome c oxidase subunit II (MTCO2), expressed ZO-1 both in the cell suspension and in the epithelial sheet (Fig. 3, E and F). Whereas most of the hESC-RPE cells in the epithelial sheet expressed human MERTK, no cells positive for human MERTK were detected in the hESC-RPE cell suspension (Fig. 3, F and G, and movie S3). Specific staining for human collagen IV indicated the hAM basement membrane in the transplanted tissue-engineered product (Fig. 3G).

Transplantation of hESC-RPE cell sheets improves visual acuity in RCS rats with retinal degeneration

The functionality of the tissue-engineered hESC-RPE cell sheet was then tested in the Royal College of Surgeons (RCS) rat, a rodent model of inherited retinal degeneration. These animals are characterized by defective phagocytosis by the RPE layer due to a mutation in the Mertk gene (46), which is also mutated in some forms of autosomal recessive retinitis pigmentosa (47). This defect results in an accumulation of photoreceptor cell outer segments in the subretinal space leading to degeneration of photoreceptor cells beginning at 3 weeks of age, with an almost complete loss of the outer nuclear layer at 6 weeks of age. Four-week-old RCS rats were transplanted in one eye with the tissue-engineered product in its final formulation (gelatin embedded) or with the same batch of hESC-RPE cells as a suspension or with an empty gelatin block (sham). The contralateral eye did not receive any transplant and served as a retinal dystrophic control. To evaluate basic visual performance after surgery, we took advantage of the optokinetic head movement reflex of rats in response to a visual stimulus. This optokinetic test allowed evaluation of the visual acuity of the transplanted rats by reducing the width of the moving stripes in the rat’s visual field (Fig. 4A). Whereas the visual acuity of untreated and sham-treated eyes of the dystrophic rats was markedly reduced compared to wild-type animals, we observed that the eyes of dystrophic rats transplanted with hESC-RPE cells as either a suspension or a sheet did not show any further reduction in visual acuity for up to 3 months after transplant (Fig. 4B). At 13 weeks after transplant, visual acuity in dystrophic rat eyes transplanted with the hESC-RPE cell sheet (0.514 ± 0.031 cycles/deg) was significantly preserved compared to sham-operated eyes (0.268 ± 0.043 cycles/deg; ***P < 0.0001) and to dystrophic control eyes (0.117 ± 0.016 cycles/deg; ***P < 0.0001) (Fig. 4B).

Fig. 4 Improved visual recovery after hESC-RPE cell sheet transplantation into RCS rats.

(A) Visual representation of the optokinetic test. Animals were placed on an elevated platform and exposed to a rotating stimulus (on four screens) consisting of vertical black and white lines of varying widths. (B) Bar graphs showing the quantification of visual acuity using the optokinetic test in RCS rats at different time points after transplantation (4, 6, and 13 weeks). (C to H) Electroretinogram responses were recorded at 5, 9, and 12 weeks after surgery in RCS rats transplanted with hESC-RPE cell sheets (yellow), gelatin alone (sham, black), or hESC-RPE cell suspensions (green) compared to untreated RCS rats (blue). Results are presented as average response (b-wave) curves (under dim light exposure) to flashing lights of increasing intensity (C, E, and G) and as area under the curve (AUC) measurements for the corresponding graphs for each group (D, F, and H). Six to nine animals per time point. Analysis of variance (ANOVA) followed by Fisher’s protected least squares difference (PLSD) test; *P < 0.05, **P < 0.01, and ***P < 0.001. Kruskal-Wallis test followed by Dunn’s multiple comparison post hoc test.

