Development of Stem Cell Therapies for Retinal Degeneration

Abstract

Degenerative retinal disease is the major cause of sight loss in the developed world and currently there is a lack of effective treatments. As the loss of vision is directly the result of the loss of retinal cells, effective cell replacement through stem-cell-based therapies may have the potential to treat a great number of retinal diseases whatever their underlying etiology. The eye is an ideal organ to develop cell therapies as it is immune privileged, and modern surgical techniques enable precise delivery of cells to the retina. Furthermore, a range of noninvasive diagnostic tests and high-resolution imaging techniques facilitate the evaluation of any therapeutic intervention. In this review, we evaluate the progress to date of current cell therapy strategies for retinal repair, focusing on transplantation of pluripotent stem-cell-derived retinal pigment epithelium (RPE) and photoreceptor cells.

The loss of sight has a profound effect on everyday life and it is predicted that by 2020 blindness will afflict 38.5 million people worldwide (Flaxman et al. 2017). Disorders such as age-related macular dystrophy (AMD) and inherited retinal disease are two major causes of severe visual impairment in the developed world, affecting around 15 million people (Resnikoff et al. 2004; Klein 2007). The economic burden associated with vision loss is substantial, and with increasing longevity and the high prevalence of age-related disorders, this is predicted to rise further (Cruess et al. 2008). For advanced degenerative disease, which is characterized by extensive loss of retinal neurons, there are currently no effective treatments. Several therapeutic strategies are currently being explored including optogenetics, electrical prosthetic implants, and, the subject of this review, cell-based therapies.

Optogenetic approaches for the treatment of outer retinal degeneration utilize the remaining retinal cells and circuitry. In general, these approaches aim to convert the secondary or tertiary neurons that remain in the degenerate retina to light-sensing cells through viral vector–mediated gene delivery of light-sensitive proteins, such as the microbial channel rhodopsins. A number of preclinical studies have demonstrated retinal function can be improved via various optogenetic strategies. However, they also indicate the efficacy of these approaches may currently be limited by the low sensitivity or the slow kinetics of the optogenetic molecules used and by limited spatial resolution and retinal processing when secondary or tertiary neurons, rather than photoreceptors, are used as light sensing cells (for review, see Baker and Flannery 2018; Simunovic et al. 2019). A phase I/II clinical trial is currently underway in patients with advanced retinitis pigmentosa (RP) to determine whether some visual function can be restored using a viral vector to deliver a microbial channel rhodopsin to retinal ganglion cells (RGCs) (NCT02556736). An alternative strategy for the treatment of advanced disease is to replace photoreceptors with photodiode arrays. The implantation of these light-sensitive prosthetic devices (e.g., ArgusII, Alpha-IMS, IMI, IRIS, and EPI-RET 3) that generate and pass electrical stimuli to the remaining retinal circuitry have been tested in several clinical trials (Luo and da Cruz 2014). Because of very limited spatial resolution achieved with the current electrical devices, this approach is only suitable for those rare patients who have complete lack of light perception. Nevertheless, trials have demonstrated that such devices are able to restore light perception and basic shape recognition, demonstrating the plasticity of the brain to interpret new input from the retina. Although optogenetic molecules and electrical prostheses are continuing to be refined, considerable improvements are still needed to recreate the range and sensitivity of light responses achieved by the highly evolved cells and circuits of the human retina.

Attempts to develop cell-based therapies have been boosted in the last two decades by the development of pluripotent stem cell (PSC) technologies that have enabled the generation of potentially unlimited quantities of specific human cells from either embryonic stem cell (ESC) or induced pluripotent stem cell (iPSC) sources. Cell-based therapeutic strategies to restore vision following degenerative retinal disease have focused on the replacement of three key retinal cells, RGCs, the retinal pigment epithelium (RPE), and photoreceptors. With the support of the RPE, photoreceptors convert light to neuronal impulses that, via the inner retinal neurons, are sent to the brain by the projection of RGC axons (see Fig. 1 for schematic of retinal circuitry). Given the indispensable role of each of these cells in the visual process, their loss ultimately results in blindness. Conditions such as glaucoma result in blindness through the loss of RGCs but achieving functional connectivity following transplantation of these cells poses a major challenge because their axons make very long connections to the brain. Although impressive progress has been made with regard to RGC transplantation and integration (for review, see Laha et al. 2017; Miltner and La Torre 2019), replacement of RPE and photoreceptors may be more feasible in the near term. In this review, we explore the current cell-based strategies to restore vision through PSC-derived RPE and photoreceptor transplantation.

Figure 1.

Structure of the mammalian retina. (A) The retina lines the back of the eye and is a highly organized laminated structure. Supported by the retinal pigment epithelium (RPE) cell monolayer, the neural retina is composed of three layers of cell bodies separated by two plexiform layers that are comprised of the neural processes and synapses. The photoreceptor cell bodies form the outer nuclear layer (ONL), whereas the bipolar, amacrine, horizontal, and Müller glia cells form the inner nuclear layer (INL). The outer plexiform layer (OPL) contains synaptic connections between the photoreceptors, bipolar, and horizontal cells, whereas the inner plexiform layer (IPL) comprises the processes and connections between the retinal ganglion (RGC), bipolar, and amacrine cells. The upper ganglion cell layer (GCL) consists of the RGCs whose axons form the nerve fiber layer and exit the eye through the optic nerve, which projects to the visual structures of the brain. (B) Photoreceptors (PRs), bipolar cells, and ganglion cells form the principal retinal circuit that mediates light stimuli to neural impulses transmitted to the brain. However, the signals are modulated by the other retinal neurons, horizontal and amacrine cells. Müller glial cells span the entire retina providing both structural and metabolic support. The photoreceptors of the mammalian eye can be subdivided in to two different cell types, rods and cones. Rod PRs are the predominant photoreceptor cell type in the human retina and contain the visual pigment rhodopsin. They are sensitive to dim light and are required for scotopic (night) and peripheral vision. In contrast, cone photoreceptors contain cone opsin visual pigment and can be further divided into subtypes dependent on the wavelength of light they detect. Human retinas contain L cones, which detect long wavelengths of light (red light), M cones, which respond to medium wavelengths (green light), and S cones, which detect short wavelengths (blue light). The RPE cells form a monolayer beneath the neural retina, and the proximity of the photoreceptor–RPE complex is essential for both the structure and function of the overlying photoreceptors. Tight junctions between the RPE cells form a physical barrier that prevents the free passage of molecules and ions maintaining the outer blood–retinal barrier. (OLM) Outer limiting membrane, (SRS) subretinal space.