Transplantation of hESC-RPE cell sheets improves retinal electrophysiology in the RCS rat retina

Analysis of full-field electroretinograms 5 weeks after surgery showed that transplanted groups responded to increasing light intensity as measured by the b-wave amplitude (Fig. 4C). Calculation of the area under the curve confirmed that transplantation of hESC-RPE cells improved the response of the dystrophic retina to increasing light intensity [analysis of variance (ANOVA) F3,49 = 59.78, ***P ≤ 0.0001] (Fig. 4D). By 9 weeks after surgery, dystrophic retinas transplanted with the tissue-engineered hESC-RPE cell sheets responded to increasing light intensities with greater sensitivity than did the hESC-RPE cell suspension transplantation group (Fig. 4, E and F). At a light intensity of −0.7 log cd⋅s/m2, mean b-wave responses of dystrophic rat retinas transplanted with the tissue-engineered product (198.86 ± 11.12 μV) were double the intensity of dystrophic retinas transplanted with the hESC-RPE cell suspension (95.78 ± 22.56 μV; ***P < 0.0001). At 12 weeks after transplant, the b-wave amplitude was decreased in all transplanted groups; however, the response to light stimulation was greater in dystrophic retinas transplanted with the tissue-engineered hESC-RPE cell sheet compared to dystrophic retinas transplanted with the hESC-RPE cell suspension or gelatin alone (sham) (Fig. 4, G and H).

To determine the specific impact of the hAM scaffold, we also transplanted RCS rats with the hAM scaffold embedded in gelatin but without any hESC-RPE cells. A transitory improvement was observed in the electroretinogram and the optokinetic test 5 weeks after surgery (fig. S11, A and B). The mean b-wave observed at 9 weeks after surgery was improved by 41% in the tissue-engineered product transplant group compared to the group transplanted with hAM in gelatin alone (flashing light intensity of −0.7 log cd⋅s/m2; **P < 0.01; fig. S11, C and D). This effect remained for up to 12 weeks after surgery (improvement of 32% in the tissue-engineered product transplant group compared to the hAM-in-gelatin only group and the gelatin-only sham group; *P < 0.05; fig. S11E).

Transplantation of hESC-RPE cell sheets improves photoreceptor cell survival in the RCS rat retina

The improvements in visual acuity after hESC-RPE cell sheet transplantation suggested that either photoreceptor cell survival or function was improved. Three months after transplantation, OCT imaging was used to scan the temporodorsal quadrant near the optic nerve of transplanted animals at an advanced stage of retinal degeneration. The heat maps reflecting outer nuclear layer thickness in the OCT scan revealed that hESC-RPE cell sheet transplantation resulted in preservation of a larger area of retinal outer nuclear layer compared to transplantation with hESC-RPE cell suspensions or gelatin alone (sham control) (Fig. 5A and fig. S12). Further examination of a horizontal temporonasal (Fig. 5B) or vertical dorsoventral (Fig. 5C) region around the area of recovery suggested that the increase in outer nuclear layer thickness was greater after transplanation of hESC-RPE cell sheets compared to hESC-RPE cell suspensions. This finding was confirmed by quantification of the mean outer nuclear layer thickness per treated rat eye (Fig. 5D). Mean outer nuclear layer thickness was increased in the group receiving hESC-RPE cell sheets compared to hESC-RPE cell suspensions or gelatin alone (ANOVA F2,14 = 8.56, **P ≤ 0.01; Fig. 5D and fig. S13). Histological evaluation confirmed that photoreceptor cell nuclei were preserved in rat eyes transplanted with hESC-RPE cell sheets compared to hESC-RPE cell suspensions (Fig. 6A). Retinal outer nuclear layer thickness was not increased beyond the site of transplantation (Fig. 6A). Immunohistochemical staining for the photoreceptor cell markers recoverin and rhodopsin confirmed more photoreceptor cell rhodopsin-positive outer segments in rat eyes transplanted with hESC-RPE cell sheets compared to hESC-RPE cell suspensions (Fig. 6, B to D).

Fig. 5 OCT analysis of RCS rat retinas 12 weeks after hESC-RPE cell sheet transplantation.

(A) Reconstructed OCT image of a retina from a wild-type rat (top left). The heat maps illustrate the thickness of the ONL for each condition. Wild-type and dystrophic RCS rat retinas are shown in the top panels, and dystrophic RCS rat retinas transplanted with gelatin only (sham), hESC-RPE cell sheets, or cell suspensions are shown in the bottom panels. White squares represent the optic nerve of each animal. The horizontal and vertical lines (denoted a′ and b′, respectively) correspond to the location of the two widths selected to illustrate ONL thickness. (B) ONL thickness in the temporonasal axis [indicated by double-headed arrows in (A, a′)]. (C) ONL thickness in the dorsoventral axis [indicated by the double-headed arrows in (A, b′)]. Five to six animals per condition. Scale bar, 100 μm. (D) Histogram showing the mean ONL thickness in RCS rat retinas after transplantation with gelatin only (sham), hESC-RPE cell sheets, or cell suspensions (mean of 420 individual measurements per eye; five to six animals per condition). ANOVA followed by Fisher’s PLSD test; *P < 0.05 and **P < 0.01. Kruskal-Wallis test followed by Dunn’s multiple comparison post hoc test.

Fig. 6 Photoreceptor cell survival in the RCS rat retina after transplantation with hESC-RPE cell sheets.

Immunofluorescence staining for recoverin (green) and rhodopsin (red) in sections of RCS rat retinas 3 months after transplantation with gelatin alone (sham), hESC-RPE cell sheets, or hESC-RPE cell suspensions is shown. Nuclei were counterstained with DAPI (white). White boxes in (A) indicate area of enlargement shown in (B) (recoverin), (C) (rhodopsin), and (D) (merge + DAPI). The asterisks in (D) indicate the position of the ONL. Images correspond to maximal projections of z stacks. Scale bars, 50 μm (top row) and 10 μm (bottom row).

We evaluated whether transplantation of an exogenous agent, in this case gelatin alone, could affect the activation of macrophages/microglia in the retina. Using an antibody against ionizing calcium-binding adaptor molecule 1, a protein specifically expressed by activated macrophages/microglia (48), we observed immune cell activation in untreated dystrophic rat retinas and those transplanted with gelatin alone (fig. S14).


Here, we asked whether the delivery of hESC-RPE cell sheets provided greater efficacy than hESC-RPE cell suspensions for retinal dystrophic rescue in the RCS rat model of retinal degeneration. We engineered growth of hESC-RPE cell sheets on a pretreated hAM scaffold and demonstrated improved photoreceptor cell rescue and retinal function after transplantation into the eyes of RCS rats with retinal dystrophy. Transplanting hESC-RPE cells as epithelial sheets rather than as cell suspensions led to improved rescue of photoreceptor cells and improved visual acuity in the RCS rats.

Comparable protocols, some with GMP-compliant components, have been developed for the generation of RPE cells from hESCs and hiPSCs (23, 4955). For example, Schwartz and collaborators (1) used GMP-compliant hESCs cultured on mouse feeder cells to generate human RPE cells, and Kamao and colleagues (29) used GMP-compliant hiPSCs to generate human RPE cells. Here, we used a feeder-free hESC line cultured and banked under GMP conditions to generate hESC-RPE cells for transplantation into RCS rats with retinal degeneration.

We demonstrated that the hAM scaffold supported hESC-RPE cell growth as epithelial sheets. We cultured hESC-RPE cells on hAM scaffolds derived from different batches and routinely obtained mature epithelial sheets. The hAM scaffold has been used clinically, for example, in the treatment of corneal ulceration (31). We prepared our hAM scaffolds in small batches and cryopreserved them in tissue banks for future use to demonstrate scalability (31).

Our preclinical data in RCS rats demonstrated that transplantation of tissue-engineered hESC-RPE cell sheets improved visual acuity when compared to transplantation of hESC-RPE cell suspensions. The early photoreceptor cell preservation elicited by gelatin alone (sham control) may reflect debris, that is, shed photoreceptor cell outer segments that were washed into the space created during subretinal surgery (28, 56, 57). The accumulation of debris causes the dysfunction and death of photoreceptor cells, and this transitory washing away of the debris might have delayed photoreceptor cell death. Alternatively, insertion of gelatin alone into the subretinal space could have promoted photoreceptor cell survival by enabling the sustained release of neurotrophic factors (58). Here, we observed that gelatin alone did not correlate with better visual performance on the optokinetic test, highlighting the crucial role that transplanted hESC-RPE cells played in the improvement in visual acuity in the transplanted rats. Electroretinogram responses indicated that some photoreceptor cells were still present in the RCS rat retina and were sensitive to light. A few remaining photoreceptor cells could explain the weak protective effect observed after transplantation of gelatin alone.