RPE CELL TRANSPLANTATION

The human eye contains approximately 3.5 million RPE cells that form a monolayer between the overlying neural retina and Bruch's membrane (BrM), an acellular structure that separates the RPE from the choroid below. The RPE is essential for maintaining the health of the neural retina and enabling photoreceptor function. Of note, there are relatively few primary disorders of the RPE, given the complexity of this cell and its diverse role (Marmour and Kent 1998), and most of these are recessive monogenetic disorders resulting in deficiency of RPE65, LRAT, MERTK, or Bestrophin (Gu et al. 1997; Veske et al. 1999; D'Cruz et al. 2000; Marmorstein et al. 2000; Thompson et al. 2001; Sun et al. 2002). Despite the relatively few primary RPE dystrophies, there are numerous photoreceptor dystrophies, including Stargardt disease and RP, that result in secondary RPE cell loss. RPE cell dysfunction is also implicated in the pathogenesis of acquired retinal disease such as proliferative vitreoretinopathy (PVR) and AMD (Witmer et al. 2003; Zarbin 2004; Xu et al. 2009), which are characterized by a loss of RPE cells. In such conditions, photoreceptor transplantation alone will fail to rescue visual function and so RPE transplantation has a broader potential role beyond that of treating primary RPE dystrophy.

Preclinical proof-of-concept studies for RPE cell transplantation have been complicated by the lack of appropriate animal models. Although widely used as an animal model for testing RPE transplantation, the Royal College of Surgeons (RCS) rat only provides a model of recessive RP caused by defects in the MerTK gene (D'Cruz et al. 2000). The mutation results in the RPE being unable to phagocytose photoreceptor outer segments and the progressive loss of photoreceptor cells because of a buildup of outer segment debris between the outer nuclear layer (ONL) and the RPE (Bourne et al. 1938; Dowling and Sidman 1962). One limitation of this model for testing RPE transplantation is that because of trophic effects resulting from cell transplantation itself, substantial rescue of photoreceptors can be achieved irrespective of the cell type transplanted (Lawrence et al. 2000; Gamm et al. 2007; Pinilla et al. 2009; Huo et al. 2012). Furthermore, despite the rapid loss of photoreceptors by 2–3 months of age, the RPE monolayer remains intact, even at 1 year (Dowling and Sidman 1962), making this model very dissimilar to most conditions such as to AMD and Stargardt macular dystrophy (SMD), in which the RPE is lost. More conclusive proof-of-concept studies for RPE cell replacement have come from autologous RPE transplants and macular translocation surgeries in AMD patients. These experimental surgeries involve either harvesting a patch of autologous RPE from the periphery of the patient's eye and transplanting it under the macular or repositioning the macular over healthy RPE cells (MacLaren et al. 2007; Treumer et al. 2007; van Romunde et al. 2015; Oshima et al. 2017). Improvements in visual acuity were observed and maintained in some patients. However, because of the complicated surgery involved there has also been a high rate of postoperative complications (MacLaren et al. 2007; Treumer et al. 2007; Parolini et al. 2018; van Romunde et al. 2019). These studies confirm that the replacement of healthy RPE cells, beneath the diseased macular, can, in principle, restore vision.

Clinical Studies Investigating PSC-Derived RPE Transplantation

The RPE was the first human PSC (hPSC)-derived cell type to be transplanted into humans, and since the first trial was initiated in 2011 many more trials have been initiated. This is partly a consequence of the relatively facile differentiation of bone fide RPE cells from PSCs in a highly reproducible manner as well as the ability to generate large-scale quantities of clinical-grade RPE cells (Carr et al. 2009; Idelson et al. 2009; Osakada et al. 2009; Vaajasaari et al. 2011). Trials were also facilitated by the previous development of advanced surgical techniques and imaging modalities, such as optical coherence tomography (OCT), which enable the transplanted RPE cells to be imaged in situ following transplantation.

The first trial using human embryonic stem cell (hESC)-derived RPE was initiated in 2011 by Schwartz and colleagues. In this FDA-approved, phase I/II trial patients with advanced forms of either Stargardt disease or the atrophic form (dry) of AMD received hESC-derived RPE cells transplanted to the subretinal space (Schwartz et al. 2012). The hESC-derived RPE cells were delivered as a cell suspension in escalating dose cohorts of 50,000, 100,000, and 150,000 cells, with each cohort consisting of three patients. The primary aim of the trial was the assessment of safety and tolerability, and no major adverse events were reported. Although minor improvements in best-corrected visual acuity (BCVA) were reported for 6/14 subjects that completed the 12-month follow-up, this was not correlated with the number of cells transplanted and there was no clear evidence of efficacy. A further follow-up at 22 months confirmed no major adverse events (Schwartz et al. 2012, 2015, 2016). A further trial transplanting suspensions of hESC-derived RPE cells into patients with dry AMD and Stargardt disease also reported no serious safety issues after a 12-month follow-up of four patients (Song et al. 2015). More recently, the first European trial using hESC-derived RPE for macular repair in patients with advanced Stargardt disease was reported (Mehat et al. 2018). This involved escalating dose cohorts of 50,000 100,000, 150,000, and 200,000 hESC-derived RPE cells transplanted into a total of 12 patients with detailed retinal sensitivity mapping pre- and posttransplantation. In this study, focal areas of hyperpigmentation were reported for all patients in a dose-dependent manner. No serious adverse events were observed and no uncontrolled proliferation or inflammatory responses were identified. Borderline improvements in visual acuity and retinal sensitivity were unsustained or matched by similar improvements in the untreated eye. In one of the patients that received the highest dose of cells, localized thinning and reduced sensitivity in the area of hyperpigmentation was reported, suggesting the potential for harm with the highest dose. An extended follow-up is necessary to determine whether there are any effects on the progression of degeneration following hESC-derived RPE transplantation in these patients.

An alternative strategy to the use of cell suspensions is the transplantation of hPSC-derived RPE sheets, with or without artificial scaffolds or membranes. Preclinical studies have suggested that RPE cell survival and function are better maintained following the transplantation of sheets of RPE cells, compared with RPE cell suspensions (Tezel and Del Priore 1997; Diniz et al. 2013). Another concern is the rejection of the allogeneic RPE cell graft, because although the subretinal space has a level of immune privilege, healthy RPE cells have a role in maintaining this. Therefore, in diseases that affect the RPE cells and compromise the outer blood–retinal barrier, acute immune rejection of allogenic cells is a possibility. As the majority of adverse responses in the initial hESC-derived RPE safety trial were the result of long-term immunosuppression, possibly exacerbated by the aged patient population, the use of allografts or major human leukocyte antigen (HLA)-matched hPSC-derived tissue would alleviate these concerns. To this end, researchers in Japan developed autologous iPSC-derived RPE cell sheets to determine the feasibility of generating individual autologous grafts for patients. Initial preclinical studies in monkeys, transplanting iPSC-derived RPE cell sheets held together by their extracellular deposits and without the use of immunosuppression, were encouraging (Kamao et al. 2014). In 2014, two patients with exudative (wet) AMD were enrolled in the trial; however, only one received an autologous transplant before the trial was suspended, because of a change in Japan's regenerative medicine law. No serious adverse events were observed throughout the 4-year follow-up. Although the RPE cell sheet appeared intact and the macula showed signs of recovery, no improvement in visual acuity was reported (Mandai et al. 2017b; Takagi et al. 2019). The iPSC line generated from the second patient was found to contain three aberrations and it was decided not to transplant the autologous hPSC-derived RPE cell graft into the second patient, partly as a safety precaution because of the unknown effects of the mutations and partly because of the patient responding to anti-VEGF treatment (Mandai et al. 2017b). It was subsequently concluded that the cost and complexity involved in generating patient-specific iPSC-derived tissue for transplantation was prohibitive for larger scale trials. Therefore, further trials will involve the use of major HLA-matched iPSC lines, available from iPSC banks that have been validated for genomic stability; this may facilitate efficient and affordable generation of HLA-matched hiPSC-derived RPE tissue for transplantation (Nakatsuji et al. 2008; Okita 2011). Although the presence of minor histocompatibility antigens may trigger rejection in humans, preclinical trials in monkeys with major histocompatibility complex (MHC)-matched iPSC-derived RPE sheets without immunosuppression have been encouraging (Sugita et al. 2016).

Other ongoing trials using sheets of hESC-derived RPE cells are investigating the use of artificial scaffolds and support membranes. As changes to BrM can occur as a result of both age and disease, this may reduce its capacity for RPE cell attachment and have adverse effects on both transplanted RPE cell structure and survival (Gullapalli et al. 2004, 2005). The use of hPSC-derived RPE cell sheets with artificial scaffolds or membranes may be able to prevent the transplanted RPE cells coming into contact with the native BrM and maintain the integrity of the cell sheet. Two clinical trials, one in the United Kingdom using polyester membranes and one in the United States using ultrathin parylene membranes, are currently ongoing to investigate the use of nondegradable membranes to immobilize the RPE cells and replace the existing basement membrane (Lu et al. 2012; Carr et al. 2013). The UK group recently published data from their preclinical studies and the 12-month follow-up for the first two patients with severe exudative AMD (da Cruz et al. 2018). This initial safety study also investigated whether the transplanted RPE cell sheet could be sustained with the use of long-term local immunosuppression in the form of an intraocular steroid implant. No severe adverse events were observed in this initial 12-month follow-up and the transplanted RPE sheet was still apparent by OCT imaging, suggesting that, so far, local immunosuppression was sufficient to maintain the allograft. Both patients demonstrated sustained improvements in visual acuity. However, 50% of patients with severe exudative AMD recover vision following surgery to remove membranes, and without enrolling many more patients it is impossible to know whether the improvement was the result of the transplanted RPE cell sheet (da Cruz et al. 2018). Similarly, the U.S. study has published a preliminary report on the first five patients enrolled in the phase 1/2a study to test the safety and potential efficacy of a composite implant in patients with advanced, nonneovascular AMD (Kashani et al. 2018). So far, one out of the five patients had a marginal improvement in visual acuity, with the remaining patients maintaining vision and demonstrating the short-term safety of this approach. Further follow-up and enrollment of up to 20 patients will help determine the continued safety and efficacy of RPE sheet studies. The use of degradable poly(lactic-co-glycolic acid) (PLGA) scaffolds as a temporary substrate to improve RPE cell sheet attachment to the existing BrM is also being investigated, prior to clinical trial (Liu et al. 2014; Song and Bharti 2016).

Combined, these initial safety studies are reassuring and support the potential of PSCs as a source for RPE cell therapy applications. Further follow-up will confirm the long-term safety of this approach, and as more trials investigate the use of various scaffold and membrane materials, as well as iPSC-derived RPE cells, this will help to determine the optimal strategy for RPE transplantation. However, the replacement of RPE cells alone will not be sufficient to improve or reverse visual impairment if the photoreceptor cells have been lost, which is the case in advanced AMD and Stargardt disease. As such, a therapy for end-stage retinal degeneration involving the RPE would also require the transplantation of photoreceptor cells.

PHOTORECEPTOR CELL TRANSPLANTATION

Once photoreceptors are lost, they do not regenerate. However, the rest of the retinal circuitry remains, even in late stages of disease, albeit synaptically remodeled (Marc and Jones 2003; Strettoi et al. 2003; Jones and Marc 2005; Jones et al. 2012). As photoreceptors are afferent neurons, they need only make a single short-range synaptic connection to the host interneurons to restore the visual circuit. There are two main strategies for cell transplantation to the retina, the injection of a suspension of single retinal cells or the transplantation of a sheet of cells into the subretinal space (see Fig. 2).

Figure 2.

Strategies for retinal cell replacement. There are two main strategies for cell transplantation to the retina, the injection of a suspension of single retinal cells and the transplantation of a sheet of cells into the subretinal space (SRS). Both approaches position cells between the host retina and the host retinal pigment epithelium (RPE) cell layer.

The Transplantation of Donor-Derived Cell Suspensions

Early studies in the 1990s, transplanting postnatal photoreceptors to the degenerate retina, reported the survival and maturation of these cells in the subretinal space for up to 3 months (Gouras et al. 1991a,b,c; Kwan et al. 1999). Our own studies in 2006 examined the optimal developmental stage of the transplanted donor population in more detail. We found that whereas the transplantation of GFP⁺ retinal progenitors resulted in the formation of rosettes in the subretinal space, the transplantation of early GFP⁺ postnatal cells (∼P1-7) resulted in the presence of GFP⁺ photoreceptors correctly located within the host ONL, following transplantation in both the developing and adult retina (MacLaren et al. 2006). By utilizing a transgenic Nrl.GFP^+/+ mouse, which expresses GFP only in postmitotic rod cells, we demonstrated that this donor population resulted in the correctly located GFP⁺ photoreceptors within the host retina; these findings have been replicated by a number of other studies (MacLaren et al. 2006; Bartsch et al. 2008; West et al. 2008, 2010; Lakowski et al. 2010). To determine whether photoreceptor precursor transplantation could restore visual function, studies in GNAT1^−/− mice, a model of stationary night blindness, were performed. The photoreceptors in this model lack rod α-transducin and are therefore nonfunctional; the photoreceptors, however, remain intact. Following photoreceptor precursor transplantation to the subretinal space, GFP⁺ photoreceptors were observed in the host expressing rod α-transducin (Pearson et al. 2012). Improved visual function was demonstrated at both the cellular and behavioral level, as assessed with dim light levels, and was indicative of a rod-mediated light response (Pearson et al. 2012). Other studies have also demonstrated the expression of photoreceptor-specific proteins that are lacking in host cells, but present in GFP⁺-labeled photoreceptors in the recipient ONL following precursor transplantation (MacLaren et al. 2006; Bartsch et al. 2008; Lakowski et al. 2010; Barber et al. 2013).