In our study, hESC-RPE cell suspensions elicited similar or better rescue of electroretinogram responses and restoration of visual acuity in RCS rats compared to cell preparations used in other studies (29, 5961). RPE cell sheets secrete more growth factors and cytokines than do RPE cell suspensions (62) and are less sensitive to oxidative stress–induced cell death (26). This may explain the improvement in visual acuity observed after transplantation of tissue-engineered hESC-RPE cell sheets compared to hESC-RPE cell suspensions. Different substrates and scaffolds have been proposed to generate RPE cell sheets (15, 20, 6365) in a way that increases the survival of transplanted cells and helps to reconstruct the damaged Bruch’s membrane of the choroid, an injury common to different forms of age-related macular degeneration.

A limitation of our study is the risk of immune responses to the transplanted hESC-RPE cell sheets in human patients. The survival of transplanted cells is limited in xenotransplantation settings (20, 63, 66). Transplantation into genetically immunocompromised athymic nude rats did enable longer graft survival (27). For most experiments, we used RCS rats that were not immunocompromised, and therefore, we had to add the immunosuppressive drug cyclosporine to the rats’ drinking water (19, 67). In addition, the transscleral surgery may damage the blood-retina barrier, potentially allowing immune cells to enter the retina. Addressing immune rejection in our xenotransplantation model will not provide information about how hESC-RPE cell sheets will survive in the human eye, an allogeneic transplantation setting. Banking of hESCs and hiPSCs with different haplotypes may help to mitigate this problem (68, 69). Subretinal transplantation of monkey iPSC-RPE cells did not result in immune rejection when the major histocompatibility complex of the donor iPSC-RPE cells matched that of the recipient in the absence of immunosuppression (70). Local application of corticosteroids (for example, intraocular corticosteroid capsules or transitory immune suppression) may also help to mitigate the possibility of immune rejection (2).

Another limitation of this study concerns the larger delivery injector device that would be required for use in human eyes. The head of the injector device used in this study for small rat eyes would need to be elongated to enable it to access the subretinal space of human eyes. For successful clinical translation, a surgical technique ensuring accurate delivery of hESC-RPE cell sheets into the human eye is required. Notably, the transscleral route used in rodent eye transplantation is not appropriate for cell therapy delivery to the human eye. Rather, a subretinal delivery route via a vitrectomy followed by a retinotomy performed by ophthalmology surgeons is more appropriate (4, 71, 72).

On the basis of our preclinical results, we are currently generating GMP-compliant hESC-RPE cell sheets with the goal of launching a phase 1/2 clinical trial to treat patients with retinitis pigmentosa caused by mutations in RPE65, lecithin retinol acyltransferase (LRAT), or MERTK. Our strategy may also be useful for treating other retinal diseases where Bruch’s membrane is damaged such as age-related macular degeneration and Bietti’s corneoretinal dystrophy (73, 74).


Study design

The goal of this study was to generate GMP-compliant hESC-RPE cell sheets and to characterize the effects of their transplantation into the subretinal space of RCS rats, compared to hESC-RPE cell suspensions. Predefined end point criteria included electroretinogram responses, histological evaluation, OCT analysis, and optokinetic testing. No prespecified effect size was used to determine sample sizes. Sample sizes were estimated from preliminary experiments. No outliers were excluded from data analysis. Operators were blinded to the treatment during acquisition and analysis of data.

RPE cell differentiation and banking

The differentiation medium was composed of Dulbecco’s modified Eagle’s medium with 20% knockout serum (KSR, Thermo Fisher Scientific), supplemented with 50 μM β-mercaptoethanol and 1× minimum essential media–nonessential amino acids (Thermo Fisher Scientific). When pigmented patches appeared in the differentiating culture, these patches were dissected under a stereomicroscope with a fine 15° ophthalmic knife. Only pigmented areas are collected to ensure RPE purity. They are seeded (P0) in 24-well plates (about 10 clumps per well-P24) coated with CTS CELLstart substrate or L7 hPSC matrix, and the medium was then switched to 4% KSR. Both matrices (CTS CELLstart substrate or L7 hPSC matrix) are equivalent and one could be substituted by the other. When hESC-RPE cells reached confluence, they were passaged via enzymatic cell dissociation (TrypLE; Thermo Fisher Scientific) and were banked frozen using Cryostor medium (BioLife Solutions) in liquid nitrogen tanks. The release quality controls that we selected for the future clinical-grade banks are listed in tables S2 and S3.


Adult nude rats (Crl:NIH-Foxn1rnu; Charles River Laboratories) and pigmented dystrophic RCS (rdy−/−, p+) rats were kept in a 12-hour light/12-hour dark cycle on plastic cages and allowed to eat and drink ad libitum in the animal facility of the Vision Institute (agreement number A 75-12-02). All experiments were carried out in strict accordance with the Association for Research in Vision and Ophthalmology statement for animal research in ophthalmology. Moreover, protocols were approved by the local ethical committee (Charles Darwin ethical committee for animal experimentation C2EA-05) in strict accordance with French and European regulation for animal use in research (authorization number 01483.02).

hESC-RPE cell sheet preparation

hESC-RPE cells were directly thawed, washed, and plated over stabilized hAM on Cellcrown for 4/8 weeks before grafting. At the time of injection, the tissue-engineered product (hESC-RPE cells on hAM) was embedded into a GMP-compliant gelatin excipient (Merck). Briefly, a block of 20% gelatin was placed in a vibratome (Leica), and the tissue-engineered product was positioned on top of this block. Then, a solution of 8% gelatin made with CO2-independent medium (Thermo Fisher Scientific) at 37°C was added over the tissue-engineered product. After polymerization, the gelatin block containing the tissue-engineered product was then cut with the vibratome to precisely adjust the thickness of the tissue-engineered product.

Cell transplantation

From 2 days before transplantation to the end of the follow-up, dystrophic RCS rats were maintained under immunosuppression through cyclosporine treatment (210 mg/liter) in the drinking water (75). Twenty-eight-day-old RCS rats were anesthetized with intraperitoneal injection of ketamine (50 mg/kg) and xylazine (5 mg/kg). Adult nude rats (Crl:NIH-Foxn1rnu) (8 weeks old) received injections of ketamine (100 mg/kg) and xylazine (10 mg/kg). An ophthalmic gel was placed onto each eye (Lacrinorm 0.2%, Bausch & Lomb). A few drops of a local anesthetic (oxybuprocaine; Thea) and tropicamide (Mydriaticum 0.5%; Thea) were added onto the cornea of the eye to be transplanted. After a small scleral incision using a small hook, a local subretinal detachment was created through air injection with a Hamilton syringe under an operating microscope (Leica F18). A hole into the cornea was created with a needle to reduce the intraocular pressure. The incision in the sclera was then enlarged with an ophthalmic knife (WPI) to allow the entry of the head of the injector (Viscoject, Medicel). The tissue-engineered product (2 to 3 mm2; up to 5 × 104 cells) was slowly delivered in the subretinal space, and the hole in the sclera was then immediately closed using a 10-0 suture (Peters Surgical). Sham-operated animals received the same protocol with an injection of the equivalent quantity of gelatin. For the cell suspension group, hESC-RPE cells [4 μl; 5 × 104 cells/μl in CO2-independent medium (Thermo Fisher Scientific)] were slowly injected using a Hamilton syringe just after the formation of the subretinal space. After the surgery, a vitamin A ointment (Allergan) was placed on each eye and animals were kept on a thermostated cage until awakening. Animals from each litter were distributed randomly to each treatment condition. For each litter/session of transplantation, every effort was made to treat at least one animal per condition. Animals with signs of major intraocular hemorrhage during the surgery or with disrupted retina/absence of graft based on OCT evaluation were excluded (about 20% of animals).