Until recently, it was thought that transplanted GFP⁺ photoreceptor precursor cells migrated into the host ONL, where they integrated with the host retinal circuitry and formed outer segments that protruded into the subretinal space (MacLaren et al. 2006; Warre-Cornish et al. 2014). However, this interpretation of the results was revised by two studies in 2016. We and others independently verified that the majority of the GFP⁺ photoreceptors present in the recipient retina were in fact host cells that had exchanged either protein and/or RNA material with transplanted donor photoreceptors (see Fig. 3; Pearson et al. 2016; Santos-Ferreira et al. 2016a). The mechanism by which this process occurs is still unclear, as is the exact type of material transferred between cells. It is known that this is not a classic cell-fusion process and it appears restricted to the transplanted photoreceptor cell population, as GFP⁺ fibroblasts, retinal and neural progenitor cells have not resulted in the same transfer of GFP reporter (West et al. 2012; Pearson et al. 2016). Neither has the presence of recombinant GFP, when injected directly into the subretinal space, suggesting it is not the uptake of free protein (Pearson et al. 2016). Material transfer appears to be a transient process that allows for the exchange of a range of proteins and/or RNA expressed by the donor cell population, including rhodopsin, Peripherin-2, and rod α-transducin (MacLaren et al. 2006; Pearson et al. 2012, 2016; Barber et al. 2013). In light of these findings, previously reported improvements in visual responses were most likely achieved by the transfer of functional levels of the phototransduction proteins missing in host photoreceptors, rendering them light-responsive (Pearson et al. 2012; Santos-Ferreira et al. 2015; Neves et al. 2016; Zhu et al. 2016). Further studies have corroborated the findings of cytoplasmic material exchange following photoreceptor precursor transplantation to the intact host retina and demonstrated that it is a developmentally regulated phenomenon (Singh et al. 2016; Decembrini et al. 2017; Ortin-Martinez et al. 2017; Waldron et al. 2018). Importantly, a minority (∼1% ± 0.8%) of GFP⁺-labeled cells were identified as truly integrated donor-derived cells, by the use of Y chromosome FISH to identify male donor photoreceptors in female wild-type recipient retinas (Pearson et al. 2016; Waldron et al. 2018). Further investigation has demonstrated that the host retinal environment effects the proportion of true cellular integration events, as well as the extent of material exchange. In models with disrupted outer limiting membranes (OLMs), the proportion of true Y⁺/GFP⁺ donor photoreceptors in the female host ONL were significantly increased (∼23% ± 3%) compared to models with intact OLMs (∼6% ± 1%) (Waldron et al. 2018). This confirms previous findings that OLM integrity influences transplantation outcome and that it may be possible to improve true levels of integration by manipulation of the host retinal environment (West et al. 2008, 2009; Pearson et al. 2010; Barber et al. 2013).

Figure 3.

Schematic representation of photoreceptor cell transplantation into various models of retinal disease. Photoreceptor cell transplantation into (A) nondegenerate retina or into retina at early stages of degeneration can result in transfer of donor RNA and/or protein to host photoreceptors (material transfer) in addition to donor photoreceptor integration. Even in degenerate models that rapidly lose photoreceptors (B), cone photoreceptors can remain for extended periods. Following photoreceptor cell transplantation into these models, the presence of remaining host photoreceptors can be a confounding factor that complicates the interpretation of any therapeutic effects observed, either because of material transfer from the donor photoreceptor population or residual function from surviving host cells. In end-stage retinal disease (C), in which there are few remaining host photoreceptors, it is possible to make rigorous assessments of whether donor photoreceptors make functional connections with the remaining retinal circuitry. However, additional barriers to the integration of donor photoreceptors are present in this stage of retinal disease, such as increased gliotic scarring and inner retinal remodeling.

The Transplantation of Mouse PSC (mPSC)-Derived Retinal Cell Suspensions

Although the majority of investigations have transplanted murine donor-derived photoreceptor precursors, the equivalent developmental stage in humans would require the use of second-trimester fetal retinal tissue, which is ethically challenging, of limited availability, and logistically difficult to source. Despite the reinterpretation of transplantation results because of the discovery of material transfer, the postmitotic photoreceptor precursor is still the optimal donor stage, and the use of PSCs as a source of photoreceptors for cell replacement remains the ideal choice for future clinical translation. Therefore, protocols to derive photoreceptors from mouse and hPSCs have been widely investigated.

Initial studies using mouse ESCs (mESCs) established the stepwise differentiation of rostral neuronal progenitors and RAX/Pax6-positive retinal progenitors using serum-free floating embryoid body (SFEB) cultures (Ikeda et al. 2005; Watanabe et al. 2005). Further defined conditions were developed based on this method for human, mouse, and monkey ESCs and used to generate rhodopsin positive rod photoreceptors in adherent cultures (Osakada et al. 2008). Our own investigations utilized this method for mESCs; however, despite the presence of some photoreceptor markers, the levels of Nrl transcript were relatively low, and following transplantation typical GFP⁺ photoreceptors present in the ONL were not observed. Even with the reinterpretation of these results as a lack of material transfer to the host photoreceptors, the findings remain the same, that the differentiated retinal cells were not equivalent to donor-derived postnatal retinal cell populations (West et al. 2012).

In 2011, a landmark study by Sasai and colleagues was the first to demonstrate three-dimensional morphological recapitulation of retinal developmental processes from mESCs. This protocol demonstrated the ability of the retinal neuroepithelium to form spontaneously from SFEB cultures and suggests that the physical forces and signaling required for optic vesicle invagination to form the bilayered optic cup are intrinsic to the neuroepithelium (Eiraku et al. 2011, 2012; Sasai et al. 2012). In addition, they reported the interkinetic nuclear migration of dividing retinal progenitors and the production of all retinal cell types in a time-dependent manner, resulting in a laminated retina-like structure. However, these structures lost integrity by 35 days and photoreceptor maturation and segment formation were not reported (Eiraku et al. 2011). We adapted this method to enable neural retinal epithelia to be further cultured without optic cup dissection, thus providing sufficient photoreceptor precursor cells for transplantation studies (Gonzalez-Cordero et al. 2013). In addition, we determined that Rho.GFP⁺ ESC-derived photoreceptors expressed genes enriched in P4 donor-derived rod precursors by microarray gene expression profiling. Following the transplantation of Rho.GFP⁺ photoreceptors into GNAT^−/− mice, GFP⁺/rod α-transducin+ photoreceptors were observed in the host ONL. The proportion of true integration events, relative to the material transfer of GFP and rod α-transducin to host photoreceptor cells, was undetermined in this study. However, it is clear that the msESC-derived photoreceptors generated by 3D retinal differentiation methods were equivalent to donor-derived photoreceptor precursors (Gonzalez-Cordero et al. 2013). Further to this, other studies have observed similar results following the transplantation of msESC-derived photoreceptor precursors (Decembrini et al. 2014; Lakowski et al. 2015; Santos-Ferreira et al. 2016b; Waldron et al. 2018).

Transplantation of hPSC-Derived Neuroretinal Cell Suspensions

In parallel to the investigation of differentiation methods for mPSCs, a number of protocols have been established to differentiate hPSCs into retinal cell lineages (Lamba et al. 2006; Osakada et al. 2008, 2009; Meyer et al. 2009, 2011; Nakano et al. 2012). Lamba and colleagues (2006) reported one of the first hPSC-derived retinal differentiation methods, whereby, following a brief suspension culture to stimulate neural induction, cells were plated onto Matrigel and further cultured for 3 weeks to give 80% retinal progenitors, of which 12% were Crx⁺ photoreceptor precursors. Of note, this protocol generated human cells within a relatively short time frame, compared to normal retinal development and other methods of hPSC differentiation (Osakada et al. 2008; Meyer et al. 2009). Following the transplantation of hPSC-derived retinal cultures, GFP⁺-labeled cells were observed throughout the host retina (Lamba et al. 2009, 2010). It seems likely that these GFP⁺-labeled cells were the result of cytoplasmic material transfer rather than true human photoreceptor cell integration. The electroretinography (ERG) recordings from treated animals did not have normal waveforms and appeared to be recording artefacts. Furthermore, recent transplantation studies into nonhuman primates did not result in any reporter-labeled photoreceptors within the host retina (Chao et al. 2017).

Studies by Meyer and colleagues determined that retinal differentiation could be achieved by the endogenous secretion of DKK-1, Noggin, and FGF following a combination of suspension and adherent cell culture conditions. Progression through all the major stages of retinogenesis were required for the complete differentiation of retinal progeny and in a timeframe that reflected normal human development (Meyer et al. 2009, 2011). Furthermore, studies by the Sasai group established a 3D protocol for hPSCs that, as in the mPSC method, self-organized to form stratified neural retina-like structures that contained all retinal cell types (Nakano et al. 2012). In a later study, they developed a stepwise induction-reversal method that resulted in the formation of RPE adjacent to the neural retina, and using this method they determined the presence of a ciliary margin-like growth zone at the RPE-neural retinal boundary (Kuwahara et al. 2015). Although the efficiency with which hPSC-derived neural retinal epithelia were produced was far less than in mPSC cultures, this method improved the long-term maintenance of continuous retinal progenitor cell sheets that could be utilized for transplantation (Kuwahara et al. 2015; Shirai et al. 2016; Iraha et al. 2018; Tu et al. 2019). In light of these pioneering studies, the majority of established retinal differentiation methods since then have been focused on generating 3D neural retinal structures, in a time frame consistent with human retinal development.

Subsequent studies based on the Meyer differentiation method, and incorporating 3D culture techniques, determined the 3D retinogenesis of all retinal cell types from self-organizing optic vesicle-like aggregates. They demonstrated for the first time the long-term (up to 180 days) culture and maturation of human photoreceptor cells that exhibited a few primitive outer segment-like structures (Phillips et al. 2012; Zhong et al. 2014). In addition, perforated-patch recordings determined that a small proportion of hPSC-derived photoreceptors (2/13) were responsive to light (Zhong et al. 2014). Another study has also demonstrated the formation of rudimentary outer segment-like structures by adaptations to the Sasai method (Wahlin et al. 2017). However, it remains to be determined whether more developed outer segment-like structures can be formed without direct contact with the RPE.

All retinal differentiation protocols described thus far have relied on the generation of embryoid bodies or cell clump suspensions to initiate neural differentiation, and the presence of Matrigel or serum. Reichman and colleagues developed a method that bypassed this step and instead allowed optic vesicle-like neuroectoderm to form spontaneously in confluent hiPSC cultures, by the removal of FGF2 and the endogenous presence of Dkk-1 and Noggin (Reichman et al. 2014). Serum-free proneural medium was used to culture the developing vesicles for 2 weeks, before they were isolated from adherent cultures and maintained in suspension as 3D retinal neuroepithelia, similar to other protocols (Nakano et al. 2012; Phillips et al. 2012; Zhong et al. 2014). These neural retina-like structures maintained laminated organization until 21 days in culture, whereby they developed rosettes with the photoreceptors positioned toward the center of the structure. The timing of photoreceptor cell marker expression was similar to that observed by other methods of retinal differentiation and typical of normal human development, with rhodopsin and cone opsin protein present by day 112 of culture (Osakada et al. 2008; Nakano et al. 2012; Phillips et al. 2012; Reichman et al. 2014). The benefits of this method for hPSC-derived differentiation are the ease and scalability with which neural retina-like structures can be generated. We therefore utilized this approach to generate photoreceptors for transplantation studies. Optic vesicle–like structures were isolated by manual dissection at 4–7 weeks of adherent culture and maintained as 3D neural retinal vesicles, similar to other studies (Zhong et al. 2014). Detailed analysis of photoreceptor development in hPSC-derived vesicles found that it recapitulated human fetal development (Gonzalez-Cordero et al. 2017). In addition, lamination of the retinal neuroepithelium was observed, with brush-like protrusions demonstrating primitive and disorganized membranous discs at 28 weeks, similar to those observed in other long-term cultures (Zhong et al. 2014; Wahlin et al. 2017).