Full-field electroretinographies were performed 5, 9, and 12 weeks after surgery. After 24 hours of dark adaptation, animals were anesthetized by intraperitoneal injection of ketamine (100 mg/kg) and xylazine (10 mg/kg). Pupils were dilated with tropicamide (Mydriaticum 0.5%; Thea), and the cornea was locally anesthetized (oxybuprocaine chlorhydrate; Thea). Body temperature was maintained at 37°C using a heating pad. Gold electrodes were placed on each cornea, which were humidified with an ophthalmic gel (Lacrinorm 0.2%, Bausch & Lomb). Ground and reference electrodes were placed on the back and cheeks. Full-field electroretinograms were recorded simultaneously in both eyes using Visiosystem (SIEM Bio-Médicale). A Ganzfeld stimulator generated light stimuli (SIEM Bio-Médicale). Each individual response is the mean of five measures. At the end of the procedure, a vitamin A ointment was added in each eye, and animals were placed in a thermostatic cage until awakening. The operator performing electroretinography was blinded to the treatment of each animal.

Optical coherence tomography

Spectral domain OCT was performed 12 weeks after surgery on rats anesthetized with ketamine (100 mg/kg) and xylazine (10 mg/kg). Tropicamide (Mydriaticum; Thea) and phenilephrine (Neo-Synephrine; Europhta) were used for pupil dilatation. The OCT imaging device (Bioptigen 40 nm HHP) was coupled to InVivoVue software (Bioptigen). Rectangular scans were used to acquire b-scans near the optic nerve. En face OCT images were reconstructed from b-scans using ImageJ software (reslice and maximal projection tools). Four hundred twenty measurements of the outer nuclear layer thickness were taken with ImageJ near the optic nerve in the temporodorsal quadrant, corresponding approximately to the region of transplantation. Data collected were represented as a heat map. The operator performing OCT was blinded to the treatment of each animal. A mean of these 420 measures per eye was also used to evaluate the mean recovery in this quadrant.

Behavioral testing

Optokinetic tests were performed 4, 6, and 13 weeks after surgery using the automated OptoMotry system (CerebralMechanics Inc.). The animal was placed over an elevated platform in a chamber composed of four screens, creating a virtual cylinder centered on the animal head. The software automatically modifies the width of the moving stripes according to the animal responses pointed out by the experimenter. Animals that did not stay in the platform were excluded from the test. At the end of the test, the OptoMotry system gave a value of visual acuity for each eye evaluated. The operator performing the test was blinded to the treatment administered.

Statistical analysis

Statview 4.5 software (SAS Institute Inc.) and GraphPad Prism 5 (GraphPad Software Inc.) were used for statistical analysis. Normality was checked and evaluated with the Shapiro-Wilk test. F test was used to assess equality of variances. For normally distributed data, ANOVA followed by Fisher’s protected least squares difference test was performed; significance was set at *P < 0.05, **P < 0.01, and ***P < 0.001. When data were not normally distributed, Kruskal-Wallis nonparametric test was used. Dunn’s multiple comparison post hoc test was then performed for comparisons among groups. Data are expressed as means ± SEM, except if cited differently.


Materials and Methods

Fig. S1. Characterization of the RC-9 hESC line.

Fig. S2. Process for hESC-RPE cell generation.

Fig. S3. Expression of RPE markers by hESC-RPE cells.

Fig. S4. Purity of hESC-RPE cell preparation by expression of PAX-6 and MITF.

Fig. S5. Purity of hESC-RPE cell preparation by expression of cytokeratins.

Fig. S6. Evaluation of hESC-RPE cell preparation by FACS.

Fig. S7. Visualization of hESC-RPE cell polarity on the hAM scaffold by immunofluorescence.

Fig. S8. Analysis of hESC-RPE cells on the hAM scaffold by transmission and scanning electron microscopy.

Fig. S9. Phagocytosis by hESC-RPE cells.

Fig. S10. Absence of apoptosis in the hESC-RPE cell monolayer after extrusion from the injection device.

Fig. S11. Functional assessment in RCS rats after transplantation with hESC-RPE cell sheets compared to hAM or gelatin alone.

Fig. S12. OCT analysis of transplanted retinas.