Having established rod and cone photoreceptor development in hPSC-derived retinal neuroepithelia, virally labeled M/L opsin cone precursors were selected from 17-week cultures and transplanted into adult mouse models. The hPSC-derived cone precursors survived in the subretinal space for at least 3 weeks, with a minority of GFP⁺/hNUCLEI⁺ cells (55 ± 38 cells) present within the Nrl^−/− host retina, demonstrating correctly orientated bud-like protrusions positive for M/L cone opsin. Importantly, as well as the entire nucleus labeling positive for hNUCLEI, the size of nuclei was significantly greater than in the host retina, suggesting true incorporation of hPSC-derived photoreceptors. However, a small number of GFP⁺ cells were negative for the hNUCLEI marker (30 ± 34 cells), positive for a Y chromosome probe (male host), and had murine Nrl^−/− typical photoreceptor morphology and nuclear size (Gonzalez-Cordero et al. 2017). These cells could reflect carryover of viral particles used to label the PSC-derived cones, although this is unlikely, as following transplantation into the wild-type host retina, very few GFP⁺ cells were observed (2 ± 3 cells). As previously mentioned, it has been reported that true murine donor-derived photoreceptor cell integration and material transfer to host photoreceptors are increased in the Nrl^−/− retina (Waldron et al. 2018). These findings suggest that hPSC-derived photoreceptor precursors may also be able to engage in material transfer with host murine photoreceptors, following transplantation to the subretinal space. Therefore, the presence of human-specific proteins may not be sufficient to identify hPSC-derived photoreceptor cells following transplantation. Thus, to ascertain whether there is restoration of vision because of the connection of hPSC-derived photoreceptors with the host retina, it is necessary to utilize end-stage models of retinal degeneration, with no remaining endogenous photoreceptors. In addition, the host retinal environment present in late-stage retinal degeneration models may be more in keeping with potential clinical applications that are likely to involve transplantation in regions of the retina with very few remaining photoreceptors.

Rescuing End-Stage Retinal Degeneration

An established model of end-stage retinal degeneration is the PDEβrd1/rd1 mouse model (known as rd1), which is homozygous for a null mutation in the gene encoding the β subunit of rod photoreceptor cyclic GMP phosphodiesterase (Pde6b). This results in rapid rod photoreceptor cell loss by 3 weeks of age and an absence of residual cone photoreceptor function by 12 weeks (Chang et al. 2007; Barber et al. 2013). The rd1 mouse model has been utilized for many decades. However, recently, we reported the presence of an additional mutation in the gpr179 gene, found in C3H/HeN mice, the most commonly used rd1 mouse strain (Nishiguchi et al. 2015). This naturally occurring mutation results in nonfunctional ON-bipolar cells, including rod bipolar cells. Gene supplementation therapy, using AAV vectors, was shown to rescue visual responses in the rd1 model, but only after backcrossing the line to remove the mutation from the C3H/HeN background (Nishiguchi et al. 2015). Further characterization of multiple C3H substrains has determined that the presence of this additional mutation was confined to the C3H/HeN mouse strain (Chang 2015). However, transplantation studies have been performed in this model using P3-4 mouse donor-derived photoreceptors, with the suggestion of improved light-mediated visual responses, as shown by a variety of tests including light/dark chamber tests, pupillary reflex, and cortical imaging (Singh et al. 2013). As the majority of transplanted cells are rod photoreceptors and rod bipolar cell function is abolished in this strain, it would be interesting to know whether functional restoration is occurring through atypical synaptic connections formed with cone bipolar cells. It is known that in late-stage degeneration there is retinal remodeling (Marc and Jones 2003). Alternatively, the presence of any remaining endogenous cone photoreceptors must be considered when interpreting the functional restoration data, even under scotopic light conditions (Cehajic-Kapetanovic et al. 2015).

Further studies using the rd1 mouse model have investigated the transplantation of hPSC-derived photoreceptors into the degenerate retina (Barnea-Cramer et al. 2016; Collin et al. 2019). Both studies demonstrated the survival of hPSC-derived photoreceptors and the expression of photoreceptor-specific proteins, indicative of further photoreceptor maturation in vivo. Transplanted cells resided next to the host inner retina and the presence of some synaptic markers was demonstrated, in close proximity to hPSC-derived photoreceptor processes (Barnea-Cramer et al. 2016; Collin et al. 2019). In addition, improved optomotor responses in transplanted eyes were reported in the Barnea-Cramer study, suggesting some restoration of light-mediated responses. However, the possibility of neurotrophic effects as a result of the transplanted cells preserving residual cones could not be excluded (Barnea-Cramer et al. 2016).

Our own studies investigating PSC-derived photoreceptor transplantation to the degenerate retina have focused on the severely degenerate Aipl1^−/− mouse model of Leber congenital amaurosis (LCA4). This model is deficient in the photoreceptor-specific gene encoding aryl hydrocarbon receptor–interacting protein-like 1 (Aipl1), resulting in the rapid degeneration of both rod and cone photoreceptors by 4 weeks of age, because of the destabilization of cGMP PDE (Ramamurthy et al. 2004; Kirschman et al. 2010). Initial studies investigated the transplantation of mESC-derived cone photoreceptor precursors into 8- to 12-wk-old Aipl1^−/− mice. Because human daylight vision relies on cone photoreceptor function, we adapted our previously described protocol to enable sufficient numbers of mESC-derived cone photoreceptor precursors to be generated and isolated (Kruczek et al. 2017). Transplanted mESC-derived cones appeared to make contact with host inner retinal neurons and expressed components associated with synaptic transmission alongside phototransduction related proteins, suggesting advanced differentiation and maturation (Kruczek et al. 2017). Similar results were obtained following the transplantation of isolated hPSC-derived cone photoreceptor precursors in this model (Gonzalez-Cordero et al. 2017). Transplanted hPSC-derived cone photoreceptors were able to survive in the degenerate retina, even without immunosuppression, for 3 weeks posttransplantation and formed a distinct layer adjacent to the host inner retina. The presynaptic protein ribeye was present in hPSC-derived cone photoreceptor processes that extended toward host retinal neurons, alongside the protrusion of bud-like structures that contained human mitochondria and the presence of phototransduction-related proteins throughout the cells. A key difference between transplanted mESC-derived and hPSC-derived cone photoreceptors was the increased presence of Peripherin-2-positive buds in the former cells compared to the latter (Gonzalez-Cordero et al. 2017; Kruczek et al. 2017). This most likely reflects the increased time required for human cone cell maturation compared with mouse cells and indicates that time periods posttransplantation need to be set accordingly. Although some hPSC-derived cone photoreceptors were still present 6 weeks posttransplantation, this would most likely be improved and prolonged with the use of immunosuppression. It remains to be determined whether PSC-derived cone photoreceptor precursors are able to functionally mature and connect with the recipient degenerate retina, enabling the restoration of light-mediated responses in this model.