Fig. S13. OCT analysis of retinas after transplantation with hESC-RPE cell sheets compared to hAM or gelatin alone.

Fig. S14. Analysis of inflammation markers in the retina of RCS rats after transplantation of gelatin alone.

Table S1. Quality control data for the hESC bank.

Table S2. Quality control data for the hESC-RPE bank.

Table S3. Targeted regions for virus detection.

Table S4. List of quantitative reverse transcriptase polymerase chain reaction (qRT-PCR) probes.

Table S5. List of primary antibodies.

Movie S1. 3D reconstructed images of hESC-RPE cell sheets after phagocytosis of photoreceptor cell outer segments.

Movie S2. Method developed to load the hESC-RPE cell sheet into the injection device.

Movie S3. 3D reconstructed images of a thin section of normal retina transplanted with the hESC-RPE cell sheet.


Acknowledgments: We thank the staff from Animal, Phenotyping and Imaging Facilities of the Institut de la Vision and I-Stem for help and members of the Goureau and Monville laboratories for helpful comments. We thank E. Nandrot for the gift of FITC-labeled photoreceptor cell outer segment preparations from pigs and V. Bazin and G. Toutirais from the electron microcopy facility (IBPS/FR3631-UPMC-Paris 6) for performing transmission and scanning electron microscopy. We thank Roslin Cells for providing RC-9 hESCs. Funding: This work was supported by grants from the Agence Nationale pour la Recherche (ANR) (GPiPS: ANR-2010-RFCS005) and the Fondation pour la Recherche Medicale (Bio-engineering program—DBS20140930777) to O.G. and C.M. and by the European Research Council ERC Synergy (Helmholtz) to J.-A.S. It was also performed in the frame of the LABEX LIFESENSES (ANR-10-LABX-65) supported by the ANR within the Investissements d’Avenir program (ANR-11-IDEX-0004-02) to O.G. and supported by NeurATRIS: A Translational Research Infrastructure for Biotherapies in Neurosciences (Investissements d’Avenir—ANR-11-INBS-0011) and INGESTEM: the National Infrastructure Engineering for Pluripotent and Differentiated Stem Cells (Investissements d’Avenir—ANR-11-INBS-000) to C.M. K.B.M. was supported by fellowships from DIM Stempole and the LABEX REVIVE (ANR-10-LABX-73). A.T. was supported by a grant from the Fondation de France (Berthe Fouassier). I-Stem is part of the Biotherapies Institute for Rare Diseases supported by the Association Française contre les Myopathies-Téléthon. Author contributions: A.P., W.H., and K.B.M. performed cell culture and RPE cell differentiation experiments. M.J. and W.H. prepared hAM scaffolds. W.H. and K.B.M. performed RPE quality control evaluations. F.R. and K.B.M. performed FACS purity experiments. K.B.M. performed in vivo experiments. A.T. contributed to electron microscopy. Y.Y. contributed to surgery experiments. L.C. performed TaqMan Low Density Arrays gene card experiment; S.D. and Y.M. performed karyotyping. J.-A.S., M.P., O.G., and C.M. provided resources. M.P., J.-A.S., K.B.M., O.G., and C.M. designed the study and discussed the data. K.B.M., O.G., and C.M. wrote the manuscript. Competing interests: J.-A.S. and O.G. are inventors on pending patent number WO2014174492 entitled “Methods for obtaining retinal progenitors, retinal pigmented epithelial cells and neural retinal cells.” J.-A.S. is a consultant for Pixium Vision, GenSight Biologics, and Gene Signal and has personal financial interests in GenSight Biologics, Chronocam, ChronoLife, Pixium Vision, Tilak Healthcare, and SparingVision. All other authors declare that they have no competing interests. Data and materials availability: The hAM scaffold was provided by the Tissue Bank of Saint Louis Hospital (AP-HP Paris; contact: M. Jarraya). The RC-9 hESC line was provided by Roslin Cells (Roslin Cell Therapies Ltd., United Kingdom; contact: J. Downie).

Stay Connected to Science Translational Medicine

Navigate This Article