A further rod and cone degenerate model that has been established is the Cpfl 1/Rho^−/− mouse, which has no functional photoreceptors and a rapid degeneration of both rods and cones, with one row of photoreceptor cell bodies remaining at 10–12 weeks of age (Santos-Ferreira et al. 2016b). The transplantation of mESC-derived rod photoreceptors into this degenerate model demonstrated the survival of cells and the presence of rod-specific phototransduction proteins, suggesting photoreceptor maturation in vivo. However, reduced contact between transplanted rod photoreceptors and host inner retinal neurons was reported, as well as relatively few examples of presynaptic markers, ribeye and pikachurin, in the mESC-derived photoreceptor cell processes (Santos-Ferreira et al. 2016b). It remains to be determined whether the differences observed in this study compared to our own reflect differential abilities of rod and cone photoreceptor precursors to form presumptive connections with host retinal neurons, or if the models of advanced retinal degeneration used differ in terms of the remodeling of inner retinal circuitry that occurs with rapid photoreceptor degeneration (Santos-Ferreira et al. 2016b; Kruczek et al. 2017). Further studies in late-stage models of retinal degeneration are required to determine whether transplanted photoreceptor precursors are able to restore light-mediated responses to the recipient retina. Furthermore, the use of various degenerate models will establish by what means the host retinal environment influences functional restoration and how this will translate to the human degenerate retina.

Retinal Sheet Transplantation

An alternative strategy to the transplantation of retinal cell suspensions is retinal sheet transplantation, whereby an organized sheet of fetal retina or PSC-derived retinal organoid is dissected and transplanted as a unit into the subretinal space. A potential advantage of this approach is the improved organization of grafted cells, although the presence of other retinal neurons may limit host interneuron connectivity with the grafted photoreceptors. Transplantation of an organized structure, which can include an RPE layer, may also be beneficial to long-term cell survival and maturation, especially at later stages of retinal degeneration (Seiler and Aramant 2012; Seiler et al. 2016; Lorach et al. 2019). Studies involving the use of fetal retinal sheets have previously demonstrated photoreceptor cell survival and maturation in the subretinal space and several clinical trials have been performed with no adverse effects reported (Kaplan et al. 1997; Radtke et al. 1999, 2004, 2008; Humayun et al. 2000; Berger et al. 2003). Because of the limited availability and ethical implications of using human fetal tissue, many groups are now investigating the use of hPSC-derived retinal sheets, using 3D retinal organoid culture methods.

Preclinical studies have been performed with mPSC-derived retinal progenitor cell sheets being transplanted into the end-stage retinal degeneration rd1 mouse model (Assawachananont et al. 2014; Mandai et al. 2017a). Following transplantation, mPSC-derived retinal sheets developed into laminated retina-like structures in the subretinal space, containing both inner and outer retinal cell layers. However, in some regions of the mPSC-derived graft, the inner retinal cells were absent, resulting in direct contact between the host interneurons and the donor photoreceptors. Ultrastructural analysis of the mPSC-derived photoreceptors demonstrated the ability of these cells to mature in vivo forming typical inner and outer segments, either in rosette-like structures or in close contact with the host RPE (Assawachananont et al. 2014). Some indications of light-mediated responses in the retina were detected by ex vivo microelectroretinography (mERG) and RGC recordings, using a multielectrode array (MEA) system, in the vicinity of the retinal graft. In addition, light-mediated behavioral responses were observed in a proportion (9/21) of the treated animals (Mandai et al. 2017a). A further study from the same group has shown similar histological results following the transplantation of hESC-derived retinal sheets, at an equivalent retinal progenitor stage (∼60 days), into an immune-deficient rd1 mouse model (NOG-rd1-2J) (Iraha et al. 2018). The transplanted hESC-derived retinal grafts demonstrated photoreceptor maturation with inner and outer segment formation and an indication of synaptic contact in some areas where the hESC-derived photoreceptors were in direct contact with the host inner retina. However, light-mediated responses were only detected in 3/7 retinas with substantial hESC-derived transplants, by RGC recordings only, compared to all retinas that received mPSC-derived transplants (Iraha et al. 2018). Although in both studies the MEA responses were greatest in animals that received transplants and in the area of the graft, it is not possible to exclude the possibility that the transplants improved the function of residual cone photoreceptors or melanopsin-positive ganglion cells through trophic effects.

In another study, day 50–60 hPSC-derived retinal sheets were transplanted into an immune-deficient end-stage retinal degeneration rat model (SD-Foxn1 Tg(S334ter)3LavRrrc), as well as a laser-induced photoreceptor degeneration monkey model (Tu et al. 2019). Histological analysis confirmed the development of mature photoreceptors in both models and regions of direct hPSC-derived photoreceptor and host bipolar cell interaction, with the hPSC-derived retinal grafts surviving for 10 months in immunocompromised rats and >2 years in the monkey model with cyclosporine immunosuppression. A modest recovery of light perception was suggested by the visual field test, associated with the grafted area, in the induced monkey model of photoreceptor degeneration. However, light-mediated responses could not be confirmed to originate from the grafted tissue by other functional examinations. In addition, SD-Foxn1 Tg(S334ter)3LavRrrc rats had substantial light responses remaining in the host retina, even at 10 months. This resulted in difficulty in conclusively distinguishing the hPSC-derived graft-originated responses from the residual sensitivity of the recipient retina in this model (Tu et al. 2019). Other studies have also utilized this rat model to examine hESC-derived retinal sheet transplantation and reported electrophysiological recordings in the superior colliculus, as well as improvements in visual acuity based on optokinetic head tracking (OKT) (McLelland et al. 2018). However, it has been reported that the use of OKT in a number of degenerative transgenic rat lines may be limited as a therapeutic outcome measure (McGill et al. 2012). These studies highlight the difficulties in confirming that the light-mediated responses originate from the donor photoreceptors rather than residual endogenous photoreceptors and also highlight the need for multiple testing methodologies to establish the restoration of light-mediated responses at multiple levels of visual function.

Similar to the strategies previously mentioned for RPE cell sheets, a number of studies have investigated the use of various scaffold biomaterials with which to transplant retinal progenitor cells, to increase cell survival, improve cell organization, and promote host–graft cell interactions (Young 2005; Redenti et al. 2009; Tucker et al. 2010; Kundu et al. 2014; Lawley et al. 2015; Yao et al. 2015; Park et al. 2019). In addition, 3D micropatterned polymers have been designed that could act as polarized delivery substrates to enable the transplantation of retinal progenitors or photoreceptor precursors in an organized array, potentially improving donor–host connectivity in models of end-stage retinal degeneration (Worthington et al. 2017; Jung et al. 2018; Thompson et al. 2019). If an ideal biodegradable or porous material can be found that can maintain orientated photoreceptor precursors and position them in close apposition to the degenerate host retina, this may provide the advantages of pure photoreceptor cell suspensions with the organization of a retinal sheet.

Considerations for the Clinical Application of Photoreceptor Cell Transplantation

As photoreceptor cell transplantation progresses toward the clinic, there are a number of key considerations. The first is the optimal cell population to transplant that will result in the replacement of functional photoreceptors with the ability to connect with the recipient retinal circuitry. This differs depending on whether retinal cell sheets or retinal cell suspensions are the preferred strategy. For retinal cell sheets, earlier stages of development, ∼day 60 of hPSC-derived retinal neuroepithelia, appear optimal when the majority of cells are retinal progenitors (Shirai et al. 2016). For retinal cell suspensions, despite the reinterpretation of earlier findings because of cytoplasmic material transfer, the optimal cell population is still considered to be postmitotic photoreceptor precursors, once photoreceptor cell fate has been specified but before the development of structures that reduce cell viability upon dissociation (Gust and Reh 2011; Pearson et al. 2016; Waldron et al. 2018). To enable the selection and isolation of a defined population of photoreceptor precursors for clinical use, a number of studies have investigated the presence of cell surface antigens, CD markers (Koso et al. 2009; Lakowski et al. 2011, 2015, 2018; Kaewkhaw et al. 2015; Welby et al. 2017). Furthermore, many studies have demonstrated the transplantation of enriched photoreceptor precursors from donor and PSC sources, using CD73 marker selection, into various retinal models (Eberle et al. 2011, 2012, 2014; Lakowski et al. 2011, 2015, 2018; Santos-Ferreira et al. 2015; Gagliardi et al. 2018). However, the restoration of daylight vision may be of more use to patients; therefore, the isolation and transplantation of cone photoreceptor precursors may be preferable to rods. Although it was initially assumed that cone precursors had a reduced capacity to integrate into the adult rod-rich murine retina, we now know this was because of material transfer to host photoreceptors and not necessarily an indication of true integration (Lakowski et al. 2010; Santos-Ferreira et al. 2015; Smiley et al. 2016; Waldron et al. 2018). It remains to be determined whether there are any differences in the capacity of rod or cone photoreceptor precursors to integrate into the recipient retina; however, CD markers that enable the enrichment of cone photoreceptor precursors have been studied (Welby et al. 2017).

In addition to selecting the desired cell population for transplantation, the removal of proliferative cells may be of equal benefit, as a safety measure to prevent tumor formation that can arise from the incomplete differentiation of unselected cell populations (West et al. 2012). Importantly, cell populations need to be well-defined and -characterized, as the transplantation of unknown populations can have devastating effects in patients (Kuriyan et al. 2017). With advances in transcriptomics, more data is available from developing retinal tissue with which to compare PSC-derived cell populations (Mustafi et al. 2016; Aldiri et al. 2017; Hoshino et al. 2017; Mellough et al. 2019). This helps to refine the stage of development of PSC-derived cells, as well as confirming the differentiation of bona fide photoreceptors that should be able to function in response to light once fully developed.

A second consideration is the requirement for a robust, clinically relevant differentiation method with which to generate the desired retinal cell population. The adaptation of retinal differentiation research methods to more GMP-like processes is ongoing and this has been helped by the increasing availability of defined xeno-free reagents and small molecules (Osakada et al. 2009; Sridhar et al. 2013; Tucker et al. 2013; Wiley et al. 2016; Reichman et al. 2017; Slembrouck-Brec et al. 2018). In addition, several studies have examined methods to scale up production of retinal cell types using bioreactors (DiStefano et al. 2018; Ovando-Roche et al. 2018). Because of the length of manufacturing processes involved, generating sufficiently high yields of defined and pure photoreceptor precursor populations at the same stage of development is challenging and methods have been explored to enable batches of retinal neuroepithelia to be cryopreserved and stored prior to use (Nakano et al. 2012; Reichman et al. 2017; Gagliardi et al. 2018; Slembrouck-Brec et al. 2018). This would enable multiple batches of retinal epithelia to be pooled, as well as the storage of retinal sheets prior to use. However, optimal methods to freeze the final purified photoreceptor precursor cell product have yet to be defined.

Another consideration is the host retinal environment, as the subretinal space may not be immune privileged in some diseased states, and immunosuppression may be required if allografts are used (Streilein et al. 2002). Further studies in degenerate recipient retinas are required to ascertain the long-term survival and function of transplanted photoreceptors. Clinical trials using RPE cells are currently investigating the use of local immunosuppression methods, such as intraocular steroid implants, and the results from these studies will inform further trials for retinal cell therapy (da Cruz et al. 2018). Alternatively, the use of HLA-matched iPSC-derived cells and tissue may negate the need for long-term immune modulation in patients (Kamao et al. 2017). Planned clinical studies using HLA-matched iPSC-derived RPE sheets will again inform future strategies for photoreceptor cell transplantation. However, early trials will most likely utilize both hESC and hiPSC lines, as banks of validated and genetically screened iPSC lines are produced. In addition, further preclinical studies will provide more information as to the extent of photoreceptor restoration in severely degenerate retinas and the optimal delivery method, either cell suspensions or sheets, required for the optimal recovery of light-mediated responses, which may vary with disease phenotype. Finally, a combined approach to replace both the photoreceptors and the supportive RPE cells will be required in certain degenerative retinal diseases, such as AMD and SMD.

CONCLUDING REMARKS

Major advances in the field of regenerative cell therapy in recent years have resulted in the rapid progression of differentiation methods to generate retinal cell types from hPSC sources, as well as the refinement and improvement of techniques and models to detect functional responses from these cells once transplanted. Because of its external location, the advanced imaging techniques, and the functional tests available, the eye remains at the forefront of regenerative medicine strategies. Ongoing clinical trials are establishing the safety and efficacy of RPE cell therapy, and preclinical advances over the last decade have opened the way to trials of hPSC-derived photoreceptor cell transplantation in the next few years. Combined, these strategies may eventually provide the potential to restore sight in many blinding conditions that currently have no therapeutic options.